Information and Organizations [PDF]

Preferred Citation: Stinchcombe, Arthur L. Information and Organizations. Berkeley: University of California Press, c1990 1990. http://ark.cdlib.org/ark:/13030/ft338nb1zq/

Information and Organizations Arthur L. Stinchcombe UNIVERSITY OF CALIFORNIA PRESS Berkeley · Los Angeles · Oxford © 1990 The Regents of the University of California

Preferred Citation: Stinchcombe, Arthur L. Information and Organizations. Berkeley: University of California Press, c1990 1990. http://ark.cdlib.org/ark:/13030/ft338nb1zq/

xi

Acknowledgments As I look back over my files, I find old syllabuses labeled "organizations as information," but in those days my desktop computer did not date my syllabuses for me. What that means is that many students for many years have put up with me thinking about the subject of this book. Some of the materials that went quite directly into the book were developed in lectures to undergraduate classes in "industrialization" and "organizations" at Northwestern University, and students' vague impression that I was reading an article from an encyclopedia in those classes may have been due to my writing on an encyclopedic effort at the time. I co-taught one of these courses with Arnold S. Feldman, which held me to enough higher of a standard as to make it believable that the lectures might someday be the basis for a book. My first debt, then, is to the students who put up with the ambiguous advantage of getting much of Chapters 2, 4, and 5 before they were ready for the rest of the world, and to Feldman for comments and stimulation. Carol A. Heimer read the entire manuscript with an eye to the argument that should have been there rather than to the failures of the argument as written. She also thought that the book should be written in English, which agreed with my convictions better than with my behavior. Being married to a person who criticizes one's work has many advantages, but the most important to me was that she knew me well enough not to try to get me to change things that I would never change. Tempering the wind to the shorn lamb may not be the best thing for the reader, but it helped me to accept more improvements to the text than I would otherwise have done. Mayer Zald read the first eight chapters of the manuscript in an earlier version and was the primary cause of there being a conclusion. An anonymous reviewer for the University of Chicago

xii Press provided sufficient detail about why the manuscript was beyond redemption to help me repair some parts. Others who have read and commented on parts of the manuscript include Alfred D. Chandler, Per Heum, Robert K. Merton, Charles Payne, Paula Rossow, James Sheridan, John Walsh, James Zetka, and Harriett Zuckerman. I suppose the general tone of these acknowledgments already demonstrates that I will often defend my faults rather than graciously accept correction, so these people will not be expected to take responsibility for the remaining imperfections. I have tried to estimate the number of working hours for the manuscript that different sources have supported. Most of roughly 1,500 hours supported many years ago by the Stanford Graduate School of Business with Mellon Foundation funds went into Chapter 6. The Institute of Industrial Economics of Bergen, Norway, with funds partly from the Norwegian State Oil Company (Statoil), supported about 1,000 hours for the research reported in Chapter 3, and supported previously published work used throughout the book. I figure that the four-course teaching load at Northwestern University College of Arts and Sciences gives me about 200 hours per quarter for research; I have used five quarters, or 1,000 hours of research time, and in addition was provided with a course off one quarter, for a total of about 1,200 hours. This time was used especially for Chapter 9 and for general rewriting work. A couple of summers supported out of our household resources supplied about 1,000 hours. The Center for Urban Affairs and Policy Research at Northwestern University supported about 400 hours of work. The total amounts to about two and a half years of full-time work, with office space, overhead, help in making research contacts, and the like thrown in. Since the salary and other expense involved are probably a good deal more than the gross sales the book will have, this level of support helps provide evidence for the basic argument of Chapter 9 about the rent of reputations. I am grateful to all these institutions, to their movers and shakers, and to the members of my household for support of my research. Aside from whatever value the research in this version may have for the scholarly community, it was great fun to do it. Chapter 6 has been previously published by Norwegian University Press in my and Carol A. Heimer's Organization Theory and Project Management . I am grateful for permission to reprint it here. Chapter 3 is revised from a paper published in photocopy form by the Institute of Industrial Economics of Bergen, Norway.

1

1— Information, Uncertainty, Structure, and Function in Organizational Sociology Rationality Rationality necessarily involves an analysis of the future, because the consequences that give purpose to acts are necessarily in the future. Thus all rationality is based on predictions of one kind or another, not on knowledge. Assuming that actors are perfectly rational, of course, implies that they are certain what the future holds, that in all relevant respects our notions about the future constitute knowledge. The assumptions of neoclassical economics are that various financial quantities (e.g., the interest rate and the savings and investment rates if distinct) summarize all the relevant information about the future that one needs in order to make economic decisions, and that everyone else can have the same information that an individual actor uses to work out a strategy (the investor, for example, presumably compares a concrete opportunity [e.g., to invest] with the relevant financial quantities [e.g., the interest rate]; the assumption is that anyone else with resources has access to the same information about the future relevant to that investment as the investor does). The assumption of neo-Keynesian economics (and apparently of Keynes himself) was that the idea that investment can operate with only financial and universally available knowledge assumes a kind of knowledge about the future that human actors do not and cannot have (see Weintraub 1979). Two main traditions of organizational sociology start from the Keynesian or neo-Keynesian assumption that the future is uncertain. The older one has as its stellar figures Herbert Simon and James G. March (e.g., March and Simon 1958). Its main thrust has been to try to describe what people acting in organizational contexts (and to some degree acting in organizational roles in the light of organizational purposes) actually do to deal with the imperfection of their knowledge or predictions of the

2 future, what decision procedures they in fact use to fill in what neoclassical economists require them to know. The other tradition, whose stellar figure is perhaps Oliver Williamson (e.g., Williamson 1985), looks at what organizational devices might be good for dealing with various sorts of uncertainty about the future (e.g., small numbers bargaining, which involves uncertainty about the future behavior of a partner who is not replaceable in the market—or is replaceable only at a higher cost). The organizational devices solve specific types of departure from perfect markets. The actual distribution of such organizational devices, then, is to be predicted from the distribution of the problems they are good for solving. The Williamson alternative is, in short, microeconomic functionalism. The approach to the microeconomic ideal of perfect efficiency in spite of uncertainty is the survival criterion for the practice in a functional argument. To put it another way, the Simon-March tradition is concerned with the causal explanation of departures from rationality. The Williamson tradition is concerned with the functional explanation of organizational structures, such as hierarchies. The Simon-March objective is to explain why organizations use less than rational methods. The Williamson approach is to explain how organizations can come near rationality, if not quite there, by adopting organizational devices that deal with uncertainty. The clue to structures needing such explanation in the Williamson tradition is that they occur in the context of market behavior but depart in significant ways from the structure of decision making known to be optimum in a perfect market. I propose in this book to follow a slightly different strategy from both of these, though the difference involves a minor shift from the received tradition. I propose that information about the uncertain future becomes progressively available in distinct social locations , depending on what sort of uncertainty it is. What resolves the uncertainty of particular actors, then, is the earliest available information that will show what direction the actor ought to be going because of the way the future of the world is, evidently, turning out. Structures that depart from idealized markets (none of the organizational structures we explain in this book work like ideal markets do) are then to be explained functionally, by the growth of the organization toward those locations where information for resolving uncertainty is chiefly located. That information then has to be processed quickly, both to adapt the previous tentative strategy and to work out the tentative strategy for the next period. The core of structure of organizations, then, is information processing, and the core information to be processed is the earliest information that indicates what sort of world (i.e., future) the decisions are being made for.

3 The explanation is, however, causal as well as functional. The organization controls neither where the uncertainties it is confronted with arise, nor exactly what sort of information is crucial. Those are determined by what the organization is trying to do and varies from one part of the organization to another, depending on what that part is trying to do. That is because different things one is trying to do have different bottlenecks, different main loci of uncertainty about the future, and consequently different information that will bring success. It is, then, the factual distribution of information relevant to crucial uncertainties (crucial in particular, in economic organizations, for the net value of the object or service produced by the organization) that determines the actual structure of the problem confronting the functional structure. Structures of organizations, and of parts of organizations, vary according to the sorts of uncertainties they confront, and so according to what sources of information they depend on and to how that information is best got to the decision-making units. The board of directors is organized as a committee, the assembly line as a hierarchy, because the financial and trustworthiness uncertainties faced by the board are different from the labor cost uncertainties faced by the assembly line. The crucial thing for an organization from this point of view is to be where the news breaks, whenever it breaks. Information is "news" for the organization when it is a first appearance of some sign of how the future is going to be, in a respect crucial for the organization. The argument then is that news about one sort of contingency facing an organization (e.g., fluctuations in the quality of raw materials) is likely to be found in a different place than news about another sort of contingency (e.g., the risks of a hostile takeover). One therefore needs a type of structure to gather and process information from raw materials sources to manufacturing different from that needed to gather and process information about suspicious movements of stock sales. In both cases, however, the organization does not know what kind of world it is making decisions in, because it cannot predict the future very well—not well enough to stay afloat. It needs the news, and it needs to respond to the news as it breaks, not after the news gets to all the participants in the market in the form of price changes. It is in fact such reactions to news that inject information about what is happening to corporate futures into the price system. Much of this book will apply the above approach to economic organizations, partly because the classical literature on economic organizations has been sensitive to risks and to information processing and partly because economic organizations is what I know most about. However,

4 the language above may suggest more restriction of the problems of uncertainty and information processing to economic organizations than is appropriate. Certainly a university scientist, in order to establish that brief moment of monopoly reflected in his or her name as author on the first publication of a new finding, is highly uncertain beforehand about where new knowledge is to be found (see, for example, the discussion in Watson's The Double Helix [1968] 1981) of getting the x-ray diffraction results from Rosalind Franklin, which limited the possible structures of DNA) and how likely it is that someone else will get there first. Such a scientist (and therefore the university) must be very attentive to the news of scientific results related to his or her line of research. That uncertainty of research is compounded as deans and department heads bet on who is going to make the correct bets about where new knowledge is to be found, without themselves being able to make those bets as well as the people they are hiring (for example, see the evidence of wrong personnel bets by Watson and Crick's administrative head, Sir Lawrence Bragg, in Watson [1968] 1981, 37–40, 60–62). Universities also tend to grow toward sources of news that tell them as soon as possible what the future of science and scholarship is going to look like, and to organize their "administration" so that decisions on research and hiring are taken primarily by people who have efficient channels to that news. We will treat a few problems in university administration in Chapter 9 to show that the approach taken here extends beyond economic organizations.

Uncertainty The basic idea here, then, is that uncertainty is not an indelible characteristic of a particular decision. Instead it changes over the course of the decision. That is why people take decisions tentatively. They enter on the first branch that might lead to an investment (e.g., a "conceptual study" in engineering) in order to get part of the news that will reduce uncertainty (see Marschak, Blennan, and Summers 1967, 49–121, on the process of decision to follow or not follow a line of development in designing jet engines; see also Crecine 1969). In general, the further a commitment to a line of action extends into the future, the larger the proportion of all the information entering into the final decision will be that is news coming in over the course of implementation of the decision. People do not decide to drill exploratory wells until after geological studies have shown a promising formation; they do not drill the first production wells until exploration shows that the find is

5 "commercial" (i.e., that its rate of flow, costs of development, transportation costs, etc., combined with anticipated oil and gas prices over the course of the wells' lifetime, show "adequate" anticipated profitability); and they do not develop the whole field until the first production wells come in as anticipated (see Stinchcombe 1985e, 41–65, for a more detailed analysis of the uncertainties in oil field development). That is, they are not nearly as blind to the nature of the future at the time a major investment is made as they were when they first set out with sonic equipment to explore the shape of the rocks a couple of miles down from a given pasture. But of course, at the very end OPEC may fall apart and the field turn out not to be commercial after all (Herskedal and Kristiansen 1987). At this point, after one has made tentative decisions to go a bit further, waiting for the final commitment until as much news as possible comes in, then one has to estimate the probability distribution of oil prices and of true rather than estimated construction costs, and "take risks." Uncertainty is reduced through news; then finally the residual uncertainty is transformed into risk, and people make their bets. Or perhaps better, people make small investments and build a small structure to collect relevant news; if the news is good, they make bigger investments and develop a larger structure to collect relevant news; and so on. Uncertainty is transformed piecewise into risk, with a large part of the risk at first being a guess concerning the value of the news that a news-collecting structure will bring in. This orientation to news must partly explain the fact observed by Mintzberg (1973) that top business managers prefer their information "live," from conversation, telephone calls, inspections, problems brought in by subordinates. Top managers are primarily responsible for dealing with uncertainties, because they are nearest in the hierarchy to profit takers who ultimately take the financial risks. So the parts of the total communications structure they are likely to be most attentive to are the parts that follow the fast-breaking news, the news that reduces uncertainty. It will be important for them to use fast communications channels, voice and telephone, rather than waiting for a memo to be drafted, revised, and forwarded with appropriate signatures up the line.

Uncertainty about What? The primary task of this book is not to follow the drama of taking risks in the light of the last possible reduction in uncertainty. Rather, it is to analyze the structure of organizations as determined by their growth

6 toward sources of news, news about the uncertainties that most affect their outcomes. The basic notion is that those uncertainties are distinctive in different parts of the organization, depending on their distinctive tasks and environments. A way to conceive of this is as follows. By analogy with the findings of linear programming methods for optimizing complex decisions, we can imagine that any particular process has from two to a few "constraints" on its success. At any particular instant, there are likely to be only one or two variable factors that will improve short-run success (such as persuading all the workers to come in Monday morning on time), but those may change from instant to instant over quite short time periods. Over longer periods there may be several constraints which may be overcome or removed, at greater or lesser expense. We can for convenience call these the "bottleneck" factors in a given type of work with a given set of historically derived assets. Many of the things that are essential for the process to go on are not constraints at a given time (ordinarily, for example, air for workers to breathe is not a constraint; at times in underwater construction, however, it is the crucial constraint). The basic generalization is that only uncertainties about the constraints on success matter, and so only information about those constraints on success is crucial to the operation. Because the constraints are likely to change over time, the uncertainties that are relevant and the news that resolves those uncertainties are likely to change. It may often be, however, that the changing information is structurally located in the same place, so that the same structures will carry it back to the organization. For example, the information relevant to figuring out the structure of DNA changed as the alternative hypotheses under consideration changed with the gusts and flurries of facts and ideas (Watson 1968). But most of the information relevant to any of the alternatives was located among physicists and physical chemists interested in biological problems, especially the "phage group" organized around Delbrück and Luria, the crystallographers working on proteins organized around Wilkins and Franklin, and the physical chemists organized around Linus Pauling. The informal network bringing in the news did not have to be rebuilt with each shift in hypotheses, because basically the same sources of news had the facts about each of the alternatives. Similarly, as we will examine in more detail in Chapter 5, information about needed improvements in an innovation that will make it more marketable comes from those who are in contact with the innovation users. In the early days of computers, given the structure of maintenance agreements and software development at the time, these were likely to be

7 the maintenance people or the "client representatives" of the vendor of a computer, the folks who would be in contact with the computer center of the vendor. So it was essential that information from the client contact people get back to the software and hardware developers in the vendor organizations, whether the design problem was with the software, the mainframe computer, or the peripherals, whether it was in the early days of the computer model or the last days before it was replaced by new models, and so on. It is such repetitive use of information from a given source of news, as problems and constraints on success change and new uncertainties therefore affect decisions, that tends to cause dense and fast information collection and information processing in organizations to be oriented to particular sorts of uncertainty and particular sources of information. It therefore follows that different organizations, having different main constraints, hence uncertainties, and hence news sources, will need to grow different types of structures of information collection and processing. The basic functional postulate of the argument of this book is that such need produces a tendency to grow such specialized structures for processing information about the relevant uncertainties that differ in different organizations. But the same logic applies to the uncertainties of different tasks in different parts of the organization. Different uncertainties matter for different products, for reaching different clients successfully, for arranging the flow of borrowing and repayment so as to give stockholders their optimum risks and rates of return (and so the maximum value of their stocks) in different states of the financial markets, for maximizing university prestige in molecular biology versus other parts of physics or other parts of biology, and so on. We will therefore expect that the structures for information collection about crucial uncertainties (and especially that part of information processing that involves rapid use of fast-breaking news) will in general tend to be different in different parts of the organization.

Information Up to this point we have emphasized the temporal features of that sort of information that reduces uncertainty, that what is uncertain at one time becomes predictable (or becomes calculable risk) at another time because new information (news) comes in. But many other features of information besides its recency make it functional for reducing uncertainty. Consequently, many other features of an information-

8 collection and -processing apparatus—besides being located where the news about the relevant uncertainties is breaking and getting that information fast to the relevant decision makers—determine the degree to which they "serve the function" of controlling uncertainty, turning it into calculable risk. We will discuss these secondary features under the rubrics of: (1) units of analysis; (2) noise reduction; (3) level of temporal abstraction; (4a) trustworthiness and error estimation and (4b) agency problems in information trustworthiness; and (5) expertise of analyzers. 1. Units of analysis. The kinds of causal unities about which an organization needs coherent information are those that make several activities subject to the same constraints: then units made up of those activities have the same conditions of success. For example, in studying production delays in a steel mill in South America, I found that the hot rolling section of a tube mill collected and analyzed delay information but that the cold processing parts did not and their delays were not treated as part of the problem of the efficiency of the plant. The reason was that on hot rolling operations there is a high degree of technical interdependence of the different operations, and they have to be done in tight sequence because otherwise the steel will cool down so much that it cannot be worked. That meant that the success of any part of hot rolling depended on delays in every other part, so any constraints that caused delays in one suboperation were a constraint on all the others (Stinchcombe 1974; Stinchcombe and Harris 1969). It was very uncertain whether at any particular moment the line would run, because many interdependent difficult operations all had to run at once for the line to run. Similar inputs (e.g., preventive maintenance for the whole hot line) would tend to remove all the constraints on operation, so information about maintenance problems of any part of the hot line would be useful in forming a remedial policy. Incentive systems that encouraged one shift not to leave steel hot, ready to be rolled, depressed productivity of the whole line on the following shift. In short, the technical unity of the line meant that it was subject to common constraints, that there were common remedies for those constraints, and consequently that common information would be valuable in increasing productivity of each of the parts. For another example, we treat below in Chapter 4 the question of what Chandler means when he says that different lines of merchandise will require separate administrative divisions. One of the examples he gives is Pittsburgh Plate Glass. The firm added several lines of merchandise that tend to be used with plate glass, such as paints, brushes, putty,

9 and the like, which have no technical unity with plate glass in production; they do not need to be heated and formed while hot or shipped in special ways because of their fragility. But the information about what is selling to finishing contractors in lumber yards and hardware stores around the country is located in the same place for glass as for putty, and for both of these as for paints and brushes. Consequently, the lack of technical unity in manufacturing did not require divisionalization of Pittsburgh Plate Glass. When the firm started to sell some products to other manufacturers on a large scale, however, those products and their manufacturing and marketing staffs needed to be divisionalized because they had to respond to a completely different structure of uncertainties and information-processing requirements. The news about what was selling in lumber yards was not useful in forming an inventory policy about such industrial goods. The market constraints on plate glass and paint brushes, then, tended to be similar, and to be solvable with information from the same structure of contacts with lumber yards. The market constraints on industrial chemicals were not similar, and so they had to be placed in a different division. For Sears, analyzed also in Chapter 4, the statistical system for inventory control in stores requires each distinct commodity to be treated separately, since all the thousands of commodities that Sears sells in its department stores are bought separately, reordered separately, inventoried according to the expected rate of sale and the reorder interval, and so have to be treated as separate units of analysis. The "unit control" system developed in Sears retail operations provided such a set of statistics on the sale and inventory of separate products. There are two general points to be made about this. First, one can support a functional argument that an information system has its origin in a need to respond to a given sort of uncertainty, if it has the units subject to that sort of uncertainty as its units of data collection and analysis. Conversely, one can assume the functional argument and infer from the units of the information system what the uncertainties are. We infer from the units of a preventive maintenance system that the hot rolling section of a tube mill has as a principal constraint the maintenance of all its parts, while the cold processing operations are not subject to the same constraints and so do not need information about the same uncertainties. 2. Noise Reduction. The odds at the race track and the average price of a stock over a week or so are good estimates of the long-run relative merits of horses or of the expected earnings and risks of a corporation. But almost all the bets of particular people, given the odds, and almost all the minute-

10 to-minute or hour-to-hour fluctuations of stock prices are error noise around that good estimate. Race track touts and investment analysts cannot ordinarily beat, respectively, the odds or the market, so their advice about what horse to back or what stock to buy is random noise. What one wants in dealing with uncertainty is a system of fast analysis (as the track itself has for odds, for example) that will eliminate the noise and error and provide only the valid information about the constraint. IBM, in pricing a personal computer, does not want to know that a particular retailer always undercuts the official price by 3 percent unless it can learn how the retailer can afford to do that and then use that information to its own profit. But the company does need to know if all retailers are cutting the price by 3 percent rather than 10 percent, because 3 percent means it is meeting the competition in its pricing policy, 10 percent that it is not. A central kind of noise is irrelevant detail in a report. A twenty-page output from the accounting department for department or project administrators has to be reduced, often by pencil calculations (usually by administrative assistants or the equivalent), to about three or four numbers that tell where they stand. In analyzing an information-processing system, then, one wants to know the four numbers that the best administrators of departments or projects have calculated for them, rather than what all is included in the twenty-page biweekly output that is stacked in a closet (after those four numbers are abstracted) until the auditors say it can be thrown away. Such detail is comparable to knowing how much each bettor has bet on each horse rather than knowing the summary (the track odds). Another kind of noise is, of course, systematic error. Deans complain, for example, that all professors they talk to earn less than average for people of their rank and distinction. The fact about which they complain comes about because professors have reasons to select the set of people or institutions to average so as to make the average as high as possible without seeming absolutely ridiculous (see 4b below on agency). We will deal with the general problem of systematic error in the section below on trustworthiness, and reserve the term noise reduction for reduction of random error. The basic device for reducing random error is to take a mean (or median or percentage); the larger the number of observations over which the mean is taken, the smaller the residual random noise. But a larger sample of fast-breaking news is always late. Consequently, the dilemma is to produce the most recent mean possible, compatible with its being sufficiently noise free to indicate what is happening to the crucial uncertainty, to indicate how the world is developing. If an information system

11 shows signs of this tension between noise reduction and newsworthiness, for example by using a four-week projection of demand updated each week (as General Motors developed—see Chapter 4 below), this shows that it is oriented to getting noise-free information about that uncertainty that is recent enough to adapt to a changing and uncertain world. The alternative strategy for dealing with noise occurs when each case has to be dealt with in the way it in fact turned out. In the textile industry, the mean strength of a thread being spun or woven does not tell when it will break; the weaver-loom system has to be set up, then, to deal with each break in the thread after it happens, and taking the mean time to failure only helps in buying thread, not in running looms (Blauner 1964, 59). The machine has to be stopped and the break tied up when it in fact happens; one cannot set up the work on the shop floor based only on the fact that an average stretch of thread in a spinning machine will not break. Skills in the division of labor are primarily for dealing with variation in the situation: in the raw materials, in the design requirements, in the previous errors (for instance, previously built walls not being quite plumb), in the weather, and so on. The reason one needs the skill of a carpenter is that one gets the news that the wall is out of plumb when one is hanging the door, and so one needs to deal with it at that point. In Chapter 2, we deal with the skill composition of the labor force as a response to substantive noisiness of the work process, the amount of random variation that has to be dealt with rather than averaged out, drawing heavily on the work of Charles Sabel (1982). 3. Level of Temporal Abstraction. Different parts of an organization deal with uncertainties that develop over different time spans, and different organizations have crucial uncertainties that vary over different time spans. The abstractions built into the information-collection and information-processing structures have to be adapted to the time spans over which adaptation takes place. For many manufacturers, much of their capital investment is in buildings that are reasonably easily turned to other purposes, with the old machines moved out and new machines moved in. But oil and chemicals firms build special-purpose capital equipment (often roughly 5 percent only is for the building, 95 percent being for the processing installations); such special equipment has to pay for its entire cost by a long time doing the single thing it is built for. Ordinarily in the oil and chemical industry, the total amount of labor that goes into building the equipment is of the same order of magnitude as the total amount of labor for running the equipment for twenty years. If the price of the product falls

12 below the operating cost of the installation, that installation, and consequently that half of the total labor that built the installation, cannot be used for anything else. One reflection of this is that a large proportion of all the middle-class people employed in the oil and chemical industries are accountants, specialized in estimating financial outcomes into the far future under various long-term uncertainties about costs of raw materials, labor, and product prices. Ordinarily the response to random variations in the raw materials, in the degree to which a particular wall is straight up and down, in what the latest client to walk in the door seems to have on his or her mind, in just what about quadratic equations a particular high school student seems to be having trouble with, is organized into an entirely oral information system—the informal world of skilled manual work, sales work, or classroom teaching. So the shortest temporal abstraction level is reflected in an information system within the individual, or at most in the interaction of a small group, and appears in our organizational analysis as human capital, as "skill." As one goes up the hierarchy of most organizations, the time span of responsibility increases (Jaques [1956] 1972), which means in turn that the uncertainties one is responsible for responding to develop over longer periods. While the skilled worker or salesman responds from minute to minute to random variation, the manufacturing division responds to shifts in demand for automobiles that are updated each week, and the general office responds to quarterly or yearly profits, yearly shares of the market, and other indicators of where long-term investments should be made or where shifts should be made among executives who are not, evidently, responding effectively enough to weekly variations in demand. Research and development departments tend to be oriented to the very long run, though they are occasionally brought back to earth by a computation of profits on recent innovations by the central administration (Lawrence and Lorsch 1967; for critiques of this research, see Tosi, Aldag, and Storey 1973; Blau and Meyer [1956] 1987, 111–116). The result is that information-processing systems have to aggregate information over different time spans, and when different parts of the organization respond over different time spans their information systems need to be segregated. We discuss in Chapter 5 how information systems for feeding market information back to engineering and manufacturing have to be faster and richer for innovations than for products from which the bugs have already been worked out. Consequently, the information systems for an innovation in general have to be segregated into an autonomous "divisional"–type structure.

13 Similarly, in Chapter 3 we discuss how the "well report" for past oil wells and the "well plan" for a well to be drilled in the same field bear an intimate relation over the very short run from the point of view of drilling operations. But all that absolutely critical detail disappears from long-term plans for connecting drilled production wells to the processing apparatus that prepares the oil and gas for shipment, for here the crucial question is whether the connections between the two have any leaks that could create a danger of explosion. From the point of view of commissioning the processing operation (starting production), the dates and mechanical details of the "completion" of the well are all the temporal detail one needs about drilling. The level of temporal abstraction of any information system, then, serves as empirical support for an argument that the system is supposed to deal with an uncertainty developing at that time scale. Conversely, again we can assume the functional connection and use the level of temporal abstraction to find what sorts of constraints a given piece of the administrative apparatus is oriented toward—the constraints that operate at the time scale of the abstraction of the system. Such information flows with different levels of temporal abstraction are central to defining "decentralization" in Chandler's work, analyzed in Chapter 4. 4a. Trustworthiness and Error Estimation. Scientific papers are required to present the evidence on which the argument they make is based. Textbooks do not generally have to analyze the evidence, nor do they have to say what sorts of errors such evidence is subject to and what protections against error one has built into the experimental procedure. This difference is due to the difference in the degree of uncertainty about the arguments in question. In general, no competent chemist would disagree with any argument in a competent textbook in chemistry (for an exception, see Watson [1968] 1981, 110, 122). Scientists do not have to give their readers the basis on which to disagree with their reading of the evidence unless there is some uncertainty about how to read the evidence. It is assumed that if something is sufficiently uncertain to be worth publishing a paper about in a scientific journal, then it is sufficiently uncertain that one has to specify the evidence on the basis of which one makes a particular argument about it. Consequently, there are routinely descriptions of experimental procedure (in highly codified form, of course) in the scientific literature but not in the typical textbook. A system of discourse that attaches qualifying information to assertions, so that the reader can go behind the assertions if necessary, according to the degree of uncertainty generally thought to attach to the conclusions, is a "trustworthy" system of information. Science, then, is a finely tuned trustworthy system because the relatively certain assertions

14 of the textbook have little data that allow the reader to assess the uncertainty, exactly because there is not much uncertainty; the more problematic assertion of a new discovery has to have evidence presented and evaluated in the standard scientific way. Hence, the detail of the evidence is proportional to the uncertainty of the assertion. Such systems do not provide useless information about certainties, nor do they present uncertainties as revealed truth. This does not mean that there are no errors in the information. If there was no chance that scientific papers were wrong, scientists could just publish the conclusions without the evidence. And of course, the textbooks of a generation ago have in them things that were consensual in the scientific community at the time but that are now "known" by consensus to have been disproved by research in the meantime. A trustworthy system of information, in short, is one that does not routinely deceive its users about the uncertainty of the conclusions reached on the basis of that information but at the same time does not burden them with information about stable features of the environment. One does not ordinarily watch the news from the scientific community to see whether a given paragraph in one's physical chemistry textbook is wrong; one does watch that news to see whether conclusions of experiments reported in the most recent scientific literature stand up to different procedures in different laboratories. One is not surprised to have a Nobel Prize winner write a sentence like "The letter was not in the post for more than an hour before I knew that my claim was nonsense" or for him to be advised by a mentor, who had forwarded some problematic work for publication, that "in this way, I [Watson] would still be young when I committed the folly of publishing a silly idea" (Watson [1968] 1981, 110). Scientific research is not worth doing unless it is in an area where the right answer is uncertain, and consequently scientific work is required to give others the basis for assessing its own degree of uncertainty. It is exactly because Watson was working in an area where he was quite likely to be wrong that he was required to give the evidence and the details of the argument. Similarly in the oil business, everyone knows that many "promising structures" on the basis of acoustic exploration have no oil or gas in them and that one knows for sure only when oil and gas comes into an exploratory well. Further, one has a reasonable estimate of the recoverable reserves only after several exploratory wells have "delimited" the reservoir. A report of acoustic exploration thus has attached to it a professionally well understood degree of uncertainty, the first exploratory well a different (smaller) degree of uncertainty, and an estimate of recoverable

15 reserves based on several exploratory wells a still smaller degree. No one professionally concerned is likely to mistake one report for the other. The information system is "trustworthy," even if uncertain to different degrees after each estimate, because each estimate has attached a professionally understood assessment of its uncertainty. If, however, an incentive system exists in which the executive in charge of estimating recoverable reserves is punished severely for producing an overestimate, while no serious consequences flow for him or her from an underestimate (presumably because one does not know one has lost money by not investing but always knows one has lost money by investing), that executive may not be rewarded for attaching the correct degree of uncertainty to the estimate. Then he or she may report lower reserves than the evidence justifies. The system becomes untrustworthy at that point, because there is a systematic bias in the estimate of the probable error. (It may be deliberately arranged to be biased downward if it is known that the decision based on the information will be taken by amateurs and so will not take account of the evidence on its degree of uncertainty; it would, however, be an unusual—and probably short-lived—oil company in which this was the case.) 4b. Agency Problems in Information Trustworthiness. One can tell that the travel voucher and entertainment expenses accounting system is oriented to contingencies of fraud and extravagance by the safeguards built into it. It is very hard to collect information on entertainment expenses that tells accurately how much entertainment expense it was rational to expend on the company's behalf, that provides a reasonable estimate of the marginal productivity of the last dollar of entertainment expense. In this case the low trustworthiness of the information is due to the fact that the person who has to report on the value of the entertainment to the company is also a person who benefits from a more expensive evening out. Organizations write multiple rules about the external features of entertainment and travel expense (those governing travel by federal employees are a wonder to behold) that, one hopes, limit the degree of extravagance and fraud. Protections against error that an organization builds into its information system tell a lot about what the organization really wants to know. Many of the sources of untrustworthiness of information are faults in the incentive system for providing accurate information and accurate estimates of how likely one is to be wrong. These are generally dealt with theoretically under the heading of "agency theory," which deals with the broad class of problems in which the agent has more information—later

16 or better news—than the principal about the uncertainties involved in a decision that the principal has to take (or that the agent has to take on the principal's behalf). Agency theory deals with such things as "moral hazard" in insurance, in which the policyholder has information about strategies for reducing loss that the insurance company does not, but is not motivated to use those strategies because the insurance company pays the losses; or "adverse selection," in which the buyers of insurance policies know that they are worse risks than the company has estimated, so the insurance is a very good bargain for them, while those who fail to buy policies know that they are very good risks and so can better self-insure. In general, information sources are people, who have motives to distort information so that they look better or worse than they really are. Such motives make information untrustworthy. The multiple devices that insurance companies have developed to deal with these problems (see Heimer 1985a, 11 – 17 and passim) indicate that trustworthy information about expected losses requires special structures to shift the incentive system. These devices also indicate that getting accurate information on the risks being insured, against the possibility of moral hazard or adverse selection, is a crucial uncertainty for insurance companies. As Shapiro (1987) has pointed out, devices to make information more trustworthy can be simulated to make fraud more effective (for case material and evidence on how hard it is to find fraud, see Shapiro 1984). For example, a scientist's conclusions look more certain than they are if he or she reports faked data than if he or she merely reports an "expert opinion." Simulating the devices that we use to make scientific reports trustworthy makes false results look more trustworthy. Every device for making information trustworthy provides an additional reassuring cover for a con game when the device can be simulated; financial insurance companies with slim reserves may have the initials FDIC, for example. A baroque structure of reviews, signatures required, committee consideration, and auditing usually indicates that the information system is endangered by agency problems, problems of motivated distortion of the content of an information flow; the tenure process in a leading university is a good model for what an information-processing and decision-making system with severe problems of agency looks like. But those same structures, by the functional argument, are indicators that agency-induced errors in the information will tend to lead to important errors in the decisions. 5. Expertise of Analyzers. The maintenance of the quality of the flow of information is in general achieved by constant minding by people who

17 know what uncertainty is to be analyzed, where the news about it is to be got, what causal unities determine which units are subject to that uncertainty, how to make the tradeoff between currency of information and noise reduction, what temporal span the decisions cover, what degree of uncertainty different sorts of information are subject to and how to indicate the level of uncertainty correctly, who is motivated to distort the system, what auditing or control procedures will work, and who are the other experts to consult on all these questions. Data on the expertise of those who mind the information flow supports a functional argument about what sort of uncertainty, and which constraints on success, the information flow was developed to control. Conversely, assuming the functional argument, one can tell that if high-priced accountants are minding the information flow, the long-term financial outcomes are the crucial things being estimated; if engineers in charge of maintenance are minding it, the constraints are likely to be reliable performance of machines; and if auditors whose chief qualification is suspiciousness go over the accounts frequently, the validity of claims and the likelihood of fraud is a crucial constraint on success.

Structure and Function Many functions never get fulfilled: getting accurate information on the quality of a used car when the place where the news is located is the used car salesman is not a social structural possibility. Conversely, many functions that are carefully fulfilled solve problems that were never there in the first place: certifying that high school teachers understand curriculum theory or audiovisual methods of instruction has a clear manifest function, though it seems to do no good (the movie projector in the school gives both teachers who have and those who do not have audiovisual training a few minutes peace and classroom order, a true and important function). Functional arguments are always suspicious because they attempt to explain things by their consequences, and this obviously requires a reverse, consequence-governed, chain of causation, a causal loop so that a consequence can explain its cause. Such arguments are particularly suspicious when one is explaining the structure of information flows in a situation in which rewards depend on what information is given. Teachers and teacher professional organizations, through their certificates, create an information system that looks as if expert judgment is used to select instructional movies. Of course, the one way to select good instructional movies is actually to look at a lot of movies in the subjects

18 one teaches and pick out the best ones. But no school board (that I know of) is willing to pay teachers a skilled worker's wage to watch movies all summer. The function of the certificate is to monopolize positions as teachers for those who have gone to education schools (where they teach audiovisual methods) and completed liberal arts programs, and to provide something to justify having people around that education school who produce or market instructional films. Since the manifest function of selecting good instructional movies would be very expensive and no one believes in such selection enough to pay that much for it, and since the latent function of this certificate is important enough to some people to keep the system going, an information system is created with very little information in it (certification systems for skill are treated as information systems in Chapter 6; the origins of skill requirements are treated in Chapter 2). In this case we can perhaps identify the causal links back from the latent function to the structure and content of the information system, having to do with the fluctuating power of education schools (which have more luck requiring such certificates when teacher training is popular and many teachers are required by the system). Consequently, we do not have to depend on the manifest function for an explanation. But the first evidence we used to suggest that all was not as it seemed was to check whether the information system had the characteristics that would be required for its manifest function: is the information in the system about the principal source of uncertainty on instructional films—namely, whether the film is appropriate to the students and to the instructional purpose of a particular teacher? Since the judgment of whether a film fits is clearly best made by a teacher seeing the film, the characteristics of the required information system are clear. One can very often disprove functional hypotheses without examining the reverse causal chain, by examining internal evidence about what the structure would have to do for the functional hypothesis to be true. The field we are mining in this book, however, is shot through with situations in which people's rewards depend on the information flowing in an organizational system. Because a sociologist's fundamental inclination is to look at the seamy side of life for explanations, the big sociological fact may seem to be that people are motivated to misinform. To convince the sociological community that explanation by manifest functions of getting the organizational job done is fruitful is therefore rhetorically difficult. Our main hope is to do so by showing that many details about the setup of organizations that are otherwise inexplicable

19 are explained by the proposition that information systems tend to grow toward sources of news about the central uncertainties of that part of the organization, and that the structure of the information-processing system can often be predicted from the assumption that its design gets the job done. Thus, the alternative hypothesis that the information system is a sham, designed to give cushy jobs to its perpetrators, can sometimes be refuted by showing that that alternative does not explain the elegant fit between what an information system does and what is needed to get the job done. A second problem with functional explanations is that they take such things as organizational constraints, and uncertainties bearing on those constraints, as real, as causal variables. These causes work not only by functional mechanisms, already suspicious, but at supraindividual level of explanation as well. Since, of course, all the significant organizational acts that we will be explaining are done on purpose by conscious individuals, any explanation at the social level has ultimately to be shown to be adequate at the level of intentions of individuals. I address this problem in detail in Chapter 4, in connection with Alfred D. Chandler's (1962) analysis of the origin of the multidivisional structure in Du Pont. But we can sketch here the broad lines of attack that this book takes. The first line of attack is that in fact, empirically, we often find that an organized structure of intentions of a lot of different organization members exists, and that that organized structure is such as to get the organizational job done. The interconnections of those intentions are of such a character as to cause organizational information systems to grow toward the sources of crucial uncertainties for organizational success, and the substantive content of the structure of the information processing system as created by those intentions is in fact such as to collect information on the right units—to reduce noise, to abstract in the right temporal frame and to segregate different temporal features of uncertainties, to assess the degree of uncertainty accurately and to control agency problems of information quality, and to put the right sorts of experts in charge. If in fact individuals make up functional structures, this is surely an important dependent phenomenon that people who specialize in explaining individual behavior ought to address. This individual-level explanation is God's work, but not my own particular vocation. The second line of attack is to assume that people as individuals confront the problems posed to them in their individual work, and that a long series of solutions to such problems of how to do one's individual assignment are such as to make the organizations "climb the gradient" and grow a structure that responds to the crucial uncertainties that

20 govern organizational success. We assume organizations climb that gradient faster if the people in charge of creating and maintaining information are truly experts, if the general organizational tradition is one that discourages corruption, fraud, and manufacturing evidence and the like. This leaves the intellectual challenge of specifying how organizational problems become problems assigned to individuals who have the "obligations, rights, incentives, and resources" (Heimer 1986b) to solve them. A third line of attack is an evolutionary one, particularly relevant to Chapter 5 where we specify some features that administration of innovations needs to have. A careful reading of that chapter should suggest that many organizations are not well set up to introduce innovations, and may fail even if they are because introducing innovations is organizationally difficult. The uncertainties that confront an innovation are severe in all sorts of areas. But the high rate of failure of innovations means that the features we find in those organizations that innovate successfully are likely to be functional for innovation. If in the long run business firms can be profitable above the interest rate only if they introduce innovations (as Schumpeter 1942 argued), and if organizations last longer if they are profitable, then organizations whose innovating arms have grown faster and more accurate information systems that reduce the uncertainties connected to the innovations are more likely to be around for sociologists to study (cf. Nelson and Winter 1982). Similar evolutionary arguments may be used to suggest that administrative structures in General Motors, Du Pont, or Sears may be more functional than those of Kaiser-Fraser, Lydia Pinkham, or Montgomery Ward.

The Plan of the Book The overall organization of the book is to ascend the levels at which information and decision systems are organized, starting with individuals' skills and ending up with class relations in whole societies. We pass from individual skills to technical departments in manufacturing, then to divisional organization, to innovations that cause reorganization of these first three, to networks of contracts among organizations, to segmentation of labor markets in the economy as a whole, and to the formation of workers' information on their class and political interests. We then hop sideways from these materials (largely focused on economic organizations and their offshoots) to illustrate the application of the theory to university research administration. In Chapter 2 we start with variations in the basic information-pro-

21 cessing mechanism of modern social organizations: the employee. The main social structure to be explained in this chapter is the division of labor, the main information-processing mechanism is skill of individuals, and the main social effect is the stratification of employees by level of skill. This definition of the problem is implicit in Charles Sable's Work and Politics: The Division of Labor in Industry (1982), so we base our analysis on that book. Our main problem is to specify precisely enough what skill consists of so that we can build a theory of when it will be necessary. We argue that all work is characterized by an inherent dilemma: that productivity is highest when all the activities necessary to production or delivery of a service are highly routinized but specialized human work is most valuable (and hence the incomes of organizations highest) when there is uncertainty about what to do. Routinized work is more productive partly because people work faster when they are doing something that is routine for them, but mainly it is because, once routinized, the work can be mechanized or computerized or otherwise improved, and the improvement will stay in the productive system as part of the routine rather than as a sporadic flash of genius. But routines are in general effective only if used under the appropriate circumstances: only if the parts to be assembled are standardized, only if the clients in a restaurant come in with standardized wants and select from a limited menu, only if the pilot can depend on his (or her, if airline piloting were different) counterpart workers in the control tower to behave in a routinized fashion so that the information they give him has an absolutely clear meaning. This means in turn that uncertainty in the environment of work—unstandardized parts, unstandardized clients, or unstandardized fellow workers—undermines productivity by undermining routinization. We will analyze skill as the capacity to routinize most of the activity that comes to a given work role in an uncertain environment . We will argue that skill is a repertoire of routines which the workers can do accurately and fast, as well as a set of selection principles among routines, such that the complex of routines and selections among them deals with most things that uncertainty brings to the worker. Thus, we will expect to find skill when a great many different things must be done to produce the product or service but when each of those things has to be done in several different ways depending on the situation. For example, more complex products are likely to require higher skill levels in the work force that produces them, but the highest skill levels should be found in industries with complex products that have to be customized for particular clients and constructed in varying environments, as in the building construction

22 industry. Much of the chapter, then, consists in the analysis of variability and uncertainty in exactly what work has to be done at a particular time, uncertainty about which of the routines in workers' repertoires have to be pulled out. Chapter 3 turns to the variation in structures of administration in different sorts of activities within manufacturing, by showing that the system for extracting, treating, and shipping crude oil has many different types of information problems, and so many different structures for processing the information. The data for this chapter come from a study of building such a manufacturing administration in the Norwegian State Oil Company (Statoil), the first such organization built by a Norwegian oil firm (other than the Norwegian branches of multinational oil firms, which of course are legally Norwegian oil firms). The different things one has to do, say, to build an information system adequate to buy a stock of spare parts rather than to develop accounting software or to drill a production well a couple of miles deep without killing anybody are especially clear. (Much of the material for Chapter 3 has been previously published in photocopy form by the Institute of Industrial Economics in Bergen, Norway [Stinchcombe, 1986d], which sponsored the research.) Chapter 4 reanalyzes Alfred D. Chandler's classic Strategy and Structure: Chapters in the History of the American Industrial Enterprise (1962). The central concern of that book, we argue, is how information coming from different sources, especially different markets, is combined to make different sorts of decisions. Chandler argues that when the information needed to make money off a company's products or services comes from several sources and has to be related to several distinct systems of decisions, one has to form more or less autonomous divisions to integrate the information from a given market and then arrange for the division so created to pass on to the central office only abstract and general information relevant to financial and investment decisions. Chandler's argument then becomes a functional one: that in fact organizations that need divisional decentralized structures because they are in multiple markets tend to get them. First we undertake to define more exactly, in terms of information-processing and decision-making structures, what Chandler means by his dependent variable, decentralization into divisions. We then examine in detail how Chandler builds this functional argument for a case where he has the most historical detail, the origin of the divisional structure of Du Pont. Chandler has a good deal of information on how individual actions inside Du Pont added up over the course of several years to a shift to a divisional structure. This gives a hint of how the methodological

23 objection to the organization-level functional analysis (that is the basic strategy of this book) can be disaggregated into individual human action that is understandable. The core of the analysis of Chapter 4, however, is the explication of Chandler's implicit definition of when a firm is in multiple markets. Our argument is that, for Chandler, a market is a matter of where the information about the central uncertainties about a flow of products or services lies, and how this information can be got to decision points and integrated in a sensible way, rather than primarily about who the firm's competitors are. Chandler uses the economic competition definition of what a market is only occasionally and tangentially; his main, implicit definition is that a market is a number of phenomena outside the organization that produce a common administrative problem—problem of relating information to decisions—within the firm. In one sense, then, the subject of Chapter 4 is the origin of the overall variation among firms between divisional and centralized organizations. But the core explanatory principle is that within the firm the problems of different product lines are different enough to need different information-processing structures. One needs to separate into different divisions administrative dealings that involve different sorts of uncertainty, because different product lines have to attend to different sorts of news, collected in different ways, evaluated according to the particular situation of that line of goods, and integrated with engineering and manufacturing in a distinct way. One needs one information-processing system and set of routines for making decisions to sell explosives to a few mines or the War Department, and another for making decisions to sell paint to thousands of householders or housepainting firms. The central administrative mistake, according to Chandler's argument, is to administer news about one sort of uncertainty with a structure built to handle another kind. Thus, Chandler's argument is exactly in tune with the main thrust of this book. Chapter 5 deals with the special kinds of uncertainties that are ordinarily associated with innovations and with the sorts of information-processing and decision-making problems that they pose. The analysis can be seen as an elaboration of Joseph A. Schumpeter's examination of the relation between economic innovation and routine administration (e.g., Schumpeter 1942). Schumpeter identified routinized administration with large bureaucratic organizations. But in order to have the economic effects Schumpeter urged, innovations must be produced by efficient—which means routinized—production processes and must reach the market in routinized channels. The temporary monopoly over

24 a valuable product or service that in Schumpeter's analysis creates the profits from innovation does not in fact produce profits unless the innovation can be produced on a large scale and marketed to collect the extra margin created by the monopoly. Our argument will be that the introduction of an innovation involves a higher level of uncertainty than do the production and marketing of goods that have been on the market for a long time. The news needed to improve the innovation, to find its market, to introduce elaborating innovations, and to penetrate new niches is likely to be found in a different place than those which the information system of an existing organization reaches. And it has to be dealt with simultaneously and rapidly by engineering, manufacturing, and marketing in much the way that Chandler argues for distinct product lines. Thus, Schumpeter is right in stating that it should ordinarily be hard to introduce innovations by using an existing bureaucratic structure. Yet a structure with many of the characteristics of a bureaucracy is needed to get the profits out of the monopoly created by innovation. An innovation poses an administration with a problem similar in form to the dilemma that we argue in Chapter 2 produces a demand for skill. It also poses a problem of differentiated information collecting and decision making similar to that which Chandler argues produces divisional decentralized administrative structures, as we argue in Chapter 4. The sociological answer to Schumpeter's argument about the unfitness of bureaucracy for innovation (aside from the empirical observation that most important innovations are produced and marketed by large bureaucracies) and the consequent necessity for an individual heroic entrepreneur is that the administrative problems posed by innovation are of the same sort that skill and decentralization ordinarily solve, though perhaps more extreme. Thus, the first functional argument here is that innovations should tend to create pressures for divisional decentralized administration, such that the innovation has a separate but integrated news-collecting, information-processing, and decision-making structure. The second functional argument is that, because the level of uncertainty of the work involved to introduce an innovation will ordinarily be higher than that of other production, such a division is likely to have a higher skill mix than a division with an equally complex product or service that is no longer an innovation. A good portion of Chapter 6 has been published previously as "Contracts as Hierarchical Documents" (Stinchcombe 1985b). The argument is that many contracts between organizations are actually formal organizations themselves, with news collection, information processing, and

25 decision making built into a joint social structure consisting of the two (or more) corporate parties to the contract. Thus, the contract creates a social structure to deal with uncertainty, just as the labor contract does. The chapter then challenges Williamson's argument (1975, 1979) that dealing with uncertainties produces hierarchy. (Williamson's later work [1985] is compatible with the argument presented here.) The argument is that anything one can do with social structures constructed by labor contracts, one can do with social structures constructed by contracts between firms. Williamson's argument is of the same functional form as ours, namely, that when information and decision problems create certain kinds of uncertainties, then a social structure that tends to resolve those uncertainties (that is, a hierarchy) tends to grow to manage them. The only problem with his argument, we will allege, is that the functional requirements of hierarchy itself involve continuity of production, such that the heavy investments in building a hierarchy and training the people to do their roles in it will be productive long enough to pay for themselves. Consequently, when one is only going to be, for example, in the business of building oneself a factory or an office for a couple or three years, one does not want to erect a hierarchy able to build factories and offices on into the twenty-first century. Rather, one wants to hire organizations that know how to build factories and offices, and that have skill structures and incentive systems appropriate to the uncertainties of that business, to put together a temporary social structure to build the building. Then, if it turns out that the site is sand rather than rock, one wants to change some of the specifications for the building in the middle and still use the same system of contracts and subcontracts to build the new design. We will show that when, for one reason or another, it is hard to build a hierarchy to do the job but when the problems that (Williamson argues) tend to produce hierarchies occur, the structures of hierarchy will tend to be built into the contract itself. Thus, we are supporting Williamson's functional argument by breaking down the variable of "hierarchy" into its components, so as to avoid identifying it with the legal unity of a firm or another organization. In short, when functional pressures toward hierarchy exist but there is difficulty building an ordinary hierarchy made up of employment relations, we find instead contractual means for creating hierarchies among corporate actors. Chapter 7 returns to the problem of skill of Chapter 2 from a different point of view. If the division of labor in a set of activities produces a set of skills to deal with the uncertainties, as outlined in Chapter 2, then the personnel system of the organization has to create a status system

26 to motivate the acquisition and utilization of those skills, as well as a recruitment system to fill the jobs with people who will turn out to have the appropriate skills. One of the most complex uncertainties that organizations deal with is uncertainty about whether people will be willing and able to do the work. The information system inside the individual is just as complex as the information system of the organization. Measuring what that system is capable of, then, involves analyzing the relationship between the particular complexities and uncertainties of the work to be done and the complexities and skills of the mind and body of the worker or recruit. It turns out to be very difficult even to measure how well people are doing their jobs because, in general, line management does not fully know the complexity of the workers' jobs, and personnel management knows even less. Unless one knows a job very well, it is very hard to differentiate explanations of low productivity that blame true variations in the world from those that involve lack of competence or will on the part of the worker. But even when one can measure performance accurately because one knows the uncertainties involved in the work well, it is very hard to predict what quality of skill and what discretion in the internal decision-making system a worker will develop over the course of his or her years of experience with the organization and, hence, what kind of productivity, in the environment to be confronted over those years, the organization will get. Further, it is exactly this sort of information about employees' performance that is hard to collect from the people who know most about it—namely, the workers themselves—because they have an interest in getting the rewards that come from giving good news about the uncertainty about their competence and avoiding the punishments that come from giving bad. We deal with four main structures for dealing with the fundamental information uncertainty of the labor market: continuity in the job (seniority in a job as a basis for holding that same job), internal promotion systems ("internal labor markets" in the strong sense), certificates from schools, and certification by a body of peers. We argue that reliance on such imperfect information systems provides the structural basis for segmentation of the labor market, because the fundamental privilege in the labor market is the capacity to give reliable certification that one can do a job. We do not, however, show why it is that some groups (e.g., white males) are more able to give such certifications than others (e.g., blacks or females). Chapter 8 uses the materials from Chapters 2–7 on how organizational structures originate to examine the problem of class consciousness.

27 It is very clear that working-class consciousness is a consequence of the formation, with industrialization, of new kinds of organizations in the economy. Much of the argument in Chapters 2–7 can be conceived as an analysis of how capitalist (and socialist) economic organizations differ from peasantries, agricultural villages under feudalism, and other preindustrial forms of organization. In particular, we argue that such reorganization of the economy tends to produce reorganization of the labor contract, such that the same labor contract is imposed on whole categories of people. Labor contracts in preindustrial social structures tended to be more individualized. From this point of view, class consciousness is a product of collective contracts. At first, collective contracts are unilaterally imposed on the workers by urban industrial and craft employers, creating categories of workers subject to the same working conditions, the same measurement of performance, and the same wages. Worker class consciousness is simply organizing those categories to negotiate their side of the collective contracts already imposed on them. This is a caricature of the argument in David Lockwood's The Blackcoated Worker: A Study in Class Consciousness (1958); a more nuanced development is given in Chapter 8. But class consciousness, in the way it is ordinarily used, is the projection of one's position in an employing organization onto the society at large, so that one's place in that society, one's interests in the economy and polity, are seen "as through a class darkly," in Kenneth Burke's phrase. We therefore reformulate E. P. Thompson's great book, The Making of the English Working Class (1963), to specify under what conditions a position in an employing organization will be projected onto the larger canvas of labor market organization and political movements as a definition of interests there. We then apply this combined Lockwood-and-Thompson theory to the decline of class hostility in the more class-conscious (industrial and transportation blue-collar) part of the economy and the low level of class consciousness in many parts of the service sector of that economy. Chapter 8, then can be thought of as an extension of an analysis in Chapters 2–7 that is oriented dominantly toward economic employing organizations, to the explanation of some features of unions in the labor market and of left political organizations. It does so by showing how the structure of organizations may provide information to workers about their interests, information that is then interpreted by the culture of the larger system. Chapter 9 carries out another sort of extension, into a field involving employing organizations whose purposes are education and research. We

28 give a couple of case studies of administrative structures in universities, namely those that administer space and those that set teaching loads of professors. A prestige university confronts the uncertainty about where new knowledge is to be found and who will find it. Clearly, unless a piece of knowledge has been highly uncertain beforehand, it is not much of a research accomplishment to discover it. Thus, as in professional athletics, research tends to be completely uninteresting if the outcome is certain and most interesting in cases where nobody knows for sure how a given line of investigation will come out. From an administrative point of view, the main implication of this situation is that the people who know both what work is to be done and who can do it are concentrated at the lowest hierarchical level of the organization—the professors—though they may be paid more than people with many subordinates. What these subordinates hopefully know how to do is to have good bets about where new knowledge is to be found, better than those of fellow scientists at the lowest levels of their organizations who provide the baseline of uncertainty. Scientific results, and to a certain extent results in other scholarly disciplines, change the degree of uncertainty generally held in the scientific community. Those findings or theories that are most fruitful, in the sense that they most change other scientists' bets about where new knowledge is to be found, are the most valuable. That is why citations turn out to be such good measures of the subjective estimates of scientists of who has done the most important work. But this deep uncertainty—in which other competent people cannot bet as well as the future Nobel Laureates one wants to hire, retain, and promote—makes management in the ordinary sense extremely difficult. Even the knowledge of what sort of work the organization will be doing in chemistry or physical anthropology is concentrated toward the bottom of the organization, and people in chemistry and physical anthropology at other universities can probably usually bet better than the dean what will occupy the space in the labs and the research time of the dean's subordinates next year and the year after. But this means in turn that the information the dean needs in order to make organizational tradeoffs between departments, for example to decide on allocations of laboratory space or teaching loads of various faculty members, have to be wrested from subordinates who have no interest in providing the basis on which space might be wrested from them or in having students and courses added to their teaching load. Thus, the commitment of a university to research distinction is in essence a com-

29 mitment by the university administration not to know how to manage the organization, not to know whom to hire, retain, and promote, and not to know how to organize the flows of the most crucial information; their subordinates, the professors, and professors in other universities, are the ones who know. In the administration of the allocation of space we analyze why this inherent situation of university authorities tends to lead to departmental sovereignty over space, with only occasional invasions from above. In the administration of teaching loads we study the pressure for university administrators to ignore measurements of individual research productivity in setting teaching loads, even though the dominant reason for having few courses in the (more or less uniform) teaching loads of universities is to give time for research. The question to be answered, then, is why universities so seldom try to allocate more time for research to those who do it best and why instead they depend on the National Science Foundation or the National Institutes of Health to make the judgments about whose teaching responsibilities should be lower, whose higher, by judging whose time those institutions are willing to buy from the university (at cost plus 50 percent or so—the percentage of overhead is higher at some leading universities). The objective of this brief exploration into the organizational aspects of the sociology of science to show that uncertainty and the information needed to reduce uncertainty shape the structure of administration of universities as much as they do that of economic organizations. The objective of the book as a whole is to illustrate in a variety of contexts how the social structure of organizations can be explained by the structure of the information problem they are confronted with. The idea is that organizational principles differ radically from one situation to another because what is functional for an organization differs similarly. What individuals have to learn; what distinct types of information structures must be built into manufacturing management; which departments processing which information should group together into divisions; how the administration needed for managing innovations and that needed for managing more routine products differ; how management tells whom to hire, retain, and promote; what the content of contracts among organizations will be; what features of organizational position will be taken by an organization's members to define their fundamental economic and political interests; and why scientists go their own way regardless of their administrators—all are to be explained by the nature of organizational uncertainty and the form of information processing needed to reduce it. Broadly speaking, we are trying to explain why the formal structure of

30 organizations varies depending on what those organizations have to do. All the elements of this theory exist in the organizational literature. For example, the centrality of routines in organizations, emphasized in Chapter 2, is central to both Cyert and March (1963) and Nelson and Winter (1982). The dependence of skill levels on nonroutineness and the location of skills in those parts of manufacturing with much market uncertainty are analyzed by Charles F. Sabel (1982). But those pieces have not been put together into a systematic theory of the skill levels of organizations. In particular, it has not been done in such a way that the same theory can be used to explain why innovations not only create such administrative difficulty but also require such high skill levels, as we do in Chapter 5. To make the transition, we have to use the complex theory implicit in Chandler's interpretations of various concrete business facts in Strategy and Structure (1962). But that theory is not systematic enough to use for our purpose. For example, Chandler writes as if the explanation for divisionalization of firms is the same as that for general office structure. But divisions respond to different uncertainties than does a general office, and so are differently explained. This distinction helps to untangle what Chandler's theory is and so to show why his analysis of Sears is not convincing. Because Schumpeter had not read Chandler and had not studied Du Pont, he could not imagine making bureaucracies flexible enough to be innovators. He therefore placed all the burden of economic advance on heroic entrepreneurs. But Chandler's theory shows more exactly what kinds of bureaucracies can innovate, can turn an invention into a going concern making monopoly profits. Thus, the work of this book is to make an overall theory of uncertainty and information sufficiently strong and exact to show where our geniuses in organization science went wrong. It does so by a systematic attempt to explain variations in the formal structure of organizations. These variations are often between parts of the organization. For example, to explain the structure of Du Pont, with product line divisions and a general office, we need to explain why decentralization to divisions involved first centralizing those divisions. But we also need to explain what uncertainties the general office responds to and therefore why it needed more abstract communication flows to it and a nonhierarchical committee structure of communication within it. The structure of the general office in Du Pont was a committee structure much like one that had been tried and rejected for internal administration of the divisions. In some sense, then, the theory of this book is old hat. But by explain-

31 ing variations in organizational structure by variations in the type of uncertainty dealt with, we can show the relations among the classics of organization theory. In the process, we make the classical theory empirically richer and make it stretch from the micro level inside individuals as they develop skills to the macro level of class relations in whole societies, from the General Motors assembly line to the university laboratory for research in biophysics. We hope, in sum, to make the theory we have used both more concrete and more general, so as to explain why the forms of organizations are at once so wonderfully various and yet so obvious once we see the uncertainties that they have to deal with.

32

2— Individuals' Skills As Information Processing: Charles F. Sabel and the Division of Labor Introduction The basic argument of this chapter is that when we say a person is "skilled," "semiskilled," or "professional," we are describing what sort of an information-processing system he or she is. If organizations have to deal with uncertainties, then someplace in the organization there have to be people who bring information to bear on those uncertainties. The flow of unpredictable events to a worker's or professional's area of responsibility sets problems for that worker or professional. The capacity to use the news about what uncertainty has come in, to decide what to do and then to do it or arrange to have it done in a fast and effective way, is what a skill consists of. Thus, to describe a skill is to specify what sort of information-processing mechanism a given worker is. We will argue that the best way to describe the individual as an informationprocessing structure is by the routines he or she can use, and then by the principles he or she uses to decide which routine to invoke as chance brings in now this, now that task to be accomplished. If we then develop a theory of the conditions under which more complex information processing by individual workers is required, then we will have a theory of the skill distribution of different sorts of organizations, and of different parts of organizations. To develop such a theory we first develop the notion of routines, of routinization as part of what skill is, and the relation of the number of routines in the repertoire of a worker and the complexity of that worker as an information-processing system. We also discuss the routinization of the relations among workers' routines in "fordist" production. Fordist production allows the use of semiskilled workers in the production of complex products. The theory of routinization and skill will then give us a theory of the conditions under which complex production requires complex in-

33 formation processing by individual workers. This provides the basis for describing what technical, raw materials and market conditions are necessary for fordist production, and conversely what conditions will require a high skill mix in the labor force of organizations. Thus, the central dependent variable, the thing to be explained, is the skill mix of an organization or of parts of an organization, such as why maintenance departments cannot be deskilled the way mass production departments can. The ultimate independent or causal variables are technical, raw materials and market uncertainty, concretely enough described so that we can tell where that uncertainty is higher and where it is lower. The core mechanism that connects causes to effects is the place of routines, discretion, and information about how to respond to uncertainty that have to be trained into the worker; these determine his or her skill level. So it is to the theory of routines in skilled and semiskilled work that we turn first.

Relations Between Routines and Skills The key thing to get out of the next few subheads is a picture of routinization within the individual , that is, a skilled person becoming really expert and fast at doing some number of distinct tasks, which enter in different combinations into different jobs. The basic argument is that skilled workers' skill consists of a set of routines , a set of smaller skills for particular tasks that they do very well, and many principles of decision which tell workers when to use one routine, when to use another. In preindustrial society, the skill that artisans had was by and large the ability to do all the things necessary to carry on a given business, such as baking or shoemaking or goldsmithing. What distinguished artisans, such as blacksmiths or wheelwrights, from outworkers, rural people doing one task in one routine like weaving, was that they had many skills, all that were required to produce some complex object or set of objects, and could switch among these task skills in an intelligent way. First we need to specify more exactly what we mean by a routine, and what we mean by a complex set of routines tied together by worker discretion. We will do that by making an analogy between work and computer programs. The part of the total task in use of a computer that is programmed into the computer is a part that is completely routinized —all decisions are prespecified. The parts of an individual's skill that are completely routinized are those that he or she does not have to think about—once a routine is switched on in the worker's mind, it goes to its end without further consultation of the higher faculties. Furthermore, a whole factory's work is like the work of a skilled

34 person, a combination of decision-making parts, requiring human discretion, and more or less completely routinized parts, parts that can go forward without consultation of either the higher faculties within the individual workers or higher authorities for decisions. These completely routinized parts of a social organization consist not only of routines within the individual workers that do not require decisions (meaning that the worker can be either semiskilled, as an assembly line worker is, or highly skilled, as a symphony musician is) but also of relations among workers that do not involve decisions. The model of such a social structure is the early Ford assembly line, and Sable (1982) has used the term fordism to refer to such socially organized routinization. We treat briefly the variation across organizations of the total amount of routinization. For example, in universities we find highly routinized work in the registrar's office; in hospitals we find fairly routinized work in pathology laboratories. Overall, however, both types of organizations have many highly skilled workers using discretion in their work. At the opposite extreme, an automobile assembly plant has a high ratio of semiskilled production workers hooked together by completely routinized connections among their tasks. But there are still unroutinized parts of such assembly plants, in management and in maintenance departments or tool and die shops. We will analyze why some parts can be more completely routinized than others, and consequently why only a small part of a university is like an assembly line, only a small part of an assembly plant is like the professional staff of a university, but why small parts of each have an atmosphere much like the main part of the other. Routinization within the individual without individual decision making (i.e., semiskilled work on complex products) is thus the outcome of a social process, the routinization of production. Historically, this has involved two causal determinants, which are analyzed separately later in the chapter: (1) the movement of work into factories and the winning by factory management of the right to specify the content and performance standards of jobs (the specification is often done by engineers, so we study in particular the establishment of authority of engineers) and (2) the finding of the kinds of markets in which firms using such fordist production systems are viable.

Two Relations Between Routines and Human Decisions Let us start by deepening the analogy mentioned above between the structure of work and the basic structural features of computer programs.

35 A computer program is made up of two main parts, the part done by the machine and the part done by humans. The part done by machine has to be completely routine . That is, when the machine starts running, every single step that the machine is going to do before showing some output has to be completely specified and has to depend only on things that are already in the machine (by machine I mean the whole system, all the machines controlled by the computer). This part of the program is unforgiving; if the wrong directions are put in, the wrong results come out. The human part can be arranged in two main ways. One is so-called batch processing, in which the humans try (and almost always fail) to put in everything necessary at the beginning to specify every part of the machine routine. There is only one set of human inputs, and one set of outputs that tells the programmer he or she was wrong or (more rarely) right. The other arrangement is called "interactive computing." The basic notion is that the program asks the person at the terminal to specify the first step or first few steps, then shows the results and asks him or her to specify the next step(s). Again, a person always makes mistakes but can go back to the previous step and correct each mistake as he or she goes along. So when people get to the end of a series of human inputs and computations on them, they are fairly likely to have corrected most of their mistakes already, and thus to have usable outputs. The contrast in structure between batch and interactive computation in outlined in Fig 1. Similarly, the routines of an organization can be thoroughly pre-

Fig. 1. Batch vs. Interactive Structures of Programs

36 specified, or they can have many stages at which human discretion can be applied. For an example of a system that is "interactive," consider the whole process involving the registrar's routines in a university. That system is built so as to have multiple points at which the human input of the professor (sometimes a committee of professors) is entered: in the title of the course, in the schedule of class meetings, in the flunk notices, in the grades; the input of the department and the school on what is required for graduation; and some more limited input of students at registration time and drop/add time on what courses they choose. But most student input is fed not to the registrar but to professors, who translates pages of messy prose or check marks on a test sheet into the few alternatives allowed by the registrar's routines. So the sixty or so course grades that are most of the sixty or so professors' human input to the registrar about a given student are actually the result of the professors turning a mass of written and oral work into quarter grades, to be output as GPA's, and so on, which in turn qualify a student for this or that major or for remedial classes or get a student advised to take "poet's science." These multiple human inputs from multiple professors make this routine a more or less "interactive" one. This interactive structure is closely related to the use of skilled professionals in the "human" part of the routine. The thing that the professors as "complex human translators" at intermediate points in the registrar's routines makes possible is an enormous variety of outputs—especially in the various "majors" or degrees that qualify people for different lines of work. That is, the process makes it possible to have simple routines or "programs" in the registrar's office, yet to produce a large variety of outputs that would ordinarily require great complexity in the registrar's routine. If it were the registrar, rather than the professor of biochemistry or biophysics, who had to decide what should go into a course on membranes, the registrar's routines would have to be much more complex. But let us look now at the routines for producing biochemists (or sociology majors, or whatever). We do find some guidelines in the requirements in the catalog, but those are the result of pulling and hauling within the department faculty—occasionally with input from student complaints. It will turn out that if an important professor is working on membranes (and if, mirabile dictu , he or she likes to teach), then a course in membranes at least will be an option in the training of biochemists. (Some general difficulties of management of people who know more about the work to be done than their managers are treated in Chapter 9 on university administration.)

37 The content of the courses may "improve" to keep up with new scholarship, but the professor-hours per student graduated shows no general tendency to go down over time. Quite often the routine parts of the teaching—say, elementary calculus for engineers if taught in the mathematics department—tend to be slighted. People in mathematics may want to get rid of such courses because they cannot teach new and interesting things there. A university may get more and more efficient registration but no real "improvement" (i.e., rise) over time in the student/professor ratio, because the nonroutine parts do not "improve" much over time. When an organization has routines that are "interactive," we usually find a very complex structure with many parts that do not seem very efficient. Hospitals, research laboratories, universities, law firms, and so forth tend to have most of their routines with such human inputs built into the middle; they therefore look inefficient and anarchic to a modern business school eye. But let us go to the automobile assembly line, which more nearly resembles a "batch" computer program. A "batch" in this case may be a year's production of a few models of cars. The engineering of the car and the tooling up for model changes are supposed to be the final input for all this year's cars. Of course, there are more bugs in the production process at the beginning, so more cars do not pass the inspection at first and line speed is slower. And maybe the manager will not be able to be as flexible about producing several models in short runs on the same line at first because, for example, the supply and inventory system for parts for running several models interspersed is not fully developed. Still, the basic structure of an assembly line is the "batch" program. The parts of the program are fixed, and the management have enough control so that even if they have a carburetor attacher who is widely respected among his colleagues, they do not increase the amount of carburetors to be installed the way a university increases training in membranes. In particular, the manager on an assembly line can really work on making each component part of the routine efficient. Taylorism—improving the pieces of the routine—will not work in a university (or it mostly will not), because a university does not want to teach the same courses efficiently year after year and end up producing 1950 model biochemists forever. Deans and college presidents cannot "engineer" the product and the production process in detail. Similarly, when in a study of a hospital—a relatively unroutinized organization—someone finds that it takes, say, ninety minutes on average to get a neurosurgeon to an emergency room to operate on an acci-

38 dent victim, a time-study engineer specialized in improving routines would have difficulty knowing what to do about it. One does not know that a neurosurgeon is needed until someone makes a diagnosis of severe head or spine damage; a hospital is not trying to produce as many brain operations per day as possible, but only as many as are necessary (and the exact ones that are necessary) given the automobile accident experience and stroke or tumor damage on that day. On an unroutinized job, timing the elements of a routine may not answer the crucial questions. (One thing hospitals may do is to start trauma centers that can afford to have a neurosurgeon there on duty all the time; what can be routinized is the presence of a neurosurgeon, not how to move him or her from place to place or get operations performed faster [Schwartz 1975; 1978a,b]). Notice that quite often in an unroutinized organization like a hospital there will be a big enough flow for some routines to warrant "assembly line" structures. In a modern hospital the pathologist (usually a man) is more an engineer for an assembly line of tests (usually carried out by women) than he is a person who does diagnoses. Pathologists do keep a monopoly over writing out the "interpretation," especially of things like biopsies. Similarly, in a university multiple choice test grading can be routinized, and an industrial engineer would be a lot of use in designing procedures for that. But an industrial engineer usually would not help much in turning a heterogeneous group of term papers into grades acceptable to the registrar. That is, there are bit of work even in a hospital in which the physician specifies a lot of decisions for a laboratory worker's routine (check marks on a lab form), the routine runs, and the output comes back the next day to the physician. The "compute" boxes on the interactive diagram that would represent the production process in a nonroutinized organization, then, can be fairly complex batchlike processes, much like assembly lines. For example, the compute process in an American university for turning a transcript into a certification by the registrar that a given student has satisfied all the requirements for graduation is a pretty complex routine, but it does not require the discretion of a leading biochemist even when the person being certified is a student graduating in biochemistry. So the first big distinction we want to make among routines is, Are there large elements of human discretion in the course of the routine (interactive) or are all the decisions preformed (batch programs)? Keep in mind here the model of the registrar's relation to the information coming from a student's term paper versus the assembly line. This first variable essentially asks, is there complexity outside the routine that is input (in simplified form) into the routine?

39

Complexity of the Routine A second variable has to do with complexity in the routine . Compare the old Ford assembly lines that made black Model Ts or black Model As and nothing else to the modern assembly line that makes the same model in different colors and makes several variants of a given model. Now one may have several different bodies that are mounted on the same chassis, a couple of alternative engines on the chassis, and an automatic or stick shift variant of each of these, or maybe even a couple of different lengths of chassis. In order to make the modern complex assembly line run, one has to supply the appropriate sequence of parts so the line does not produce a two-door body with front seats that do not fold down. Obviously each one of these model options on the modern complex assembly line itself has the structure of batch program. All the decisions are made in advance, so that everybody knows folding seats go into two-doors. One wants to build the routine of the assembly line so that as word comes back from the dealers that there are many station wagon liberals out there this year, one increases the frequency with which one "calls" the subroutine for station wagons, and one hopes all the links in that routine work as efficiently as all the other routines. In such a complex assembly line, one has a bunch of subroutines, all very similar, and an overall routine that selects out or "calls" the subroutines according to results in the market. When that control routine says, "twenty station wagons, one hundred two-doors, and two hundred four-doors," this entails a whole set of twenty decisions about the special placement of the station wagon spare tire, one hundred decisions about two-door folding front seats, and maybe twenty decisions about a longer chassis. But once the quantities are selected, each production run of those quantities is a batch process and entirely routinized. That is, the hundred decisions about whether to put in folding front seats are part of a hundred-unit-long batch program for producing two-doors and within that routine are completely automatic, at least if the routine is working properly. Obviously, this is a much more complex program or routine than that of the Model T assembly line and requires more flexible systems for inventorying and buying parts, for organizing the storerooms to put out the parts in the right sequence, and so on. The routine itself is more complex. This in turn means that it will be harder to make all its parts work smoothly. There are more things to go wrong with three kinds of seats than with one kind, and more color of paints to run out of when there are six colors, not just one.

40 Adding complexity to a computer program is one of the main activities of programmers, and it takes a lot of person-hours to write a modification, and even more to get it really running. The same thing is true of increasing the complexity of the routines of an organization so that they routinely yield a wider variety of options, in order to respond to a wider variety of states of the world. Batch programs with human parts need debugging, and the more complicated they are the more bugs they will have.

Artisans at the Beginning of the Industrial Revolution It is useful in specifying the relationship among the organization of work, the division of labor, and the skill of individuals to examine artisan production in the early nineteenth century in England, as presented in E. P. Thompson's great Making of the English Working Class (1963). Somewhere around a fourth of the labor force of London was made up of artisans in the time Thompson was writing about, including masters who ran small businesses in a given trade, journeymen who had all the skills to do so but not the property or managerial responsibility of a master, and apprentices who were learning the skills. This ratio was probably somewhat less than a quarter in most other urban areas, except for a few places that specialized in metal goods (Sheffield, Birmingham). The main thing that made up the skill of such an artisan, and that makes up the skill of a modern craftsman such as Sabel talks about (1982), is flexibility . The reason it took several years to become a craftsman was that a person had to be able to do many different things well in order to make the product. The reason craftsmen were so hard to replace by part-time agricultural laborers, who might also work as outworkers in a cottage industry like weaving, was that there was no one thing craftsmen did (craft workers were almost always men in those days, and still are usually ) that one could teach a rural laborer to do in a short period of time. Running a loom to make one kind of cloth is one routine, making shoes of all different sizes and styles is more craftsmanlike, and making pieces for machines with the technology of metalworking of 1790 to 1830 required the worker to master many different routines. The first job, then, is not protected from the competition of unskilled rural laborers, while the last is subject to the competition only of very skilled artisans. For instance, the way one ran a lathe in the days Thompson wrote about was to get the piece of work spinning, then to hold the tool up against it on a long, heavy lever that the machinist supported on his

41 shoulder and, with practiced judgment, to adjust the cut by varying the pressure and angle of the tool with slight movements of the shoulder (see Wallace 1972, 148). Nowadays the machine holds the tool in place, in those days a person did, and it required strength and coordination. Since machinists had to make many different pieces to make, say, a spinning jenny, they had to know how to do many different things with this primitive lathe technology, and to do them fast and correctly. This variety in the work, with a consequent variety in the routines that an individual had to master, gave craftsmen protection from some of the rigors of competition and made technical advance in their trades into something they incorporated into their skills , rather than something that drove them down to subsistence wages as they tried to compete—as technical advance did to the handloom weavers, for instance (see Smelser 1959, 245–264). A power-driven loom by about 1830 seems to have been about 25 percent more efficient than a hand loom (Smelser 1959, 205– 209), so a cottage industry could compete if the folks were willing to be about 25 percent more miserable than urban workers were. But if one introduced, say, a new sewing machine into the shoemaking trade (until the development of automatic shoemaking machinery in the late nineteenth century), it became incorporated into the skill of the shoemaker and increased the difference between such an artisan and the untrained agricultural worker. The artisan knew how to operate a sewing machine as well as having other skills that the unskilled workers could not match. This contrast in the position of artisans versus the more single-routine trades gave the former a collective capacity to resist disruption of their system of work social relations by capitalism and technical change. A modern carpenter does relatively few of the detailed jobs that an eighteenth-century carpenter did. Technical change has almost completely changed the content of the craft. But he (98.8 percent of carpenters were men in 1985; Statistical Abstract of the United States, 1987 , 386) still occupies about the same place in the general labor market, union or nonunion, because as one after another element changed it got incorporated into the skill of a personally flexible craftsman. So it still takes about three years to learn to be a carpenter, though now an apprentice, rather than learning how to smooth logs into beams with an adze, learns to form a frame of a house with nails, glue, and plywood. The structure of the program that represents the carpenter's skill is still interactive, with many subskills that are almost entirely habitual. Carpenters are intelligent human decision makers who allocate their time among those subskills. The subskills have been almost entirely replaced, or at least very substantially modified, since the time Thompson wrote about.

42 To give a more concrete picture of what artisanal skill consists of, consider the example of a housepainter (because I have been a housepainter). The skill consists essentially of three elements: (1) speed and accuracy of doing pieces of routines (or subroutines); (2) efficient and accurate switching among routines; and (3) creation of routines for new purposes. (Roughly speaking, in computer systems the people responsible for these aspects of running the machine are, respectively, the machine operators , the users , and the programmers or systems architects .) For a housepainter, skill consists first of all in being able to paint windows so that the paint does not get all over the glass, to enamel or varnish a door so that the paint does not "sag" or "run," to slop paint evenly on flat surfaces at a high speed, to do small plaster-patching jobs so they are imperceptible when painted, and so on. Speed and accuracy in doing the subroutines, the parts of the job, are the first things an apprentice learns. A second part of the skill is to adjust the trade-off between speed and accuracy, between cost and quality, in a way appropriate to the job. For example, if upper-middle-class people are going to be living with a window a few inches away, sitting in a soft chair and contemplating it, one wants the paint to go just enough onto the glass to make a clean smooth line, but the overlap onto the glass should be very narrow, very straight, and very neat at the corners. In a restaurant they want a considerably faster job, because extra accuracy will not pay off in restaurant profits except in very tony places. People do not contemplate windows in commercial establishments. In a factory they want to let the light through the glass—at least through the middle—and to make sure the metal sash is covered so it does not rust. They want to do it as cheaply as possible just so it looks good from the road. College students' dormitories or apartments were very likely painted at a "commercial" standard, while the homes they came from were likely painted to a "residential" standard. So painters really need three different routines—residential, commercial, and industrial—for painting windows, at three different speeds (industrial fastest), three different qualities (residential highest), and three different prices (residential highest). In the construction industry, nonunion firms in general use less skilled labor than union firms. By hiring nonunion painters one is likely to get a commercial standard in a residence, because the painter cannot do residential-quality painting. A union painting company can pay about 30 percent more per hour and still compete with nonunion painters. For a given quality of work, people seem to increase in speed for about three to five years; thus nonunion contractors by and large employ these workers for the first three to five years, the union contractor after that. This discussion of painters brings out the contrast between the first

43 two components of skill: speed and accuracy within the routines and ability to switch among routines depending on the situation. Being able to paint windows at the three different speeds, qualities, and prices is the first kind of skill; knowing when each is appropriate is the second kind. These two components of skill correspond to the distinctions among computer programs that we talked about, between batch and interactive programs; the more elements of painter discretion built into a painting role, and the more routines painters are required to switch among, the more the role has the structure of an interactive program with decisions between one routine part and the next. Skills embedded in the role of an individual painter depend on the overall routines of the painting contractor. In a big city one can find industrial painting contractors that never do commercial or residential jobs, and so their painters do not have to switch levels of speed and accuracy. And there will be commercial painters and residential painters (sometimes even outside versus inside residential). So the structure of the overall routines of the contractor, the degree of specialization, will determine what kinds of skills he or she needs in a painter. Linoleum laying and wallpapering are also (or at least were thirty years ago) in the painters' union jurisdiction. The larger the city, the more likely it is that there will be specialized linoleum and wallpapering firms. The more specialized the contracting firms, the more narrow the specialization of painters will be, on average—the less variety there will be in each individual painter's skill, the faster and better each will be with those subroutines or task skills he or she actually uses, and the less switching there will have to be between those task skills. But in the big city painters are still "artisans," in that they know all the skills and the switching routines that are necessary to produce their firm's product. The division of labor among occupations in construction is still basically the division of labor among firms, but the firms are more specialized in the big city. The discussion above may seem to imply that only skilled jobs have multiple routines and involve worker discretion in choosing among them. Yet operators of simple machines and bank tellers, jobs that take less than two months to learn to do "adequately," have to use much technical knowledge and discretion too (Kusterer 1978, 45–62, 75–94).

The Division of Skill Between Workers and Professionals We will now discuss three ideal types of ways people can learn to make the decisions involved in a skill. I will call them craft, young professional, and senior professional. I will discuss each of these in terms of three

44 topics: (1) what kind of basic training each type of worker needs for the exercise of his or her decision-making skill; (2) what sort of knowledge this leaves him or her with; (3) how the "jurisdiction" of the occupation, the set of activities and decisions over which it normally has control, could best be described. After analyzing these three ways of socially organizing to divide decisions about work routines within roles and to learn how to do those roles, we will turn to a big historical change in the social structure of decision making about manufacturing work, which Reinhard Bendix has studied in his Work and Authority in Industry (1956, 274–287), namely the introduction of time and motion study, or "industrial engineering," especially in the steel industry, between about 1890 and 1920. Industrial engineering was central to the establishment in American industry of the right of management to enter into the jobs of the workers and to reorganize the work itself, to redefine its skill level, to break it up into more specialized occupations. This assumption of authority over the organization of each person's work was involved in the transition to a whole new way of organizing the definition of "skilled worker." The new skilled worker in the steel industry—or later, say, in the automobile industry—was not a person who knew how to do everything in the business of making steel (or automobiles). Instead, after the transition to a modern factory organization of production the highly skilled worker was either an engineer or manager on the one hand or a craftsman on the other, a person who was master of various routines or tasks that have to be done occasionally, adapting to the situation. The skilled craftsman in an industrial plant nowadays can almost be defined as a specialist in everything that has not yet been routinized by industrial engineers. The point here, then, is that defining the skill of a skilled worker is no longer like the process that defined artisan skill in prefactory manufacturing (which still is found in construction). For an artisan enterprise, every skill needed to produce the product of the firm was part of the repertoire of the most skilled workers. Modern factory management got authority to define routines, to assign them to different workers, and to assign to factory-trained skilled workers tasks still requiring many routines used occasionally at worker discretion. After engineering authority was established in factories, skilled workers had skills created by managerial authority rather than by market processes. Further, routines or tasks now down by factory skilled workers could be reorganized, at any time it became economical, so that they could be done by semiskilled workers in mass production.

45

Three Organizations for Learning Routines and Decision Skills Table 1 gives a brief outline of the main ways different skilled workers learn their roles, what their knowledge consists of, and how the jurisdiction of the craft or profession is typically defined.

Craftsmen Craftsmen, in ideal circumstances, learn their job by supervised experience , the supervision often being one-on-one. First they learn the bunch of skills that hang together practically in a certain job, whether these involve the same principles or not. For instance, the thing that connects painting and paperhanging into the same trade union jurisdiction is that they come at the same stage of the finishing of the building and have the same purpose—looking good. They have to be coordinated with each other, so that the ceiling is not painted after the wallpaper is on, putting splashes on the paper. The knowledge of an all-around painter thirty years ago would have included both being able to paint (and to paint various things at different speeds) and being able to hang wallpaper; wallpaper has pretty much gone out of fashion since then. So the training of craftsmen is in the first instance to learn these skills one after another from someone who knows how to do them, in the course of doing the work the journeyman or master assigns to them according to their developing skill. When they have worked enough so Table 1. Ideal-Typical Ways Craftsmen's, Young Professionals', and Senior Professionals' Roles Are Defined and Learned



Craftsman

Young Professional

Senior Professional

Basic Training

Supervised experience

Taught principles

Own experience analyzed in the light of principles

Knowledge

List of routines and "indications" for use of routines

Principles without routines

Analyzed routines

Jurisdictions

Bodies of routines connected "practically" to each other

Areas to which principles apply (e.g., mechanical vs. electrical engineering)

Creation of new routines and "higher management"

46 that they will have learned almost all the routines involved in the job, the apprenticeship will be over. There is nothing "intellectual" that connects the job of wallpapering to the job of painting, the physics of making the paper adhere, the skill of making the pattern of one strip of paper match another, and so on are very different from the physics of making a paint film cure into a viable wall covering or the skill of "cutting in" an edge where the woodwork meets the glass so that it looks good. It is possible to train for the intellectual professions by an apprenticeship method. For instance, the Inns of Court in Great Britain have traditionally trained barristers (lawyers who appear in court and who are likely to become judges) by apprenticeship (Flood 1982; Abbott 1988). What lawyers end up knowing through such training is the law connected to a large number of practical situations that appear in the courts, and what painters end up knowing is how to cut in window sash, how to align wallpaper so the patterns match and so they still match when they get around the room and attach the last strip of wallpaper to the first one, how to get paint evenly on the ceiling in a hurry but without lap marks, and so on. That is, in both cases what people learn is a list of routines. The basic training of craftsmen, then, is supervised experience in practical work. What they end up learning is a list of routines and knowledge of "indications" of when this routine applies, when that one; and the set of routines they learn are those that connect practically to get a given kind of job done, not a set connected by intellectual principles.

Young Professionals Young engineers just out of engineering school are perhaps the ideal type of school-taught young professionals of a modern system. They are taught a series of principles in a set of courses. Courses in engineering schools are not so much a list of things engineers ought to know how to do as a set of principles that engineers will likely find useful in doing the various things (which we cannot very well anticipate now) that they will end up being assigned to do. So if the knowledge of craftsmen is a set of routines unbound by intellectual principles, the knowledge of young professionals is a set of principles they can apply. Young professionals go to work without much practice in the routines in which these principles are in fact applied practically, and certainly without practice in routines enough to cover all the sorts of things they have to do to get an engineering job done. So the job of young engineers is defined in some sense by what the school has taught them—by whether they have been taught as electrical engineers to solve problems with electricity or as civil engineers to solve problems with strength of materials for supporting heavy weights.

47

Senior Professionals Senior professionals have, of course, had the abstract training of young professionals and the hands-on experience of many years so they command the routines of working in a given field. In some sense senior professionals have the three years of actual coursework of the young engineer, as well as three or four year of apprenticeship of the electrician or plumber, and the advantages of both. But the crucial thing that a senior professional has, which craftsmen usually do not, is the experience of many years in applying abstract principles to various routine situations, and so improving the routines . In the civil service they sometimes say that a person "hasn't got twenty years of experience—he's got one year of experience repeated twenty times." A crucial thing about a senior professional is that one year of experience repeated twenty times is unlikely: each added year of experience of engineers add (on average) to their value, because they can do routines that have been developed and improved in their own experience—improved by the application of a deep abstract understanding of the objectives and the causal processes involved in the profession. So professionals' wages keep going up for a long time with experience, while the craftsman's wages level out after three or four years.

Earnings Curves for Craftsmen, Professionals, and Managers In general, brand new professionals just out of professional school are not as valuable as craftsmen, simply because they have not really mastered any routines. But experienced professionals are a lot more valuable than craftsmen, and are eligible for positions in higher management. The two (or really three, with the upper management branch for professionals) wage curves look something like those in Fig. 2. (These curves are not based on systematic research, but on happenstance data seen in the course of paying attention to the problem for many years.) The general picture is that craftsmen become more valuable at their craft for the first three to six years of experience but (except perhaps for promotion to foreman) do not increase in value after that time. They generally cannot do a lot more different kinds of tasks at fifty than they could at twenty-four, nor can they do them any faster. Engineers, in contrast, cannot really do anything until after graduation and are rarely hired at a skilled worker wage rate even after they have had as much training as craftsmen have. Young professionals may get their first job at

48

Fig. 2. Wage-Age Profiles (Idealized) of Craftsmen, Professionals, and Professionals Who Go into Management below the wage rate of craftsmen (for example, an assistant professor of sociology often makes less than a fully qualified plumber). But engineers do not (as they say in the trade) "burn out" until about thirty-five years of age. They keep increasing in competence because the routines they learn by experience are placed in an abstract framework of engineering knowledge and are improved by further applications of engineering principles to the engineer's experience. At around thirty-five one starts to perceive which engineers are "on the fast track," headed for managerial responsibilities. If they move into management, their incomes tend to increase substantially up to about age fifty. The flat places on these curves are not really flat, because people keep getting increases to keep up with inflation and with the general growth of real

49 incomes (the curves in Fig. 2 represent wages relative to the median of a given year). In general, the jurisdictions of professionals increase in size up to about age thirty-five as they "get more professional experience," and of course managers have more general responsibilities than specialized professionals, increasing further after age thirty-five. So one feature of senior professionals is that their jurisdictions, the areas of competence within which they operate, may have been substantially modified since their education (Faulkner 1983). It is important to keep in mind that these differences, say between young professionals and senior ones, are analytical rather than concrete. For instance, in graduate school in chemistry the work of a graduate student is more like the work of a beginning chemist than it is like the work of a student in a first-year undergraduate chemistry class. Graduate students do take classes; but to a considerable extent they are practicing the routines used in chemistry so that they can do them right every time, analyzing those routines in relation to their own scientific problems, finding new techniques, and improving a technique for their particular application of it. Already they are in the process of becoming "experienced professionals," and they are paid in graduate school at a rate that is in general already about equal to that of a craftsman (of a low level) rather than of an unskilled worker. When chemists with newly minted Ph.D.s go on the market for the first time, they will already be part way along the curve of "experienced professional." A rather similar thing happens in medical education, with its division into the basic science years, the clinical years, and the residency years: by the time specialists get their board certification, they are quite experienced professionals.

Jurisdictions of Occupations The jurisdiction of the craft union is not just a matter of union cantankerousness or self-interest, though it is certainly that also. It really is essential that it be defined which craftsmen should learn a given skill, so that at least somebody will know how to do the work. For example, is the list of routines for finishing plasterboard joints to be with the finishing trades (painters or plasterers) or with the ones who put up the plasterboard (carpenters), and if it is a finishing trade, should it be with plasterers or painters? The final decision is arbitrary. It went one way (carpenters) west of the Mississippi, the other way (painters) east of the Mississippi. The main point is that the bundle of things that fall in a craft's purview is determined by practical connections, not theoretical

50 ones. No abstract principle says whether plastering and sanding the joints of wallboard should be considered plastering (because one uses a plasterer's trowel first), sanding (because one uses sandpaper, a traditional painter's tool, next), or carpentry (because it is part of installing things put up mainly with hammer and nails). Similarly, both plumbers and electricians install pipe, one to put wires in (it is then called "conduit") and one to run water through (it is then called pipe or tubing). But pipe and wires are connected to the jurisdiction of electricians in one case, pipe and waste lines to the jurisdiction of plumbers in another, because of installation connections (one puts in conduit at the same time and in the same place as wires and connects them to switch boxes; the other puts in tubing at the same time and in the same place as sinks and connects these to waste lines). It is important to put the skills of pipe installation in both places because otherwise technically interdependent things (wire and conduit, tubing and waste lines) would not be done by the same people. Construction management would have a mess trying to coordinate plumbers with electricians (neither of whom ever come when they say they will). Of course, there are also political resolutions of jurisdictional questions, as when the compromise was worked out saying that the painters would finish plasterboard in the East, the carpenters in the West. Once one has decided where the jurisdiction goes, then a whole series of social arrangements follows: painting contractors will bid on the job of finishing and painting plasterboard in the East, general contractors employing the carpenters will bid on it in the West, so the competence of the businesses follows the lines of the jurisdictions of the workers. The apprentices of painters (or workers for nonunion painting contractors) will learn to finish the joints of plasterboard in the East, the apprentices of carpenters in the West. When a big job involving plasterboard comes up in the East, the placement of finishing workers will take place through the painters' union; when in the West, through the carpenters' union. The jurisdiction of a craft is a central principle in the social organization of firms, of apprenticeship programs, and of the placement of workers. The jurisdiction of an applied science tends to be some compromise between the reach of a given cognitive discipline and the practical unities of a job. For example, while from about 1900 to 1940 the best definition of the applied-science aspect of medicine was applied bacteriology (medicine did whatever one could do by killing germs, by preventing infections, and by quarantining sick people), there was also a sort of patient-educational task as well. A doctor had to learn the "normal course" of a lot of diseases in order to tell people what was happening

51 and when they would get over it—not because we then knew scientifically the immunological response to bacteria or to viruses, but simply because the patients wanted to know. So the patients to whom physicians were applying what we knew then about the applied science of bacteriology happened to come to the doctor with a series of questions about the disease: Do people usually die from it? When will the fever go down? When will I be able to go back to work? When does the rash go away? How sure are you that this is really chicken pox? Doctors thus had to learn the course of various diseases, even though that was not particularly a part of bacteriology that made medicine into an applied science. Now, in fact, the normal course of infections is a part of medical science. Physicians specializing in particular diseases write in Scientific American about the mechanisms that produce the fever, the mechanisms that produce the continuing tiredness after the fever goes away, the end of the contagious period when the free viruses (that can infect other people) will be killed inside the body and the only remaining function of the immune system will be killing the cells of the patient's own body that are infected (the fever is connected with the first phase). This is because now medicine, as an applied science, has a lot of knowledge of how the immune system works, and the immune system determines the course of most diseases. Nevertheless, a lot of medical learning takes the form of learning a bit about a lot of disease and about a lot of parts of the body—a sort of jurisdiction defined the same general practical way as craftsmen's jurisdiction. So medicine is (and even more was , between 1990 and 1940) a combination of an applied science and a "craft" of healing. Young professionals tend to be bad at the "accidentally" attached tasks of their profession, for these are not in the applied science taught in classes. The jurisdiction of senior professionals tends to be much more vaguely and individually defined. Within the broad range of, say, surgery, one surgeon specializes in the thyroid. He (or she) can accumulate a wide variety of experience in the alternative ways that that organ can lie or can malfunction, guess the likely meaning of oddities on the X-ray of that organ, relate alternative treatments to the symptoms various patients show after surgery, and so on. A physician can only specialize solely in thyroid surgery if he or she has sufficient reputation, and the only way to learn enough to get that reputation is by analyzing a lot of experience, so it is somewhat a matter of chance who gets to be a world expert on surgery of the thyroid. To some degree, the jurisdiction of senior professional is self-defined. But a general tendency is that experienced professionals' jurisdiction is in

52 those parts of the professional practice that have the most uncertainty. Senior professionals in engineering are more on the preproject roughing out of the plan, less on the detailed work, then more again on the management of the actual building or starting up of the assembly line, where technical knowledge has to be combined with managerial experience. Similarly, senior surgeons are more likely to be referred difficult and esoteric cases, while junior surgeons do more routine surgery.

The Determinants of the Division of Labor Between Engineers and Skilled Workers So far we have not said anything to indicate that one rather than the other of these methods of creating skill would be better for a particular economic operation, except for some obvious disadvantages of a newly minted degree; if average new engineers can get 70 to 80 percent of the questions right (which might be the percentage for a passing grade in an engineering class), then 100 percent of the bridges they design will fall down. The big difficulty is that senior professionals are too expensive for operations involving a rapidly changing mix of quite routinized suboperations. For machining a die, senior engineers are too expensive, young engineers do not know enough about metal working, so it goes to craftsmen, to machinists. Frederick Taylor himself, the main originator of "industrial engineering" and an engineer with a lot of experience supervising workers in a steel plant, could figure out how to do skilled operations faster than skilled craftsmen—eventually. But someone with a new degree could not. Often no one can tell early on which way of organizing a given kind of work is better. An example is the recent history of numerical control (Sabel 1982, 63–70; Noble 1984, 79–192). In numerical control a built-in computer controls the machine. That computer has to be programmed for each different piece to be produced. The question was, should it be programmed by "record and playback" or by an engineer programming a big computer? In the United States, the air force decided to have engineers do the programming before a skilled worker ever saw the operation. In Germany and Japan, in contrast, users of numerical control and machine designers mimicked the activity of a skilled worker by having their most skilled workers operate the machine, recording all the behavior of the machine during that person's operation, and then making that recording tell the machine to repeat those motions. Such tools provided for workers to correct certain bits of the recorded program (to go to the correct point, to rerecord a section, then to go to the

53 end of the section and go on from there) as they learned to do that part of the operation more efficiently. In the United States many managers and engineering departments put obstacles in the way of "machinist-programmed" numerical control. This was partly because they wanted the work for themselves, partly because they did not want to give the unionized skilled workers the power advantage of being the only ones able to run the machines, and partly because they really believed that programming by engineers would be the best in the long run ("technological utopianism" I call it in Chapter 5). Part of the air force's reason for choosing the engineer-programmed version of numerical control was that no machinists knew how to produce the complex shapes of aerodynamical surfaces that they wanted, so machinist-programmed numerical control simply would not have worked for what they wanted to do. The result is that nowadays most of the machines are being programmed by less expensive skilled workers rather than by senior engineers who, much more expensively, know both the abstract language of numerical-control programming and the concrete routines of metalworking. This means that now the Japanese and Germans control the international sales of most machine tools, whereas thirty or so years ago the United States was the dominant force in that market. Obviously, someone in the United States made a big mistake in what kind of skill system to build numerical control into, and we are all paying for it in our negative balance of trade. Part of the problem here is a general danger of trusting engineers too much. If managers go with what an engineer thinks is "state of the art," then generally no one can make it run all the time. And at any rate, they need engineers to run it. For instance, fighter pilots are trained to a considerable extent as aeronautical engineers, as well as in the routines of piloting. That makes a fighter pilot a very expensive product indeed. But even having engineers with years of experience flying our fighter airplanes does not work. Twenty-three percent of navy pilots, for instance, in an average career of twenty years, are killed in flying accidents (Perrow 1984, 126; citing Wolfe 1979, 17). Even senior engineers cannot always manage a "state-of-the-art" airplane landing on a "stateof-the-art" aircraft carrier.

Manufacturing Artisans in the Early Industrial Revolution In this section I want to discuss how authority is organized in a typical artisan trade (above we treated the organization of skill in such a trade)

54 and what this means for the structure not only of work but also of class consciousness. Then I want look at how the factory changed that organization, by discussing the erection of the authority of senior professionals and managers over the content of jobs. Authority over job content was the central thing that remained pretty much in the hands of the worker in a preindustrial "artisan" mode of manufacturing. Because early industrial artisans (I will call such workers "he" in what follows, for the reasons discussed briefly above—women in manufacturing in the early nineteenth century mostly worked in unskilled work in the textile and apparel trades) had all the skills necessary to complete the manufacture of a product, artisans could, if they did not like the authority system or the wages where they worked, go elsewhere or organize their own firms. In many respects the artisan was invulnerable to employer sanctions. This meant that there were tough social structures that could carry working-class culture, structures that did not get disrupted, that young apprentices were exposed to for years in a one-onone supervisory relation in which they learned the trade. The government could not suppress the carriers of this working-class culture without suppressing the craft. Many of the traditions carried by these artisans in 1790, when E. P. Thompson (1963) took up his story (a very similar story is told for France in Sewell 1980, 1986), were sort of "trade union" traditions, about wages and working conditions. But there were also technical traditions, including in particular notions of what good work really ought to look like, tricks of the trade, an idea of how fast a good workman reasonably ought to be expected to work, and the like. There were also political and philosophical traditions. For instance, the radical "Enlightenment" values of scientific belief (which in those days was more often antireligious than is true nowadays), republicanism, freedom of speech, and the whole complex of notions that Thompson connects with the writings of Tom Paine were part of the culture of machinists in England, France, and the United States at least from about 1800 to 1850 or so (on the United States, see Wallace 1972, chap. 5, esp. 211–219; on France, see Sewell 1980, 64–72, and chap. 9, 194–218). A key fact in the stability of these deviant systems of working-class culture was that capitalists, the "masters," were themselves virtually always craftsmen as well. They really had to be, because otherwise they could not glance at the hundred different things their workers were doing and see whether they were being done right and at a reasonable speed. Another key fact was that in general the number of people in a given artisan trade even in the big cities was reasonably small, and that

55 people moved around from shop to shop or from one construction job to another. There was therefore a dense social network in a given trade. The network crossed class lines (it included masters) but did not include the nonmanufacturing upper classes, which had a different culture from the artisan workers. Cross-class contact was still within craft culture. Informal culture supported craft or artisan guild and union organizations when there were any and kept the traditions of organization alive during times of repression (Thompson 1963). Thus Sheffield, a metalworking center north of Birmingham, could come out of the period of wartime authoritarianism (ca. 1800–1815) with a lot of its radical tradition intact. When he was doing politics from about 1807 to the mid 1830s, Francis Place, the London working-class reformer, could still talk the language of the people he had helped organize in the 1970s (Thompson 1963, passim). Sewell (1980) shows for France that artisan corporate culture from the old regime still existed, in modified form, in the revolutions of 1830 and 1848, after periods of Napoleonic and restorationist repression.

Economic and Technical Threats to Artisan Organization An artisan trade is threatened by one of two things: such big growth in the market that it pays capitalists to break the elements of a craftsman's skill into separate jobs and train a large number of semiskilled workers to do those jobs—the mass production threat; or radical technical change in the product or, more rarely, in the process of production. When E. P. Thompson talks about how the son of the boss wheelwright had to learn from the workers what was good work on a wheel (1963, 235–236), we as modern readers hardly know what a wheelwright is. The reason for our ignorance is that today the job of making wheels is broken down in an auto factory into its component parts, and these parts are done by semiskilled workers who would never tell their boss what the customary way to do things was because workers no longer have authority over the content of the job. This threat due to growth of the market is perhaps most dramatically manifested in what happened in the U.S. steel industry in the 1890s and the first decade of this century (Stone 1975). Through bitter strikes and reorganization of management, a well-organized craft method of production was reshaped into a system of mass production using semiskilled workers. The steel industry still has a high proportion of workers classified as skilled in the census, so not all the complexity of work was ex-

56 tracted from the work force and concentrated in management. In the 1890s, control of the definition of jobs was wrested from workers and relocated in the management of the biggest steel companies. From the point of view of the argument here, a crucial fact is that these were the companies making rails for the railroad boom (cf., e.g., Chandler 1962, 53; 1977, 258–269), companies with a large market for uniform goods. In both wheels and steel, a huge growth of the market for uniform goods made the flexible wheelwright or the flexible steelmaker replaceable by a division of labor organized along rigid bureaucratic lines. We will return to this point later when analyzing the market conditions for the success of "fordism" (Sabel 1982). The other main threat to artisan trades is a radical technical change that makes the whole product archaic. The craft of sailmakers was destroyed in the period between the decline of sailing merchant ships and warships around 1900 and the rise of the middle-class leisure sailor after World War II. Technical change made sails archaic for freight and passenger traffic, and they did not become modern again until a mass of middle-class people could afford leisure sailboats. More recent examples of technically archaic crafts (since World War II) are the plasterer and the marblesetter crafts, which both disappeared with the growth of new wall materials such as gypsum board and ceramic tile. From the point of view of working-class culture, it is important not only that craftsmen or artisans can carry a deviant (working-class, trade union, or radical) culture and that they were relatively invulnerable to capitalist and governmental oppression. They were also respected in the working class. They could do things that other workers could not do. They had often traveled to far places because their skill was transportable (Tom Paine, for example, was a craftsman—a corset maker—when he left England, and eventually he became a printer, also an artisan trade).

Authority Reorganization and Artisan Skill Let us go back to the problem of the disappearance of wheelwrights. Everybody started buying wheeled vehicles, the market expanded, and the wheelwright shop with artisans each of whom could make dozens of kinds of wheels was replaced by a factory with separate divisions of labor for each kind (cf., e.g., Buick Motors creating a wheel factory in Flint, in Chandler 1962, 117–118). Within that factory, management broke up the dozens of things a wheelwright did into specialized jobs for semiskilled workers. So the skill of the wheelwright became the skill of the factory as a social unit. The individuals within the factory may have a socially defined skill

57 anywhere from unskilled worker to engineer (engineers have a college education about equivalent to the apprenticeship of a wheelwright—three years of actual experience in classes—but, as we have discussed, they keep gaining valuable experience up to about age thirty-five). But the crucial thing to notice is that the deskilling of, say, the person who makes the hole in the wheel fit the axle (once a wheelwright, now a semiskilled machine tender) is an active managerial process. Deskilling is creating a semiskilled job out of a piece of a wheelwright job by management effort, authority, and thought. The job of the semiskilled workers is designed by someone (or a group) of higher skill. For instance, an engineer may specify what has to be done by the lathe (and by the foundry before that), and a tool and die maker may shape the tool to put into the lathe to do it. The engineers, die makers, and machine tenders are tied together by a social structure, a "management," which defines their jobs, their skill levels, the standards to which they are supposed to work, performance measurement to see whether they are working fast enough and well enough, the incentive or pay system, and so on. This proliferation of managerial functions and relationships is the crucial development that turns a craft shop into a mass production factory. And the authority of management is absolutely crucial to making the whole thing work. Reinhard Bendix (1956) is interested in the development of the ideology that justified managerial authority. That authority has to take from workers the right to define their own job, their own skill level, their own standards of quality, and put these rights into the hands of an engineer or manager. I am speaking historically here—of course the particular worker who is now a semiskilled lathe operator was not in general the same worker who was once an all-around wheelwright, so the process Bendix is talking about took the authority from a social group and transferred it to another social group, not actually (usually) degrading the work of someone actually on the job. An engineer's skill is, so to speak, half managerial, half technical. Engineers are not going to be making precise holes in wheels; rather, they will be designing the jobs and skill levels of folks that do. Engineers often move out of engineering into pure management (see, e.g., Chandler 1962, 317–319).

The Ideology of Mass Production Management The idea of efficiency was the basic thrust of the early-twentieth-century campaign (Bendix wrote about the period from about 1900 to 1940) to establish the authority of management over jobs that used to be artisan

58 jobs. A fine quotation from the Review of the National Metal Trades Association (comprising steelmaker and machinist managers) shows this clearly: "The employer is willing to pay for results, and in fact results are the only thing on which the value of labor can be based" (Bendix 1956, 271). The right to measure and determine results, and to specify what results are satisfactory and what they are worth, is the central component of managerial authority based on the idea of efficiency. This ideology of efficiency reserved to management not only the right to determine the value of labor, and hence what it should be paid, but also various other elements that went into the definition of a job. The most developed paradigm of this conception of efficiency justifying the complete control over job definition by engineers and managers is the "Scientific Management" movement led by Frederick W. Taylor. Taylor, an engineer who worked as a manager in the steel industry, helped build up the bureaucratic version of steel management after the breaking of the craft system in the 1890s. The modern version of the same ideology is represented in the field of "industrial engineering." The movement was based on time-and-motion study. This involved breaking down the job of the worker into its components and describing the motions that went into those components. Then, after redesigning the machine, the material, the work flow, or the worker motions to make the motions most efficient, the time-and-motion study engineer (or manager) tried to estimate what the efficient minimum time is for each of these motions, and hence for each task, and hence for the unit of output. This is the "result" referred to above by the National Metal Trades Association writer. Time-and-motion study gave management a measure of result on which to base their authority. The first thing to notice about this process is that it theoretically reduces worker skill from a matter of the worker selecting from a repertoire of skills to do a certain job to a matter of making the motions he or she is instructed to make. Anyone who has tried to learn to do the motions required to play a Bach piano piece or do the butterfly stroke in a way acceptable to a swimming judge knows that doing the motions fast and accurately can still be quite a learning task. But the cognitive part of the skill, the creation of the routines, is moved to Bach or the swimming coach, and the performer or athlete only does the motions. The same is true for the lathe operator after an industrial engineer is through with him. But for Taylor this efficiency picture of the worker's job entailed still more. He thought that this picture—not collective bargaining or craft traditions or a fair day's work for a fair day's pay— should determine the

59 compensation scheme. First, the efficiency conception would determine what an efficient day's work should produce, allowing output quantity measures to be substituted for a craftsman's pacing of his or her own work. Second, the same conception allowed one to measure extra effort above the norm accurately, where the norm was described for all jobs because the engineer had supposedly described exactly the efficient time and motions for each job—what went into given worker outputs—in minutes required. The manager could translate the number of widgets produced and the number of wing nuts screwed on them into the same metric, the same system of measurement: "the number of minutes an ideal efficient worker" would take to do them. This picture enabled management to base the compensation scheme of the whole factory on a set of efficiency models for each and every job. They could then compensate people who produced over the norm above the "competitive labor market wage" because they supposedly knew what was the absolute standard an efficient worker, hired off the labor market, could work to. In this ideology the skill and effort of the worker, and the definition of the job of the worker, come to be a "scientific" or "engineering" matter rather than a matter of competing interests, of class conflict over the division of the surplus, or of defense of skilled workers' stake in their investments in their skill. Further, it is the industrial engineer, the scientific specialist in management, who knows how to define all these things.

Scientific Management Authority in Practice What happened in fact when time-and-motion study people went out to set up this "time needed by the ideal efficient worker" for a given job? They timed a worker, got an average for all the motions, redesigned the job by trying to improve the time-consuming motions, and timed it again. But what they now had was a measure of how long it took, on the average, for John Johnson to pick up a wheel blank, put it into a lathe, tighten the clamps, align it, cut the hole, remove it from the machine, and lay it on the conveyor. This was a detailed picture of John Johnson, not of the pristine idea of the efficient worker. If they were going to let John Johnson define the job, they would not have needed scientific management. So the time study engineer supplied an estimate of how far John Johnson was from perfectly efficient . If he was working at a normal speed for workers, he might be marked down as 65 percent efficient. If this was a new operation (e.g., on a new design of the wheel), he would be marked

60 down at 45 percent. Now the industrial engineer found the efficient ideal by dividing John Johnson's performance by the efficiency estimate, for instance:

Of course, this is exactly as "scientific" as the time-and-motion study people's estimate of what percentage of an "efficient" worker's speed John Johnson is working. And this is obviously a subjective managerial judgment, about as good as our wheelwright master's son's judgment as reported by Thompson. If the industrial engineer is young, just out of industrial engineering school, and socially naive, an old hand in the factory will sweat and strain and move super fast and get nothing done—or, in the post office, a mail carrier will go to the crosswalk and wait for the WALK light to cross the street, which he or she would never do if the time-and-motion study person were not there—and still be marked 98 percent efficient. Then again, if the new industrial engineer is bucking to get promoted by setting high standards, he or she will not be "taken in" even by a worker who is really busting ass and will set a "scientific" standard no one can achieve. The point is that if management conceals managerial standard-setting as a "scientific" estimate of a number, it may take a fairly clever union bargaining agent to find out exactly where they have built the speedup into their science. The converse is also true. If "science" does not impress workers into working 1 1/2 times as fast for the same wage (1/.65 1 1/2), then managers have to shove it down their throats—which, especially if workers have a smart trade union specialist in time-and-motion study backing them up, may be difficult. Many managers felt that, since they had to assert their authority in the long run anyway, why do it with all these trappings? Management could just assert that it was management's job to get efficiency, ride roughshod over worker objections, and fire them if they continued to work at 65 percent of what management thought they might get out of them. Arbitrary authority of the "efficiency hero," not the science of efficiency, was the Henry Ford tack on the problem of speedup on the early assembly lines. In the Ford plant, engineers had authority to redesign and set time standards for the jobs because otherwise the assembly line would not run—one could not tie hundreds of jobs to an assembly line without con-

61 trolling all the workers in considerable detail. So the Ford management controlled them, science or not, and made a lot of cars—and that , not scientific management, justified their authority.

Conflict over the New Authority System For all the reasons specified by Thompson (1963), a lot of craft workers were not very enthusiastic about this destruction of the social definition of their skill within the factory. The trade union organizations in the United States (Taylor worked in the United States) were quite generally, up until about 1937, powerful only in the crafts. When the steel industry was reorganized from a craft to a mass production industry in the 1890s, it became in many ways a model for the mass production factory of the twentieth century (and the place where Taylor began to develop the ideology that defended such a reorganization of the authority to say what a job was). The big steel companies explicitly beat a strike in order to get the right to define jobs and skill levels managerially, and then to set up hierarchies of jobs and training for those jobs by the factory rather than by other skilled craftsmen. This reorganization eventually resulted in about a 30 percent increase in productivity (Stone 1975). This is the opposite of what usually happens. Union plants are almost always more efficient (Freeman and Medoff 1984). In fact, in the construction industry union workers usually do about 30 percent more work per day. The increase in productivity by breaking the union must have been due to going over to a system of job description and control more compatible with mass production industry: managers do not need flexible workers when they are producing one ton after another of identical rails for a railroad boom. Many further innovations in management, such as those described by Alfred Chandler (1962, and esp. 1977), depend on the system for defining work by management described above. Many of these techniques were first introduced into American managerial practice in the first large-scale industry, railroads. For example, Andrew Carnegie moved into making steel rails from the railroad industry (Chandler 1962, 406; Chandler 1977, 258–269; Beniger 1986, 240), bringing with him cost accounting (Chandler 1977, 267–269) and an orientation to precise specification of what time it should take to get a given bit of work done. This was part of his attitude in being willing to take on an expensive and violent strike in order to reorganize production management in the steel industry. Pierre du Pont learned the cost statistics approach to financial measurement of work in the steel industry before bringing it to Du Pont

62 and General Motors. Cost accounting depends on managerial standardization of costs, and a very important part of costs is labor cost. So management definition of adequate work standards is intimately bound up with modern cost accounting and quantitative management technique. But the engineering ideology of what a manager and a worker were like did not resolve all industrial conflicts. Not every worker believed his work was captured in fully by a cost-per-piece figure in the cost accountant's statistics, or in the motions and times associated with specific tasks by a time-and-motion study expert. In fact, like E. P. Thompson's workers, steel workers often thought that depriving them of their status, self-control, and traditional levels of remuneration so that the corporation could make more money was a heartless way to proceed. And they eventually said so again with unions in the 1930s, with the "soldiering" on the job that Taylor complained about, and with cutting no corners when the time-and-motion expert measured their work. People were hard to manage heartlessly.

"Fordism" A principal purpose of the conceptual development above is to have the concepts to define precisely "mass production," or "fordism" in Sabel's more evocative terminology (1982). "Fordism" we will define to be a productive system characterized to a high degree by five characteristics: 1. Batch programming mode. The main productive activity, the one that employs the most people, is one that has the structure of a "batch," rather than an "interactive," computer program. That is, for a great many activities in the firm, all the relevant decisions—and so virtually all characteristics of a given activity—are prespecified. The batch program may have a number of subroutines that are called with different frequency, depending on what the market wants; thus an assembly line, for example, may produce more four-doors, fewer two-doors, more black, fewer red, and so on, and still be running in batch mode. The key is that the discretion is only in the orders for the whole line, and once the numbers of red and black two-doors and four-doors are specified, the actions of almost all the workers are specified. Very little discretionary human input informs the decisions of what activity a given worker should carry out at a given time. 2. Semiskilled production workers. The skills of many

63 or most workers, then, include only a few routines, which they develop to a high degree so that they can do them fast and make very few mistakes. But their task skills do not form a large repertoire. This is what we usually call a "semiskilled" worker. After only a short time in the plant the workers can produce as fast as they are ever going to, and they know all the routines they are ever going to use, at least until the new batch program (the "model changeover") is brought in. Such workers are inflexible and know few routines, and are in that sense far from craftsmen. 3. Management with authority to create jobs. Management has the authority to create jobs, to specify the routines of jobs, and (quite often by way of collective bargaining) to specify the system of incentives and pay for the job and the measurement of performance (e.g., the speed, or the proportion of rejects, that will be satisfactory). 4. Engineering staff as designers of work. This authority to specify jobs that is "taken from" worker social structures like craft unions and put into the hands of management is used by professionals, especially engineers, to make the routines as efficient as possible, to rewrite the "program" of worker actions for a model changeover, and the like. The cognitive functions of designing a job are to a large extent located in a group of professionals rather than, for example, in supervisors who are flexible workmen promoted from the ranks because of their exceptional skill. The higher management tends to be recruited from senior professionals, and they design systems for creating routines for the workers, for enforcing them, and for measuring, rewarding, and punishing workers on their performance. 5. Skilled maintenance and related workers. The work of correcting difficulties, introducing changes in the productive process, and so on is not routinizable. This work includes maintenance, making the tools for stamping out metal parts, programming numerical control machines, "setup" of machines for producing different products, and mechanical drafting for the new products. Because many routines are involved, because the routines are not highly repetitious, and because the use of the routines has to be decided on according to the situation (they amount to "interactive" programming struc-

64 tures), they will tend to be done by skilled workers or semiprofessionals. To put it another way, the remaining sources of uncertainty in the production process will tend to be "buffered" or "controlled" (J. Thompson 1967) by skilled workers or semiprofessionals as well as by professionals and higher management, because uncertainty is too great. Dealing with the uncertainty so that routine production can go on involves learning a lot of routines. There will generally be a separate set of skilled manual work departments (maintenance, tool and die making, and special departments that vary with the technology, such as the crew who lay firebrick inside steel furnaces) and skilled staff workers at the managerial levels (engineering, quality control and inspection, scheduling and inventory), besides the whole routinized structure of the production line and the "line" supervisory structure that keeps it running. So this is what a "fordist" factory's social structure looks like: a large number of semiskilled workers doing many different jobs, each worker using a few routines and having little discretion; a smaller group of skilled workers in maintenance and related work; engineers designing the product and the production process, improving the routines in that process, and hoping to be promoted to management; and management, mostly former engineers, organized in a strong hierarchy ("line" management) to drive the people on the production line to work as fast as they can. Management also coordinates the engineering and maintenance process with the requirements of production. It is characteristic of such production processes that, except for maintenance and related workers at the lower levels who have a great deal of discretion, the higher people are in the hierarchy the more discretion they have, the more time they spend planning how to do things in the more distant future, and the less hierarchical relations—the more "committee meetings"—they have in their own work situation. Authority and discretion are arranged very hierarchically toward the bottom, more democratically toward the top. The way such a system looks from the bottom is described in Walker and Guest's (1952) account of a General Motors assembly line; the way it looks from the top is described by Chandler (1962, 114–162).

The Impact of Certainty and Uncertainty on Fordism For such a massive investment in developing and integrating routines to produce a fordist system to make sense, the production process must be

65 protected from uncertainty. If we contrast such a process as that described by Walker and Guest (1952) with, say, building construction, in the latter the workers, as much as management, deal with the variability in what the firm does one day as opposed to the next. It is construction foremen who adjust the amount of labor the firm hires to the amount of work there is to do. It is construction workers as well as foremen or managers who coordinate the work of different crafts. There is no particular "buffering" of the work of the skilled workers in their routines to take out the uncertainty; after all, the firm hired skilled construction workers in the first place because it was in an uncertain work environment. There are basically two ways of getting low uncertainty so that the technology will not be disrupted and routinization can be made to pay. The first is to select that part of the market that has demands with low uncertainty (we will come back to exactly what that means), and the second is to "buffer" the routinized part from the fluctuations of the environment (Thompson 1967). The registrar does not need to know that Sociology C-15 at Northwestern University is a fairly different course when I teach it than when Arnold S. Feldman teaches it, so the registrar can treat the grade that a student receives from me exactly like the grade a student receives from him. This means that the registrar can completely routinize the processing of C-15 credits and grades. As long as we keep the course basically the same so that students would not be permitted to take it twice, it will create no trouble. The registrar's routines are completely "buffered" by the professors giving the same simplified output, even though scholarship and the tastes of the different professors change the content of the course, the types of assignments that are graded, and so forth. That is, the professors "buffer" the uncertainty deriving from new scholarship, from differences in approach to a problem within the discipline, from differences in teaching philosophy, and the like, and provide the registrar with a standardized or "leveled" input, regardless of variations in the environment. The routinization of pathology work in a hospital relies on similar buffering by physicians of the variety of patients that come in, changing their various conditions into a few checks on lab test requisition forms. One can get the same buffering from outside the work organization itself. For instance, the outworkers in weaving that E. P. Thompson (1963, 269–313) talks about were given their work by merchants, who turned the variety of demands in the market, the difficulties of getting a supply of raw cotton or wool, and the troubles of getting the wool or cotton spun into yarn into parcels of routinizable tasks that could be given

66 to part-time rural weavers in small towns who had a loom at home. The "absorption of uncertainty" was done by merchants who did not run the weaving operation themselves.

Sources of Uncertainty in the Market But the big determinant of routinization of the whole productive process of an organization is the lack of uncertainty in the flow of demands on the organization; only with low uncertainty of demand is routinization possible. We therefore need to analyze where uncertainty in the demands on an organization comes from. I will argue that there are four main sources of uncertainty which tend to prevent the development of fordism, or to undermine it later: (1) unstable markets; (2) unstable specifications of the product (short production runs or "made-to-order" goods); (3) unstable technology; and (4) variable raw materials, parts, or environments. The basic argument takes the form that we will find fordism in one form or another more often when there is (1) stable demand (2) for a standardized product (3) whose technology of production is not changing rapidly and (4) which uses raw materials or parts that are easily available in standardized form and a technology that can be used in a stable environment (e.g., inside a building). It is in such conditions that firms can best use batch-program-like production processes, where there may be many decisions that have to be made once (such as for a complex product like automobiles) but once they are made the firm needs relatively low levels of discretion—there are few new decisions per unit produced. Conversely, we will expect to find organizations with a lot of skilled workers or senior professionals when (1) the market has wide fluctuations, or (2) each product produced is more or less unique or is produced in small batches, or (3) there is rapid technical change, so that either the firm has to produce a new product because there are new possibilities that someone else will exploit if they do not or the firm has to change their production processes a lot to keep costs competitive, or (4) the work has to go on in very unstable environments or use very variable parts or raw materials. 1. There are very stable markets in industries like electricity production, gasoline, beer, food products (especially those such as dairy products that are not seasonal), some kinds of textile goods such as sheets or towels; there are very unstable markets in contract construction or in producing capital goods such as machine tools or airplanes, and so to a

67 slightly lesser extent in industries that market to those industries, such as building materials or steel. One finds markets of intermediate stability in "consumer durables" such as appliances, automobiles, furniture, or rugs. In electricity and gasoline production, a lot of the decisions that need to be made for each kilowatt or gallon are actually built into the machines, so there are not very many semiskilled workers in those fields. Automation has gone furthest where production is most routinized. The main place where manufacturing firms still need flexible humans (and also where robots are most rapidly replacing humans) is in "quite routinizable" markets of intermediate stability. In such markets many decisions have to be made to produce each unit of the product, so routinization pays off even if one can only routinize for a year or so, and even if after a year one has to lay off half of the semiskilled workers, each of whom has been trained in a single routine. It is also important that there be a big stable market for the product. When the oil industry was mainly producing kerosene for lamps and each consumer did not use much kerosene, oil refining was much more of a small-batch process, with workers moving the stuff from one stage to the next. When routinization goes far enough so that the firm can completely automate the routine decisions, then they do not have so many semiskilled workers. But to do that the firm has to build the electricity plants or the oil refineries to be useful for twenty years or so of a steady market for gasoline or electricity. Where there is a stable market but the product itself is reasonably simple (that is, fewer decisions, routine or not, need to be made because there are so few parts), as in some parts of the textile industry that produce standard white goods, then one often finds a lot of automation and semiskilled workers buffering the uncertainty (e.g., getting the loom running again when the thread breaks; see Blauner 1964, 59, 62–65. 2. Industries with unstable specifications for the final product include contract construction; the movie industry; some building of very big machines such as oil platforms (Alvarez 1986a,b; Stinchcombe 1985c); ships and airplanes (Newhouse 1982), where it pays to adjust the machine specifically to the environment or the size of the market of the firm buying it; art products or the artsy end of craft products (jewelry, for example); academic publishing (Powell 1978); and the fashion end of the apparel industry (Vernon 1960). In all these cases each customer, or each small batch of customers, wants something somewhat different, and so one has to adjust the work process to produce exactly that. In contract construction (Stinchcombe 1959), the building has to be fitted to the site and to

68 local building codes and has to be the right size for the market of the firm it is being built for (or the size that the family buying it can afford). To some degree there is a new design for every building, and the contract for each building will have a unique combination of tasks for the building work process to accomplish. Similarly, one does not want to produce the same movie, even a smash hit like Beverly Hills Cop , over and over again until one can produce it in a month instead of a year and produce it with semiskilled actors (Faulkner 1983). Fashion goods need to be produced mostly after "the market" finds out what is going to sell in fancy shops this season, what will sell in the intermediatelevel "women's department store" market, and what will sell in the mass ready-made market (Vernon 1960). So the specifications for what will be produced have to be adjusted to the latest market information. In contrast, men's clothes, especially work clothes, or white textile goods, or work shoes, have a much slower changing set of specifications. Automobile specifications are set more by the factories than by the consumers, and at any rate mainly change once a year, and models of refrigerators or stoves often stay on the market for years. Quite a few basic industrial goods have very stable specifications (e.g., various grades of plate steel, steel I-beams, cement, bulk aspirin, or fertilizer). We would expect more skilled workers (for a given complexity of product) in ladies garments than in other apparel, in contract construction than in the construction of mobile homes, in rugs than in other textiles, in the construction of big machines than in the production of standard machine tools or hand tools, and so on. Conversely, we would expect more fordism in appliances that sell in the same model for a year or more, in textiles other than rugs, in industrial chemicals or other industrial raw materials than in the finished goods made of them, and so on. 3. Unstable technologies produce unstable specifications when the developing technology makes new products possible. We would expect that at present computer manufacturers, as compared with builders of other machines of the same complexity, would have more skilled workers and professionals not only because their production processes are always changing, but also because their product this year is twice as powerful as it was last year (Fishman 1981; Kidder 1981). Changing technology in general produces changing specifications. Especially when there are powerful reasons for the consumer to be up with the "state of the art," as in weapons of a high-tech sort, rapidly changing technology will tend to produce rapidly changing specifications of the product and consequently

69 short production runs. By the time a company routinizes the production process for a fighter airplane, the order is canceled (Coulam 1977). But a changing technology also produces changes in the production process, even aside from those involved in the changing specifications of the product. Computer chips are produced by a different technology this year than last year, while the locks on a car door are produced using pretty much the same technology this year as last. So one needs more skilled workers on the new production line of computer chips than on the old production line of car locks. One expects, then, to find a high proportion of skilled workers and professionals in the high-tech industries: weapons, computers and computer software, telecommunications, pharmaceuticals, medical hardware, or scientific instruments manufacturing. Conversely, in industries with relatively low rates of technical change, such as canning, beer production, steelmaking, and most appliance manufacturing, one would expect to find more fordism. 4. Finally, variability of the inputs into the production process produces a need for a lot of decisions that are very hard to routinize. For example, fishing on the North Atlantic, at least in the old days, was done in boats small enough that the weather was a big determinant of whether one could stay afloat. Fishing is still one of the most dangerous industries in most counties where there is any substantial fishery, because a firm needs a small and economical boat to be economically viable, since in any given area of the ocean one can find only a relatively small quantity of fish. Fishermen do not want to sail an enormous boat around to be safe when they are only going to catch a few tons of fish before they have to take them back to shore. So the weather is a variable input into the fishermen's production process because the main way to control it nowadays, building bigger and more capital-intensive boats, is not economically feasible. Similarly, underground mining, especially of soft rocks such as tend to be found in coal areas, is done in an environment that is very unstable, with variations in the width of the seam, the physical stability of the roof rock, or the underground water conditions. So although coal is, of course, a very simple product in both underground mines and strip mines, underground mines need more professionals and skilled workers. In addition to such natural variability, inputs can vary greatly too. For instance, one reason factories are so much bigger in the Soviet Union than in the United States is that Soviet manufacturers cannot manage to motivate parts suppliers to supply highly standardized parts (Granick 1967, 144–147). Consequently, instead of depending on other firms for

70 parts, a factory builds an annex and makes the parts itself so it can institute quality controls. The more routinized they want to make the assembly process, the more they have to be able to depend on standardized part. One can go into an assembly plant in some underdeveloped countries and see people sitting and filing off the imperfections in the parts before assembly (I happened to see this in mainland China in 1975, but it occurs elsewhere). In these circumstances, a factory has to have workers with files making decisions about when a part is ready to be assembled, because it cannot trust the supplier to supply standardized parts. Finishing parts in a highly standardized way to eliminate workers with files may therefore require larger factories. In particular, when the parts than an organization has to work with are individual people, it is in general quite hard to standardize them. And the more complex the thing the organization wants to do with them, the more difficult it becomes. Hospitals, psychiatric institutions, schools and universities, and other institutions that perform complex tasks on people have to take account of more variability in the qualities the individuals input to the process than do supermarkets, prisons, census takers, and the like. One can routinize the treatment of people more in the latter cases, less in the former (Leidner 1988). And the more one wants to influence individual decisions by clients, the more one needs to adapt to the clients' varying vulnerability to influence; thus it is harder to routinize the selling of expensive goods, especially goods that people buy only occasionally, like automobiles and houses, than of cheap repeat purchases like groceries or dimestore goods.

Conclusion Let us now look back at what we have been trying to do. We start off with the fact that some things that people want to produce in the economy are inherently more complex, in the sense that somewhere along the way someone had to make a lot decisions for it to turn out right. But fordism consists in making such complex products cheaper by routinizing not only all the decisions but also the production process to a high degree. The key here is that the more a fordist factory can routinize the whole process, the fewer decisions have to be made anew for each unit produced. So the real dependent variable here, the thing we are trying to explain, is, How many new decisions, nonroutinized decisions made by human intelligence and discretion, are there per decision that has to be made to make the product? Thus a formal measure of fordism would take the following form:

71

This is the level of routinization or degree of fordism of production. The more complex the product, in the sense of the more total decisions required to produce it, the more time and expense will be saved by routinizing. Mass production in the fordist sense saves more on a complex product like an automobile or refrigerator than it does on a simpler one like a sheet or towel. We have specified four variables that produce uncertainty in a production process, and have argued that the more uncertainty there is, the less the production can be routinized and so the less one can turn skilled work into semiskilled work by many different workers with different tasks (a "collective skilled worker with semiskilled parts"). We have also argued that the most opportunity for routinization will exist when there is less uncertainty from (1) unstable markets, (2) unstable specifications, (3) unstable technologies, and (4) unstable natural conditions or unstandardized parts that affect the process of production. The ratio of skilled workers and professionals to semiskilled in any given industry should be raised by all these kinds of uncertainty, because there will be more decisions (out of the total necessary) that cannot be routinized, and therefore have to be built into the skill of the worker:

where f is a decreasing function of fordism and an increasing function of complexity. The basic argument of Sabel (1982) is that there are many more industrial situations than we have imagined that are very uncertain, where both firms and individual workers have to adapt to a changing situation and yet do the tasks they do very efficiently to remain competitive. Further, the ratio of work settings that can be highly routinized to the numbers with irreducible amounts of uncertainty has not been going steadily upward, in the fashion argued by, say, Braverman (1974) and Noble (1984), because there are still just about as many fluctuations of the market, just as many people who want unique goods, just as rapid change of technology, and just about as much variability in the inputs to productive processes as there always was. There is, then, no steady tendency to turn all jobs into semiskilled assembly line jobs. If anything, Sabel would argue the reverse. The total skill level of the work force is not going down. In industries with large stable markets for standardized goods, with stable technologies, and with stable qualities of raw materials and working environment, we will expect

72 an approach to an equilibrium with a minimum of skilled workers and engineers, relative to the inherent complexity of the product, and a large number of semiskilled workers doing a few things each, with the routines of one intimately connected with those of another. Or in the extreme, we will expect that completely routinizable decisions will be built into automatic machinery. The first variable determining the information-processing structure of an organization is how complex the information processed by each worker needs to be. This determines how skilled the individual worker needs to be: how many different routines he or she will have in his or her menu of things that he or she can do fast and well, and how complex his or her reading of the environment needs to be to select the right routine. Stability of markets, of product specifications, of technology, and of raw materials produces an opportunity to routinize. Routinization, or "fordism," lowers the skill level by simplifying the information that needs to be processed by the average worker to make decisions; it does this by decreasing the number of decisions that have to be made anew for each unit of product.

73

3— Manufacturing Information Systems: Sources of Technical Uncertainty and the Information for Technical Decisions Introduction The main argument of this chapter is that middle managers in manufacturing set up specialized information systems and keep them running. If most of the decisions in most organizations are "programmed," in March and Simon's language (March and Simon 1958), someone has to program them. In particular, they have to program them in the light of the uncertainties that affect the departmental part of the operations. In this chapter we deal with lower level, "departmental," information and decision systems in manufacturing. We will find that there are different kinds of information systems in place in the different departments of a manufacturing firm. In particular, we look at the Norwegian State Oil Company (Statoil), which extracts oil and gas from under the North Sea and prepares it for shipment. We will be concerned with where different sorts of technical uncertainty come from, what sort of information is news about those uncertainties, and consequently what sort of information has to be fed in to make the routine decisions correctly. "Correctness," as we will see, changes over time, as technical development (e.g., in software design), organizational experience (e.g., such as shows that the inventory of a spare part does not get depleted as fast as was at first estimated), or new knowledge about the environment (e.g., where there are soft rock formations that may collapse into the well being drilled) change the criteria of correctness. Thus, information systems have to be built to be adaptable to uncertainties about which baseline information changes over time. The departmental routines of information and decision are, then, very analogous to skills. They do common things, that have to be done fast and well, because they are routines. They do the right thing because the choice of which routine to use, and when to build a new, better one, is a matter of managerial discretion.

74 The departmental information systems of this chapter together constitute the managerial system of the "manufacturing" part of a modern organization. It is such a system of differentiated complex parts that has to be coordinated with marketing and with production and process development engineering, integrating the information systems to create divisions, that we take up in Chapter 4. Without understanding how intricate the system for "routine" manufacturing is, it is hard to understand why special structures have to be built to coordinate it with marketing or with product design. Systems of comparable complexity, but quite different structure, exist in engineering and marketing.

People Driving Versus Information Systems in Management The first idea we have to get out of our minds is the "people-driving" view of what modern management is about, the Taylorist view. The data on what bureaucrats in manufacturing organizations do with their time are not really compatible with this picture (see, for example, the tables in chaps. 3 and 4 of Stinchcombe 1974). First, much of managers' time is spent reading, writing, calculating, drawing, and otherwise relating to documents. Second, they spend a good deal of time on innovations, rather than on routine operations; this is more true the higher their status but the fewer their subordinates. Third, the mean of their own estimates of how often they report to or are inspected by their superiors is in the general region of once a day, which is a small part of their total communicative activity, this low amount of "hierarchical" communication is not compatible with the higher part of the hierarchy having a supervisor-worker relation. The general problem here is that organization theorists have very little idea of what there is to talk about (or write or calculate about), where such information has to flow and why. We therefore have very little idea of what relation the hierarchy that appears on the organization chart bears to the total flow of communication, because we cannot easily imagine what the communications are other than supervision. One consequence is that we do not know what sorts of information there are in a manufacturing administration to be connected to engineering or marketing information. The answer that is implicit in the general hypotheses of this book, as outlined in Chapter 1, is that there are likely to be a lot of different types of information in the manufacturing system, for dealing with various sorts of uncertainty and consequently for making various sorts of decisions. For example, there are likely to be millions of

75 pieces of information on the spare parts needed for maintenance work on any complex system of manufacturing machinery. The uncertainties that this information system is about are the uncertainties of how likely it is that we will need that particular part, what specifications it must have, and where one can get competitive bids on it. That information has to be connected periodically to the inventory of that part and, when the inventory is down, to the purchasing department to buy a new supply. It will be very unusual that such information will have to flow anywhere in a hierarchy, except in very aggregated form as a budget estimate. But it has to buzz around at low levels all over the manufacturing system, and all over the construction project that builds the installation, so that the relevant millions of bits of information can be put together in orderly form and kept up to date. One of the many things that spare-parts middle-level and lower-level technical managers in a manufacturing department do is set up this system, keep it running, update it, see to it that the routines are buying the parts when necessary and forwarding them to skilled maintenance workers when the time comes. But since there is almost no pyramid to this flow of communications, the fact that the managers are lower or middle level actually has very little to do with what communications they enter into. Their communications are determined by what it takes to keep a million pieces of information in order, not by the hierarchical pyramid. The functional requirements of this system will be determined by where the information about reducing the uncertainty of how many parts one needs to have in stock, what parts will be satisfactory, and where one can get them cheapest are to be found. Thus we will expect that the structure of this system will look quite different from, say, the system of information in the software in the computer center. What has to be done to ensure that people can solve their problems with the software in the system is quite different from what has to be done to ensure that spare parts will be of sufficient quality, will be ordered on time, and will be bought from the cheapest supplier of that quality with the required short delivery delay. The social and organizational features of these two systems will therefore be quite different.

Some Data on Manufacturing Information Systems I had a chance to study the buildup of the operations administration of the Norwegian State Oil Company (Statoil). The following descriptions of information systems reflect systems being built in preparation to oper-

76 ate a platform that was about to be delivered to them (by the construction part of Statoil's own organization). The platform was delivered while we, Carol Heimer and myself, were interviewing for the study. Our objective was to describe strategic aspects of getting ready for operations (the reports on this work are Heimer 1986a; and Stinchcombe 1986d). In order to design the information systems, Statoil middle managers had to know what general kinds of information had to go into them. Then they could describe the problems of determining the systems' structure, of putting starting values on the pieces of information they needed for making the first decisions, and of making sure that one was ready to update the systems as new information came in. This descriptive material on what is actually in a manufacturing administrative system's various information files turns out not to be in the organization sociology literature (cf. Cooper and Gaskell 1976). In what follows, the data on information systems' general characteristics come from only one organization, and that organization was not yet manufacturing anything. This is not the ideal sampling scheme for describing manufacturing information systems, but a sample of one is of course a great deal larger than any now in the literature, and the fact that the administrative system was being constructed meant that the people had recently thought a good deal about the structure of their part of the system. They were therefore probably better informants than people who merely operate a system handed on to them by their predecessors.

Operating Characteristics of Information Systems The general purpose of this and the following section is to describe types of information used in manufacturing operations so that they are distinguished by the operating characteristics of the bodies of information—how information is generated by and incorporated into activities. An example will make this clearer. A preventive maintenance program has the purpose of repairing all potential faults in machinery before the machines fail, by replacing or reconditioning them beforehand. These programs are especially crucial for machines whose failure might be fatal, such as airplanes or oil installations. Part of the information embedded in such a program comes from thousands of bits of experience: experience of the mean time to failure of thousands of parts, experience about how good a predictor this mean is for failure of particular individual parts (experience of the variance of time to failure), and experience of the costs of having that part fail unexpectedly before it is replaced or repaired. There is no useful general en-

77 gineering theory from which such pieces of information can be derived ("reliability theory" in engineering [e.g., Green and Bourne 1972] is instead about combining that information for different parts of an interdependent system to get the temporal distribution of failures for the system; it does not generate the basic numbers about failure times for the parts). Variations in operating conditions mean that generalizing from, say, the general behavior of the parts of pumps to the behavior of a particular pump and its parts is not very useful. In addition, the incentives of the group planning a preventive maintenance scheme are strongly affected by their not usually being responsible for the budget for person-hours to be used in opening up a perfectly good pump just to make sure it is all right. An operating characteristic of this information, then, is that it tends to be quite bad at the beginning of operations of a system and to improve rapidly with experience. So preventive maintenance plans are hard to create before operations begin and are not at first very good approximations to the plan that is in fact best for operation because the information in them is poor. Plans created before operations start, then, have a certain fictional or utopian quality and need to be adjusted drastically to reality in the first months of operation. This dependence on experience makes the information contained in a preventive maintenance plan distinct from, say, the accounting information from the project that gives the capital value of the installation. The dollars paid for welding convey the same economic information as the dollars paid for compressors, as far as the tax authorities or insurance companies or financial auditors are concerned, and it is they who have to be satisfied with the numbers in the capital accounts. The capital value estimate for an installation does not improve with experience, and in fact perhaps gets worse, since depreciation is not usually calculated in a way that uses experience to reestimate an asset's economic value. The current economic value of an installation might be equal to the investment required now to create an equally efficient productive facility with the same capacity, and would not be estimated by the experience of the installation at all, nor by depreciating the cost of an old installation by any of the common accounting schemes. British Gas attempts to estimate the "current cost" of their capital equipment to assess their profitability and to set the appropriate rates for gas service, probably generally rational in the oil industry, where capital equipment is so long-lived. But tax authorities and credit institutions, who have learned to use historic cost figures, resist such innovations. The main dimension that differentiates the information embodied in

78 a preventive maintenance plan from that in a capital account is the degree to which operating experience improves the information. Operating experience gives information on—to give one example —the mean time to failure of a general class of pumps in the operating conditions of an oil installation. If a pump is left until it stops, the time to failure of that particular pump at that particular time is known with certainty, rather than from an estimate of a mean. The way we keep our financial accounts, the cost of a capital installation is not any better estimated after operating experience has been gathered, though of course the stock market will react to the profit outcomes of operating experience regardless of the book value in the accounts. Such a dimension for differentiating bodies of information tells us something about how we would expect the respective information systems to be reorganized and updated as the transition from the project or building phase to the production or operating phase of a manufacturing establishment takes place. It also tells us to look at other systems of information to see whether change of organization of such systems as one moves into operations is affected by their degree of improvement by operating experience. The following examples of types of information systems in manufacturing organizations and other similar organizations identify dimensions of differentiation that will be useful in analyzing the problems of creating information systems adequate to the manufacturing operating phase.

Types of Operating Information Systems Well Reports The report of experience on a given well provides a great deal of detail about the geology in a particular field, and about the technical solutions that have worked to solve the problems that geology poses. That means that the well plan for the next well in the same field will be based, in great detail, on the reports for surrounding wells: where shallow gas with its danger of blowout is likely to be found; where the rock is solid enough to anchor the casings with cement; where the formation is weak enough that the weight of a mile or two of heavy drilling mud is likely to break out of the hole into the formation instead of returning to the top to carry the drilling chips and be reprocessed; and so on. I may of course have the technical details of what needs to be learned about the geology and what it implies wrong (an amateur's introduction to drilling can be found in Alvarez 1986a, 58–60), but my list gives the flavor of the information a well report contains that can be used in subsequent well plans.

79 This information is to a high degree environmental information about a particular field, is unique to that field, and is obtained within the drilling operation rather than from any outside source. Further, it is this unique internal information on which almost all crucial technical decisions depend, so that almost everything important that a driller or a drilling engineer has to learn to make his (or her, if drilling were other than what it is) operation succeed is internally generated. To exaggerate, everything a drilling engineer needs to know to plan a new well is generated within the drilling department from experience of previous wells. Drillers talk only to each other because other drillers (especially drillers in the same field) are the only ones who know anything of much importance. The respects in which this is an exaggeration are crucial for analyzing how drillers relate to the company's construction and operations organizations and information systems. For from one point of view, the drilling of wells is only the first stage of operations of a manufacturing installation (an oil well) that will produce oil, gas, or both in a condition ready for shipment. A closed circle of drillers' communication does not serve the larger organization well. In the special case of the typical North Sea platform (until 1986, at least), in which the drilling equipment and the oil processing equipment are installed together on the same platform, the platform as delivered by the construction project is at first an extremely expensive drilling rig. The main operations that go on on the platform are drilling operations, until the wells are ready to produce gas and oil. So the first "operators" on the platform are primarily drillers, and drillers have to coordinate their schedule with the schedule of completion of construction and installation of equipment. They have to make sure the drilling equipment designed into the platform will in fact drill production wells efficiently and safely (or else they have to modify them so they will). In general, they have to collect information sufficient for their operations from the design and construction organization, to produce operations manuals for the drilling part of the platform, to prepare themselves to maintain the equipment, and so on, just as any other operations group has to do. From this point of view it is really the drilling division that "takes delivery" of the platform from the project organization that built it. From another point of view, drilling is the last "investment" activity that prepares the platform for producing oil. Of the total investment preparatory for operations in the North Sea, roughly 60 percent is for the platform and the production equipment and 40 percent for the drilling and well completion. Drilling goes on for a relatively short period (roughly a quarter of the life of a platform), and after a well is drilled all

80 that the production organization has to do is let the well flow. (If anything else has to be done, the drilling department comes back in to "open the well" and to "do a workover," producing a well with new characteristics.) The completion of a well (that is, turning it from a hole in which drilling is going on to a hole from which oil and gas can be produced and delivered to processing equipment) has to be coordinated with the readiness of the production organization to handle the gas, oil, and water produced. The well has to be technically connected to the production system so that it has no leaks and can be controlled and shut down in emergencies. The product of a drilling operation is not simply a hole, but a production hole, a hole that can be operated to produce gas and oil profitably; thus it bears the same general relationship to operations as does the construction project that builds the installation. The specifications for the output of the drilling operation ultimately come from production, and the hole that drillers drill has to be technically coordinated with the production system. The "specifications" for the characteristics of the production well need to be those that the operating or producing organization can work with, so the relations of production with drilling are much like their relations with the project that builds the installations above the seabed. The withdrawal of drillers into communication only with each other is framed on each end by organizations with which they have to coordinate schedules, technical criteria, and economic criteria: the project or manufacturer that builds the drilling equipment, and the production operation that uses the hole drilled as a capital asset in its work. Drillers in Norway celebrated having made it through another week alive with coffee and whiskey on Friday; they had a rubber stamp to mark "Bullshit!" on the documents from other parts of the organization that distract them from the important matters of dealing with an open hole into which shallow gas may come at any time. These are informal reflections of drillers' dependence on the information generated by drilling operations, for they strongly mark the boundaries within which it is valuable to pay attention to what others say: the rest of the world does not know what is really going on. The delicate negotiations to get drillers to specify what spare parts they will need somewhat in advance so they can be ordered and inventoried with other spares by the production organization, or to get them to coordinate the design of the well completion with the project's design of the apparatus the completed well has to connect with, reflect the fact that drillers' withdrawal can be only a temporary phase of the total development of the oil field. The dimensions that differentiate well reports (and well plans) from

81 most other information systems, then, are that the information that is crucial to drilling operations is fairly unique to a single operating organization and that it is generated from experience in operating a particular installation in a particular environment. No one else needs to know exactly where the shallow gas or the weak formation that will let the drilling mud out of the hole are, but drillers cannot do their work without that information, and that information is generated within drilling. This is an extreme case of a common feature of technical operations information systems, that much useful information is to be obtained only from operations and is of interest only to the people operating particular machines. It is called "plant-specific skill" by Charles Sabel (1982, 58–62). Operators and their supervisors all learn and use information about the peculiarities of the particular machines they use, or have useful classifications of the detailed variations in the raw materials they use, or notice variations in the environment useful to their work that are not noticeable to others. (The many words for snow in Eskimo languages is the standard anthropological example of such "local knowledge"; the many words in Norwegian for types of rain and rain mixed with snow or ice would serve as well.) When a large share of the information used in a given activity is such "local knowledge," a subculture grows that is more or less isolated from the rest of the organization. That subculture is organized in large measure around an information system that is of little use or interest to anyone else and so is adapted to particular concrete features of the environment, uses an arcane language or system of notation, and resists invasions by standards from larger and more uniform information systems. Other extreme examples in the oil business are the culture of deep-sea diving, helicopter piloting (for both, see Alvarez 1986a,b), and the "systems" part of computer programming (see Stinchcombe and Heimer 1987). Work based on such specialized and bounded information systems is quite often subcontracted for rather than built into the hierarchical plan of the main manufacturing company organization, and may often be found in a section of an organization chart or a project plan under some such title as "specialized contracts."

Software Systems The distinctive feature of applications software systems in a large organization is that after they are put into operation they usually grow by about 10 percent a year (in lines of code or in the number of different reports they put out, two highly correlated measures; the revised program may not be integrated into a single file, and some of the additions may "stand

82 alone"—these are included in the 10 percent [Lientz and Swanson 1980, 34]). The information embedded in a software system then grows rapidly as it is operated; programmers constantly write code to apply the power of the computer to produce new forms of information from the same data base or to increase the complexity of the data base. The new subroutines' output is used in operations, in decision making, in reporting on activities to outside authorities, and so on. In the computer business as a whole, about the same rate of growth characterizes "systems" programming, such as operating systems or programming languages, with the main growth being in the number of "options" in the system. We can guess that two sorts of processes go on here. The first is related to the fact that computer programs are still innovations. For all innovations, the first developments are directed at the main purpose of the innovation, or to functions where the innovation is highly productive. As the innovation persists, more and more "elaborating" innovations are added to it, fitting the innovation to occupy niches in the productive system that are more peripheral to its central contribution—where the market is smaller, the innovation less productive, or the value produced smaller (Grilliches 1957). One writes a program for the main financial reports, for payrolls, for accounts receivable and accounts payable before one writes subprograms for cost analysis, which are in turn written before subprograms to take care of the special features of travel vouchers. One writes a programming language to translate algebraic expressions into computations before writing programming languages for matrix algebra (or before adding matrix options to the algebraic language), which are in turn written before packages convenient for solving systems of nonlinear equations. Thus we expect that all innovations when they are new become more complex as they add progressively less productive complexities to the original major innovation. Since the hard part of making these supplementary innovations in software is apparently describing their specifications (which is often done by research about what the users really need) rather than inventing a way of satisfying those specifications, innovations in a particular software system tend to be produced at a more or less steady rate. Software is one of the few fields in which we have regular undergraduate educational programs for training "inventors" of new devices, so the movement to progressively less productive innovations in a given software system is more routine, regular, and predictable than in most invention. A second possible explanation for the growth of software systems dur-

83 ing operations is that user understanding of what they would really like to have is at first shaped by what they could have when they had to do it all by hand. As the cost of producing new types of differentiated reports decreases, people discover what they would really like to have because for the first time they can think about it realistically: what report format would be easier to read, or what further calculations in the report would involve less work with a pencil to get at the number that really tells them where they stand. People do not bother to think about what they might want until it becomes possible; the existence of an operating software system with most of their data already in it makes many more things easily possible, and so encourages the development of new specifications. Probably many of the new specifications of reports or other information products of the software system do not actually increase the productivity of the operating organization by much, but their marginal cost, once a basic information system and data base are computerized, is small (and the estimate by the software department of the cost to develop the elaborating innovation is often even smaller). The situation with software systems is in some ways similar to that with the cost reduction use of cost statistics described below. The beginning of operations for both systems starts a systematic serial change of the goals for which the information system is being used. Just as one goes from the most obvious and productive cost-saving projects to the less productive ones after a detailed cost analysis system is introduced, so one goes from the most standard and useful reports from a data base to those that will be useful only for small audiences or will only slightly improve decision making. Similarly, just as one finds out by experience with a production line what cost-saving innovations need to be analyzed, so one finds out by experience what reports other than the standard ones are that one might want to get out of the data base by improving the software system. This constant shifting in the goals of an information system requires detail, complexity, and flexibility—increasing "options"—to be built into the information system itself. Flexibility, detail, complexity, shifting purposes, and shifting applications characterize software systems, and the shifts of purposes and applications come more rapidly as the software system begins operation. The dimension that differentiates software systems, at least in degree, from other information systems is that after a decade of operations about half the information structure in the system has been added since operations began. The continuous innovation that characterizes software systems makes the development phase of the system blend into the operat-

84 ing phase, because both involve a steady stream of capital investment in the capabilities of the system by a steady stream of technical innovative work.

Cost Accounts, Cost Analysis, and Financial Accounts The primary value of a system of cost statistics for an operating organization is to suggest where it is strategic to look for cost improvements. This is quite different from the main value of cost statistics for project or "building" organizations (see Stinchcombe 1985e), which use costs of component activities of the building process to construct estimates of the total cost of building proposed installations, depending on the amounts of the activities that will be involved in construction. Cost statistics are very little used in project organizations to minimize costs (except in choosing between installation designs involving differing amounts of different building activities and different materials). The use of cost statistics for operations can be broken into two main parts: (a) comparing activities or products produced by a given organization with standards derived from productivities elsewhere or market prices elsewhere and (b) causal analysis of particular activities to seek modifications that will most reduce costs. I will call the first use "Cost Comparisons," the second "Cost Reduction." Then there is the traditional use of accounts for tax and financial purposes: (c) ledgers and financial accounts. Cost Comparisons . Costs at a more gross or aggregate level can be used for comparison of lines of activity within a company with comparable lines of activity elsewhere, or with market price. These can be used to suggest lines of business a company might emphasize because of its comparative advantage or might go out of because of comparative disadvantage. In a rather similar way, they can suggest which components of a productive process it would be strategic to subcontract or buy (wherever the market price is lower than the costs of internal production) and which to internalize. There are sometimes technical or economic reasons to keep activities within the firm, for example where there are strong technical or scheduling interdependencies with other activities (e.g., where prompt maintenance is required to keep production lines running—more generally, see "The Decoupling Principle in Project Administration," pp. 68–71 in Stinchcombe 1985e), where contracting would give the contractor access to sensitive information (this is common in defense industries and in software development), or where a contractor or supplier would be placed

85 in a partial monopoly position (e.g., in running pipelines to a gas or oil field). In these cases cost comparisons can suggest low-performing activities among those which, for strategic reasons, have to be within the company—places in the organization where administrative energy and information about industry practice need to be injected into the operating system. High costs of a line of activity in the firm compared with industry standards show that such information and energy will be productive. The principal operating characteristic of such a use of cost statistics is that the categories of costs have to be comparable to generally available cost standards. Costs may be aggregated into units that can be bought in the market, to allow comparison with available market alternatives; or they may be aggregated into units in common use in cost accounting systems elsewhere so that accounting consultants with wide industry experience will be able to locate costs that are above standard. Thus they depend on what Carol A. Heimer has called a "negotiated information order" (Heimer 1985b), in which a local information system must use the categories and standards accepted in a larger system of organizations rather than those that might be convenient for their own decisions. In particular, cost comparisons in an enterprise in which a firm goes into a new industry requires using the categories commonly used in the same business, which may not be the same as those used in the firm's original business. Cost Reduction . To use statistics for cost reduction, one needs statistics that are broken down by significant causal components of costs (see the analysis in Stinchcombe 1974, 3–31). That is, one wants to know what costs can be manipulated, and costs subject to distinct methods of reduction need to be distinguished in the statistics. If an investment is required to change a production process, either formally, as with investment by buying a new piece of equipment, or informally, as with investment of managerial attention or retraining time for workers, the payoff can be evaluated if the cost contribution of the improved process to total costs is known. In general, as the opportunities for cost reductions with one piece of cost statistics analysis are used up, new analyses are appropriate, so it is important to build flexibility into the system so that different detailed categories can be used for analyses at different times. The operating characteristics of cost statistics used for cost reductions, then, are somewhat in conflict with those used for cost comparisons. Rather than industry-standard categories at a quite aggregated level rigidly enforced throughout the cost system, one needs a detailed dis aggregated system of statistics, with easily changeable levels of detail so that one can analyze exactly the costs one is hoping to reduce with a

86 new program. As one moves into the operations phase of a manufacturing enterprise being newly started, one wants to adopt more detailed and more flexible cost categories and to have an increasing part of the system of analysis not be standard reports but instead specialized studies dealing with changing cost reduction strategies. An operating system is subject to a "learning curve" of cost reduction. That curve consists in reality of incorporating a long series of cost-reducing projects, involving improved equipment, improved procedures, improved skills, and the like, into the routine functioning of the production lines. It is information support to a series of such cost reduction projects that constitutes the cost reduction aspect of a system of cost statistics. Cost reduction uses of cost statistics are therefore distinguished from the cost comparison use by having the detail and flexibility that makes them useful for a shifting mix of cost improvement projects. The dimension differentiating the two uses of cost statistics, then, is from rigidly enforced agreement with industry-standard aggregated cost statistics to shifting and flexible categories corresponding to the causal system that has to be manipulated to reduce costs. (An account of a similar problem in Norwegian oil insurance loss experience analysis—loss experience data are of course cost accounts for insurers—is Heimer 1985c, 216–221.) Ledgers and Financial Accounts . Much the same dimension as distinguishes cost accounts for cost comparisons from those for cost reduction also differentiates all cost statistics from accounting for tax reporting and reports to owners—"ledgers" or "financial" accounts. Financial accounts have to provide statistics that cannot be fudged: legally enforceable numbers for taxation and auditable numbers of financial condition for use in the capital markets and for the assessment of management. There is an industrywide or economywide information order, though it might better be called "enforced" than "negotiated" (Heimer 1985b). Accounts are carried at a very aggregated level, so that capitalized values to be depreciated by the rules of the taxation system are comparable across all firms and profit and loss statements mean approximately the same things about the financial condition of different companies to banks, stockbrokers, and investors (Meyer and Rowan 1977). But in addition to the dimension of rigid comparability across the economy rather than flexibility for internal use, all the data that go into financial accounts have to be auditable. There has to be a system for legitimating each figure of costs (e.g., an invoice that appears in some other company's accounts under accounts payable) and of profits or capi-

87 tal accumulation, which an outside auditor can check to see whether the figures have been produced by legitimate means. Financial accounts have to ensure financial discipline and honesty, as well as provide comparability between enterprises. (Of course, financial discipline and honesty are essential to the comparison process as well.)

Project Design The drawings, specifications, quality control procedures, and the like that make up the description for a complex manufacturing installation are supposed to be a description that the fabrication and construction contractors who build the facility can bid on and use as a guide in their activities. Of course, the specifications have to be oriented to the way the equipment will be operated, since the purpose of the design is to produce commodities during operations. But the primary problem to which this information system is dedicated is the construction and delivery of the installation on time and within budget, according to the functional specifications and the broad strategy of design and construction specified (ultimately, at least) by the buying organization. We will call the information and decision system that guides fabrication of the capital installation "the project" (see Stinchcombe and Heimer 1985 for more descriptive material), and the one that operates the completed installation "operations." But after the installation is constructed, some parts of the operations depend not only on broad functional specifications (such as "closing the fire alarm manual controls should shut down the processing train and close off the supply of oil and gas from the wells safely") but also on details (such as "for maintenance the screen in this instrument should be replaced with 90% fine platinum screen of the same size"). Broadly speaking, the maintenance activities, which in the case of highly automated production like that of oil and gas or chemicals may constitute up to 70 percent of the work to be done during "operations," depend on many of the details of the design, while operating activities depend largely on functional specifications. For example, the spare parts system needs to have categories of electric motors with feet versus those with flanges to distinguish those that can be mounted in different kinds of structures, so that the maintenance craftsman can mount the replacement motor where he has taken out the motor installed by the project. The operator, however, need know only that shutting a given switch turns on the motor that runs a given pump; he does not care whether it has flanges or feet. Even in the case of maintenance programs (and corresponding spare

88 parts buying programs), much of the information in the design for fabrication and construction will be useless, and some new information will be needed (e.g., mean time to failure of an electric motor in occasional use, alternative sources to buy particular spare parts). And particular pieces of information "needed" in the design may have much less importance for fabrication or construction than they have for maintenance. Which of several models of motors with flanges produced by a given supplier is used in a particular place is of very small importance in construction; it may be of great importance in sending out a supply of parts when the motor will be overhauled in preventive maintenance. A particularly important difference in information usefulness for the project versus for the operations organization in that toward the last stage of building, what the construction or fabrication people need to know about a change is only what activities they need to do to change it; what the operators and especially maintenance people need to know is how it looks at the end—the "asbuilt" drawings and technical data sheets and parts lists. Thus, while much of the technical information in the design is needed in operations, different parts of operations depend on different amounts and kinds of detail in that information, and need different amounts and kinds of supplementation to that information. The design information has to be reorganized in several ways to provide the technical information needed for operations. Some of that reorganization is ordinarily done formally by the construction and fabrication project, such as the production of "as-built" drawings and documentation. However, since the adequacy of this documentation is crucial for operations, but for the project it is merely one last bit of paperwork, there is a problem in getting the project to produce the quality needed. Similarly, some of the reorganization, such as finding competitive bidders for supplies of spare parts for maintenance, is ordinarily done by the operating organization, who need detailed supplementary information, partly not needed for the construction process but most easily obtained from vendors and contractors with whom the construction project has most contact. Since construction and fabrication do not depend (or at least not much) on knowing competitive bidders for parts, it is hard to get the project interested in supplying that information to the operating organization. In short, while most of the information that went into the design of the project will be buried in archives, never to be retrieved, much of it has to be massively reorganized to be useful to operations. Yet the project that has the information often has neither strong motives nor a strong sense of responsibility to provide quality information. The dimension that differentiates design information from the other

89 information systems we have considered is that in design the information guides one kind of activities, fabrication and construction, and is therefore organized subject to project imperatives—to have the information ready for use only once in the construction process but that once to have it ready on time. It has to be reorganized and supplemented to guide quite different activities (especially maintenance, but also operations) such that these can be repetitively performed and can be improved over time. So it has to be obtained for reorganization from project people who are geared by the incentive system to have information ready only once, and then on time; a core problem of such information systems is the organization of project responsibility and motivation for doing a good job on someone else's behalf (a question addressed in Heimer 1986a). This problem, however, occurs in other information systems as well, as a leading software designer makes clear in the illuminating statement, "It takes only about a third of the total work on a system to get it to the point where we can make it run" (Allman and Stonebreaker 1982, 31). Unless software designers are strongly motivated by user-friendliness, one gets awkward solutions and indecipherable software manuals.

Information about Authorizations Several information systems are distinctive because the information in them has been approved, so it can be used as a basis for action. Perhaps the most important systems producing approved information in oil operations, for example, are the budgeting process, the design approval process, and the complex of certifications and approvals involved in safety. The design approval process for an installation in Statoil consists of two main parts: the approvals leading up to the "design freeze," perhaps embedded in the project manual with supplementary design documents; and the approvals that result in a "design change proposal," for a change slated to take place after the design freeze. The budgeting process, resulting in the set of information that approves of expenditures in the production organization often stretches out over much of the year (though most people involved do not do budgeting full time for very long). The safety systems in oil operations or chemical operations are comparable to those in airline operations, though they are generally more complex than in other manufacturing. In oil operations, the safety approvals system involves such things as the classification of areas according to the possibilities of the presence of explosive hydrocarbon and air mixtures. The classifications then vary over time—for the construction phase before drilling, from the start of drilling but before the startup of

90 gas and oil processing, and for the normal operations phase of actual production. Other safety authorizations include special approvals for doing "hot" work (that presenting ignition risks, such as welding) or for performing lifting operations on platforms at sea while divers are under the water. For all these systems there is a radical difference in the implications of a piece of information depending on whether it has been approved or not. The process of transforming information from proposed to approved is central to the whole formal part of the organization (Barnard 1946). The transformation from information for decision making to approved plan is much of what the authority system is for, what meetings among authoritative and high-ranking people are mainly about, and what causes much of the bureaucratic delay characteristic of decision making (Stinchcombe 1985c). It often takes quite a while from the time the information becomes adequate to show what the decision will have to be until that decision can be acted on, because the information about what we will (in fact) do has to be transformed into an approved decision, with an approved design, an approved budget, and a safety clearance, all attested by the correct number of signatures and meetings, all auditable by a quality assurance team, and perhaps by ordinary financial auditors as well. At any given time one could say that the formal organization consists of the body of information whose approval has made it a valid guide for action. (Occasionally people will in fact describe their responsibilities as seeing to it that people follow the approved procedures.) The operating characteristics of approvals systems depend a great deal on the particular structure. One description of such a system (at a lower level) was "as a yo-yo": "I make a draft, it goes up to my supervisor, he returns it to me with comments, I revise, the revision goes to my supervisor, who sends it to his supervisor, who returns it with comments, I revise. . . ." Another description of an approvals system at a very high level in the oil industry is for design change proposals: after informal circulation in both the production and project organization for comments, leading to an informal approval to send the suggested change to be sent on for formal approval, the proposal may go through a premeeting and then a meeting of a cooperation committee made up of production and project people, then perhaps (for high-cost proposals) to a meeting of the general management committee for the whole organization. It is of course much easier to change a decision taken in a process of draft-comment-redraft-comment at a low level of the organization than it is to change one that has already required hours of higher management meeting time.

91 Such systems of approved information (or of "records of decisions and decision-premises") are "auditable," by ordinary financial auditors, by internal or external quality assurance auditors, by an "internal investigations department," by superiors, and so on. It is convenient to explain what goes on in modern quality assurance programs, for example, by making analogies with police versus contract law enforcement, with quality assurance more similar to contract enforcement. The main limitation of ordinary police enforcement is that police must act within given uniform rules, so that everyone stands in a position of "equality before the law." The great contrasting advantage of contract law is that in each area of commerce, people can set up rules convenient for their purposes and write them into their contracts. Then if the courts enforce the contracts, they will be enforcing arrangements set up to be convenient for a particular line of activity, and the obligations that they enforce will be different for, say, the typical insurance company than for the typical manufacturer. Financial auditors are perhaps the first example of enforcement organizations for formal administrations that are set up more like courts enforcing contract law than like police departments. Auditors try to see that the accounts of a firm adequately represent the financial implications of whatever the firm does and that those accounts are honest—though they represent quite different sorts of activities and legal situations for, say, a local government than for a manufacturer, and still different ones for an insurance company. For instance, an auditor concerned with enforcing honesty in an insurance company's maintenance of reserves, on the one hand, and a manufacturer's inventory evaluation, on the other, faces two very different problems. What the auditor actually enforces varies with the situation, as reflected in the accounts kept in a firm or other organization. Some of the responsibilities of higher management are of this "auditing" sort, seeing to it that the procedures that reflect efficient manufacturing are enforced in the manufacturing division, those that reflect effective marketing are enforced in the marketing department, and those that reflect effective R&D are enforced in the laboratories. In recent years some highly complex and differentiated companies, such as IBM, have developed internal management audits, which try to ensure that procedures in different parts of the organization adequately reflect the true management problems of those parts and that the procedures are adequately implemented in reality. One does not want a police system to enforce uniform ways of getting quality performance regardless of the actual managerial and technical situation. Instead one wants

92 enforcement of what the best management in a given area thinks is the best way to proceed. What is being audited in such cases is the system in a particular part of the company for changing "decision information" into "approved plans," and for changing "approved plans" into "implementation." The main dimension that differentiates such systems of information from the others we have considered is that attached to the specialized information system of a given part of the company is a supplementary information system that tells which pieces of information have been turned into approved plans. These are the routines of decision making, with their corresponding records of approved decisions. They are supported by a system for seeing that such approved decisions are either carried into action or else challenged by approved routines for changing decisions. This dimension is, then, the degree to which there is an authority system (both decision-making and enforcement aspects) attached to the information system. A second dimension of such systems is that these decision-making routines have to vary with the situation. To take a very simple example, some decisions will need a technical review to see whether the engineering will work, while others will not. Likewise, some will require formal bids by letting specifications out for tender, others will not; some will need clearance with outside bodies such as insurance companies (and classification societies), governmental authorities, the work inspectorate, the aliens office of the police, others will not; some standing routines will need constant revision in the light of experience, others will not; some will need to be checked for evidence that people have padded their expense accounts, others will not. And so on through the relevant variations in decisions in different parts of the operation. The variety of problems in different parts of the organization requires different systems of authority, so that one needs not only one organizational plan for the whole company (a broad sketch of the main divisions of authorizing responsibilities), but also routines and procedures in each of the subparts that differentiate approved plans from information in an appropriate way for that part of the administration. It is only at a very high level of abstraction that a whole organization "has" a formal organizational structure, for in the relevant respects each significantly different kind of activity has its own formal organizational structure. An example of an authority system to be used only once, adapted to its particular task, is the system for startup of the process train in an oil processing plant. After each of the parts of this system is "commissioned" (tested and inspected as far as possible as an independent entity), the system needs to be connected together and then tested and inspected as a

93 whole (especially for leaks) before letting gas and oil into it. This testing will never be done again. But the pressure test mainly involved here (in 1986) is still not a perfect simulation of operations. It is done with helium rather than gas and oil; connections to the completed well and control of that well are not used; heating and pressure control of the oil-gas mixture is not carried out so as to reduce the gas content of the oil to what will remain dissolved in the oil at atmospheric pressure and environmental temperatures; the resulting gas is not compressed, dried, and shipped out; and so on. A lot of things can still go wrong on the first startup of the whole process. Thus, one needs to set up a series of stages, each with a special set of safety provisions, each with a measurement of results, each with a way to stop and go back if something is wrong so that faults can be repaired, each with an appropriate approval apparatus to decide to go ahead to the next step. That is, the formal organization needed for startup is different from the one needed for building and commissioning the separate parts beforehand, as well as from the one needed for operating the system after it has been started. There will be many more engineers and managers per operator for the startup, many more special tests of how things are going, many more safety clearances, much more bureaucracy before going on. (See Gouldner 1954, 67–72, for an account of intensified supervision with the startup of a new gypsum board—producing machine system.) The information that goes into this special system is much the same as that which goes into production operations; consequently, much of it must be extracted and reorganized from the information in the project design, just as an operations manual must be. What is different is the special structure of approvals that governs this part of the operations. Running the routine production organization with this many delays for special tests and approvals would, of course, be an economic disaster. The structure is entirely inappropriate for either building or operating the installation, though many of the people and most of the information used in the two structures are the same. The approvals system for commissioning is exactly—and only—appropriate for startup.

Summary of Dimensions That Differentiate Operating Information Systems The following list summarizes the seven dimensions of operating characteristics of information systems we have described.

94 Dimension

Examples

1. Quality of information improves dramatically with operational experience or information remains of same value.

Preventive maintenance plans, operations manuals, versus capital values of installations.

2. Information generated in operations is only useful for further operations of the same kind or operating information enters other systems.

Well reports, operator experience with particular machines (helicopters, diving equipment, process equipment), versus project design, cost statistics on operations.

3. Supplementary innovations in the information system add to the system rapidly during operations or information systems are stable.

Software versus invoice processing for payment.

4. Flexible, detailed, shifting categories are needed for cost reduction projects or constraint is exercised by categories of systems outside organization.

Cost reduction use of cost statistics versus financial and cost comparison accounting statistics.

5. Each figure is auditable by a "trial of paper" or figures are useful for decision purposes only.

Financial accounts versus cost accounts.

6. Information is generated to guide one kind of activity crucial for guiding other activities, requiring reorganization, or information is in a "closed loop" within a department.

Project design for construction useful for spare parts buying, maintenance plans, operating manuals, versus well reports, helicopter pilot experience.

7. Approvals systems for decisions are adapted to differing information and problems or authority is applied uniformly across cases.

"Audits" versus "policing"; budgeting versus ledgers; procedures for various special situations (e.g., startup) versus organization manual for production organization.

Each concrete information-processing system will, of course, have a value on each of the seven dimensions. For example, a spare parts information system is distinctive in having to be created mainly from project design information, an information system that responds primarily to different uncertainties. But corruption in parts buying is a risk; fraud, while not as routine as it is in entertainment expenses, may involve more

95 money when it happens. There is therefore a careful system of approvals in the purchasing sector of the spare parts system, nowhere near as elaborate as the tax regulations for financial accounting, but there nevertheless. In the above description I have chosen the most distinctive aspects of manufacturing information systems, having often found them out because they create problems for other people in the organization. The turning inward of the drillers, the rigidity of financial accounts from a cost reduction point of view, the lack of enthusiasm of the construction project for producing adequate documentation for operations, were all frequently mentioned as practical problems. The fact that one has to prevent corruption in spare parts buying is a routine and unproblematic aspect of the system (that is, of Statoil in 1986—this is an idiographic fact, not a law of organizational life).

Conclusion Cross-sectional studies of occupational composition of organizations always show that the number of clerical workers is very well predicted by the number of technical and of sales workers. The computer programs for routine processing of organizational data are generally under the control of middle and junior managers in charge of this or that technical or sales function. The reason large human and electronic information-processing systems are connected to technical and marketing functions is that both of these require a great many decisions, each one of which is minor, but in the aggregate they all have to go reasonably well for the business to go well. The thousands of decisions that have to be made in a spare parts system under a technical department about buying and inventory, or in a marketing department about which goods will be shipped to which clients, by which transportation system, with what insurance coverage, with what tax liabilities accruing, with what discounts on the "normal" price, with what dates of payments due, all require routine information systems that can be adapted readily to meet changing technical and market conditions. The fact that they have to be adapted means that the routines are not, in fact, entirely routine. A technical or a marketing manager is primarily responsible for seeing that the routines are set up to respond quickly enough to changing conditions, by entering new information in the system, by basing decisions on that information in different (but still, for some period of time, routine) ways, and the like. While the discounts, shipping routes, insurance levels, and so on all have to be authoritatively resolved, there is no par-

96 ticular difficulty in getting people to consent to that authority within the organization—though there may be difficulties in getting them all to know what to do under various contingencies (to control errors, respond to complaints, and the like). But these authoritative systems are not, in any strong sense, hierarchical. Not everybody can say what discounts are available, and not everybody can decide which discount should go to a particular buyer, but the great majority of the people who have to do something about a particular discount, once it is decided, have no inclination to rebel against the decision. The problem instead is to see that once the discount is decided on, everybody who needs to know about the decision to settle other things (such as the size of the invoice, of the insurance cover, etc.) in fact gets it right. There has been a great proliferation of these intermediate-level information systems in recent years. Organizations employ engineering and business school graduates to set up such systems, to adjust them to changing circumstances, to see that people follow the procedures with appropriate levels of discretion, and the like. In sociology the basic notion of what managers do is derived from a vague memory of Taylor figuring out first how to embody the skill of manual workers in a routine and then how to deskill those workers. That is, we have been interested in managers primarily as "people drivers" (cf. Stinchcombe 1983, 114–115, for an analysis of where people drivers are actually found in manufacturing). But many modern managers are not concerned with improving the productivity of manual workers, except indirectly. The productivity of many workers is determined mainly by whether the machine they run is working, and sometimes by whether neighboring machines are working. This is especially true in continuous process plants, but also in such manufacturing operations as rolling steel (Stinchcombe 1974, 11–31). The quality of the information in the maintenance system—which determines, for example, whether or not the repair part for a machine that has just broken down is in stock—is likely to be a more important determinant of productivity than how hard either the operators or the maintenance workers work. The purpose of this chapter has been to give a portrait of what else managers in manufacturing operations do besides driving people. The reason there is so much to that portrait is that the central uncertainties in productivity in a modern factory often are not whether the workers will work hard enough, but whether they will have the appropriate spare parts, or whether the cost accounts are flexible enough to focus attention on those parts of the line that should be improved first because they are bottlenecks in production. These determinants of productivity cannot in

97 general be manipulated by driving workers harder. Instead they have to be manipulated by creating an information system that will routinely produce the right administrative details at the right times so that spare parts can be bought, or so that a decision to buy a particular machine for a particular place in the line to reduce costs can be evaluated correctly. Of course, how hard workers work also sometimes affects productivity, so middle managers do some people driving as well. The system for collecting information on worker performance and attaching rewards and punishments to different levels of performance is the subject of Chapter 7. Because these different systems respond to different sorts of uncertainty, their relations to information generated in operations varies. The information in a preventive maintenance system is largely fictional until corrected by operating experience; the information in the present version of a software program is at a temporary resting place in a continuous process of innovation, which produces a software system with ever increasing numbers of options in what reports one will get; the information in a well plan comes quite directly from well reports on the drilling experience in nearby wells. In each case the operating characteristics of the system, and consequently what a middle manager has to do to make the system run properly, depend intimately on what sorts of uncertainties it has to control and where the information about those uncertainties is available. But all the responsibilities of middle managers in maintaining these various kinds of information systems generate clerical labor, computer time, routings for paper flows, data files, files containing catalogs of files, and all the other features of modern manufacturing bureaucracies peopled by engineers, programmers, and business school graduates. The theoretical point of these examples is that the social structure connected with different manufacturing information systems, the technical skills embedded in them, and the problems of keeping information separate from approvals are all determined ultimately by what sorts of technical or economic uncertainty they are designed to deal with. The tendency of drillers to form a separate subculture in the organization and to be somewhat contemptuous of the information requirements of the rest of the organization is a product of an information system that responds to the central technical uncertainties of drilling. The constant tension between the demands of the cost accountants working in particular operating departments to produce more, and shifting, detail in the system and the demands of the financial accounting people for auditable and entirely standard figures reflects the different uncertainties of a cost reduction use of financial statistics and a taxation or profit measurement use of those statistics.

98 The conflicts between and within departments that arise from differences in what information systems are for and what uncertainties they respond to produce a "politics" of information systems. Information systems in departments are not merely organizational reflections of what objective reality is like; instead they are the outcome of much pulling and hauling about what the information systems should contain. The pulling and hauling may, of course, arise in part because this is one of many ways to get the responsibilities of one's department regarded as more important to the organization. But it arises also because a driller really has to pay attention to those who have the information he needs to drill a hole efficiently and safely—other drillers who have drilled nearby and specialized subcontractors who service drillers because they in turn have an expert and internally oriented subculture about, say, drilling mud. The general theoretical point is that different parts of the information system of manufacturing are oriented to different types of technical uncertainty, ranging from the concern of software writers about what the next innovation should be to the repetitive buying of spare parts that have to be here on time; from the once-only system for safety approvals before the next stage of commissioning can begin to the repetitively auditable financial accounts for tax authorities and owners. This means that large parts of the information system that are responsive to one set of uncertainties are unresponsive to uncertainties that concern others. This is most obvious when divergent needs are built into the same information system in the same department, as when cost comparison functions, cost reduction functions, and ledger and financial functions are all under the accounting department and have contradictory "needs," different requirements for flexibility versus comparability versus auditability. To see why such a manufacturing information system, with all its ill-fitting parts, needs to be coordinated with systems outside manufacturing, let us suppose that a manufacturing enterprise with a system something like the one just described has to decrease production because of a downturn in the market. Adjustments will have to be made in the spare parts inventory system to take account of fewer hours of operation causing less wear and cut parts inventory costs to correspond to the new requirements, while different adjustments will have to be made in the cost comparison part of cost accounting to spread overhead costs (downsizing usually does not lower overhead costs as fast as production costs). Thus, the authority to demand adjustments in all the different parts of manufacturing information processing has to be located in a place that is highly responsive to shifting market demand, to sensible predictions of how long the market slump will last, and the like. It is precisely because

99 the manufacturing information system is a ramshackle affair doing different jobs with the same basic informational materials that authorities over manufacturing who are responsive to the market need the authority to intervene throughout the manufacturing system. The creation of such authority is the core problem that Chandler addresses in his classic Strategy and Structure (1962), which we take up in the next chapter.

100

4— Market Uncertainty and Divisionalization: Alfred D. Chandler's Strategy and Structure Introduction A central theme in organizational sociology for the last couple of decades has been the growth and distribution of "decentralized" management. Broadly speaking this has divided into two branches: the "contracts" branch that says that decentralization is often achieved by autonomous firms tying themselves together for specific purposes, and the "decentralized administration" branch started by Alfred D. Chandler (1962) and Peter F. Drucker (1946). We will treat the contracts tradition more extensively in Chapter 6. In this chapter we treat Chandler's argument. Since the richest source of data on the development of the multi-divisional structure is still Chandler's own, this chapter is in effect a close reading of the theoretical aspects of Chandler's great historical monograph and of the relation of the theory to the data Chandler himself presents. Our first task is to describe as exactly as we can what the dependent variable, the thing to be explained, is. It turns out that according to Chandler the central trouble with a centralized administration was not that it was centralized, but that it made centralization impossible where it was needed. Chandler's argument, especially in the treatment of Du Pont, is that in order to centralize the response in each of several markets that required markedly different tactics, one had to decentralize the firm as a whole. The main difficulties he points to that kept people at organizational reform derive from the firms' lack of centralized response to particular markets in which they were losing money. The reason they did not respond in a centralized way to those markets was that the firm had centralized at the firm level, which implied that the various functions (e.g., manufacturing, marketing, and engineering) that needed to be coordinated to respond to a particular market were each centralized above

101 the level at which response to a particular market had to be organized. A central manufacturing executive would coordinate all manufacturing together (explosives and dyes, for example), rather than explosives manufacturing with explosives marketing and dyes manufacturing with dyes marketing; similarly for a central marketing executive and a central engineering executive. This part of the argument, and the historical analysis of Du Pont, is also the place where Chandler best shows the connection between the firm-level problem of inadequate response to the various markets the firm was in and the resulting solution of administrative reorganization. In particular he treats in considerable detail how individuals in the firm came to see that they had a problem of market response, how they came to see that reorganization of the administration was a solution, and how they stuck to administrative reform until they came up with the solution: centralized administration in each market, decentralized administration in the firm as a whole. Consequently, it is in this analysis that we can best see how functional problems in large firms come to be posed as problems in the work life of executives, the problem that has come to be called "methodological individualism." Briefly, methodological individualism is the principle that explanations at the social level are not adequate unless the individual actions that are required for the social explanation to be true can themselves be explained adequately. This idea has been treated in the literature primarily as a question of the structure of theory. We will show in the first section below that Chandler solved the problem of methodological individualism for this part of his argument without having a theory of the general individual rationality kind that methodological individualists generally demand. The argument for the functionality of the general office, detached from operating responsibilities and responsible for investment policy, is made most intensively in Chandler's chapter 3, in connection with Sloan's reorganization of General Motors. The basic idea is that only if the firm has better knowledge than the stock market does of the market projections and cost picture of each of its divisions can it do better than a completely decentralized, holding company structure. What is lacking in Chandler here, first, is a demonstration that such general offices do in fact do better than the stock market as a whole in predicting the future value of investments in their "subfirms." Are the profit rates of divisionalized firms with a general office in fact greater than those of holding companies with investments in various firms, each of which must respond separately to stockholder interests in profits?

102 Second, we need to be shown that this low predictability of the rates of return on investments in the subparts of firms without a general office was posed as a set of problems to the executives, who then set up the general office working with abstracted projections of demand, costs, and capital improvement proposals, and that those executives had the power to institute their proposals. In the second part of the chapter we will try to show that Chandler's analysis of General Motors almost satisfies those requirements. General Motors in fact expanded a great deal and remained the most profitable firm in the industry for many years after its reorganization into a multidivisional structure with a general office. The individual-level analysis required turns out to be, in large measure, a biography of Sloan himself, based largely on Sloan's own accounts. It is clear that Sloan had an adequate theory of his own behavior. We will have shown in these first two analyses that the explanation of why a company will invent divisions is quite different from the explanation of why a company with divisions will invent a general office. The divisions are to facilitate different responses in different commodity markets; the general office is to provide information to satisfy bankers and other investors, to provide them with better guarantees of continued profitability than they would have from a holding company without such a general office. Thus Chandler himself has two different dependent variables here, which have two different explanations. In the third section we treat Chandler's analysis of Sears. This is a particularly useful case to look at in connection with the problem posed for General Motors. Sears in fact competed with a structure of wholesalers and retailers who were organized as autonomous firms, with no centralized office and no centralized buying. We can pose the problem Sears had when it tried to use a mail order firm's buying structure to supply an appropriate inventory to a dispersed set of department stores in automobile-oriented shopping centers, then, as one of building a bureaucratic structure that could do what a market-organized structure of wholesalers and retailers was already doing very well. This is not a functional problem peculiar to Sears, because during the period Chandler analyzes here chain stores with multiple commodity lines were coming to prominence in the novelty trade (the "dime stores"), groceries ("supermarkets"), auto accessories, and eventually hardware and furniture, as well as in department store retailing. Our analysis of Chandler's study of Sears, then, involves a systematic treatment of what functions the structure of wholesalers in various commodity lines nationwide in fact served and still serves, which parts of those functions were already served by the extremely decentralized struc-

103 ture of buyers in the nineteen or so commodity lines that characterized the Sears mail order regime before the reorganization Chandler discusses (described at that time as "a coalition of merchants"), and where in the new Sears structure the remaining, unfulfilled, functions were located. Our argument here is that the decentralization into commodity lines, already present in the old regime, remained the central functional structure in the new regime as well. It was apparently the addition of detail to the inventory control system in commodity lines, rather than, as Chandler argues, decentralization into autonomous regional divisions, that made Sears (and other chain stores) able to do the jobs that specialized regional autonomous wholesalers were already doing. We argue that Sears had to begin with an organization in many different commodity-line markets, that had already created "divisions"—the autonomous buyers for the mail order business— to respond to those markets. This divisionalized organization already had a general office, the one that decided to invest in department stores rather than expand the mail order business. In short, it started with the very structure to be explained, and had that structure for the reasons that the theory says it should have had such a structure. It entered very few new commodity lines. So what it needed was regional offices as branches of the general office to decide on investments in stores, and it also needed new modifications of its commodity-line information systems to feed store inventory and sales data into the offices of the commodity-line divisions run by buyers. We argue, in short, Chandler in fact has little to explain—and mistakes what there is to explain—because he believes the regional divisions of Sears are "operating" divisions. He thus has not interpreted his theory of the first two cases correctly in analyzing Sears. In the final part of the chapter we try to unpack the empirical content of Chandler's main independent variable, "being in several different markets." The overall functional argument in Chandler's book is that a multidivisional administrative structure is required if a firm is in several different markets. We provide a detailed definition of a distinct market as market uncertainty ramifying into the manufacturing and engineering of a commodity line in a way different for other commodity lines. In Chandler's chapter 7 he analyzes a large number of manufacturing industries in terms of his overall functional theory. If Chandler's book had been a sociological article, say in Administrative Science Quarterly , chapter 7 would be the article, and a page or two about the histories of Du Pont, General Motors, Jersey Standard, and Sears might have been in the

104 original draft as motivation, to be cut by the editors as not science but anecdote. But I believe it is more useful to treat the chapter as a careful analysis of what the phenomenon, as perceived by the corporation, of "being in several diverse markets" looks like as an "objective" variable spanning a number of industries. We will treat the variable of being in several markets as a matter of the social organization of uncertainty, of the organization of information necessary to adapt to that uncertainty, and of the amount of uncertainty. This will constitute the explication of the implicit analysis in Chandler's chapter 7 of the conditions under which market uncertainty ramifies into the manufacturing and engineering of a commodity line, so forcing marketing, manufacturing, and engineering together into a division. We will find that the most successful parts of the case studies, as well as the crosssectional analysis of industries in chapter 7, are neatly summarized by five variables describing how uncertainty of demand is organized in different sets of commodity lines.

The Concepts of Centralization and Decentralization The first big conceptual trouble in understanding Chandler's argument at the structural level is that "centralization" and "decentralization" as Chandler uses them are really two different kinds of centralization: (1) centralization for administering one product (or a small group of related products) for one market, with market control (losses and profits) over the central authority; and (2) centralization for administering one product for one market in a division of the company , with both market and administrative controls over that division (the administrative controls are by a central office that also controls over divisions). To see what is going on, let us start with the Du Pont firm making explosives or U.S. Steel making rails or Ford Motor Company making Model T's. In each firm there was a marketing or sales department—Chandler tells about the development of the marketing department in Du Pont, with explosives experts advising clients (for example, mines or tunneling projects) on efficient and safe blasting. In each firm there was some sort of engineering or chemistry department (or both). And of course there was a manufacturing department. For any one product, the firm needed the engineering force to design the product to fit the tastes of the market and to design the production line so it could make the product cheaply, yet so production could be adjusted to variations in demand; the engineering had to be adapted to

105 both marketing and manufacturing. The firm needed to arrange for feedback from marketing to manufacturing so that manufacturing would not make too much and get committed to an enormous inventory, or would not have delays in the delivery that could kill the market, or so manufacturing could produce the right mix of sizes and colors. And obviously the firm needed engineering to design a line that manufacturing could in fact keep going, work on bottlenecks in the line, or redesign the product so as, say, to eliminate a weld by stamping the piece differently or otherwise contribute to cheap or flexible production. A big problem of all industrial administration is to get each of these three functional departments to work together, to do what the others need instead of what they are inclined to do. Manufacturing can always save money by making long runs at a steady speed, with inventory taking up the fluctuations in demand. But if it turns out that a market fluctuation is not just a fluctuation, if clients are really stopping building railroads with the firm's steel or if folks are really poor in the depression and cannot buy a Model A, then the firm is stuck with the inventory. Then (as Ford did) the firm could force dealers to take the inventory and pay for it, and the dealers (the smart ones) would switch to General Motors so they would not go bankrupt—and Ford would have screwed up its marketing. Marketing wants each order treated as a special case: stop production and make us this thing for this good customer. But when that happens, manufacturing gets disorganized, costs go out of sight, and the firm cannot sell (and it is "manufacturing's fault"). That is, there is a true dilemma between the requirements of cheap manufacturing and those of effective marketing, a dilemma someone has to solve in the interest of overall profits on that line of goods. Some things that sell cars better make manufacturing more expensive; some things that make manufacturing cheap really ruin the market. Since engineering (e.g., design for ease of maintenance versus design for engineering efficiency) has to depend on what the market wants, on how much the dealers complain about the costs of warranty service, and so on, engineering has to be coordinated with marketing. Similarly, the tradeoff between "getting it out the door" and making it "state of the art" determines engineering delays, and those delays are reflected in manufacturing costs. There were, then, in such centralized firms true conflicts among departments, because there were true technical and economic tradeoffs between what one department and another were supposed to maximize. And this meant that somebody (or some group—but in fact, Chandler says, a firm needs somebody) had to tell engineering to do what man-

106 ufacturing said instead of what they wanted to do, to tell manufacturing not to treat each customer as a special case but to control their production by an inventory depletion time criterion, and so on. Note that the right decision about which department ought to give way on what questions depended on the product. Eventually Du Pont would sell enough blasting powder to use up a shipload of nitrate, but they might not sell enough of that awful artificial leather Chandler talks about to use up a shipload of wood pulp. The decisions Du Pont needed to make for a standard manufacturing ingredient like blasting powder and an ingredient of novelty goods like artificial leather were very different. There was a different combination of the risk of being stuck with an inventory the company could not sell (low with blasting powder, high with artificial leather), manufacturing cheapness (very important for blasting powder, not so important—as long as it is cheaper than leather —for artificial leather), marketing flexibility (not important for blasting powder, of the essence for artificial leather), and so on. That is, the contradictions between marketing, manufacturing, and engineering had to be resolved in different ways to make money on different products. The centralized firm, the structure with which Du Pont started after Pierre reorganized it to look like a steel plant, puts one man with a staff of accountants, marketing specialists, production line managers, and so forth in charge of all three departments. He (it is always a he in Chandler's book, so this "he" is a historical fact) needed expert engineering managers, sales managers, and manufacturing executives under him, because it was hard enough to do engineering, marketing, and manufacturing in the first place. But he also needed to be able to order them to give up their own professional standards in the interest of one of the others. And he had to do it in the light of whether his firm was selling blasting powder or artificial leather, Model A's or Cadillacs, steel rails or "special shapes." Centralized administration is represented by diagram A in Fig. 3; decentralized by diagram B. The decentralized firm looked exactly like the centralized one, except the thing in the decentralized firm that looks like the centralized firm is the division—and there was, in addition to the market controls over the division, also administrative control by the general office. The reasons for the similar organization for administering a single product were the same. For a given product, the engineering and the manufacturing boss had to be forced to resolve their different priorities in the light of how one really made money on that product , and the same held for marketing and the other two. Forcing departments to resolve their conflicts in a way appropriate to the product requires autono-

107

Fig. 3. Comparison of the Organizational Form of Centralized and Decentralized Firms in Chandler mous operating divisions. Before we use Chandler's materials to develop our argument, it will be useful to consider a methodological issue: the role of individuals in organization-level functional arguments.

A Definition of Methodological Individualism By "methodological individualism" we mean that every individual action that enters into some collective pattern must itself have an adequate explanation. The collective pattern may be, for example, that corporations that enter multiple markets with markedly different contracting patterns,

108 spans of temporal adjustment, quality or service requirements, and so forth will tend to develop differentiated divisions to respond separately to those markets, and a higher-order integration mechanism (a "general office") to coordinate this divisional structure, because such a decentralized divisional structure is more efficient in responding to multiple markets. This is a functional connection at the level of the corporation: corporations in multiple markets "need" separate responses to the separate markets, and a decentralized divisional structure "fills that need." This is a way of stating the central argument of Chandler's Strategy and Structure . The principle of methodological individualism, specified to this functional argument, says (1) that the actions of individuals in corporations that create the new structure should be adequately explained, in the sense that they should ultimately be shown to be a response to multiple markets; (2) that those actions should continue and aggregate in a way that ultimately creates the multidivisional structure; and (3) that such individual actions arise because the separate responsiveness to markets was "needed." It has frequently been alleged that many functional explanations do not—and, in general, cannot—satisfy the requirements of methodological individualism (e.g., Homans and Schneider 1955; and Elster, e.g. 1983, 55–68). But obviously, if the explanations are true they must satisfy those requirements, since all social action is carried out by individual human beings. Often the plea for methodological individualism is coupled with a preference for rational actor explanations: it is a concealed plea for utilitarianism. There is, however, no inherent reason why the model of individual action in a methodologically individualist account must be a rational one. For example, John Padgett has offered an account of how the Department of Housing and Urban Development (HUD) budget-making process "serves the function" of adjusting the departmental budget to the macroeconomic objective of the president's office, as transmitted through the Office of Management and Budget (OMB) (Padgett 1981, 85–92). The models of individuals in his treatment are specifically designed to take account of limitations on individual rationality, and have many random and nonoptimizing elements in them. The central point, though, is that the requirements set by the president's office enter as variables and constraints in the nonoptimal problem solving of the bureaucrats in HUD. Such models of a collective process satisfy the requirement of methodological individualism but are the opposite of any appeal to utilitarianism or to rational actor models. Chandler's book is an especially interesting case in this regard, because he has followed some part of the strategy of deriving predictions

109 from the collective-level theory and documenting them. His chapter 7 (1962, 324–382) tries to show empirically that in industries that function in single markets the multidivisional structure has not been adopted, while in those that function in multiple markets it has been. But Chandler thought it necessary for his purposes to write some three hundred—odd pages of history, mainly of four firms (Du Pont, General Motors, Jersey Standard, and Sears), to see how and why the multidivisional structure was invented and adopted. As is typical for historians, many proper names of individuals appear in the accounts. Clearly Chandler does not believe the functional argument at the collective level is adequate with only collective-level evidence, for he includes additional evidence involving a large number of named individuals as well. If these people were optimizing, they took a long time about it and lost a lot of money in the meantime, and still disagreed about what was optimal when they adopted the new structure. That is, the Chandler book looks like a case of a functional argument with individual-level explanations, which is not the same as a simple rational actor model at the individual level. We will next examine Chandler's explanation of the origin of the multidivisional structure in Du Pont, with some side glances to his own theoretical statements and to the more formal models in Padgett for guidance on the logical and theoretical role that statements about named individuals in organizations can have (Chandler 1962, 52–113 on Du Pont; and Padgett 1980, 1981).

Individuals in Du Pont: Organizing Information Flows In Chandler's discussion of the old-regime centralized administration at Du Pont, he emphasizes the systematic effort to create reliable flows of information, especially by Pierre du Pont. Pierre du Pont thought of the legal and financial changes that incorporated previously independent businesses into E. I. du Pont de Nemours Powder Co. as "only necessary preliminaries" to reorganizing the administration of all of them (44; all page numbers cited in the following with no further specification are to Chandler 1962). In that reorganization a central point was "to establish a system of costs in order that an economical manufacture could be installed throughout the business." Cost accounting introduced first and foremost in manufacturing (55, 58). Once sales were taken from jobbers and agents and done by salaried technical sales people, cost statistics were also created in sales, as were market share statistics (59–60). The treasurer's department, headed by Pierre du Pont, a member of the top management, came to develop a system of cost, performance, and sales

110 statistics and projections (60, 66, 101–102). Since Pierre du Pont later became the power in the board of the company, the place of cost and performance statistics in defining the realities that confronted the company was secure. At several points in Chandler's narrative, locating the difficulties facing the company depended exactly on the flow of differentiated cost, profit, and other performance measures (86, 92, 95, 100, 104). That is, a flow of differentiated statistics determined, in a general way, where people thought problems lay at various points throughout the administrative evolution. In one case, the fact that the statistics were inadequate to prevent problems of overstocking of supplies or of manufactured products (which were partly detected and diagnosed by statistics that came in later) was a crucial argument for reorganization (102). Thus the environment in which administrative discussion went forward was shaped by information-collection technologies that could, for example, show that the problems after diversification were concentrated in lines where package goods were sold directly to the ultimate consumer, rather than in lines where tonnage goods were sold for further processing (92, 95, 100). In short, the problems people thought they had, and the fact that those problems persisted even after the company had reorganized to solve them (104), were features of an organizational intelligence system created largely by Pierre du Pont, and the results of that system were sufficiently persuasive that a negative decision on reorganization by the president, Pierre's brother Irénée, after Pierre had gone to General Motors, was overturned by the data (99–100 for the rejection in 1920; 110–113 for its acceptance in 1921), despite Irénée's continued skepticism. The flow of detailed performance measurement information thus shaped the perception of the problem by individuals. In particular, it connected a systematic pattern of failure to individual perceptions in top management of that failure in a way that kept management looking for a solution even after rejecting the one they finally adopted (they obviously cannot have been optimizing in both 1920, when they rejected decentralization, and 1921, when they adopted it). The statistics to facilitate centralized administration thus ultimately helped cause decentralized administration. They also then became the basis of the general office's capacity to assess the work of the divisions they helped to create.

Individuals in Du Pont: Organizational Theory In order for a problem, perceived through an organizational data collection system, to induce an organizational solution, it must also be ana-

111 lyzed by an organizational theory. The standard theory at the time of the innovations at Du Pont was one that ordinarily gave centralization and specialization as its answer (100 for Irénée's statement of these principles). But at least the theory's concrete manifestation at Du Pont was one that could diagnose a pattern in the data (e.g., the fact that other companies were making money in lines where Du Pont was losing money: 95) as requiring an organizational solution. The most general guide to division of responsibility that could give rise to operational divisions specializing in a product line as stated by Harry Haskell (then vice president in charge of the high explosives department, but detached to head a subcommittee on organization): "The most efficient results are obtained at the least expense when we coordinate related effort and segregate unrelated effort" (69). He had already used that notion as an argument "to carry on the dye business as a separate entity. I think it would [be better] because it is a developing, unstandardized industry and should merit independent attention just as the Parlin chemical mixtures business was better by itself until standardized—when it was merged with the regular sales and operating departments" (68). The dye business needed "one individual in control of both production and sales, because the relation of the product and its qualities is so mixed up with the demands of the market for the product that to divorce them . . . would be detrimental to the business" (70). This analysis meant that later evidence of interdependence of marketing, manufacturing, and raw materials purchasing (as shown, for example, in projections by manufacturing that were uncoordinated with marketing experience, resulting in inventory losses on raw materials or on manufactured goods in "merchandise goods" lines) was an argument for integration of those three functions. Further, it meant that the sales manager, Frederick W. Pickard, could explain the problems with the sale of package goods to the ultimate consumer by saying that he had products with "no logical sales connection with one another" (101). In analyzing these sales problems further, a subcommittee (which included Pickard) of the executive committee pointed out that (successful) competitors had one person in charge of both manufacturing and sales of each line and responsible for the profits of that line (96). Further interdependence provided a theoretical category that made the experience with "minidivisional" management of several new products (dyes, artificial silk) relevant to these problems of interdependence of sales and manufacturing on larger lines of products (96). As is usual when evidence requiring a "paradigm shift" accumulates, the facts were at first explained away as a series of individual mistakes in a system of functional departments and centralized administration

112 that had functioned brilliantly in producing munitions (96–100) and was still making much money on tonnage goods (104). In particular, "[we] have carried excessive stock . . . [have made] several unfortunate guesses on the purchase of raw materials . . . have been working against orders rather than setting up appropriate stocks. . . . Rearrangements, repairs, renewals, and replacements of plants have been high. . . . [There have been] miscellaneous shortcomings in factory operations such as poor routing, and inefficient piece-work, pay schedules, short-runs, etc." (98). But the theoretical category of interdependence, and the proposed general solution in the theory that problems arising out of interdependence should be put under a single authority, maintained cognitive focus on the difficulty with the reigning centralized organizational paradigm. The category existed in the organizational theory of the old regime, and this made it possible to codify the experience to which it was relevant, so both the category and the experience would be available to diagnose the new problems in multiple markets.

Individuals in Du Pont: Responsibility for Inventing and Adopting a Remedy The process outlined above provided the environment in which individuals could see functional need, which Chandler identifies at the collective level, as a problem whose answer was probably organizational. But individuals become aware of thousands of problems over the course of a couple of years (say, 1918–1920 in the Du Pont company). These problems have to become the responsibility of someone, of some committee, or of some routine process. Committees were central in Du Pont's higher management. By the time Chandler began his detailed story, the Du Pont company had an overlapping set of problem-allocation mechanisms, by status and by committee jurisdictions. The status distinction was that each department had two chief executives, the higher of whom (a "vice president") was responsible for long-range policy matters, the lower (a "director") in charge of day-to-day or operational matters (57). There were both standing committees and task force committees (usually created as subcommittees of a standing committee). The committees had a shifting membership whose most active members were usually the vice presidents responsible for long-term planning. The main work of writing proposals for reorganization, writing critiques of those proposals, and passing or

113 vetoing the final proposals was done by committees, though often one member of the committee was fairly clearly the author. The mechanisms for creating committees, determining the charge to the committee (or the jurisdiction of standing committees), and specifying the membership are not clear in Chandler, probably because they were not clear to the people in Du Pont. There seems to have been more "constitutional" thought given to the composition and jurisdiction of standing committees with some decision-making authority, and much more ad hoc arrangements to give a charge and a range of authority to more specialized and temporary committees (58, 65, 93–96, 98–99, 103, 105, 113). Decision making was also apparently much influenced by rank; when Irénée du Pont, the company president, was being overruled on the issue of divisional organization, the executive committee was expanded to include the financial committee as well as Pierre du Pont and Raskob (who came back from General Motors to attend). In deference to the opinion of his colleagues, Irénée du Pont did finally vote for reorganization (110). Two things are central to explaining how individuals got responsibility for devising solutions to the problems identified by the data flow and the existing organizational theory. The first is that the structure allowed for cutting the flow of "problems of subordinates" to the high executives, so that the latter could be detached from day-to-day problems to study, and restudy, problems of organization (303–304; see also Stinchcombe 1974, 24–25, 81–93). An executive directly in charge of a group of subordinates has to divert his attention to their problems when they need it, and so is not a deployable resource for analyzing long-run problems. The second factor is the existence of a flexible structure to tie together responsibility for solving problems and authority to implement the solution at the top levels of the organization. The top executives have to be not only free to study reorganization but also authorized to do so and permitted to put the resulting proposals on the agenda of a decision-making committee . The right to have one's proposals put on the agenda is the core of what we mean by "authorizing" a committee to study a problem, though of course the authorized committee also gets access to the intelligence apparatus of the top executives (58). Although we do not get the details from Chandler of how this worked in Du Pont (the structure at the top seems to resemble that described by Cohen, March, and Olsen 1972), responsibility for the analysis clearly ended up being pinned on particular individuals (sometimes the individuals were drafting a committee consensus), and the proposals then usually went onto a decision-making committee agenda. (For a tempo-

114 rary hiatus in the second, see 99, where the first recommendation of a multidivisional structure was deleted from a subsubcommittee report by the subcommittee; this hiatus was only overcome in the face of a company financial crisis.) What we have, then, is not a description of the structure that authorized individuals to propose solutions to problems, but evidence that such a structure effectively existed in the top management at Du Pont—a sort of incomplete methodological individualism. It is incomplete because the collective-level solution is located in the behavior of specific individuals operating in a given structure, with their individual actions in fact linked in such a way as to create the collective-level pattern. We are not told how they came to be so linked, or why those individuals rather than others were authorized to act. The submission by minorities to the decisions of the majority of the (expanded) executive committee is also not discussed in Chandler, though the lucky fact that one of the minority was the president, Irénée du Pont, gives us some evidence on the question. Aside from the uninformative phrase "in deference to the strong opinion of the majority of his colleagues," we are also told that he was appointed to the new executive committee, whose structure he had objected to. In addition, we are offered the speculation (in the final report to the board of directors of the corporation) that, after having experienced the operational running of market segment–oriented divisions by divisional councils (one member from sales, one from manufacturing, one from purchasing), managerial morale will improve by fixing responsibilities—which may be an explanation for minority acceptance. The failure to explain the submission of minorities is not unique to Chandler; the explanation of the acceptance of decisions made by authorized bodies is a weak part of social science generally.

Individuals in Du Pont and HUD: How Decentralization Works What we mean by decentralization, presumably, is that the controls of a superior level of the organization over an inferior level are highly abstract. The performance measures that the general office gets about the inferior level in a decentralized organization carry few details that would enable them to penetrate the veil shielding the division manager from close inspection of his or her subordinates. If the general office knows too much about the performance of subordinates of the divisional manager, the knowledge that the general office is ultimately in command of

115 their careers will lead those subordinates to try to make a good impression on the higher management rather than on the divisional manager. The goals that the general office specifies for the division must thus be somewhat abstract, so as to leave discretion with the division manager. Padgett describes the supervision of OMB by presidential-level bureaucratic politics as follows:

Presidential decision making about fiscal policy and defense may be exceedingly complicated and convoluted. However, from the point of view of OMB, all that is produced from this within-level conflict is a single number—either of the form "your overall expenditure target is 500 billion" or of the form "cut 2 billion." The very top echelon of OMB may have some idea where this number came from, but whether they do or not is largely irrelevant to their task. Politics has been compressed into a single piece of information. (Padgett 1981, 80)

Similarly, Chandler points to the general office working mainly on "estimates and forecasts of anticipated conditions rather than on past or even current performance" (292). Clearly they cannot inspect many details of subordinate-unit behavior at any considerable distance into the future. They have to focus on one or a few numbers measuring performance (e.g., costs, market share, profits) and to supervise with a few numbers measuring constraints or goals (especially capital allocations and the description of which measures of performance will be used). Thus the key feature of decentralization is the degree of abstractness of the information flow between the divisions and the general office. When that flow is, in Padgett's words, "compressed into a single piece of information," decentralization is maximal. It is the responsibility of the divisional manager to combine that one piece (or a few pieces) of information with the details about the world brought in by his or her subordinates and by the information-collection system of his or her division. Padgett has shown that simple models of individual behavior in budget making at each hierarchical level are adequate to explain, on the one hand, the statistical dispersion of cuts across programs within HUD and, on the other, how the cuts come to add up to the totals specified by OMB, and therefore (more hypothetically) how OMB itself allocates cuts across cabinet departments so that they add up to presidential fiscal and macroeconomic objectives (Padgett 1980, 1981). By analogy, we might construct a model of problem solving within di -

116 visions of Du Pont which could coordinate flows of information about purchasing, manufacturing, and sales of a particular product line from mines in Chile to the ultimate consumer. We could incorporate in this model investment and cost constraints imposed by the general office and specify the performance measures that office will use as objectives for the divisional manager to try to maximize. For our model to be workable, flows of information about what the problem is and flows of control communication about what the solution has to look like must be correctly segregated by degree of abstractness, so that each level has the right information to solve its level of problems. For example, in the abstraction process that turns divisional information into a flow to the general office, much information relevant to measuring subordinate performance within the division must be deleted, but information relevant to making market size, price, and cost projections must be retained. The rather ordinary men at the general office deal with large time spans easily because information about daily and weekly time spans has been deleted from the information flow to them and because they allocate capital authorizations over large time spans rather than approving purchase orders. The rather ordinary division managers likewise have a manageable number of subordinates, given how often they have to confer with each or to force them to confer with each other, and the directions they get from the general office are simple enough that these can be integrated with the flow of communications with subordinates. A key indication of how well decentralization is working is the proportion of all communication of top executives that occurs in committee meetings (Stinchcombe 1974, 79–84); if they are talking to individual subordinates rather than to committees (often made up of the subordinates of the committee chair), they are probably not making long run policy with sufficient attention to data abstracted to reflect long-run trends. To explain how abstraction and corresponding decentralization add up to the effective functioning in the U.S. federal government, Padgett uses the image of differentiated ecological control systems.

Each level of organizational aggregation [of lower-level budget data] is embedded in a distinctive cultural context of "ecological control" premises which reflect, in highly compressed form, the historical residues of past political struggles—[at the lower level], program controllability [i.e., the degree to which future budgets are subject to executive discretion rather than fixed], whose roots lie in the legal structure; organizational priorities [of particular government departments], whose roots lie

117

in institutional roles and constituency relations; and presidential fiscal targets, whose roots lie in macroeconomic and defense issues. Such ecological control premises change on different historical time frames. (1981, 121–122)

The point here is that through segregated levels of abstraction and the time frame of responsiveness, different types of problems become problems to different hierarchical levels of the government. The same applies in Chandler to Du Pont and General Motors. The solutions to the problems at each level go to the other levels in highly abstract form. For example, the presidential level sends out a target cut and gets back a proposal to cut certain programs by certain amounts; it need never know how controllability and constituency relations were balanced in arriving at these cuts. Similarly, Du Pont's new general office need not know how much of the capital and operating expenses of a division went into making manufacturing respond more quickly to market changes, how much into producing the product more cheaply; they need only know whether whatever the division did worked to preserve market share. But they need to know this over the course of years, not weeks, as one does if one must keep the inventory of unsold goods of many types within bounds. The statistical techniques for abstracting performance records to the appropriate level to detect problems for the general office that require for their remedy yearly shifts of investment are different from those for detecting problems with too high production of one type of paint, too low of another. The hierarchical segregation of information is essential if individuals are to be able to respond to shifts in the environment by manipulating resources under their control. This sketch of Chandler's theory of what decentralized administration does (considerably sharpened by a reading of Padgett) shows that his main functional connection can also be disaggregated into individual behavior. Through decentralization from the central office and recentralization at the divisional level , authority to respond on different time scales is coordinated with information adequate to detect problems on those time scales, so each person's job is doable and yet the organization responds correctly enough at all time scales. The sketch also shows that individual rationality depends on appropriate decentralization. Only if a general office executive gets information appropriately abstracted for making projections, but not for promoting a division chief's subordinates, can he or she respond strategically in making investment decisions.

118

Is It Still Sociology? The historical chapters of Chandler's book are convincing partly because they disaggregate the functional connection to the individual level. But they are also convincing because they show that even if we do not always understand exactly why the individuals acted the way they did, they in fact acted in such a way as to create a multidivisional structure. So far his discussion could be either psychology or history, but nothing at the collective level, the level of the corporation, shows that the connections are social. Chandler's chapter 7 furnishes the evidence, in a form familiar to sociologists, that the histories are not merely unique paths leading from wherever it was to wherever the company happened to get to, as illuminated by what the individuals said they were trying to do. Chapter 7 shows that there is a strong correlation between being in a single market and having a centralized structure and between being in multiple differentiated markets and having centralized divisions—but divisions decentralized from the central office. A sociologist would not ordinarily be satisfied with the four administrative histories in the first part of the book because there are more variables and contingencies than cases. Chapter 7 convinces that, almost invariably, firms that succeed after having entered many markets have a multidivisional structure, while those that stay in a single market have a centralized structure. This, then, is a functional connection that is decomposable into individual actions, yet it is clearly not the outcome of any simple aggregation of individual rationalities. If decentralization is an aggregation of individual rationalities to all, the majority of top Du Pont management who were against it for a year when they were losing a lot of money by not having it shows that it is a more complex rationality than we generally meet in microeconomics. If a centralized firm, without changing structure, has one central executive coordinating engineering, manufacturing, and marketing of both explosive powder and artificial leather, the executive's job will be within his or her capacities only if he or she does the job by rule. If the executive is a powder person, he or she will say to engineering: make it cheap; to manufacturing: build up inventory, there will be rock to break up forever; to marketing: teach the customer how to do it right—get in there and help them run their blasting operation. But those rules will make the firm go broke in artificial leather for novelty goods. The solution is to invent subordinate rule makers when the firm goes into many businesses: the bosses of divisions. The central management have to teach themselves not to meddle because they do not know artificial

119 leather, only powder. Creating these subordinate dispute-resolving and rule-making authorities—the division heads—and getting the general office head of engineering, manufacturing, or marketing not to interfere are the first big achievements that Chandler talks about in Du Pont.

The Causes of Divisionalization We have used our analysis of the details of Chandler's history of Du Pont to explicate his theory of the causes of divisionalization. The dependent variable in this story, then, is the centralization of the management of product lines in distinct markets . The first part of Chandler's explanatory problem is to explain why centralization at the divisional level, rather than at the firm level, is required for manufacturing in lines with distinguishable markets. We have not explained why Du Pont wanted to maintain itself as a single firm, nor does Chandler (but see Chandler's later work [1977] for an explanation). Consequently we, and Chandler, have left aside the question of why there needed to be a general office at Du Pont. We have said only that if they had one, it needed to have abstract nonoperating information fed to it by the divisions. Since Du Pont was already centralized, and since this fact did not change, we get no information from Chandler's history as to what brought about a general office, or what brought about firm unity when unity was not demanded by operational unity. To explain the multidivisional form, we need to explain the causes both of divisionalization and of a general office. But the second set of causes is better explored by looking at the creation of a general office than by looking at the creation of divisions. In Du Pont that process can be explained by what we now require the general office executives to do, now that we can no longer let them run operations. To guess what general office executives would do after decentralization, one has to realize that they were the ones who controlled investments and had to answer for their profitability. But investment is a future-oriented activity, which means that an investment center like the general office has to form a picture of the future; this, then, will tell them returns from an investment made now in powder or in artificial leather. The general true answer they ought to be giving, of course, is, "Damned if I know what the future will be like!" But they cannot run a business on that truth, because they have to put their money somewhere or they will get no profit. So what the general office does is make projections—of market demand, of Japanese competition, of costs, of competitor effi-

120 ciency. Then they calculate how these projections would influence profitability if, say, more money were to be put into artificial leather than into blasting powder, and vice versa. Then they decide where to expand, and how much. The advertisements for Lotus 1-2-3 or Visicalc give an idea of what general office executives are doing. What the firm usually gets is an "exact" calculation (Lotus 1-2-3 does not give standard errors) of expected profit with a margin of error a mile wide; then it takes a decision. On average, the decision taken is the wrong one. Apparently the general office does not make its investment decisions any better than the stock market as a whole does, for the profits of multidivisional firms are not ordinarily any bigger than the average return in the stock market. Many firms with general offices that are very expert in highly abstract projections of future demands and costs go bankrupt. Chandler would argue that if the central management were not engaged in making such projections, they would be meddling in the business of the divisions and so would do worse than the stock market as a whole. But constructing an argument about what a general office is for, when, as in the case of Du Pont, it might well be a fossil left from the days when the company was a single centralized firm, is not a satisfactory way to proceed. Let us turn, then, to General Motors and its creation of a general office.

General Motors Creates a Multidivisional Structure by Centralizing In Chandler's account of what happened to General Motors, the basic structure of William C. Durant's carriage business was just what Sabel (1982) says is the alternative mode to mass production, to "fordism," in organizing industrialism—what in Chapter 2 we called the "interactive" program structure. But Chandler does not describe how the skills of the workers in, say, the wheel maker's firm were taught and organized. By subcontracting for the parts, Durant himself (or rather, Durant's company itself—Durant was off in New York talking to bankers) managed assembly, and especially marketing, directly. That way the carriage firm that preceded General Motors could respond rapidly to shifts in fashion. In particular, the whole structure could more or less turn around and make cars instead of buggies, using more or less the same marketing structure and the same set of subcontractors, but with different technical details, different volumes, and so on. The flexible structure in Flint, Michigan, was similar to that described by Sabel for flexible manufacturing in many classical industrial cities and was sufficiently adaptable to

121 switch from routines for producing carriages to routines for producing automobiles. If all these independent craft firms were doing fine, why did Durant buy out his suppliers to form General Motors? First, he didn't always. But when he did, the reason seems to have been risk management and capital flow problems; these led to vertical integration—to assemblers buying out suppliers or marketing firms. Small business owners could not afford to build to ten times their previous size, going deeply in debt, unless they knew they would have contracts. But the only game in Flint for these big contracts was (and still is) General Motors. Durant himself moved on the New York capital market, not in Flint. To borrow large amounts to finance a wheel plant's expansion, Durant could commit General Motors to buying from "GM's own" wheel plant (or from the Fisher Body plant after GM bought out Fisher Body). If Durant acquired a firm, he could afford to take the risk of making it ten times as big—which the owner could not—because he could guarantee the market for ten times as many wheels. Essentially, General Motors played "insurer" as its larger assets and secure market position guaranteed the loan or stock issue; in addition, it made the returns of the small business (say, the wheel maker or Fisher Body) insurable by providing a "guaranteed" market. The firm becoming a subsidiary sold its human and organizational capital to General Motors, had that spread over a larger capital stock, and was given a secure access to the market (as secure as General Motors', that is). There was then no reason for Durant to intervene in the administration of the subsidiary unless it lost money, had too little capacity, or priced itself out of the market. The "private" wealth of the subsidiary's old owners was freed up (by having General Motors' purchase price in the bank), instead of further embedded in their parts firm. Especially, they did not have their wealth "leveraged" with a large debt that made the leverage very risky for them (which would have happened had they expanded with capital borrowed in their own name rather than letting Durant do the expansion) just because General Motors needed more wheels than they had capital to produce. General Motors needed the new, much larger capacity, so General Motors should take the risk, and not some little wheelwright who would have had to mortgage his house (or his race horses) to expand. So what made this extreme decentralization work? In fact it is the thing that Sabel says makes fordism work, namely, a mass market for identical goods, organized into a bunch of reliable dealerships providing parts and service as well as selling cars and spread out around the coun-

122 try. This was what made Durant able to organize General Motors; he had made the carriage business work the same way. But what produced the money to invest in expanding the parts suppliers by, say, ten times was mass marketing. Durant's own innovations were not in fordism, but in mass marketing supported by craft skills; and that mass marketing provided backing for capital flows that eventually allowed parts of General Motors to fordize their production. The basic generalization in Sabel (1962) is that mass marketing encourages the development of technical routines (see above, Chapter 2). But here we see that mass marketing first produced innovative technical firms tied together by subcontracts, with skilled, flexible workers producing whatever parts are needed this year. In short, it caused expanded craft production. Perhaps the key is that at that time producing automobiles was technically new. A mass market in a technically new system may produce a flourishing "small-firm skilled worker" system. Only when the firm has mastered the technical problems at the operative level can the new knowledge of how to make, say, a million wheels a year filter up so that engineers can routinize the knowledge of how to make a million wheels. It is important to realize that a flexible system held together by flexible contracts and occasional vertical integration to redistribute the risks could compete successfully in the same market that the Ford Motor Company itself was producing in. Subcontracting can compete with fordism on fordism's own turf.

The Centralization Revolution at General Motors To understand what the Sloan revolution at General Motors was, we have to note more specifically what functions it was that were created as new functions of the central administration (now the "general office"), what functions were not decentralized to the division, and what functions for the divisions and for the investors the general office had.

Coordinating Projected Demand The crises that created the Sloan revolution at General Motors were centered in the marketing system. The basic problem seems to have been that the parts suppliers, being far back in the system from the dealers who have direct information from the clients, could not tell whether a slackening of demand was temporary or long term. (It is indicative that Sloan himself had experience in the parts production part of General Motors.) To keep their own production efficient, they needed to keep

123 their trained technical workers on, to use their capital to the fullest, and generally to work up to capacity. So they had motives not to mistake a temporary downturn for a shift in demand. The main reason General Motors needed a general office, then, was to organize, and then universalize, an authoritative system for projecting demand, with projections revised often enough to control production and inventory policy far back into the parts and materials supply system.

Centralized Differentiation of General Motors' Markets The second big thing Durant had developed was a class-differentiated marketing system, with the makes he had bought up into slots in a market organized from cheap to expensive. The cheapest line was Ford's, and it was a long time before Durant and Sloan could figure out how to produce a car to beat Ford in the low end of the market. Their short-run adaptation was to continue to produce and market the distinctive cars of the companies they had bought. Clearly, though, one should not encourage a successful Olds division, say, to expand to competitively damage the Buick division, when maybe no one else in the market was in a position to damage Buick. As General Motors centralized, then, they did not want to go in a completely fordist direction, producing only one car. Diversified mass marketing was already a corporate strategy, an implicit result of acquisitions and letting the firms acquired continue to produce what they had been producing. In any case, decentralization as a way to deal with the administrative problems of producing several lines of cars was already institutionalized. What was new with Sloan and the general office was a deliberate plan to cover the spectrum of markets, to fill in (invest in) areas of the class system they did not cover, and to run down (disinvest or convert investments) in situations where subcorporations competed with each other for a class segment of the market. So the firm needed parallel feedback—parallel demand projections—to control production in the several lines differentiated by class .

Coordinating the Service Market with the Car Market The third general problem confronting Sloan, apparently solved in the operating divisions rather than by the general office, is that marketing cars ultimately depends on a reliable supply of repair parts. Producers of parts for, say, the Buick division had to cooperate as well with Buick dealers all over the country in supplying the repair parts needed to keep Buicks running. Having lots of Buicks in local junkyards for lack of parts ultimately depresses sales. The corporation thus had to have the market-

124 ing information collected from Buick dealers available for planning repair parts production. Still, General Motors did not want Buick parts suppliers using monopoly power (e.g., being the only producer of Buick air filters) to jack up the price of repair parts. So while GM needed to supply information to the Buick air filter manufacturer, it needed also to keep that producer from exploiting his position as the only one who could supply the dealers. Sometimes such a problem was resolved by buying the parts supplier and putting it administratively into the Buick division, but such decisions would have been strategic ones made by the general office on recommendation of the Buick division.

Centralized Functions Not Delegated to the Divisions under Sloan It is important to note that some things were not left to divisional decision. The biggest example was purchasing. There are all sorts of reasons why people generally do not want purchasing to be under divisional (or any decentralized) control: corruption, price competition between divisions, integration of orders to get longer production runs for suppliers, relevance of purchasing information to the decision of whether or not to acquire a supplier, and so on. In any event, a crucial part of the administration of divisional supplier relations was not under divisional control. Similarly, standards of cost and profit accounting, calculation of investment returns, and the like were not decentralized. That is, mechanisms central to the integrity of general office control of the divisions (for the aspects that they mainly wanted to control) were not left in the division's hands.

The General Function of a General Office The main point of the specificity of our description of what was centralized is that when a firm already has a decentralized administration, imposing a new centralized structure on top of it threatens to centralize decisions that had better be left decentralized. Decentralization was the situation General Motors was left in by the acquisition strategy of Durant. When a firm like Du Pont has a centralized structure and then goes into various businesses that require different sorts of coordination among, say, production, engineering, and marketing, the firm is likely to end up running one of the businesses by a strategy inappropriate to that business, though it may work very well for another one—thereby losing money. Such a firm thus has to learn not to make so many decisions centrally. Chandler's point is that both firms—Du Pont and General Motors—ended up with pretty much the same structure, with coordina-

125 tion of the different specialties involved in a manufacturing operation going on at the divisional level and projections (and investment strategy based on those projections) being done at the general office level. The functions outlined above for the general office are especially those in which information from one division, parts subsidiary, or marketing branch was useful for assessing the investment value of another division, parts subsidiary, or marketing branch. In such circumstances the general office has an interest in securing the following features in that information: 1. Financial integrity. The chains of information that link expenditures in a business to capital markets are more effective if they guarantee financial integrity. Centralized monitoring of purchasing, firmwide accounting standards, and coordinated projections all tend to make the standards for investment in each part of a decentralized firm as good as the standards for the whole firm. One function of the general office, needed to justify to an investment bank a loan of money to expand a wheel maker tenfold, is to guarantee the financial integrity of the information both on the wheel maker and on the buying division. 2. Guarantees of markets. If General Motors gives a long-term contract to its wheel-making subsidiary, this guarantees (within limits) a market for the wheels, which in turn guarantees (within limits) the profitability of the investment. Similarly, arranging that Olds shall not invade Buick markets, and vice versa, does not make either division secure from competition; but it does reduce the risks that completely autonomous divisions would autonomously invade each others' markets. A general office can make competitive destruction of the value of an investment less likely. 3. Reliable external economies. Reliable repair parts supplies and available retail installment credit are profitable businesses in their own right, but in addition, their reliability and availability are selling points for car marketing. Yet a completely autonomous firm supplying repair parts or installment credit might not find it profitable to supply exactly what a dealer needs. Being able to arrange reliable supplies of such externally generated selling points stabilizes the value of marketing investments, and hence of the manufacturing investments behind them. In turn, of course, stable marketing arrangements provide reliable opportunities to repair

126 parts suppliers and installment credit suppliers. Coordinated provision of complementary goods and services makes a more profitable portfolio of investments. Thus, the functions we see performed by the newly organized general office in General Motors provide information crucial to capital markets (cf. Zetka 1988). They also developed out of functions performed by Durant as an individual, before Sloan created a committee version of Durant. In this sense, Chandler's analysis of the causes of the general office reduces to one cause, of supplying information of use to investors, rather than of supplying coordination of response in commodity-line markets. There is no general reason to believe that general offices will be required wherever divisions are required.

The Theoretical Problem of Sears This disjunction of the causes of divisionalization and the causes of general offices creates a problem for Chandler's analysis of Sears, for we cannot, I believe, understand the administrative reform of Sears without a careful analysis of Sears's competition. That competition was a structure of small and medium wholesalers in distinct commodity lines, with no national commodity-line buyers and certainly no national general office. Sears had to compete with this extremely decentralized system, and could only do so if it could coordinate decentralized demands better than that competition, through national "divisions" run by commodity-line buyers. But I will argue that the central administrative innovation discussed by Chandler—Sears's regional divisions—could not do that work. The problem Sears had was not to create commodity-line divisions, as at Du Pont, because it already had them. Nor was it to create a general office to govern investment, because it already had that too. Instead, it was to build new structures of flows of information from retail stores to autonomous buying divisions, inherited from the mail order business. The regional divisions, I will argue, were not part of that flow. The central argument, then, will be that neither the causes of divisionalization by commodity lines, nor the flows of information justifying investments, changed. Hence the causes in Chandler's theory as developed above cannot explain what happened at Sears.

Regional Information in Merchant Wholesaling and Sears Let us take a problem of marketing administration from Chandler's analysis of Sears. We will try to illustrate again how Chandler argues, and in

127 particular how a functional inadequacy of a structure creates a flow of problems for individuals (or for small groups, more often) and how structural change in a functional direction is caused by the solution of individual-level problems. When we inspect Chandler's argument about Sears from this point of view, several inadequacies appear. We will look at the problem of adjusting the inventory displayed and on hand in a Sears store to the seasons and to the regions of the country. Obviously adjustments all the way back to the buyer (the person who arranges the contract between Sears and the factories that produce the goods) are needed if the overall "functional" problem of having summer goods gone and back-to-school goods on display by August, and of having different backto-school clothes in Phoenix than in Seattle, is to be solved. "The market" solves this problem by thousands of wholesalers stocking what local retailers are going to want to buy and sending travelers (traveling salesmen) around to the stores in their area with catalogs and samples, which results then in a large flow of small orders from wholesalers to manufacturers. If Sears was to have an advantage in this trade, it had to use the experience of, and the opportunity for, volume buying that it had developed in the mail order business. But buyers for a mail order house need not time deliveries and displays for each store by the season and the region. They need to know how many raincoats and how many cowboy boots children whose parents order from Sears will want this year and to place both large orders rationally, thus allowing manufacturers to run their plants year-round and to process only one order for raincoats or one for cowboy boots. Sears did not need to build the buyers' information system to be attuned to seasonal and regional variations when it was a mail order business, and the buyer did not necessarily notice that the children's cowboy boot orders in August came from Phoenix and the children's raincoat orders from Seattle. After going into the retail shopping center business, then, Sears had the "structural" problem that store-level problems were not the same problems their buyers were good at solving, namely minimizing manufacturing and wholesaling costs. This was because the functions of wholesalers—which cost so much—of adjusting inventories to seasonal and regional variations had no place to be solved in a structure created by "adding stores to a mail order house." Chandler shows that when the aggregate performance statistics of the store division were not satisfactory, many people were set to work to find out what was wrong. They located many different kinds of problems—for example, a consultant with experience in cutting costs of government administration, found problems of cost minimization that Sears could

128 work on. The structural strain of having a buyer information system that is not adequate to the new coordination problem does not necessarily mean the problem will be posed correctly. Many reforms may be instituted which do not solve the problem that Chandler (and we) locate by hindsight, that the organizational environment of Sears structures the buyers' problems in such a way that the buyers do not solve the stores' inventory adjustment problems as well as wholesalers in the open market solve competitive stores' problems. If Chandler provides lots of evidence that Sears's people did not "rationally" solve their problem (instead, they "rationally" addressed a whole lot of problems in searching for what their problem really was), he still has to explain why "structure follows strategy," that is, why a functional result came about in the long run. His basic proposition is an "absorbing Markov chain" proposition: the organization will continue to throw up problems to individuals until they correctly identify the source of the functional inadequacy and build a structure that remedies it.

Commodity Line Rationality Versus Store Inventory Rationality Note that the buyer versus store problem will recur in many different lines. In Arizona, Sears should not stock as many gardening tools in the spring as in Illinois; above the glacier line, Sears probably should stock more fishing and hunting gear in the sports department; in the Southwest, the automobile lines should be adjusted for the large population of pickups. Buyers specialize in commodity lines because efficient buying requires a detailed knowledge of the characteristics and prices of competitive goods and competitive producers. Individual-level adjustments to problems of different store inventories in different parts of the country by particular buyers specializing in commodity lines is an inefficient strategy, because too many buyers in too many lines have to learn too much about regional variations. Individual-level rationality by buyers in each merchandise line is thus not a very viable "functional alternative" to reorganizing the connection between stores and buyers on a regional basis. Sears needed to divorce some of this flow of store problems to buyers (after being reanalyzed in regional inventory terms by a regional management) from the mail order flow to buyers, because the mail order flow has to solve a different structure of problems. There are two kinds of connections from the functional outcome (adjustment of store inventories to season and region without losing all the

129 advantages of massed buying) to the structural thing to be explained (operative divisions controlling wholesale buying and some kinds of regional administration of retail stores). The first is that individuals posed with problems of inadequate competitive performance in stores hunt for solutions. The second is that until they find a patterning of overall activity that solves the functional problem, people keep on being posed with individual problems that keep them looking for solutions. So the basic problem Sears had, according to Chandler, was where to locate informational complexity in the organizational structure. The old regime, that of the mail order house, had something like nineteen decentralized divisions in the merchandise division (the number varied from time to time): nineteen lines in which the core executive—the buyer—had to keep track of what sorts of garden tractors were selling this year, of whether they could sell heavy goods like fertilizer competitively when they had to send it through the mail at a high transportation cost (they could not, when they were selling by mail order, buy carload lots locally, as a garden and lawn supply place in each town might do), of what manufacturer of law mowers could reliably produce a machine that Sears could guarantee customer satisfaction on without going broke, and so on. The old Sears structure could be described as a "coalition of merchants" in these different lines, each functioning essentially as an autonomous buyer and "selling to" the mail order division, which in turn sold retail for the coalition. There was much decentralization of decisions to operative (i.e., merchandise-line) levels in which the big decisions about the sizes and contents of merchandise flows and inventories were made essentially by buyers in the different lines. This is one part of the essential information a wholesaler must have to be successful in the market; the other part is knowledge of local and regional peculiarities of the seasonal variations in how lines of goods move. Sears's advantage was that they had a lot better buyers than commodity-line wholesalers for supplying a national market and for arranging for cheap production and response to changing fashions nationally in commodity lines. They had a big statistical base on which to estimate future demand and a developed routine for translating the necessary information from the buyer into a commodity-line section of the seasonal mail order catalogs. The code word for this system in the mail order business is "quantity buying," by which, with that information and the big flow of goods through the mail order house, the buyer in a given commodity line could make it possible for the manufacturer to smooth out production of a

130 good over the seasons and have long production runs, and so make it cheaper. The manufacturer could plan a smooth production schedule in much bigger batches before a changeover, reduce the amount of clerical work on a given quantity of goods because it was all for one buyer, and would ordinarily have no inventory risk because he or she would be producing to Sears's order rather than to what he hoped a lot of wholesalers in different parts of the country would decide to do. Wholesalers both had and created for the manufacturer a much bigger burden of clerical work than the wholesale part of the Sears mail order business did, and they had and created a bigger inventory risk as well. This was because decentralized wholesalers' information was not as reliable and their expertise not as well organized to take advantage of big flows of goods. Wholesalers were more likely to have to write off or sell at cut prices if a line did not sell to retailers, for instance, because a competitor wholesaler had a slightly better model. Sears or Penney's might guess wrong about what would sell at the retail level, but at least as a wholesaler for their own retail outlets they did not have to mark down a lot of goods because of a few big retailer decisions to go with a competitor. The Sears buyer did not have to have salespeople out convincing each department head or buyer in each department store in many different metropolitan areas and small market towns to buy from her or him, as the small wholesaler did. The clerical, inventory, and sales costs of wholesaling and manufacturing are cut down by chain stores generally, and it is these costs that are especially cut by a joint buying operation of a huge mail order house with a very big chain store. But what Sears had been sacrificing in order to make that saving (because the sacrifice was not important in the mail order business) was all the information about local peculiarities of demand that were built into the thousands of wholesalers adapting their buying and sales strategies to local markets throughout the country. Now that Sears was going into the local department store business, they had to try to perform the functions that all those wholesaler clerical and sales people and all those autonomous buyers in each retail store did in the "free market" structure. The free market structure was very expensive, but it did that work of relating local demands to centralized "commodity-line" production by manufacturers who sold on national markets (Beniger 1986). Their competition, unlike General Motors', was already decentralized a lot more than Sears would ever be, because the competition was many autonomous firms scattered throughout the country. The average wholesaler in the United States even nowadays has about twelve employees (Statistical Abstract of the United States, 1989 , 761). Sears had to compete against a structure that used a lot of clerks, a

131 lot of salespeople, a lot of special ordering from factories, a lot of experienced small entrepreneurs in particular lines of wholesale trade in particular parts of the country, a lot of folks taking inventory risks because they knew their local market—a structure that was doing a very good job at something Sears had not had to do when they were only in the mail order business. But Sears had the advantage of knowing a lot better than that decentralized apparatus of wholesalers how to use the power and efficiency of buying on a large scale and of advertising nationally through catalogs. The structure Sears had to compete with would never deliver swimming suits in midwinter in Maine or skis to a department store in New Orleans. What we see in Sears in the period Chandler studied is Sears trying to improve their accuracy and flexibility in wholesaling up to the level that their competition was already at, without losing their cost advantages from large orders placed well in advance. Note that except for a very few comments on how to get good store managers and department heads in retailing, almost all of Chandler's talk about administrative reorganization is about the wholesaling end of things: how to relate the buying and supplying of mail order with that of stores (the long-run solution was to have separate inventory and quantity controls but to use the same buyers); how to adjust overall supplies to regional market variations; how to put controls on the buyers by forecasting demand so that Sears would not be caught (again) with excessive inventories; how to get the departments in the merchandise department (buying) down to a reasonable number; how to decide who was going to coordinate transport and select suppliers for different regions to minimize transport; how much Sears was going to intervene in giving technical specifications to manufacturers rather than just picking what was offered; and so on. Sometimes (always, really) these decisions had retail implications. But these decisions were, in the competitive structure of small wholesalers spread around the country, all wholesaler decisions. Mass or chain store merchandising is mainly a reorganization of wholesaling by substituting bureaucratic or administrative control of wholesaling decisions for market control. Sears's decentralization had to cope with problems that the already decentralized wholesaling system had not had, and that Sears had not had when they had a terrifically decentralized system of buyers operating as more or less independent merchants but selling nationally through the catalog in a mail-order-only business. To see what was wrong with Sears's "buyer divisions" for coordinating inventories at many local departments stores, we need to analyze what Sears's competition, the system of tiny wholesaling firms, accomplished.

132

The General Problem of Wholesaling Where does the problem that wholesalers have to deal with come from? The analysis has to start with a contradiction between what one needs at the retail store level and what the factories can produce. Each factory makes a restricted line of goods, goods related by a common technology. What a store like a Sears branch has to have is a flow of goods into the store inventory that is shaped by the consumption bundle that customers are looking for when they go to a "men's" department store, what a given kind of person or family buys on their monthly (or so) trip to a big store. The job of wholesaling is to take the set of flows of homogeneous goods from a set of factories and to turn it into a set of flows of heterogeneous goods through a whole bunch of stores that is shaped to the current buying habits of each store's clientele. Price, style, quantity, performance criteria, and the like for the mixed flow at the store end has to come from folks who know what is moving, has to be adjusted for class and regional variations in consumption habits, and has to vary with the seasons and with what the seasons mean locally. That is where the ultimate information comes from, from the individual Sears big store responding to local peculiarities in consumption bundles that families buy on their once-a-month shopping trips. But to turn all these dribbles of information on how many sofas we sell in the spring in each store into a large order for a factory, the information has to be aggregated into one decision about how many sofas to order of what styles and prices several months in advance—and it has to be aggregated separately for factories in different parts of the country, since shipping sofas is expensive. That aggregation is how a chain store gets its price break. It is what the economics of chain store wholesaling consists of. The chain store has to transmit that information to its wholesale buyer in a given commodity line in such a way that, when it looks as if things are maybe not moving as fast this year, the buyer can cut down the flow volume without completely disrupting the manufacturer and the traffic department in a way that will cost the chain a lot in the long run. The chain wants to make the adjustment before it has a cubic mile of extra unsold sofas. It is because this problem can be solved by forecasting, by using expert buyers in each line of goods, by having regional executives manage wholesaling in distinct areas of the country, that we have chain stores that can sell at lower prices. It is because the competition of independent department stores and independent wholesalers spread out all over the landscape can do that adjustment, at a cost of more clerical and sales work

133 and more inventory risk takers, that a chain store has to keep reorganizing until it gets a responsive system that solves the inventory problem of each local store and at the same time solves the manufacturer's problem of more expensive production if it cannot get larger orders far in advance. Actually, the multiple-regional-divisional structure that Chandler focuses on was only a small part of the administrative innovation he describes. The main innovations had to be increased flows of market information from the various departments of the many stores to the wholesale buyers in specialized lines—information about what would move, how much of it would move, and at what price, style, and seasonal mix it would move. Regional top executives, even with larger staffs than they had, just could not do it. There was too much knowledge of too many lines of goods for them to administer it. In fact, Chandler says as much, but he does not emphasize this point. He talks about the reforms undertaken by Barker, the vice president of retail administration and personnel, in the 1930s. As Huber says in his critique of Chandler's analysis of Sears, "Despite the centralized retail administration in place, Barker developed a policy of 'unit control' and other 'statistical procedures to ensure . . . that merchandise would be in stock in the right amounts at the right times' (p. 264 of Chandler). By 1934, according to Chandler, Barker had created a 'cohesive and capable retail organization,' even though it was to be more or less this same organization which would later be reorganized on territorial lines" (Huber 1985, 3). It may be that the regional divisional structure helped fine-tune this overall system for feeding buyers the adequate aggregated information that ultimately came from what was moving in the retail stores. But Chandler does not actually show the interconnection between the development of the information system that did the aggregating which functionally had to be done and the decentralization to regions as well as to buyers in each commodity line. Another weakness of Chandler's evidence (Huber 1985, 3–4) is that Chandler does not actually show regional diversity. For example, the Pacific Coast territory must have stretched from Seattle to San Francisco, at least. But if one considers climatic variation, the wet mountainous market stretching out from Seattle and the desert market stretching behind San Francisco are probably no more alike than those of Seattle and Chicago. Instead, what may have been going on is that as Sears got a lot more stores in total, they needed to have lots of intermediate offices to handle the sheer quantity of decisions that are involved in running stores in San Francisco and Seattle, and then in Eureka, Portland, and Pullman. While Sears could run two stores on the West Coast with those stores'

134 big decisions being made in Chicago, it was difficult to run twenty Pacific Coast stores that way simply because there would be ten times as many decisions to review and execute. The same was particularly true of decisions involving investments in local stores. Manufacturing investments are in commodity lines; retail investments are in locations. The investment arm of the general office needed to know that stores in the Pacific region went in the right places. Once they were in the right places, the store-to-buyer information flow could adjust the store inventory. It therefore seems likely that regional offices were evaluating investments in stores in particular locations—a general office function—rather than aggregating inventory statistics—an operating division function. The more such localized investments had to be made far from Chicago, the more a regional division was required to advise on them.

Organizational Problems of the Service Sector All of this illustrates why it is that the service industries tend to have a lot of the labor in the economy nowadays. The service sector as a whole has many clerks, salespeople, commodity specialists like the buyers and department heads in department stores, and the like. These people know about the prices, quantities to be sold, qualities, fashionableness, and so on of a line of detailed commodities for a local area or a particular store. In particular, they need to know what will be wanted in one department or boutique of this product from Janzen, this one from Dior, or of this product from AC Sparkplug, this one from Black and Decker. Then all that information has to be translated into orders to a wholesaler for particular quantities at particular prices, shipping orders and bills of lading, addresses on the hundreds of pieces of mail that keep things in order so the mail carriers know where to deliver them, and so forth. A very large share of the work in the service sector, especially in wholesale and retail trade, transportation, and communications, is the processing of enormously detailed pieces of paper about how much of thousands of different commodities go to this or that store, when, and at what price. It is in this mass of clerical and routine sales work that the feminization of the service industry has progressed furthest; only about one out of five lawyers is a women (about one out of three new law school graduates is a woman), whereas something like nineteen out of twenty secretaries or clerks in retail and wholesale trade are women (Statistical Abstract of the United States, 1987 , 386).

135 Sears's administrative problem emphasizes that what makes mass production possible is the processing and aggregating of this enormous mass of detailed information about commodity flows for thousands of commodities by all these women clerical and sales workers in the service sector. They are the ones who solve the problem of turning the particular demands of the clientele of a particular store into the factory's yearly plan of production, the plan that allows the factory to have efficient long runs of production of the same commodity. The problem that Sears had to solve was this same one: how to make mass production possible when the information was concentrated originally in what was happening to this and that stock of lawn mowers in Seattle and in Phoenix, and in how much less demand for lawn mowers there was in Tucson than in Phoenix (because more folks in Tucson have desert plants on their "lawns," but people still have grass lawns in Phoenix). This is why Chandler's analysis of regional divisions is not adequate to explain all the changes made in information-processing structures using store data. Regional divisions did not usually have buyers to arrange for the production of goods with factories. What is problematic, what shapes the uncertainty, is the relation between store inventories and "quantity buying" by a decentralized "coalition of merchants," each in a single commodity line. For Sears to succeed, it had to coordinate its local store-inventory information so commodity-line quantity buyers could use it. Much of that information processing, as Chandler reports it, completely bypassed the regional divisions he thought were central. But Sears did not gain its competitive advantage in wholesaling by regional organization. It gained it by using its mail order commodity-line buying divisions, by using information aggregated by commodity line from the stores. They used less sales and clerical labor, and created less trouble for factories, not because they had regional divisions, but because they had national commodityline operating divisions. Thus, Chandler's main cause of divisionalization did not change. The divisions needed new information from commodity-line departments in stores, but their uncertainties were not substantially different from what they were in the mail order business. Sears may have made its general office more competent to invest in retail stores by having regional divisions, but it already had a general office coordinating investment in commodity-line divisions. The causes for having a general office also did not change. Chandler's theory is more useful for explaining the structure Sears started with than for analyzing the administrative changes, including regional divisions, of the period he studied.

136

What Is Chandler's Independent Variable? Chandler's analysis in Strategy and Structure is mainly about divisionalization; in The Visible Hand (1977) it is more directly about coordination and centralization. Consequently, his argument is clearest in Strategy when he argues about the causes for divisions becoming separated and internally centralized. We have only hinted above at these causes, as "being in several markets." We have proceeded so far as if it were obvious whether a company was in several different markets. But since Chandler's argument as a whole depends on markets being a source of uncertainty that has to be adapted to (and since this book is about that part of his argument), we have to define markets such that they are separable sources of uncertainty. That uncertainty in turn has to ramify through the several functional departments (which make up the multifunction divisions whose origins Chandler is trying to analyze) in such a way that a coordinated response of the different functions to a market's particular uncertainties is necessary. In this section we will be more specific about what "being in several markets" means. We will argue that in Chandler's argument being in more than one market had five elements: (1) that market uncertainties ramify into manufacturing and engineering; (2) that information on market uncertainties come from segregated social sources; (3) that those segregated social sources carry distinct information; (4) that the inventory risks that reflect the ramifying of market uncertainty are managed in similar ways within a market, perhaps in different ways in different markets; and (5) that total uncertainty is large. If all five conditions are met, then market uncertainty ramifies into the administration of manufacturing and engineering in such a way as to create divisions.

The Transmission of Market Uncertainty to Manufacturing Consider, for example, Chandler's description of the steel industry (331–337), which ends with the statement, "If the market, apparently, has not been different enough [for different products] to bring about the growth of autonomous, integrated divisions, neither have strategic decisions been numerous or complex enough to demand the building of a general administrative office with general executives and staff specialists" (337). Yet this is in spite of the fact that "the big integrated steel works produce a vastly greater amount of a larger variety of products for more types of industries and businesses than do the copper smelters, refineries, and fabricating mills" (331). The key reason why vastly larger and more dif-

137 ferentiated markets do not lead to differentiated multifunctional divisions (i.e., decentralization as defined above) is that

the complex demands of scheduling and of coordinating departments require close centralized administrative supervision if the great integrated mills, the mines, ore boats, and distributing facilities are to be operated at a fairly even capacity and if loss and waste are to be avoided by piling up inventory in one department or another or through having temporarily idle facilities and other resources. The continuance and growth of the multidepartmental division may have been checked by the fact that the demands of a single market, whether it is in one industry or one region, is not enough to keep a single great mill operating at full capacity. Therefore such mills must draw on the larger national market. (336)

That is, for steel production in great mills, (1) there are great economies in producing all the stages up through molten steel (iron ore mining and transportation; mining and transportation of coal; coke production; reduction of the ore to iron; and burning off the carbon together with other alloying and chemical changes to make steel) for all the various products in one operation. There are also large economies of scale in combined iron and steel production (see, e.g., Scherer's and others' estimates in Scherer 1980, 96). (2) There are varying but substantial economies in producing many of the heavier steel products (plate, construction shapes, rails, large tubes) from semimolten steel at the same site. (3) The users of these products are willing to pay, in some delay between their orders and their delivery, for the cheaper products. It has paid the great integrated steel mills to turn many markets in a given large region of the country into one market by administrative means. That is, they cut the connection between the order coming in and any special speedup or slowdown of the steel-making process. Instead they integrate all the orders for different products into a gross quantity of steel needed from the plant and govern the first phases of the process up through molten steel by that aggregated quantity. Then the flow of steel from the integrated part of the plant is sent to fabricating installations (especially rolling mills, but some to foundries), where it is differentiated to meet the needs of the different markets. This means that the differentiation of the market does not result in differentiation of the main part of the manufacturing organization—mining, transportation, coke production, ore reduction, and steel production. The technology is not dedicated to serving a particular market,

138 but instead is a general-purpose technology that produces a raw material for a wide variety of markets. But for a large share of steel products manufacturing in which that raw material is used, it pays to have the fabrication and marketing of that product near the furnaces. Fabrication near the furnaces in particular reduces transportation costs as well as energy costs for reheating. When the market does get differentiated in what kind of steel it wants, so that smaller (usually electrical) furnaces must be dedicated to producing it (as in the steel minimills that have been expanding recently while the great integrated mills have been losing customers), then there are great advantages in producer autonomy and integration of molten steel manufacturing with the marketing of the products of that steel. So far, that autonomy has mostly been achieved by independent firms producing specialty steel products rather than by having autonomous divisions in the integrated mills. The first requirement for there to be "differentiated markets" in Chandler's theory, then, is that the manufacturing apparatus serving the market be a dedicated resource , one of little use for servicing any other market with a different pattern and source of uncertainty. Otherwise the uncertainty from the market will not in general be transmitted to the manufacturing operation, except after being aggregated with the uncertainty that comes from other markets. Sears did not have regionally differentiated manufacturing, so its regional divisions did not amount to much. The differentiation of contact with manufacturing was already there in the nineteen commodity-line "divisions" of the mail order house. When there are differentiated markets served by common, general-purpose manufacturing, then we will expect to find differentiation of the marketing structure and considerable autonomy of each marketing department. Of course, separate marketing departments may have to be hooked onto the administration of the last phase of fabrication (e.g., the rolling mill) in such a way that uncertainties from the market can give information on what exact products need to be produced in that last phase, how much inventory needs to be carried, what engineering improvements in the product might be required (for example, for fabricated steel products, the recent development of X-ray and acoustic inspection of the finished product has allowed the detection of many more flaws, resulting in much-changed quality control procedures in the last phases of production as an adaptation to customer demands; see Stinchcombe 1985b, 157). There will then be pressures in steel to have innovations that correspond structurally to the multifunctional division only for a late stage of the manufacturing process. But these pressures will not tend

139 to penetrate to the management of the whole manufacturing operation—as happens when a given chemical plant produces only paint, another only explosives—and so will not actually result in multifunctional divisions. The first point, then, in defining Chandler's independent variable—in defining when a firm is in several different markets—is this: whether the market of one manufacturing operation is different from that of another one depends on whether manufacturing operations are technically differentiated, so that the same general-purpose technical operation cannot serve now one market, now another. Even when there are reasons to administratively connect a general-purpose technical system—such as iron ore reduction and steel production furnaces—with technical operations serving different markets—such as rolling mills for producing plate (mostly sold to motor vehicle manufacturers), tubes (mostly sold to the petroleum industry), rails, and construction shapes (mostly sold through wholesalers and brokers to a dispersed clientele)—the differentiation of these markets tends not to be transmitted backward into the general-purpose technology that serves all of them. More briefly, several markets with different sources and kinds of uncertainty may be transformed into one market by administrative means, if that seems wise and profitable. It is wise and profitable if a general-purpose technology can serve all the markets, especially when there are large economies of scale in that technology. (See 352–353 for Chandler's analysis of the same general pattern as is found in the steel industry for the early history of the petroleum industry, and 349 for similar considerations in the tobacco and meat-packing industries and for United Fruit and American Sugar.)

The Segregation of Product Information One of the ways manufacturing and engineering are connected to the market is that engineered and manufactured features like design, color or finishing treatment, chemical composition, quality control and reliability, maintainability and service cost, taste, and complementarity or compatibility with other products available in the market (e.g., software) largely determine a product's success in the market. Engineers and the manufacturing branch thus need to get market information in order to do their design, manufacturing, inspecting, or finishing work better. The question of whether some improvement in the product is worth its cost, then, is decided jointly by what the market will pay for it and what the engineering and manufacturing cost of the improvement is. The social location of that joint market and manufacturing information for a particular product depends on the social organization of the

140 selling process. The fact that information about what fertilizer sells in what sorts of packages to homeowners is concentrated in the same place as knowledge of what features sell garden tractors, while information on what fertilizer sells to farmers is concentrated in the same place as knowledge of what features sell gigantic field machines, is not determined by the technical interdependencies of fertilizer and tractor production. It may well be that the same machine manufacturers produce garden tractors and field machines, while one chemical plant produces both packaged fertilizers for the home gardener and carload lots for the farmer. A commodity line that includes home fertilizers and garden tractors will therefore tend to develop information on the gardening market this year in a given Sears department in a given store, and that information has to be aggregated properly for the gardening buyer who deals with both the machine factory and the chemical factory. Similarly, outside Sears there will be a feed store commodity line with a farm implement department, which provides information on this year's demand for fertilizer in carload lots in a given part of the country and on the availability of credit for new field machines, and that information has to be aggregated by wholesalers and brokers for those same chemical plants and machine manufacturers. The markets are different both because features that sell the products to homeowners and farmers are different and because the differentiated information is available in two different social structures. Of course, there is not much commonality in the concrete material features that sell a bag of fertilizer versus a garden tractor, and in that sense they address "different markets." But the gardening buyer in Sears (after appropriate reorganization of the Sears structure, as analyzed by Chandler) gets the information about the two products' features from the same flow of retail information, as aggregated by the appropriate inventory system and by informal transmission of sales know-how. The manufacturing commonality between field and garden tractors, or between a carload and a bag of fertilizer, does not mean they have common markets, because the information about what causes them to sell is not located in the same social place: the Sears buyer never learns what sells either carloads of fertilizer or great field machines, nor does the feed dealer learn what sells bags of fertilizer or garden tractors. Chandler often talks as if selling more different products leads to differentiation into multifunctional divisions. For instance, in discussing why the chemical industry almost uniformly adopted the multidivisional structure, he starts with a quotation from Williams Haynes, American Chemical Industry—A History:

141

Production of chemicals by the petroleum industry appeared to be economically and technically sound, but most petroleum executives could not see what appeared to them to be a tiny market for a multitude of chemicals produced by a complexity of operations and sold to a long and diversified list of customers, tasks for which they had neither the technical nor the sales staffs. (358)

Chandler's basic idea here is that the more products one sells, the more different markets one is in, and so the more pressure there will be for multidivisional organization. Presumably this generalization is approximately true for the way diversity in chemical products is produced. In chemicals, diversity is due to the knowledge needed to produce one material being useful in producing a chemically related material whose uses may be very distinct. When the uses are very distinct, then the knowledge of product characteristics that help to sell the product is likely to be organized into different social structures. This principle applies even to the same chemical product. The knowledge of what makes nylon bearings and runners sellable to the furniture industry, for instance, is not found in the same places as knowledge of what makes ladies' nylon hose sellable in supermarkets and department stores. Similarly, one could not learn in the department store that a bearing for a drawer ought to be white, whereas one could learn that nylon for hose ought to be "nude" and that nowadays "nude" is not the color of Rubens's nudes. When the information needed to perfect the manufacturing process or the product design is located in different social structures, the manufacturing process has to be organized to respond to information coming from those social structures separately. This connection between segregation of information and divisionization is shown by what happened in cases where diversity of products was not generated by research or manufacturing commonality, but by the social organization of the market itself. In most such cases, multidivisional structures did not develop. For example, Firestone "took on automobile parts, rubber and plastic items, and other lines [other than tires] that used its marketing organization more than its production facilities" (352) but did not develop a multidivisional structure. Similarly, a General Foods memo is quoted:

A single selling organization would be used in marketing the several product lines, since "the demonstrated economics of selling a line of products through a single

142

sales organization would be increased by the number of items handled by each salesman" [footnote to the original source of this quotation omitted]. To administer the other activities involved in producing such a large line, the company promptly set up a centralized, functionally departmentalized structure. (347)

Earlier, Pittsburgh Plate Glass, "to make fuller use of its marketing resources . . ., turned after 1900 to developing a line of paints, brushes, and related items that could be sold through the same channels as window and plate glass" (342) without developing multifunctional divisional structures. The example of many commodities using a single merchandising division is related to the steel one, except that in this case it is the marketing information resources that are general purpose and the engineering and manufacturing operations that are specialized and dedicated to a given product line. General-purpose marketing resources (rather than dedicated marketing resources) do not tend to produce multidivisional structures any more than general-purpose manufacturing resources do. Hence, it was not necessary for Pittsburgh Plate Glass to integrate its paint sales information and its paint production group separately from its integration of sales information about window and plate glass and its glass production group, since the sales organization could transmit product information about both from an integrated structure back to the differentiated manufacturing operations. Later, some of the separate production organizations at Pittsburgh Plate Glass developed industrial markets, and the sales organizations marketing to these customers were integrated with the manufacturing function in a divisional organization. Nevertheless, "merchandise division" sells the products of all these manufacturing divisions "to independent dealers, jobbers, distributors, and directly to small customers. Since the chemical unit sells only industrial goods, it has long been almost completely autonomous" (342). Chandler restates his own hypothesis about the number of products as follows: "Those that sold a larger variety of one major line of products in much higher volume to a greater number of industries and businesses have consistently centralized the control of their activities through developing and rationalizing their functionally departmentalized structures; while only those making and selling quite different lines for increasingly differentiated groups of customers turned to the new multidivisional form" (343; Chandler's emphasis). Our point is to define more exactly just what makes a group of products into "one major line." In short, a group of products becomes one line when a single marketing social struc-

143 ture provides engineering and manufacturing with information on what characteristics help sell the products.

Socially Organized Market Segments Carry Different Information The problem we pointed out above in our critique of Chandler's analysis of Sears's regional multifunctional divisions reappears in his chapter 7. In a number of cases regional divisions have disappeared, to be replaced sometimes by product line divisions, sometimes by centralized structures (334–335 for U.S. Steel; 341 for American Can; 348 for National Dairy Products; 354 for petroleum, ambiguously; 355–356 for Indiana Standard in particular, less ambiguously). The basic fact here seems to be that although regional marketing systems are socially differentiated, serving different sets of customers through distinct staffs, the flows of information about product characteristics or product mix demanded are not distinct. Once one knows what the ratio of chocolate ice cream demand to vanilla ice cream demand is in one section of the country, one knows the ratio in other sections. Tin cans for tomatoes and applesauce have common characteristics, even if they are sold in different parts of the country, and milk cartons substitute for milk bottles at about the same rate in California as in New York. Only when distinct social structures carry distinct information do they have to be integrated separately with manufacturing and engineering. A difference that makes no difference is no difference. If my analysis above of whether Sears's regional divisions carry distinct information about product qualities is correct, we might expect to see the regional divisions wither away there as well.

Consumer Products Versus Industrial Products Du Pont started to differentiate its divisions when it for the first time entered the "package goods" market on a large scale, especially in paint (see above), while retaining its market for "tonnage goods," especially in explosives. Pittsburgh Plate Glass was already in the paint market, and it differentiated when various of its parts started to sell to industrial markets. The multifunctionality of Pittsburgh Plate Glass's divisions consisted in the fact that their industrial sales were administered together with their manufacturing, while their package good sales were still managed through the merchandise division. What is common to both histories is that selling to industrial markets and coordinating manufacturing with such selling is quite different from selling to consumer markets and coordinating manufacturing with that.

144 Part of the problem here is that the social structures that collect and aggregate information about desirable product characteristics are different for the two types of market (see above). Yet much of the problem is instead that the producing firm's job of projecting demand and the associated job of inventory control differ greatly for the two types of goods. With industrial goods, the customer typically predicts its own future demand, signs a contract for delivery in the future, and gives some warning if it is going to change that future demand. This procedure of course minimizes the inventory risk of the manufacturer. At the other extreme, the human capital resource of a regional wholesaler in a line of consumer goods is an informed projection of what will be sold in that wholesaler's area of the country; the wholesaler can then take the inventory risk involved in buying goods from the factory in hopes of a sale to local retailers. That is, the projection is not by the customer but by the seller, and the inventory risk associated with changes in demand is likewise typically borne by the seller. There are, of course, intermediate forms. For example, building materials, though often sold to "industrial" consumers—construction firms—are not typically sold on long-term contracts signed far in advance, and so lumber yards and construction steel wholesalers carry large and risky inventories. And while Ford made the Ford dealers carry the inventory created in the depression by his continuing to produce at a higher rate than sales justified, other automobile companies took a good deal of the inventory loss themselves. Within a given industry, who does the projection, who carries the inventory risk, varies with administrative and contractual arrangements. Nevertheless, the broad distinction is whether the buyer or the seller has to deal with the uncertainty of future demand, both in the sense of projecting that demand and in the sense of taking the risk. When selling consumer products, the manufacturer (or others on the seller's side of the market) has to do the projections and take the inventory risk itself. So manufacturers had better adjust their projection methods and their production response to fluctuations in the market—otherwise they, not the customer, will pay for projection and inventory mistakes. The central rationalization of the information flow Chandler talks about in General Motors (146) is a set of running predictions of demand and of running corrections of those projections (projections for four months, updated and extended to a new month each month). The system of "unit control" introduced into Sears retail stores seems to have been similar, but with muchsimplified projections based on recent sales (and resulting inventory changes). They had to be simplified because there were thousands of demands that had to be projected.

145 The pressure to form autonomous divisions tends to be higher where market knowledge derived from a large number of small flows of goods needs to be immediately translated into manufacturing and engineering decisions. This pressure is in general higher for goods in which the seller makes the projections of demand and takes the inventory risk. So in general, the integration of sales and manufacturing is higher in "package goods" than in "tonnage goods." Where the demand is quite unpredictable at the level of the retailer or wholesaler, and so requires very rapid adjustment of product mixes and quantities (as in fashion goods in the apparel trade, novelties in dime stores, some building materials for buildings that are highly sensitive to the business cycle, and goods that are sold to secondary manufacturers in the fashion or other volatile trades, such as dyestuffs and artificial leather, information has to flow very rapidly back to the manufacturers (most of these examples come from Chandler's chapter 7; some are from Vernon 1960). No one can afford to take much inventory risk, but here, since sales are still basically from the retail inventory, the risk is on the sellers' side. The great firms whose administration Chandler studies generally stay away from these businesses, because such small flows are most efficiently managed only in entirely autonomous small and tiny firm structures, which rely a great deal on subcontracting. Where there are large economies of scale in production, as in plate glass, or where specialized manufacturing knowledge is required, as in the manufacture of dyestuffs or artificial leather, the general rule excluding large firms sometimes breaks down. In such cases, however, sales of the specialized product has to be tightly integrated at least with manufacturing, and usually with some parts of engineering as well. In commodity lines in which product differentiation is the basic form of competition, and consequently where small differences in the qualities of the product (real or merely advertised) can make large differences in market success, product design needs to be tightly integrated with market knowledge. In the sorts of enterprises Chandler mostly studied, product design (with attendant production-line design considerations) was in the hands of engineers. In such cases, engineers will also tend to be tightly integrated with sales (as well as with manufacturing, of course) giving rise to a typical integrated product division. In the publishing trade, however, the relation of product design to the market is managed by authors and (especially) by acquisitions editors (Powell 1985), so it is acquisitions editors rather than engineers who tend to be tightly integrated with market information networks (often acquisitions editors are former book salespeople) and with manufacturing. There are no "engineers" designing books to integrate with marketing. Simi-

146 larly, in the apparel trade the central design experts who respond to the market are artistic and style professionals, so it is they rather than engineers who are tightly integrated with market and manufacturing people in the same enterprise. The general point, then, is that manufacturing units oriented toward retail markets and consumer goods, or other goods with a dispersed and variable market (e.g., industrial goods for agriculture or the building industry), are pushed to integrate market information into the administration of manufacturing and engineering. Hence, when incorporated into large diversified firms, they need autonomous divisions. This is more true when competition by product differentiation requires rapid shifts in design or product mixes to the latest marketing information, as for example in selling industrial goods to the apparel trades. In particular, these sorts of retail product competition businesses cannot in general be run by an administration attuned to a market in which the customers themselves project their own demand and take the consequent inventory risks (and reduce them as best they can), as is typical in selling tonnage goods or other industrial products for concentrated manufacturing operations. If a company has manufacturing operations oriented to such industrial markets, it is very unlikely to go into consumer markets without organizing autonomous divisions to connect market information intimately with manufacturing and engineering (or the equivalent of engineering, such as acquisitions editors or dress designers).

Sources of Total Amounts of Uncertainty in Markets The more total uncertainty there is in the markets to which a firm is oriented, the more total information has to flow to govern manufacturing quantities and product mix or to institute shifts in product characteristics. The more whatever market information there is has to be coordinated with uncertainties of other kinds in the environment of production or to meet technical changes in products or production processes, the more total information has to flow from manufacturing and engineering to the people who coordinate those two functions with market information. (If costs or technical characteristics of products are uncertain, that is in effect an uncertainty of markets: costs determine prices, and technical characteristics determine desirability.) That is, the total uncertainty of a given market can be analyzed as in Chapter 2, when we were predicting routinization of production. We "specified four variables that produce uncertainty in a production process, and have argued that the more uncertainty there is, the less the production can be routinized, and so the less one can turn skilled work into

147 semiskilled work by many different workers with different tasks (a 'collective skilled worker with semiskilled parts'). We have also argued that the most opportunity for routinization will exist when there is less uncertainty from (1) unstable markets, (2) unstable specifications, (3) unstable technologies, and (4) unstable natural conditions or unstandardized parts that affect the process of production." The same four variables should apply here to predict decentralization to divisions as are there applied to predict decentralization to skilled workers or professionals, except that they should be modified by the variables of linkage of manufacturing to markets, social structuring of markets, different information in different markets, and inventory risk by the sellers in the market. In fact, we generally predicted in Chapter 2 that very high pressures for detailed response to vagaries of the market would produce so much decentralization that small firms of skilled workers would dominate a market, and there would be no decentralization of large bureaucracies to predict because there would be no large bureaucracies. Whenever large bureaucracies get into such uncertain markets, however, there would be pressure toward decentralization. Our question here is when that decentralization would tend to take the form of multidivisional organization. The first four indicators of when firms are, according to Chandler, "in different markets" should tend to produce their effects more strongly, the more market or other uncertainty there is in a given commodity line. Chandler often talks about the degree of routinization in decision making as a determinant of how complex the coordination problem of operations is, or how complex the decisions about investment and long-run firm expansion are. Cases where he mentions routinization as a variable that depresses the tendency to form autonomous divisions include steel (337: "The selling of steel as well as the obtaining of ore and its production and fabrication still follows a fairly long-established routine pattern"); ambiguously petroleum, in its main markets (362, gasoline, fuel oil, and lubricants) and as a variable whose absence tends to increase divisionalization in the chemical industry (358–362, petrochemicals; 374–378, other chemicals); and the electrical and electronics industry (363–370). There is also often an implicit argument in Chandler's analysis that corresponds to the argument in the Du Pont dyestuffs business, that more total uncertainty obtains in a new business (including a new business for the company), where closer coordination between marketing, manufacturing, and engineering is required than in more routine businesses. For instance, a proposal to incorporate the Stainless and Strip Steel multifunctional division into Jones & Laughlin's centralized admin-

148 istrative structure (333) seems to imply such a developmental routinization, after the innovation of building a stainless production and marketing operation in an autonomous division, leading to recentralization. Multifunctional divisions responding to uncertainties of the environment rather than of the market probably include the extensive occurrence of divisions in the petroleum industry centered on regional crude production. To produce crude, large risks have to be taken in exploration, and large investments have to be made in drilling, crude stabilizing installations, and transportation to refineries. Both exploration and drilling involve great uncertainties about what is under the ground from three hundred feet to as far as three to five miles down. Separate crude production divisions are found in Jersey Standard (the Humble Oil Company, 173); in Standard Oil (Indiana) (the Stanolind group, 355); and in Mobil (Magnolia and General Petroleum, 356). Considerable autonomy is generally conceded to foreign raw materials–producing operations as well as to foreign marketing subsidiaries in metals, petroleum, and some agricultural products companies. Sometimes it is not quite clear how far the presence of both marketing and producing subparts in a foreign operations division is really an integrated operation—sometimes, evidently, the foreign raw materials operations also ship to the United States, and the foreign sales divisions also sell goods produced in the United States. The uncertainty that ties separate production and marketing organizations together administratively, then, seems to come from the political foreignness of their common environment, rather than from the uncertainties of dedicated manufacturing operations being tied to the uncertainties of the foreign markets they are dedicated to.

Conclusion The core variable to be explained in this chapter is the degree to which the information-processing and decision-making systems in manufacturing, marketing, and engineering have to be authoritatively coordinated by creating product-line divisions. The basic explanation has been that when distinct market uncertainties ramify the manufacturing and engineering of distinct products, then divisionalization and decentralization are necessary. But this statement is too vague to be of much use to a student of organizations. By reworking Chandler's data and argument, we have tried to give a more concrete version of this explanation. Let us start at the back end of the chapter, with the social organization of information about market uncertainty. We argued that Chandler's im-

149 plicit definition of a firm as "in several distinct markets" was of distinct social structures bringing in information about market risks that affect manufacturing and engineering. Chandler's real independent variable might be conceived to be (1) the total uncertainty in a given market, including certain types of uncertainty of the environment of production and of technical change; modified by (2) a factor that describes how tied manufacturing uncertainties are to uncertainties in the market; modified by (3) the extent to which the information for adapting to the uncertainties is located in several different social organizations in the marketplace itself; modified by (4) the extent to which these different social structures actually carry different information, i.e., are not mere administrative conveniences such as regional arms responding to geographical distance; modified by (5) a factor that measures whether or not the projection of demand and the inventory risk are taken by the seller rather than the buyer in that market. More formally,

where MD stands for the degree of multidivisional organization, then each variable in parentheses describes a flow of commodities and is a matter of degree, varying between 0 and 1 for convenience. This description gives concreteness to the explanatory side of the general argument. The argument now says that whenever all five of these conditions obtain to a high degree, then the information-processing task of an organization breaks into separable divisional parts. Consequently, these five conditions predict divisions with distinct authority and that such divisions will be units in the profit, investment, and performance measurement systems of the general office, in the case of multidivisional organizations, or of the stock market, in the case of divisions becoming separate firms. Only if divisions do not become separate firms will we need to explain the existence of a general office and of "decentralization" to operating divisions. The theory of market uncertainty and market information segregation does not predict anything about the general office, except that if it exists it will tend to restrict itself to investment, profit assessment, and performance measurement functions, much as the stock market or investment bankers do. In sum, the variable to be explained in Chandler's central argument, as we discussed at the beginning of this chapter, is really centralization

150 within product lines . This only becomes an organizational problem when manufacturing, engineering, and marketing might not be coordinated within product lines—for example, when they are coordinated at a corporation level and the corporation is producing and marketing for several types of markets, as at Du Pont. When this is the case, internal processes may take place in the firm to produce decentralization. Information flows that focus attention, organizational theory that facilitates correct diagnosis, a structure of detachable high executives to study long-range organizational problems, a shifting committee structure allocating responsibility for solving these episodic structural problems—these are some of the conditions that facilitated such an internal evolution at Du Pont. When the problem is instead how to manage investments centrally for a firm that is already very decentralized, as at General Motors, an evolution of more abstract information flows to a general office will tend to be instituted. We argued that the way Durant built General Motors showed that he (and General Motors as a corporation) functioned as an investment service for decentralized car parts suppliers, car assemblers, car sales finance companies, and car dealers. To continue to serve that function, Sloan (and the Du Pont financial crew who supported him) thought they had to build a system of more abstract projections to evaluate internal investments without destroying the operating efficiency of the autonomous divisions already in place. Sears posed a still different problem for Chandler. This company was, in the mail order days, already decentralized (a "coalition of [nineteen or so] merchants") and already had a unified structure to coordinate the commodity-line "divisions." Sears's task, then, was not to build a Du Pont–General Motors–type of structure, because they already had one. Instead, it was to attach an information system that translated the behavior of store departmental (and specific commodity) inventories into information that the quantity-buyer divisional structure could use. Because Chandler misidentified the core of that structure (because in turn he was not looking very hard for the crucial links that ought to have existed, according to his own theory, between the regional divisions and the buyer divisions), he failed to interpret much of the administrative reform he reported at Sears. His central trouble here was not applying a principle he often used in his chapter 7 analysis of manufacturing firms, having to do with whether the distinct regional markets really carried much distinct information to the manufacturing installations or buyers. Our argument here is that commodity line–supplier relations at Sears remained distinct with dis-

151 tinct buyers but that the regional markets by and large conveyed the same information to those buyers. When Chandler's vague connection between ramifying market uncertainty and divisional organization is theoretically elaborated, it helps us to identify difficulties in his treatment. General Motors did not have the same reasons to develop investment expertise in the general office as Du Pont had to develop divisions. General Motors already had divisions. Sears's commodity-line divisions and general office were already in place; their regional divisions did not substantially change this structure, because they did not transmit much information that buyers needed. What Chandler, modified, adds to our theory that formal structure is to be explained as information structures growing toward central sources of uncertainty, then, is a specification of the conditions under which operating information systems will tend to be integrated into divisions. He also gives a few hints about when investment decisions will be taken by a general office that is fed by very abstract information flows from the divisions. We have not had much success in developing a theory of why such decisions are sometimes taken by the stock market and investment bankers, sometimes by an internal corporate general office. That problem is left as an exercise for the reader. In the next chapter we study in detail a particular kind of force that tends to produce divisionalization, namely the introduction of major product innovations. We treat this problem by analyzing how the findings of Chandler on making bureaucracies flexible fundamentally change the antibureaucratic premises on which Schumpeter (e.g., 1942) predicted that innovations would always tend to be introduced by autonomous entrepreneurs.

152

5— Turning Inventions into Innovations: Schumpeter's Organizational Sociology Modernized Introduction The general purpose of this chapter is to elaborate the fundamental distinction made by Joseph Schumpeter (1942) between invention and innovation. Since Schumpeter was interested in the transformation of the economy by the development of new technology, he was concerned only with innovations that could produce a continuous effect in the market, not with inventions that remained visions of isolated discoverers. He made the distinction between invention and innovation to distinguish those new ideas that revolutionized the economy from those that did not. The particular lines that Schumpeter drew between inventions and innovations are not very useful in a modern economy. The basic problem is that Schumpeter thought that anything that could be done by large-scale bureaucratic organizations could be done routinely. Since (according to Schumpeter's argument) it was exactly the nonroutine character of innovations that produced the economic effects of monopoly, entrepreneurial profits, business cycles, and economic progress, this specification of the problem led him to predict the disappearance of many of the relevant economic effects as innovation became routinized (1942, 131–134). Even if innovation continues, he argued, its routinization destroys the entrepreneurial function; this destruction in turn undermines the argument in defense of capitalism that entrepreneurs are needed for economic progress (if innovation is only bureaucratic routine, then socialist governments can carry it out just as well). The routinization of innovation also means that innovations can spread rapidly in the market. This creates a lowered expectation of monopoly profits, which previously motivated excess investment in boom times and so decreases the amplitude of business cycles. The argument below is that building a social system around an inno-

153 vation is still not routine, even though it is (sometimes) done in large organizations. Not all organizations successfully innovate, but when they do monopoly profit positions and a heady flow of new investment are established, because not everybody can follow their lead. Turning inventions into innovations is difficult, because inventions need a particularly complex information and decision system connected to them to become innovations. For example, all organizations must create incentives so that workers will work hard to do what needs to be done; but creating incentives for the personnel who have to solve all the nonroutine problems of a division producing an innovation is a substantially different, and more complex, problem. Innovating workers and managers have to work into the night, go out and help the salesmen when the solder comes loose, help engineers understand marketing problems, and in general be motivated and authorized to take more responsibility than is needed or tolerable in a more routine production operation. This means that an incentive system that works fine for routinized production will not serve for producing an innovation, which in turn means that the information and decision problems for an incentive system in an innovating part of an organization will need special attention. We have already argued that routine industrial administration is not so routine as all that (Chapter 3 above; and see Stinchcombe 1974, 3–39). Below we will argue that there are many structural impediments to be overcome in building the social organization that can carry an innovation—that can, in the phrase John R. Commons (1924) adapted from legal terminology, make the innovation into a "going concern." In the first sections we will ask why it is that innovations create monopolies. That is, we develop variables that determine how long the monopoly created by an innovation is likely to last and, consequently, that determine the size of the economic impact. Schumpeter argued that the monopolies produced by innovation set in motion a disequilibrating cycle of profits, investment, and reestablishment of competition. Under some conditions innovations create larger and more durable monopolies than under other conditions. Longer duration and larger size of monopoly advantage both increase the incentive for introducing an innovation and increase the duration of the market's departure from competitive conditions, making the business cycle effects larger. Many of the variables we identify have social components, or are technical and economic determinants that take social form. For example, it is quite clear that the learning curve of cost reduction after the introduction of an innovation creates a monopoly advantage for innovators, because at any given time they can expect to be further down the learning

154 curve and so to have lower costs than potential competitors. But the actual process of cost reduction seems to be mainly one of introducing minor labor-saving and capital-saving innovations into the production process. Introducing such innovations requires not just thinking of them, but introducing them into an ongoing system of activity in which people's interests and skills are invested. Social arrangements determine whether such labor-saving innovations can be introduced easily. Consequently, the rate of "learning" of a production line has social determinants. For another example, network relations between vendors and buyers that establish the advantages of the current vendor in some product markets have more stability if there are many reasons for the vendor to visit the buyer. In industries in which vendors visit for maintenance or to introduce innovations into the buyer's system, as frequently happens in computer systems, networks will be more stable. An innovator's market advantage partly consists of having stabilized networks that connect it to buyers. An innovator's monopoly advantage will thus tend to be longer lasting when the vendors typically provide maintenance or sell innovations on a system. After this sociologizing of Schumpeter's basic economic observations on the relations between innovation and monopoly, we turn to sociologizing his view of entrepreneurship. The basic purpose of the middle sections of the chapter is to describe what has to be added to a technical idea in order to make it into a continuously producing going concern. To make an innovation viable, there has to be a market, as well as a social structure that can reach that market. Otherwise the innovation will not create a monopoly that will be worth anything. An investment group with control over the necessary resources has to be convinced that the innovation will work, which usually means in particular that it be convinced there is a market the innovation can reach. A manufacturing (or service producing) process has to be created that can produce the prototype of the innovation at a cost only slightly above what people will pay. This production process has to be started down the learning curve so that, to get the item's price, one need not add a profit to the cost at which the first prototype model was manufactured but instead can add a profit to a product the innovator has learned to make more cheaply. The benefits and profits have to be divided up in such a way that the people who as investors make the innovation viable feel they are sufficiently rewarded, that workers will tolerate the introduction of cost-saving improvements in manufacturing, and the like. And personnel adequate to all these purposes have to be added to the technical group from the R&D department to complete the social system for carrying the innovation.

155 This theory of what it takes to produce a social system that can carry the innovation, then, provides the basic outline for a theory of the social conditions of innovation in organizations. It is because that theory has so many variables, because there are so many things that can fail in the process of developing that social system, that introducing innovations is not in fact routinized even in large organizations. In the final section of the chapter we turn to the question of administrative organization for innovation, using the theory extracted from Chandler in Chapter 4 above. The basic observation here is that innovations very often have market uncertainty that ramifies into manufacturing and engineering, and so need to have an autonomous "divisional" organization. Early in building the social system that will carry an innovation, many manufacturing adjustments have to be made in the light of market information, many engineering innovations are involved in solving the manufacturing problems of early production, and many customer service problems of the marketers require information from the engineering or manufacturing staff. Consequently, there is much interdependent uncertainty, requiring rapid information flow among the various parts of the innovation-carrying organization. This is the condition that Chandler argues requires an autonomous, fast-acting, productoriented administration. That is, innovations must be administered with minidivisions. But in large organizations it is difficult to create a special administration whose marketing is detached from other marketing, or whose manufacturing is not under the supervision of the vice president for manufacturing. The special reward system needed to get employees to work all night to get things going, to bet their careers on an innovation that may not pan out, is hard to justify to a personnel department trying to get uniform fair standards instituted throughout the corporation. In short, divisionalization on a large scale was not easy for the corporations Chandler studied, and it is not easy on a small scale for building up an organization to turn an invention into a going concern. As a consequence, routinization of innovation is not administratively easy, and large corporations very often fail at it.

Innovation, the Learning Curve of Cost Reduction, and Monopoly Two main processes account for monopoly advantage: advantages in production, such as cost advantages; and advantages in the market, such as having networks in place through which firm products can be marketed. A central part of the advantages in production, as outlined first by

156 Schumpeter (1942, 81–106), is the temporary advantage a firm has when it has just introduced a successful innovation. An innovation monopoly advantage consists mainly of better production performance than the earliest imitators. Let us start with the cost advantage of an innovating firm. The curves in Fig. 4 show the decline of costs with increasing numbers of units produced (the horizontal axis is the units produced by the innovator). The general pattern of decline of costs has been estimated for the production of airplanes and is known by the name "the learning curve." The basic shape is given by the fact that, for airplanes, the manufacturing cost of the second plane of a given model produced is roughly 75–80 percent of that of the first, the cost of the fourth plane is roughly 75–80 percent of that of the second, and so on, with the manufacturing cost of a unit being cut by between 20 and 25 percent each time the number of units produced is doubled. So the cost keeps going down with increasing production (or with time), but at a decreasing rate. We have imagined that this cost curve describes the experience of the innovator and the first follower. That is, they both learn from experience in production at the same rate, so that, for example, the triangles B and C in Fig. 4 are the same shape. At the time the follower enters the market, the innovator has most of the production problems well in hand and is quite far down the learning curve. Obviously, the larger the distance A

Fig. 4. Cost Reduction Curves with Increasing Production ("Learning Curves") for an Innovator and a Follower

157 (the number of units the innovator has already produced before the follower gets into the market), the larger the innovator's cost advantage from learning at that time and at all succeeding times. We might imagine that at point D , the cost advantage of the innovator over the follower is trivial, so that they are by that point truly competitive. The total size of the monopoly advantage of the innovator, then, is roughly the area bounded by the two curves and the double-headed arrow A , up to point D when the two are essentially even competitors. Between the time the innovator starts production and the time the first follower starts, any other follower may be assumed to be able to start where these two firms started, at some point on A . So up to the point of entry of the first follower, the innovator has an advantage for each unit produced, its size being measured by the distance between its own learning curve and the arrow A . When the first follower enters the market, the innovator's advantage starts to be reduced as the competitor learns to produce more cheaply than a new entrant would. The innovator's loss of advantage is progressive, because the innovator's learning goes slower after many units have been produced, allowing the follower to gain on the innovator. So the distance between the innovator curve and the arrow (before the follower enters) or the competitor curve (after follower entry) measures the cost advantage. The area between the two curves and the arrow is therefore the cost advantage multiplied by the number of units (giving total excess profit up to unit D ) or by the time (giving total excess profit up to time D ). That is, if a follower were to enter the market at the same time the innovator produces its first unit, it could presumably make its own first unit at the same cost, and there would be no monopoly advantage. Both innovator and follower would go down the left-hand learning curve and would have equal costs of production. Since, however, the follower is presumed to enter later, after a period A , the innovator reaps all the advantage of learning. It can still sell in a market where a potential competitor's price would have to cover costs at the top of the curve, even though the innovator has learned to produce the product more cheaply. As the innovator comes down the left-hand curve with no follower, its cost advantage (over a fictional follower) increases from zero on the first unit to a quantity q when the follower starts. As it produces the first unit it has zero monopoly advantage; on the second it has, say, 20 percent, on the fourth 36 percent, over the fictional new entrant, and so on up to the advantage q for the (A + 1) unit. To get the total learning-curve monopoly advantage created by having no entrant up to time A , then, we need to sum up zero times the cost

158 of the first unit, 20 percent of that cost for the second unit, 36 percent of the initial cost for the fourth, and so on, up to q for the unit produced at the time of the follower's first unit. If we imagine a series of rectangles one unit wide, each with a length from the arrow A down to the left-hand learning curve, the advantage got from any one unit will be the area of that rectangle (the cost advantage times a width of one unit). Summing up all the rectangles for the total monopoly advantage due to the follower starting after time A will give the area between the lefthand cost curve and line q and below arrow A (which is at the level of the cost of the fictional early entrant). Similarly, the (A + 2) unit of the innovator (the first after line q ), produced in competition with the second unit of the follower, has a markedly reduced cost advantage. The follower has cut 20 percent (say) off its costs, while the innovator is producing at a stage when it is harder to find production economies of the same size. This would be a cost advantage equal to a rectangle one unit wide and having a length equal to the distance between the two cost curves (extending just to the right of the line q ). This cost advantage is fairly quickly reduced to the distance between the curves at D . By summing up the cost advantage of the innovator at each point, times the number of units to which that advantage applies, we get the area between the curves—between lines q and D —as the total monopoly advantage after the entry of the first competitor. (Complications of different rates of production at different points on the learning curve are ignored here, for the sake of simplicity. They would not change the conclusions.) Clearly, when learning is fast, the monopoly advantage will be larger. This is shown in Fig. 5. Here we have added to Fig. 4 two learning curves with a flatter slope, one for the innovator and one for the follower. That is, if one compares triangles B and E , it takes more added units produced in triangle E to give the same amount of reduction of cost as in triangle B . Learning is slower. It might intuitively seem that the faster followers could learn, the quicker they would catch up with the innovator (one might, for example, guess that learning is very rapid in the computer industry, as competitors imitate each others' innovations quickly). But if we assume faster learning for the innovator as well as for the follower, the head start of the innovator is larger because it has been going down the learning curve faster. Another way to see this is to compare the total cost savings for the innovator up to the point F from faster learning (the area with vertical hatching) with the total cost savings for the follower to that same point (the area with horizontal hatching). If the learning curve is shallower, the total advantage of being farther along is not as great. Obviously in either case, the longer the "lead time" between the inno-

159

Fig. 5. Graph Showing That the Monopoly Advantage from Innovation Is Lower When the Learning Rate Is Lower vator's introduction of the good into production and the beginning of production by the first innovator, the larger the cost advantage of the innovator at any given time and so the larger the total size of innovator advantage. Many of the proposed antitrust remedies against IBM's dominance of the mainframe market have involved demands that IBM give the technical details of an innovation in central computers to possible competitors as soon as these were released to its own peripherals and software divisions, so that at least in manufacturing peripherals and developing software the followers would have an equal start with IBM's own divisions. That is, the proposal was to reduce the distance A in Fig. 4 so as to reduce the monopoly advantage of IBM's software and peripherals divisions (Fishman 1981, 233–245).

Cases in Which the Follower and Innovator Have Learning Curves of Different Shape So far we have noted that a pair of learning curves can be flatter or steeper, or closer together or farther apart on the x-axis (the axis indicating number of units produced or time spent in production). Both affect

160 the size of the monopoly advantage, with steeper curves farther apart being advantageous to the innovator. A third difference in the relationship between the curves is that the imitator may learn faster (e.g., if imitation is not as hard as innovation, or if the imitator can hire experienced labor from the innovator, leading to a steeper learning curve). Even if the imitator approaches the learning curve of the innovator as an asymptote, never producing more cheaply than the innovator, faster learning for the follower reduces the distance between the curves and so reduces the total monopoly advantage of the innovator. Ordinarily, faster learning of the follower will eventually bring the follower to a lower overall cost of production than the innovator. In Fig. 6, the right-hand curve is steeper than the left-hand one, so the follower curve approaches the innovator curve faster than in Figs. 4 and 5. This is shown by triangle G , with its shorter base, indicating that for the same amount of cut in costs (the vertical of the triangle), the follower needs to produce a smaller number of units (or a shorter amount of time has to pass). Because the follower curve approaches the innovator's curve faster, the point at which we judge the innovator advantage to be trivial comes at point H rather than, as before, at point D . Because the follower reaps cost advantages in less time, and because the follower more quickly reaches a point at which the innovator's advantage is trivial, the total area representing the excess profits of monopoly is smaller. For example, IBM apparently learned something that was essential for success faster than Remington Rand (Univac) in the computer market, since Univac started with the innovation and IBM was a follower (Fishman 1981, 29–47). Relatively soon after IBM entered the market, however, it could compete at least equally with Univac. That is, point H occurred earlier than point D (the point of follower competitiveness with identical learning curves for follower and innovator) because of IBM's more rapid learning than Univac's. Obviously, if some component of costs is lower for the follower—for example, a follower in Hong Kong may have cheaper labor, making its total curve lower from the beginning—the advantage of the innovator will be smaller after the competitor starts because the right-hand curve is in general shifted downward. Unless the firm with higher labor costs learns faster (it may be that innovation itself is a measure of the capacity "to learn," in the sense measured by the curves), the Hong Kong producer will eventually have lower costs than the innovator and should be able to drive the innovator from the market. In Fig. 7, the right-hand curve starts lower than the innovator's curve did, because of the lower cost of some factor of production, though at

161

Fig. 6. The "IBM Effect": A Steeper Learning Curve by the First Follower Reduces the Monopoly Advantage of the Innovator

Fig. 7. The "Hong Kong Effect": A Follower with Cheaper Labor Reduces the Innovator's Monopoly Advantage, Eventually Driving the Innovator Out of the Market

162 first it is still above the innovator's curve because production in the cheaper location is not yet routinized. As experience moves the low-cost follower along the learning curve, it becomes essentially competitive with the innovator at point I , rather than at point D as before. Further, with increasing experience the follower in the low-cost location has lower costs than the innovator, eventually perhaps driving the innovator from the market it created. So the size of the monopoly advantage created by the learning curve increases with: (1) longer delay between the first production of the innovation and the start of production by the first competitor (longer A in Fig. 4) and (2) steeper decline of costs with experience in production (steeper rather than flatter learning curves in Fig. 5). The size of the monopoly advantage decreases with (3) the "IBM effect," faster learning by the first follower than by the innovator (steeper right hand curve in Fig. 6) and (4) the "Hong Kong effect," cheaper labor or other factors of production by the follower (as in Fig. 7).

Innovation, the Marketing Network, and Monopoly The market advantage of the "first mover" in general consists in having a more solid relation to the client than latecomers. One main way to be a first mover is, of course, by innovating, but there are others (e.g., getting the first contract with a big client). What we want to do now is specify what determines whether a contact in the marketplace is solid (and consequently creates a monopoly advantage for a first mover) or weak (is easily entered into by competitors). The network advantages that tend to maintain an innovator's monopoly position take three main forms: (1) the vendor in place may develop a multiplicity of contacts with clients and serve multiple functions for them, with each such contact reinforcing the others; (2) clients may become technically or economically dependent on the vendor in place, meaning they would pay heavy costs if they changed vendors; and (3) clients may develop such trust in the vendor that information about that company's product is believed while that about a competitor's product is received with suspicion. The multiplicity of contacts between vendors and clients is increased when the client depends on the vendor for maintenance, for a flow of technical innovations adapted to the installed system, or when the client is involved in continuous buying (including when the equipment is leased from the vendor rather than bought). The responsibility of the vendor for maintenance builds the vendor into the technical system of produc-

163 tion, so that day-to-day operations of the client require continuous interaction with the vendor. This provides many opportunities for the vendor to find out what the client needs and to give advice about how the vendor can satisfy that need, to exchange information informally about the economic relation and about sources of client dissatisfaction, and otherwise to support the capacity of the vendor to sell the next installation. The "maintenance" relation established in the computer industry involves not only maintenance, but also access to a continuing supply of technical innovations. When one leases software from a vendor, one ordinarily gets the rights (without cost) to improvements in the system. Similarly, in the days when clients typically leased the machines, technical improvements in the hardware were often installed free of charge. But whether they are free of charge or not, innovations developed for software or hardware around which the client's operations are already built are more useful to the client than are random innovations. IBM's advantage over other mainframe computer manufacturers in selling peripherals was that any peripheral adapted to an IBM machine had a larger market than peripherals adapted to any other machine; their advantage over the plug-compatible manufacturers was that they could earlier and better adapt their improvements to their own systems, as well as make the systems harder for peripherals manufacturers to plug into. When the client is buying a continuing stream of goods (as when an assembler buys from parts manufacturers) or a continuing stream of services (as when the client leases a building or a computer), continuity of supply is an important consideration. If the present supplier provides continuity, while changing suppliers would create an interruption, then the present supplier has an advantage. Further, the continuous flow of goods tends to be accompanied by a continuous flow of information, giving the vendor an information advantage in future sales. The vendor can in extreme cases make what Oliver Williamson calls "transaction specific investments," which are especially adapted to efficient and satisfactory delivery of what the client wants. A common case is remodeling by the landlord in a leased building, but transaction-specific investments may involve innovations made especially for a client, such as modifications on software for a specific client application. The technical and economic dependence of clients on the vendor is increased when clients' operations are adapted to specific features of the product, such that they would have to reorganize many of their routines in order to go over to a competitor. For example, as long as the only people who use the computer are in the computer center, the costs of going over to a competitor system tend to be small. But when many

164 people in the accounting department use a particular software program for invoicing, accounts payable, and so on, it can be very expensive to go over to a slightly superior or slightly cheaper competitive software system. Since software itself often depends on the hardware, it is often the case that building software into an organization's routines makes the organization technically dependent on the hardware as well. One of the most common forms of technical and economic dependence is moving costs in real estate relations. Because it would cost us a good deal in time and effort, and often a good deal of money as well, we do not easily choose to move to gain small advantages in the real estate market. The costs of moving are even more substantial for a business, for it must move a social system with recalcitrant parts. (Moving a clientele is often extremely difficult; cf., on moving a labor force, Mann 1973.) There are often similar costs in changing equipment. For example, IBM for a number of years simultaneously rented two functionally equivalent memory systems, the "Mallard" and an older one, with the latter costing more per unit of performance. Although clients renting the older, more expensive system were free to change and save money, many kept it simply because finding out about the new one or moving the old one out would have been too much of a disruption (Fishman 1981, 235). If such moving costs are required when switching vendors, the first vendor in the market has a monopoly advantage in the amount of the moving costs, whether it got its first mover advantages from innovation or not. Economic dependence can take the form of special credit arrangements with the current vendor. Because the current vendor may enjoy an information advantage, it may be willing to give credit when other vendors would not. Or the vendor may have got inadvertently into an exposed credit position with a client which another vendor would not rationally get into. Nevertheless, the client now depends on the vendor. An obvious source of stability in vendor-client relations is mutual buying. IBM, of course, buys a great many telephone services, and arranges for its clients to buy many more. Bell was IBM's largest customer for computers when it was a single system, though this relation has been somewhat undermined by Bell's breakup. If either company is being difficult in its buying behavior (e.g., threatening to go over to competitors), it may face the sanction of the other being difficult in its . In continuing client-vendor relations, trust grows out of a history of trustworthy behavior. In many cases the trustworthy behavior is making the system in which the product functions run, which may mean getting the bugs out of that system (e.g., out of the interfaces between the vendor's machine and the rest of the production line). Clearly, the vendor already on the scene has an advantage in having got the bugs out and is

165 more likely to be able to deliver adequate system performance than another vendor. The long-run delivery of such adequate bug-free performance (or minor-bugs-rapidly-fixed performance) creates an atmosphere of trust. In addition, the vendor's representatives can deliver information useful to the client on the vendor's own probable future behavior (e.g., products coming out, early warning of delays in release of those products), on its financial situation, on who to call to get action in the maintenance department, and the like, which may not improve the vendor's behavior but at least renders it more predictable by the client. Insofar as trust is based on predictability, it will tend to grow in an information-rich relationship with the current vendor rather than in a new relationship. Further, because the vendor on the scene is in a continuing relationship itself, its own behavior will in general be more trustworthy. One is less likely to imagine that a continuing relationship can be maintained by lying to one's clients than one is to imagine that one can sell for the first time by misrepresentation. The client may be able to use continuing small sanctions on the vendor to improve its performance, and so to make the vendor in fact more trustworthy (Rogers and Larson 1984). The sum of these advantages in creating trust results, for example, in managers of a company or university computer center using as a clinching argument a phrase like "I feel comfortable with IBM." It is important to remember that one can destroy one's own advantages. Apparently Univac sent out engineers to maintain its machines—engineers perhaps too honest about Univac's problems. IBM sent maintenance people supervised by marketing specialists, who, whatever their other disadvantages as maintenance supervisors, were not excessively honest about technical details they did not know (Fishman 1981, 38, 44). Thus one would predict that the more the vendor-client relationships in a market involve a multiplicity of contacts, technical or economic dependence of the client on the vendor, and developed trust between the client and the vendor, the greater the innovator's monopoly advantage in that market would be. The argument to this point has been that the monopoly advantage of an innovator is also a social system advantage. To gain and preserve a monopoly advantage by being farther down the learning curve, an innovator must first build, then constantly improve, a production social system. To gain a monopoly advantage in marketing, an innovator must build and solidify ties with clients as the first vendor in the market. The next section specifies what kinds of information-processing and decision-making structures it takes to create social systems that can exploit an invention by building and maintaining such monopoly advantages.

166

The Theory or Doctrine of an Innovation In this section we hope to describe what sort of a social system an "innovation" is. Schumpeter (1942) is very clear that an "innovation" with a potential economic impact is quite different from an "invention" that might be patented. In the first place, a number of things that cannot be patented are innovations, such as the marketing in the big Eastern cities of lettuce from irrigated desert farms by packing it in ice and getting through service on the trains. In the second place, an innovation is what John R. Commons (1924, 143–213) has analyzed as a "going concern"; that is, it is a social system that can in fact regularly produce and market the innovation. An innovation, then, is the unit of analysis, and we are talking about things like introducing word processors into an office, bringing out a new computer (Kidder 1981), designing and building a wide-body passenger airplane (Newhouse 1982), developing hybrid seed-corn (Grilliches 1957), and introducing multidivisional administrative structures (Chandler 1962). I have written elsewhere about some of the variables describing administrative innovations, whose essence is social, which determine how much opposition they are likely to meet (Stinchcombe 1986b), such as whether they involve the redistribution of status symbols. These variables described innovations by variations in the impact those innovations had on the interests embedded in the old regime, and consequently predicted whether or not they would be adopted at all. But here we want to describe what kind of social system it takes to make innovations that have already got past this adoption barrier into going concerns. Obviously, the social systems created to administer an innovation differ a lot in size and complexity. When IBM embeds a number of improvements in a new board to slip into an old machine, they are very careful to design that board so that it interfaces with the rest of the machine and with the software having the same structure, so that the maintenance workers can just haul out the old one and put in the new one, and the users do not have to change a single line of code in their programs, appoint or fire anyone, change any human routines, or learn anything new. The innovation requires some modification of IBM's system, but nothing from the user. This is an innovation where the client's adaptation involves only lower costs and fewer mistakes, and it has already been paid for (in the old days, at least) in the maintenance contract that went with the lease. For the replacement board, the value of all the variables we describe below—what one has to do to make the social system attached to the innovation go—is very near to zero for the user; very little

167 social system with this innovation needs to be made to go. So what we will try to describe with the variables below is a series of ways that innovations can depart socially from this ideal-type of an innovation with zero social system requirement, the replacement board. I will organize the description of the social requirements of an innovation around six elements. Since a proposal to institute an innovation (or to implement a decision to introduce an innovation) must have a theory of the social requirements of the going concern, or be forced to develop one as it goes along, I will call each of these elements a theory of the social requirements in question. I have previously used many of these ideas to describe the theory one has to have to create an organization (Stinchcombe 1973). Here I use the same approach to describe the elements that must go into a theory of an innovation for that theory to be sufficient to build a going concern that can create and maintain a monopoly advantage. The complete theory of an innovation, then, requires the following elements: 1. A core theory of the innovation, of what is centrally involved technically in the design of the innovation, and a corresponding theory group (sometimes a "group" of one) that develops and defends this core theory 2. The theory of the investment in the innovation (and the risks and profits theory that justifies it), and the investment group responsible for collecting, safeguarding, and spending the money or other resources 3. The technical-costs part of the theory of the innovation, which is about what it will take to produce the innovation and what that will cost, and a corresponding engineering or other technical-costs controlling group that designs the technical productive system 4. The market or benefits part of the theory, about who wants the good or the effects produced by the technical-costs part, what price will they pay for the benefits, how the organization can reach those who want the benefits, and the concrete sales apparatus or clientele networks that the organization needs to build up for these 5. The division-of-benefits part of the theory, about what promises of future profits or interest payments can motivate the investor group, and about how benefits are to be distributed so as to provide careers for motivation of personnel 6. The personnel part of the theory, and in particular

168 answers to the problems of competence of personnel, trustworthiness of personnel, motivation of personnel, and the arrangements for the corresponding personnel flows.

A "Zero Resources Innovation" Described in Detail Let us describe briefly how these elements enter into our "zero innovation" of the improved slip-in board, but now looking at it from the point of view of IBM, for which this is a "small" rather than a zero innovation. (The details of the following picture come mostly from Fishman 1981, 51–151, 379–411.) 1. The core theory involved in a replacement board is built up out of a series of maintenance problems brought in by service personnel, special fixes in the operating system to get around a limitation in the old board collected from folks who consult on software (some even from IBM's plug-compatible competitors), and perhaps some engineering innovations that came after the original design. There will be a small group of engineers and others who collect these, look them over, design modifications in the board, test them to make sure the "interfaces" are the same and that new bugs are not built in. That is, there is a flow of detailed information about problems with the old board flowing to a group dedicated to improvements of the design of the hardware, a "theory improvement" group. 2. The "investor group" is of course ultimately IBM stockholders, who allow IBM to retain earnings to invest at IBM's internal rate of return. More immediately, the group within IBM responsible for the system will have allocated such "upgrading investment" funds as part of their normal budget, probably financed out of returns from the particular machine being upgraded. IBM does this because they know that one main thing users buy is the reliable operation of the system, which is why IBM sells maintenance contracts. (When technical change is as fast as in the computer industry clients cannot provide reliable maintenance for themselves; see Stinchcombe and Heimer 1988.) A system executive committee of some sort will invest in the upgrading. 3. The technical-costs part of the theory is an engineering cum software theory about how the various glitches can be alleviated by rewiring the board. 4. The market theory of IBM has changed over time, especially in response to antitrust suits. The basic IBM market philosophy is that a cus-

169 tomer buys the service of a computer system, not the hardware. For a long time they would not even sell the client the hardware. Originally the client bought (i.e., rented) the system package, which included the maintenance agreement (which included the upgrading). Part of the system package was general access to all sorts of technical innovations—that is, the client bought a flow of improvements as well as the latest system. Much of IBM's advantage in the market consisted of the faith (assiduously maintained by the marketing person in charge of the account) that an IBM user would not be trapped with a big investment in an obsolete machine. A second part of the market theory is that if IBM makes clients rewrite their software and reorganize their data-processing department when IBM introduces an innovation, that is the time they will go over to Control Data or DEC or some other company. So the interface stability criterion for the new board is part of the IBM market theory. 5. The central part of IBM's division-of-benefits theory is that 20–30 percent is a reasonable expected return on an investment. A second is that the company should not lay anybody off—the worker has a job at IBM regardless of business conditions. Market uncertainty is to a large degree put off onto the suppliers; when demand goes down, IBM is likely to start manufacturing parts that it once bought from suppliers, in order to keep its own workers busy (on this strategy in general, see Sabel 1982, 47–56). Related to the security of employment, however, is the practice of demotion. While employees are more or less guaranteed a job, they are not guaranteed the job they have now or one at the same rank (Goldner 1965). A third part is market criteria dominance in IBM's reward system. At IBM, marketing dominates over engineering much more than at Data General (Kidder 1981). People get rewarded in IBM if their product does well in the market, not if it is only technically superior. The division of benefits is then made up of administration and profit criteria and career-structure rules that are traditional at IBM. 6. The personnel part of the theory at IBM is embedded in the company's general personnel system, by which people's careers organizationally depend on satisfying the users of the systems they sell. So the engineers in IBM will be part of a group dominated by the contact people who service the big users. The contact people, not the engineers, will be close to the managers for the system involved. In IBM the engineers are closely controlled so they give the users what they want, not merely a state-of-the-art product. These engineers, even more than other IBM engineers, are dominated by market people, whose big aim is

170 to hold customers. Bell, for example, in going competitive has imported a lot of IBM marketing people, modeled their new marketing departments on the IBM marketing structure (IBM has systems people responsive to marketing people, who are divided according to type of customer and type of system). Obviously, the personnel also have to understand electrical engineering, some computation theory, and so on. Overall, then, this innovation process makes use of parts of IBM's general structure, which is set up to introduce client-serving innovations on a regular basis.

The Multidivisional Structure of Chandler As an Innovation Let us now go through these elements with Chandler's innovation (1962), the multidivisional structure. 1. We note Chandler's comment about the innovations coming faster if people took time off from operational tasks to study organization, if they had a general tradition of organizational thinking, and if an engineering-rationalizing tradition dominated the organization. These are characteristics of a "theory group" that is likely to develop the sort of theory involved in an organizational innovation. But we also note that the problems could be misdiagnosed if a company had experts from outside without access to the details of the problems. In general, though, adaptive response—or "muddling through"—was a main way of developing a theory of the new situation. For example, if Du Pont executives theorized too much about organization in the abstract, they tended to come up with the traditional theoretically optimal solution according to the organizational theory of the time, namely centralized, departmentally specialized administration. Chandler argues that a theory group with a decentralizing innovation was less likely to develop among older executives and more likely to form among younger executives (which means people about forty to fifty years old, apparently), and it was less likely to form when the organizational arrangements were strongly determined by family ties (though Pierre du Pont and Henry Ford II were leading advocates of the decentralized system). 2. "Investment" for administrative reform has to be the time and attention of authoritative people. Only a few managers are inclined to be reflective, to develop a theory. Sloan was very unusual in this respect. Top managers in general like to live in a bustle of activity, to be on the phone, to pay no more than five minutes' attention to a given subject (Mintzberg 1973). So it is not an easy thing to develop an "investment group" when the time of top administrators is the investment required.

171 3. The technical-costs theory has to break down the idea that minimizing administrative costs is the most efficient approach and also that part-time generalists (the "cabinet ministers" system) are good enough for strategic management. In Chandler's analysis, generalists who also had operating responsibility, especially responsibility for a functional department that was likely to develop goals of its own and a distinctive worldview, were not sufficiently attentive to general problems. Generalists with specialized responsibilities were less likely to depend only on market forecasts and an analysis of the distinctive competence of the organization and less likely to invest without fear or favor in different company activities; instead they tended to favor investments in more of their own specialty. Operative executives, too, were generally likely to invest more money in trying to solve their problems, rather than deciding that their area of operations was so full of problems that it ought to be scuttled. Further on in the technical theory of the innovation the innovators had to decide what sort of abstraction techniques could be built into the information system so that the general office would not find the general contours of the strategic problems buried in a mass of operative detail. A crucial intellectual innovation in all this was to see that committees of generalists are good for making some kinds of ("strategic") decisions, while individuals (with advice from committees subordinate to them) are better for making operational ("routine") decisions. Different kinds of decisions required different administrative structures. The technical-costs theory described by Chandler was embedded in many details about how particular decisions are to be made, what information about the basis and quality of decisions goes on to the general office, what controls the general office puts on those decisions, and so on. In short, as a theory, administrative decentralization was pretty helter-skelter, which is why Chandler had such a hard time explaining exactly what it was. 4. In Chandler's description, the benefits of the innovation were supposed to go ultimately to the stockholders. What the stockholders supposedly wanted was managerial control of investment in the activities of the corporation's subsidiaries by the criterion of long-run profits. Ultimately the innovators had to sell the active part of the board of directors—that is, the active top managers—that the benefits would show up in long-run profitability. The market for the innovation was, thus, internal to the company. 5. The benefits of innovation mostly went into the pot of general corporation profits. There may have been career benefits, such as becoming general officers or having full charge of a division, but Chandler does not

172 discuss this. Sloan has endowed many programs in business schools, so he must have got something out of his work of introducing the innovation at General Motors. 6. The main personnel part of the theory of administrative decentralization had to do with the full-time dedication of central office people to central office problems. The comparable thing in government is the relative role of the White House staff, especially the Office of Management and Budget (playing the role of the general office) versus the cabinet. If we imagine putting the White House staff on top of the cabinet, we have the personnel arrangement that Chandler saw as central to the innovation he was describing: a full-time top staff with general and abstract responsibilities supervising a group of heads of operative units. In general, the hierarchical superiority of the central office tends to lead to "problems of subordinates" gobbling up the superiors' time (see Stinchcombe 1974, 81–90). This had to be avoided by limiting communication between the general office and the divisions to general matters. To put it another way, the innovation theory has to figure out how to make a hierarchical superior in the general office into a "staff" advisor on the implications of strategic policies on operative decisions, rather than a "line" executive. It involves having technical executives in the general office make reports to a committee, rather than making decisions. The crucial personnel question, then, was to create "general officers." The companies Chandler analyzed already had experience producing operative executives, so that was not a new personnel question.

Social Predictors of Success in Introducing Innovations The basic point of the preceding sections is that if we want to describe an innovation in social terms, so as to bring sociological insights to bear on what happens to it, we can use this simple sixpoint outline (theory, investment, technical costs, marketing, division of benefits, and personnel) for descriptive purposes. We will next specify some variables that make it difficult to introduce an innovation. Obviously this bears directly on the question of monopoly, because the longer a successful innovator has the innovation while its competitors scramble to build the social system needed to carry it, the larger the monopoly advantage of the innovator will be. As preparation for this task, let us briefly reconceive the requirements of making an invention into a going concern as a set of information and decision requirements. The first requirement is that the information system be set up so as to

173 receive the information that needs to be processed in all six areas. The most common way this fails is to have engineers or inventors build a system that is equipped only to make technical decisions. That is, information leading to sensible investment, marketing, production management, division of benefits, and personnel decisions is not incorporated into the plan of the innovation, so these decisions get made by default. We will call the failure to satisfy this requirement "technological utopianism" below. The social origin of this deformation of an information and decision system for producing an innovation is that one needs the invention first. The social requirements for producing an invention do not in general lead to a theory adequate to produce an innovation, because the theory built by inventors does not create a structure that allows social information to get through. If the inventing group were good at processing social information and making social decisions, they would likely be managers instead of inventors. That is, the variable we will be describing as "technological utopianism" describes a specific common type of inadequacy of the information and decision system: confinement to technical matters and to administrative structures useful for getting the right technical answer. This deformation is particularly likely to happen in a part of an organization dedicated to introducing an innovation, precisely because innovations tend to start with inventions. It is a deformation because a theory adequate to describe an invention will not produce an information and decision system adequate to introduce an innovation. In the following sections we will develop a theory of the information and decision difficulties that inventions get into, by discussing the information requirements of each part of the necessary social structure and the information problems they are likely to create by ignoring the requirements of the rest of the going concern being built.

Technological Utopianism The basic argument of this section is that a common feature of theories of innovations is that they are seriously incomplete about what it will take to make the innovation a going concern. Let me illustrate this by giving a somewhat idiosyncratic reading of Engels's Socialism, Utopian and Scientific ([1882] 1935). The main thesis of that work, as I see it, is that the utopian socialists want to give a theory of how society might be organized better; Saint-Simon, for example, had scientists running everything. The utopians think of the process of introducing socialism as basi-

174 cally one of teaching people the advantages of socialism, whereupon, being convinced, they will institute it. But Engels says a "scientific" socialism has to specify the social forces that will bring this ideal state into being: specifically, the proletariat, subject to oppression in a capitalist market, and class conflict as a way to organize this group into a political force. That is, Engels's argument involves a theory of what the social system has to be to make it a likely carrier of an innovation like socialism. And that theory does not look like schoolteachers giving lectures from socialist texts. The big first variable about theories involved in innovations, then, is "technological utopianism," by which we mean the degree to which the theory is merely describing the technical way to achieve some end, with no analysis of the social and economic forces needed to bring the innovation into being. A very good example is Apple's scheme a few years back to give one computer to each of one hundred thousand schools. It is not quite, but almost, fair to say that the scheme said nothing about how six hundred children in a school were going to practice using one computer, how teachers were going to plan their science or mathematics or writing curricula around it, how the teachers themselves would learn how to do it, who would pay the costs of their training, which children would be able to do their homework on a computer at home, what it means to have an academic training program that cannot have homework, and so on. There was, in short, no analysis of the investment and investor group, the benefits and who gets them, the training and motivation of the personnel, how the social system relates to who gets promoted in the schools, and so on. It was a very utopian theory, worthy of Saint-Simon or Fourier. Of course, if it was not intended to be a theory of an innovation, but merely a way of distributing samples to schools for advertising purposes, then the technological utopianism of the theory behind it was perhaps not a business disadvantage—merely a disadvantage for the adopting school. A good many technologically utopian theories like this can be found in education, ones not based at all in the realities of thirty children in a room six hours a day for 180 days with one teacher (or five teachers one at a time) and no money for extras. People proposing innovations in education rarely say how many minutes their innovation should take, whether the time should come from the time now spent on arithmetic, reading, social science, or whatever. They are often utopian in the even narrower sense of not even having a technology for how to do it and when. When social scientists interview teachers about why they did not follow up on some jazzy program, the most common response is that after they got

175 their enthusiasm up, nobody told them how to institute it concretely. So the educational theories tend to go skating over the surface of the schools, earning education professors lecture fees (small ones) and ending up being entertainment for teachers who get an hour off to go hear them. The social significance of utopian theorizing is, first of all, that the computer sits in a closet or the reform movement disappears without a trace. The first consequence, then, is failure of the innovation to be truly introduced. It remains a technological utopia, without social flesh. A second consequence is that if the technical solution is a real advantage but is nevertheless embedded in a technologically utopian scheme for bringing it to reality, there tends to be a long period of thrashing about until a viable social vehicle for it is found (a good description of such thrashing about, which has informed the discussion here, is Smelser 1959, 50–179; see also Hirschman 1963, 12–91). In such situations the innovation in a particular organization often depends on some enthusiast working eighty hours a week to put together the pieces to make the innovation go. For all innovations there is a long period of "social debugging" of the scheme for the innovation, with someone needed to carry the scheme through when unexpected difficulties (mostly market, personnel, division-of-benefits, and investment difficulties) arise and when the benefits of the technical innovation are delayed. But there is more muddle in the muddling through for innovations introduced with utopian theories. If the plan has not specified the social forces needed to bring the going concern into being or described the market and who benefits from the innovation, how to divide the benefits that return to the organization, or who invests how much and why they do it, or the training and motivation of personnel, or the measures of success to be used, then the organization has only a "neural itch" of an innovation, not a theory. One generalization in this field comes from Karl Mannheim, in a great book, Ideology and Utopia (1929, 117–122, 211–219, 229–239). The basic argument of that book is that theories of activities already embedded in ongoing practices tend to be disintegrated theories about how to solve particular problems, emphasizing the wisdom of institutions and the value of experience, with no overall scheme of how the whole thing works. Theories associated with movements for innovation ("utopias"), in contrast, tend to cover great areas of social life with one simple theory, one master mechanism to drive it all, involving a cataclysmic crisis or revolution that will quickly create a new society that runs on different principles, with no hard work of building that society up piece by piece involved.

176 To put it another way, the conservatives (he calls them "ideologists") typically believe that no reform can ever work because it would upset all the specific solutions we have worked out to deal with the various problems of coping with human nature, while the utopians or radical reformers typically believe there is no work to do after the revolution, that things will all fall into place. If one takes the advice of checking out the theory of the innovation implicit in the above analysis, and asks the reformer about who benefits, how much it all costs and who pays, who is going to run it and how we can arrange things so that we can trust them, one is likely to be ranged on the side of the conservatives, the "ideologists." Obviously Engels did not think he was a conservative, but the anarchists and other types of utopians did. So our first big social variable about a prospective innovation is "utopian versus scientific," the degree to which the social structure necessary to bring the innovation into being is analyzed in the theory.

Investment Approval In general, an "investment group" is a set of people who have to believe in the innovation enough to decide to put money and time into it. The first variable describing investment is the relation between authority to invest in an innovation and authority to keep running the routines. The core variable is how much harder it is to get authority to innovate than to get authority to keep going. To be sure, the ultimate source of money may not be the group that controls its investment, for instance the stockholders, who never hear of the innovation until it has either succeeded or failed. But the yearly report of the corporation and whatever appears in Business Week or the Wall Street Journal have to be convincing to the stockholders, even though the power to invest has been delegated to a set of managers. An analysis more detailed than the one that may appear in the annual report has to be approved by a management committee high in the hierarchy, a strategic management group. Sometimes the plan for the innovation must go outside, as when the airline that is going to buy the new airplane has to approve the manufacturer's plans and to promise to buy a certain number before the manufacturer can borrow money to make them (Newhouse 1982). Often even quite small innovations involving "capital expenditures" have to go much higher in the organization than does approval of how to spend already budgeted resources. For example, a manager who controls ten thousand dollars worth of labor per day (in 1989 that might be roughly

177 one hundred factory workers) may have to go to a higher-management committee for all capital expenditures over, say, five thousand dollars. The higher management have already said they believe in the theory that explains how the ten thousand dollars per day is spent, but not the theory for new capital above five thousand. In the civil service or a university, the control is often much more minute, so a person with subordinates worth ten thousand dollars per day (somewhat fewer than one hundred because civil servants and professors are more expensive than factory workers) may have to ask for approval by line item of all capital expenditures whatever. In a certain sense, then, investment groups are ideological groups. The reason nonprofits often have tighter investment controls is that almost everything they do is ideological, so nearly all expenditures are "investments." One has to believe in a university before one will invest money in it, even as a client, because one is not buying a well-specified good. New research projects (presumably all innovations) whose total budget will amount to less than a year's salary plus overhead of a leading professor may be reviewed by several levels within the university before going to be reviewed by a body of peers at the National Science Foundation (NSF), while decisions to renew the contract for that leading professor for another year are not ordinarily reviewed at all, or only reviewed if he or she gets an offer from another university. But when businesses are making investments, they conduct themselves much more like nonprofit organizations and universities, with endless committees and discussions among high-level executives about small expenditures. The general point is that a clear functional distinction is made even at the lowest hierarchical levels between "operational" and "strategic" decisions, with the strategic decisions getting kicked upstairs, and innovations are almost always considered strategic if they involve any substantial capital investment at all. The approval of investments in innovations usually takes place in a committee-type decision-making structure. People ask whether they believe the theory involved in the innovation enough to put some freely disposable resources into it, resources not already committed to running the operations on which there is already consensus. The formation of beliefs in a theory quite generally goes on more in discussion groups ("congregations") than in an individual mind as a person reads a memo on the innovation. A second variable in the investment process is the relation between the theory group and the investment group. Ordinarily, an investment group will require more in a theory than a technological utopia. Suppose we examine the difference between an NSF proposal (addressed to an

178 "investment group") and a seminar paper delivered early on in a line of scientific work. The seminar paper is supposed to be a technical utopia, outlining the possibilities of a line of work in a way not much constrained by practical realities of money and time. The NSF proposal, in contrast, will have (1) a specific description of activities proposed (though the concreteness will be to some extent specious, and NSF specifically provides that within reasonable limits researchers do not have to pursue a line of activities that turns out not to be productive, but can invest the resources in another that appears more promising); (2) a budget; (3) a set of vitas of the people in charge; and (4) a statement of how this research will advance the science. The proposal is in a sense a routine extension of the technological utopia presented in the seminar paper into a theory of the social system that will accomplish this utopia, what its benefits will be, and so on. The additions are sometimes called "boilerplate" in the survey research business, are quite often already in standard files in the word processor, and are very often written or calculated primarily by administrative rather than technical personnel. In general there will be a conflict in points of view between a technical theory group and the business group that makes the investment decision, a technical-business conflict (Veblen 1919 makes this conflict central to modern capitalism) about adding to the technical theory enough social (especially financial and market) information so that the investors can judge the viability of the innovation as a social system. The technical group will often resist doing so much "irrelevant" work on the details of administration and finance. A third general variable in the investment process is that investment groups have to trust each other with large quantities of money or other resources. There is a strong tendency for people to trust only those others whose behavior they understand well—people of the same culture, often the same people they see socially (Kanter 1977, 47–63). If one knows seven of a person's friends, one knows whether he or she plays the horses too much or likes women or men who are too expensive for his or her personal income. One consequence is that people of different cultural backgrounds (for example, blacks, women, or foreigners) may have a harder time breaking into an investment group than into a technical theory group (see Stinchcombe 1986b, 255–264). There may be women executives at the middle levels (e.g., department heads and buyers in a department store) long before there are women at the top investment levels. There may be women selling houses long before there are women selling industrial real estate, because the industrial real estate broker is a central participant in an investment group. There may be minority groups in a country (for ex-

179 ample, Chinese in Southeast Asia) who control most of the country's business because they trust each other enough to carry out the investments but do not trust members of the majority (for example, Malays). Social distance between the innovator group and the investor group reduces the likelihood that the innovation will be carried out, because it undermines the process of convincing the investment group of the theory of the innovation. A big variable among investment groups is whether they are already a part of an organization (in the case of the IBM board, for example, the investment group is an integral part of the IBM organizational structure) or whether they have to be assembled from scratch to start a new company. The crucial people in the first case are decision-making executives, usually acting in a committee. The crucial people in the second case are brokers, who go around from one to another investor or approval body with a partially worked out investment coalition and persuade one after another to come in until the coalition is complete (cf. Banfield 1961). If a structure with money available for investment already approves of an innovation, it is ordinarily easier to bring the innovation into being. On the other hand, if the distance one has to go in the internal structure of the organization before reaching those with authority to invest is too long, then assembly of new enterprises, with short distances between investors and innovators, may work better.

Cost Reduction and Manufacturing Improvements Let us now turn to the creation of a technical system to carry out the job of introducing an innovation—the process of coming down the learning curve. As we have already discussed, a crucial thing that creates a monopoly position from an innovation is being ahead of other firms on the learning curve. "Learning" here is the process of developing routines; of using cost statistics and other knowledge derived from experience to improve the efficiency of a productive system; of setting up incentives and social mechanisms so that workers who figure out a faster way of doing the work will let the company know about it (quality control circles as discussed by Cole and Siegel 1979, 134–155, 160–168, 202–211, 222–223; the idea that worker cooperation in technical improvements is more crucial for workers in pilot plants is suggested in Whyte 1961, 198–217); of making opportunistic improvements in capital equipment when it is down for repair (see Stinchcombe 1974, 3–41, for a series of managerial strategies used in a steel rolling mill to reduce down time). That is, learning as described in the learning curve is a set of innovations associated

180 with the initial product innovation that improve the efficiency with which the innovation itself can be produced. This translation of theories into technical action may show the technical-costs theories on which the innovation is based to be mistaken. The engineering theories that allow people to estimate the net benefits of an innovation by estimating its costs are very unreliable, especially when the innovation itself is more fundamental. The theories may, for example, be false or radically incomplete only in a specific environment, as I illustrated earlier in discussing Granick's book on the Soviet Union (Granick 1967, 161–164): a mass production theory of machine assembly lines depends on being in an environment that can standardize parts. If a company's parts suppliers do not have adequate quality control, the company cannot run an assembly plant, but instead needs assembly craftsmen or craftswomen with files and calipers to make the pieces of the machine fit together. The bugs in engineering theories are often in the details, as when a machine tool that is supposed to work to the nearest thousandth in fact works to the nearest two thousandths, or when the clay that looks solid turns to jelly if it gets too wet (and so causes an earth dam to come sliding down the valley). Generally, though, the theory is likely not to work quite right at first not because crucial details were wrong, but because they were simply left out. A big first variable that causes variation in turning technical theories into production systems is the engineering adequacy of the theory—its truth and completeness, how many bugs it has in it. Obviously, then, the speed with which the theory improves and the adequacy of feedback about mistakes are crucial as well. This brings up a second feature of the whole startup process. The engineers and factory managers for an innovation need to be got down into the shop looking at the details. But this means that production workers who are used to running things for themselves have to get used to an engineer or manager always looking over their shoulders. If they have been used to starting work at 8:15, after fifteen minutes of joking and bullshitting, and the big boss comes down at 8:05 to a shop with no machine noise at all, the situation becomes strained. All sorts of illegitimate little arrangements (though they may be agreed to by lower management and tolerated through careful not-noticing by higher management) for making the work easier and more pleasant, the "indulgency pattern" (Gouldner 1954), tend to be undermined. Gouldner found this attentiveness of an innovating management to be a cause of wildcat strikes in a plasterboard factory he studied.

181 When the theory is known to be uncertain, the company quite often builds a pilot plant, say one-quarter of the projected efficient size. In general, the relations of workers to managers, the variety and excitement of work, the chances of promotion, the pressure for production, are better in the pilot plant. Hence, the creation of the actual production system involves making the pilot plant workers give up their close relations to engineers and higher bosses, making them do more boring work, cutting down their speed of promotion, and putting them under productivity standards that are more demanding. This transition from a sort of test laboratory for the theory on which the innovation is based to the production system that is actually going to produce the innovation thus creates strains that are the reverse of those documented by Gouldner when the innovation was introduced on the assembly line that he studied (Gouldner 1954; see Whyte 1961, 202–217).

Markets and Innovation Success The market structure with which an innovation is faced determines how much profit the firm in which the innovation might take place may expect to make. This in turn depends on (a) whether the competitive costs are within or outside the organization—that is, whether the destruction of other economic opportunities, which Schumpeter (1942, 81–86) calls "creative destruction," also destroys economic opportunities and advantages that the innovating organization has (as in the case of the cheaper disk drives that IBM developed under the code name Mallard, mentioned above), or whether only economic advantages held by other , competitive, organizations are affected; (b) whether there is an easy market niche to pay for the period when costs are exceptionally high (cf. Grilliches 1957); and (c) whether the profits of monopoly can be retained by continuous improvements . a. If the costs of competitive damage to unattractive alternatives by the greater attractiveness of the innovation are outside the innovating organization, the innovation is more likely to be introduced. Schumpeter (1942) argued that the central virtue of capitalism was that it could carry out "creative destruction" of technically and economically archaic social structures, primarily because the losses from an innovation occur in a different place than the gains, and in particular because there is no (in actual practice, very little) political communication from one to the other. People hurt by competition with an innovation cannot in general sue those who hurt them for damages, and they cannot stop those benefitting from the introduction of the innovation from introducing it (for a more formal analysis, see Commons 1924, 97–100).

182 We can immediately see some ways in which this virtue of capitalist social structure is limited in the United States, and we would expect lower rates of innovation because of these limitations. (i) Competition may be politically vulnerable. For example, innovations by the Japanese in the steel industry or by Hong Kong in the textile and apparel industry that damage their competitors in the United States are often limited through political constraints on free competition: tariffs, "voluntary" restraints, and the like. When At&T was the common carrier for telephone traffic, they defended their monopoly on the manufacture of transmitting instruments by not allowing instruments made by competitors to be hooked up to the telephone lines. Political vulnerability of an innovator can transfer costs of competitive damages to the innovator by political means. (ii) Benefits may be displaced automatically outside the organization that makes the innovation. For example, no one inside the garbage collection department would ordinarily get the benefit of a more rationalized work plan that saved labor—only taxpayers would be benefited. The "Scanlon Plan" of measuring base-rate efficiency and then giving the workers half the savings from laborsaving innovations introduced at their suggestion (or the rather similar arrangement of the West Coast Longshoremen's Union with the longshore employer's association [Finlay 1987]) retains some benefits for the workers, and so increases the rate of innovation insofar as worker resistance would significantly impede innovation. That is, the Scanlon Plan uses the benefits of innovations within public organizations to motivate innovation. When all costs of the innovation's introduction are costs within the innovating organization, then innovation is likely only if some of the benefits of the innovation can be transferred in, for example by special incentive plans. (iii) Innovations discovered by one organization that would be in the jurisdiction of a competing organization tend not to be developed. For example, navy research rarely develops weapons that would fall under the jurisdiction of the air force, although their technical expertise in aircraft to be based on carriers might make that otherwise likely. If only a competitive organization can exploit an innovation, that innovation is not likely to be pushed. b. One can classify market segments by how easily they can be penetrated by an innovation. Two sorts of market segments are often reached only by reducing the levels of profit from monopoly over the innovation. First, some segments of the market can be entered only at increasing cost, such as smaller ecological niches for hybrid corn or smaller users of computers requiring larger marketing costs than OEMs (OEMs are

183 "original equipment manufacturers," a description of computer brokers invented in government bureaucracies—they sell large systems to large computer users by integrating other manufacturers' parts into a system). Second, some sections of the market do not gain particularly large benefits from the innovation, so they buy only when the innovation is cheap. An innovation, then, may depend for early success on being able to identify and reach those sections of the market that will give quick and lush returns. (i) Government markets, especially the military, are often the first users of expensive innovations. Electronically read Hollerith punch cards (the starting innovation of IBM) were first sold to the Census Bureau; the first computers were sold to the government; the Navy Advanced Research Projects Administration is now a main client for the most advanced computers. Similarly, many transport planes for civilian freight were first developed either for military transport or for civilian passenger traffic, while the electronics for landing and taking off in bad weather was first developed for military use and only later applied to civilian air transport. The government payoff for having the innovation, especially the military payoff, may be very high, making it a lush market for innovations. (ii) Raymond Vernon has argued (1960, 1966) that rich countries and rich cities are more likely to introduce innovations than poor ones. The argument has three parts. First, many innovations can be sold as luxury goods before they become cheap enough to be sold to a mass market. In a world where "rich" feudal landowners in poor South American countries often have about a fifth the income of North American steelworkers, the real luxury market tends to be concentrated in rich countries, and within those countries in the cities where the rich and powerful live, such as New York, Paris, or Tokyo. Innovators in those countries and cities, then, have the best chance of reaching the lush luxury markets that can sustain the costs of the high part of the learning curve. Second, advanced countries are advanced in part because they have developed ways of marketing innovations and of concentrating that marketing in great metropolitan centers; the rich of poor South American countries may often be more easily reached from New York than by an innovator in the South American country itself. The Marcoses and the Shah of Iran used to come to the United States or Europe to spend their billions. Third, the productivity of an innovation in capital equipment is likely to be proportional to the total volume of production to which it is applied, and that total volume is likely to be higher in countries with more production per capita. (iii) Rank, prestige, and the network density of a consumer is associ-

184 ated in many fields with likelihood of adopting an innovation, even when someone else is paying for it (see Coleman, Katz, and Menzel 1957). If a firm or its members are close, in network terms, to the center of the network of consumers of the innovation, then the firm can reach the likely early adopters more easily. Thus we would expect innovations that are made at leading teaching hospitals to be diffused earlier among physicians, or those made at prestigious departments of a scientific discipline to become known earlier by people doing related work. Innovations connected to the centers of consumer networks should find the richest market niches more quickly. (iv) In some lines of business, geographical similarity between the locus of the innovation and that of the adopting unit facilitates adoption of the innovation. Thus, in early modern times a pattern of crop rotation, artificial fertilization, and other innovations whose center was Flanders increased the yield-to-seed ratio (a rough measure of productivity in grain agriculture) from around 4:1 to around 12:1. This complex of innovations spread across the geographically similar rich plains of northern France at a more or less steady rate of about ten to fifteen miles per decade (Slicher Van Bath 1964, 330–333), but stopped at the borders of granite and sandy soils: the hills and mountains of Brittany, the massif central , the Ardennes, and the Alps. Just as many plants native to the isolated valleys of the Himalayas or Ethiopia only spread outside them into protected gardens and houseplant pots elsewhere (e.g., many roses, African violets), so innovations can have a limited radius because they are developed in an environment inhospitable to their spread. Such innovations may not find a niche rich enough to sustain them because they are not fitted for other geographical environments into which they would have to spread (Wallerstein 1974, 1980). c. The monopoly advantage of innovators may be sustained by introducing new innovations that are tied to or elaborate on the first innovation. Boeing, for instance, often seems to come down the learning curve to manufacture a new airplane faster than its competitors by introducing production innovations more quickly. IBM manages to continue to sell its computers partly by making it easy for its customers to get new software and peripherals that will increase the productivity of their current installations. Hughes Tool in the oil business can sell innovations better than many of its competitors because it can service the tools anywhere in the world through an existing service apparatus (Boeing has the same advantage). All these companies can market their secondary innovations relatively easily because the links they have with their customers have induced

185 those customers to trust the vendor. Such trust is particularly crucial in selling an innovation, because the client cannot turn to alternative sources—such as competitors—to check out the vendor's sales information. Put another way, companies that have innovated successfully in the past are most likely to have access to that part of the market that will buy innovations. The general point here is that features of the information-processing and decision-making structure attached to an innovation can determine the market success of that innovation. If that social structure does not have to take into much account the losses created by investments in the goods it replaces; if its competition is not protected by tariff or other barriers; if the benefits of the innovation do not automatically accrue to clients; if the innovating social structure's contacts are with the richest markets in government, in metropolitan cities, in innovative and educated groups, in the geographical area where the innovation applies; if the innovation has a stream of other innovations connected to the first one—then the innovation brings prosperity. If the innovating social structure's market contacts are poor, isolated, behind a prohibitive tariff or other barriers, and uninterested in buying improvements, the innovation fails.

The Division of Benefits The market theory of an innovation says that there will be benefits to the innovating organization that must be divided up, but it does not say how. At least in economic organizations, the basic notion is that the total benefits that come back to the organization itself (the appropriable benefits) have to be divided up to motivate the activities that go into the innovation. I will analyze this under two headings: (d) whether rewards are closely tied to the growth rate of one's section or organization, and consequently generally whether the innovation is a new product (growth inducing, hence rewarding) or labor-capital saving (decline inducing, hence punishing); and (e) whether there are powerful residual claimants on the profits or benefits of the innovation—people whose interest is in reducing costs and taking the profits out rather than in motivating and rewarding everyone generously out of monopoly benefits. d. Growth dependence of rewards is actually a complex variable. It takes a positive growth in the size of the innovating organization, plus some "firm-specific human capital" or "job rights" to tie the rewards to the people already employed in the enterprise, to produce from an innovation career advantages for presently employed people. It is because civil servants in the garbage department or West Coast longshoremen

186 (Finlay 1987) cannot be fired to allow the work to be hired out more cheaply that the Scanlon Plan or the sharing of the benefits of labor-saving changes by longshoremen motivates acceptance of innovations. If the workers expected that as soon as the innovations pushed their wage rate above the market rate they would be replaced by new workers at the market rate, they would not be inclined to support innovations; hence, job rights in the innovating organization are essential to turn growth into an incentive for workers. If a whole labor market area is involved in the innovations, as Houston was (once) for offshore drilling in the Gulf of Mexico, then engineers may participate in the boom even if their own firm goes under. In such cases the job rights are in the industry rather than the firm. In some cases, labor-saving innovations may produce enough of a competitive advantage to maintain positive growth of the labor force, and so to maintain job rights—for instance, Japanese and Korean steel plants have at times expanded production when the world steel market as a whole was in glut. Product innovations may not produce growth if they are nonpatentable and easily imitated. In general, then, the connection of innovation to growth of the firm is through monopoly, either monopoly of lower production costs through labor-saving innovations or monopoly of the market niche through product innovations and established market position. And the connection of growth of the firm to career rewards for workers is through worker control of their jobs, which in turn can come through indispensability, through labor union power, or through civil service regulation or other legal or customary tenure provisions. These interacting forces on career incentives for building an invention into an innovation are outlined in Table 2. e. "Residual claimants" are those people who take up the excess of revenues over costs, and in particular those who can be expected to get most of the benefits of establishing a monopoly through introducing an innovation. These residual claimants can range in power from the anonymous and disorganized clients of a city's garbage system, who will pay one-tenth of a mill lower tax rate if garbage workers work more efficiently, to the family-owner majority of Du Pont's board of directors in the days Chandler was talking about. The basic and easy generalization is that the more powerful the residual claimants are, the more likely a profitable innovation will be introduced. The probability of an innovation's succeeding may be drastically increased by changing the residual claimants—for example, when the developers of an innovation leave the firm in which they have developed

187 Table 2. The Impact of Innovation Type and Labor Market Control by Workers on Worker Career Motivation to Favor Innovations



Innovation Type



Labor Saving a

Growth Inducing b

"Free" Labor Market

Mechanization producing unemployment (e.g., in farm labor)

Spreading career benefits in boom towns (e.g., Houston) or replacement by cheap labor.

"Internal" Labor Market

"Featherbedding" or "deals" in West Coast longshoring; effective union resistance to innovation

Career rewards for innovations (e.g., Data General)

a Without large compensating competitive impact.

b Product innovations and labor saving innovations with large competitive impact.

the competence to introduce an innovation, to exploit that competence in a firm of their own. Leaving is especially likely if the company where the developers were employed has already disapproved the innovation. Many large computer and software firms were started in this way (for example, the Data General firm analyzed by Kidder 1981; for software, see Stinchcombe and Heimer 1987). Passing out equity interests to leading technical innovators is one way computer firms have tried to stay at the leading edge of technology. Merton has argued that the capacity to establish priority through publishing makes the person who publishes the residual claimant for innovations in science—that is, the one whose fame is increased by the innovation. His argument is that priority established by publication increases the motivation to carry out an innovation (Merton [1942] 1973b, 273–275). Joseph Ben-David (1968– 1969) has extended Schumpeter's argument above by arguing that prestige competition among universities is crucial for the development of leading scientific nations, for then the costs of older discoverers whose fame is decreased are displaced outside the university introducing the innovation. He holds that nations with many regional universities, such as Germany and the United States, tend to be more scientifically productive, particularly in new fields, than countries with more centralized university systems, such as England and France.

188

Examples of Incentives for Innovation We can specify which ends of these marketing and career reward variables predict easy introduction of an innovation by considering the account given by Tracy Kidder (1981) of the design and introduction of a new computer at Data General. (a) DEC VAX, the competitor to the machine with a soul, takes the losses if the Data General computer comes up to its promised level of performance. Contrast this situation with the introduction of a new computer at IBM, where most of the users who buy the new machine will be replacing an IBM machine. (b) The original equipment manufacturers (OEMs) provided an easy market for technically better products, which they in turn marketed as systems to the military and other large users; Data General could start making money on the innovation before any general marketing was required. (c) A series of "downward compatible" and "upward compatible" related models, improved software, and the like enable computer companies in general to maintain their monopolies created by providing an element crucial to the whole system, such as a main computer, so Data General expected to benefit from the innovation for a long time. (d) Career rewards for computer folks are in general tied to their firm's growth rates. As we suggested above, innovations in the product itself tend to spur firm growth, whereas innovations on the production line tend to depress growth. The Data General innovation was a product innovation, and most of the people Kidder describes as being crucial in the innovation were employees rather than owners or stockholders; they could thus expect their rewards to depend on growth in the size of the labor force rather than, say, on increased profits from cost reductions. And (e) the powerful group in the company was the group who would collect what was left over of the benefits of the innovation after costs were deducted—the residual claimants. The chief executives at Data General were large equity holders and had killed the last innovation, one carried out by some of the people who implemented the innovation studied by Kidder. That is, the control over the introduction of this particular computer was not in the hands of those most committed to it and most motivated to see that innovation introduced—the engineers who worked on it. The top executives did not care whether this one or the one being developed in a competitive company group at another installation in North Carolina was ultimately put on the market. The technical group had to convince the money group. But once convinced the innovation would be profitable, they were strongly motivated to introduce it. In contrast we can consider a labor-saving innovation in schools, such

189 as teaching two hundred children at once by television. (a) Competitive costs are paid by teachers who are laid off. (b) There is no special market for easily automatable classes; students will not flock to required lecture-demonstration classes such as civics, American history, and English because they are automated and so cheaper. (c) Schools are already virtual monopolies, and there is no monopoly advantage of a more efficient school that could be maintained by spreading the innovation to related areas, because the school already has all the monopoly it will ever get by law. (d) Saving labor reduces the growth rate of the school labor force, and so reduces the number of supervisory or other lucrative and prestigious positions that will be created. And (e) there are no powerful claimants of the excess profits; any benefits go to the taxpayers. But since the constraining law on schools is that they should have equal expenditures , not equal effects, if they save money one way they will probably have to spend it another and there will probably not be any benefits for the taxpayers anyway. It would be difficult to set up a social system less hospitable to laborsaving innovations than American elementary and secondary schools, and it would be difficult to set up a situation more encouraging to product innovations than that described by Kidder for Data General—except perhaps by making the engineers in the innovation group shareholders in the company.

Divisionalization and Innovation Innovations not only require a complete social system with markets, division of benefits, personnel policies, and the like, just as any manufacturing enterprise does; but they require rapidly adjusting social systems as well. Because the market for an innovation is untested, new information about what will sell and at what price is always coming in, and manufacturing and engineering have to be adjusted to that rapid flow of new market information. Marketing people selling machines need, for example, machines that will actually work, and if the machines have been improperly soldered at the factory (because the soldering is also a procedure carried out under new conditions for this innovation), they need to be able to call on engineering and manufacturing people to come out and solder them correctly at the client's place of business (see Fishman 1981, 43). Since getting production costs down requires introducing innovations on the production line, and since innovations on the production line also have bugs, engineers have to come down and take a hand in manufacturing. That is, the introduction of a product innovation typi-

190 cally requires the sort of intimate relationship between manufacturing, engineering, and marketing that Chandler argued produced the requirements for divisional organization in Du Pont, General Motors, and other chemical, electrical, and automobile manufacturing organizations (1962, 52–162, 363–378). Of course, Chandler was usually arguing that once a corporation began to market to distinct markets it would be under pressure to introduce a divisional organization. Here we are asking not whether an innovation that is successful will cause divisionalization of the corporation, but whether an innovation will be introduced successfully in the first place. What interests us, then, is not why the chemicals industry, which often introduces innovations requiring responses to new markets, is divisionalized, but rather what it is that makes the chemicals industry capable of introducing many innovations. The functional argument of the last chapter can however be reversed to say that it will be easier to build a system that can introduce an innovation if it is easy in a given organization to organize a new division, largely autonomous from central manufacturing, marketing, and engineering, so as to allow rapid response as the news needed to make the new social system go comes in. We noted in Chapter 4 that the Du Pont divisionalization was based partly on an analysis of Du Pont experience in introducing innovations. Let us quote again from Harry Haskell of Du Pont (in Chandler 1962, 68) on why a minidivisional arrangement was needed for the dyes business, which was at that time an innovating organization, before Du Pont became generally organized into product divisions:

It may be that it would be better for a few years to carry on the dye business as a separate entity. I think it would because it is a developing, unstandardized industry and should merit independent attention just as the Parlin chemical mixtures business was better by itself until standardized—when it was merged with the regular sales and operating departments.

And then the conclusion of the committee based on this analysis (Chandler 1962, 70):

[Dyes should have one] individual in control of both production and sales, because the relation of the product and its qualities is so mixed up with the demands of the market for the product that to divorce them and segregate the business into a clearly defined production

191

department and an independent sales department, would be detrimental to the business. Later on when the production of dyes becomes standardized it will no doubt follow the evolution of other portions of the business.

In the final analysis, then, the dyes business required what Chandler calls (when it happens later) a divisional administrative structure because it is carrying an innovation. But this means not only that Du Pont had the category of division in its organizational analysis before it introduced divisionalization on a large scale; they were also used to introducing autonomous divisions to manage product innovations, and although this was treated as an exception to normal good centralized management, it was a "normal exception." Presumably the causal relation between innovation and decentralized minidivisions goes both ways. Innovating companies learn to create autonomous minidivisions easily; companies that create minidivisions easily are more successful at innovation. Stated more generally: one can easily create a new social system to fit a given niche only if allowed considerable autonomy, and larger social orders that allow considerable autonomy to new social units are therefore advantaged in introducing innovations. This is one of Schumpeter's main arguments in favor of capitalism—that it allows innovating entrepreneurs enough autonomy to produce economic progress. Thus Schumpeter's advice on how to organize an economy can be turned into advice to bureaucratic corporations: only routinize and subordinate an innovating minidivision of a large bureaucracy when its adaptation to its niche and its viability as a going concern are established.

Conclusion Our first task in this chapter was to show that the monopoly advantage from innovations, which Schumpeter saw as the driving force of capitalist progress and of business cycles, is increased and decreased by social structural variables. The profitability of innovations depends on how fast the innovator and potential competitors come down the learning curve, and on how solid the network connections an innovator builds to its clients are. Our second task was to show that turning an invention into an innovation—into a going concern that can regularly produce benefits for an innovating firm—is in fact creating a social system. A social system has to meet its functional requirements in order to produce benefits over the

192 long run. This means it has to nurture technical ideas, to make investments in risky situations, to build a production system that can produce effectively and can come down the learning curve rapidly, to reach the market that can afford the innovation while it is still expensive, and to arrange the division of benefits so that both investors and personnel will be motivated to develop the competences needed to do all these things and then do them. In short, because routinizing innovations is still a difficult and risky process, even in large bureaucratic organizations, it still produces excess profits when well done, creates investment booms as followers try to catch up, and induces creative destruction of archaic social forms during the recession that follows the investment boom. Furthermore, since there are special problems in administering viable innovative social systems within large bureaucracies, in that adaptation occurs on a shorter time scale than in more routine production and marketing, divisionalization serves in part as a social process for making innovation possible; it is the organizational creation of entrepreneurial social structures, taking the place of the heroic individual in Schumpeter's model. In one sense, this chapter is merely a specialized application of the theory developed in previous chapters, especially Chapter 4. It says that in introducing innovations, market uncertainty ramifies into manufacturing and engineering and so tends to produce divisionalization. But turning an invention into an innovation also involves the special task of creating information-processing and decision-making routines. It therefore has distinctive pathologies. As an invention, for example, the normal state of the information-processing system is that of a "technological utopia." In general, the people who developed this technological utopia are not good at collect ng and process ng the nformat on needed to sat sfy an nvestment group, know ng tt e about "bus ness p ans." They may we not know where the ucrat ve markets are to be found (un ess they deve oped the nnovat on for an eng neer who s a so a co one n the Pentagon). They may be bad at est mat ng how fast the bugs can be got out of a product on ne for the product. And so on. When we descr be what s ack ng n a techno og ca y utop an p an for ntroduc ng an nnovat on, we are descr b ng the nformat on-process ng requ rements for a new product- ne d v s on. Each of those nformat on cr ter a n turn requ res sk s n the peop e (as ana yzed above n Chapter 2: nnovat on s a s tuat on of ow rout n zat on and hence requ res a h gh average sk eve ) and the bu d ng up of a sorts of departmenta nformat on systems (as ana yzed n Chapter 3), respond ng to the d st nct ve techn ca uncerta nt es of the nnovat on. For

193 examp e, the cost-account ng system needed to f nd the cost bott enecks n a product on ne for an nnovat on wou d have to be much more f ex b e than wou d the cost system serv ng the overa f rm n wh ch the nnovat on s ke y to be embedded. In Chapter 6, we w ana yze some of the d st nct ve prob ems of bu d ng a system to produce nnovat ons through networks of contracts and subcontracts (see a so St nchcombe and He mer 1988); and n Chapter 7 we w focus on the nformat on system that has to be embedded n the ncent ve system for workers. As shou d be becom ng c ear, what th s book s deve op ng a theory of s the th ngs that have to be added to a techno og ca y utop an theory to turn t nto an organ zed go ng concern.

194

6— Organizing Information Outside the Firm: Contracts As Hierarchical Documents Introduction Coase (1937), Dah and L ndb om ([1953] 1976), W amson (1975), Teece (1976), and L ndb om (1977, 27–29, 237–309) have a bu t and used a contrast between market transact ons among f rms and h erarch ca adm n strat on w th n f rms. The bas c not on s that when many ad ustments w have to be made dur ng the course of contract performance, the transact on costs of negot at ng and enforc ng contract r se, and the great f ex b ty of a abor contract used to create a h erarchy saves transact on costs (see W amson 1975, 64–72). Thus, whenever t s d ff cu t to spec fy the requ red performance n advance (Marschak, B ennan, Jr., and Summers 1967, 64–72), when the costs, pr ces, or quant t es to re gn at the t me of the performances are uncerta n (Macau ay 1963), when team nterdependences do not a ow separate measurement of performances (A ch an and Demsetz 1972), h erarchy s preferab e to market coord nat on through contracts. The argument of th s chapter s that contracts are often s gned between a corporate c ent and a corporate contractor when th s theoret ca trad t on pred cts h erarch ca ntegrat on. A though research and deve opment n commerc a fe s ord nar y carr ed out by a subord nate R & D staff, as Mansf e d et a . (1977) pred ct, [1] the government buys weapons R & D by contracts under the same cond t ons of uncerta nty of performances. [2] Uncerta nty about costs, pr ces, and quant t es frequent y eads to vert ca ntegrat on, as Thompson pred cts (J. Thompson 1967), but automob e franch ses and weapons procurement often nvo ve contracts for sh ft ng quant t es, uncerta n costs, and pr ces to be determ ned (see Macau ay 1966; and Scherer 1964). Team performance of techn ca y nterdependent product on often eads to h erarch ca contro s, as pred cted by A ch an and Demsetz (1972), but a r nes penetrate deep y nto the techn ca work of a rp ane manufacturers. [3] And n rushed mega-

195 pro ects n energy product on, the nt mate techn ca dependence of eng neer ng and construct on does not prevent the r be ng sp t between contractors. Performances can be ad usted to chang ng s tuat ons by contractua means; adm n strat ons of performances can be set up by other k nds of contracts than abor contracts. If the easy way to get f ex b e cont nuous performance over t me s w th a h erarchy so ated from d rect market processes, we need to ask how peop e manage who are forced by the r s tuat on to do t the hard way. The terature that pred cts why, for examp e, weapons deve opment or bu d ng construct on w be vert ca y ntegrated shou d pred ct what k nds of d ff cu t es weapons and construct on contracts w run nto. Then we can ook for attempts to so ve such prob ems by wr t ng adm n strat ve prov s ons nto the contract. What we w be ook ng for s ways to construct soc a structures that work ke h erarch es out of contracts between ega y equa corporate barga n ng agents n a market. In the next sect on, we w ana yze very br ef y what k nds of soc a re at ons h erarch es cons st of, and what funct ons they perform. Th s then prov des an out ne of what the funct ons are whose occurrence n contractua s tuat ons w ead to h erarch ca e ements n contracts. It may a so suggest the nature of the structures that may ntrude nto the dea type of market transact on, produc ng a var ant of the dea -type contract. The dea -type contract s def ned here to nc ude a market transact ons n wh ch one f rm or person makes an offer and another accepts. An offer and acceptance create ega ob gat ons, whether or not a wr tten contract s s gned. In the dea -type contract, the performances the offer ng party offers to perform are c ear y spec f ed, and the performances t w requ re n return are c ear y descr bed, though these may be mp c t. Thus, a store s d sp ay of papayas w th a s gn, "Papayas, $2.39 each," s an offer, and carry ng them to the cash reg ster and pay ng the money s an acceptance, creat ng a contract. Th s contract s hard y ever wr tten, except as a cash reg ster rece pt. The papayas are spec f ed by be ng on d sp ay; the performance of the buyer has a mean ng def ned by aws of ega tender and sa es tax aws and by the s gn. Lega ob gat ons created nc ude the store s ob gat on to et the customer have the papaya and the customer s ob gat on to take t away, to suffer the consequences f t s dropped on the way home, and the ke. A contract w th h erarch ca e ements departs n many ways from such an dea type; those br ng t c oser to the descr pt on of a h erarchy we offer n the next sect on. Some cond t ons make t d ff cu t, uneconom ca , or mposs b e to spec fy the performances to be requ red at the t me a contract s s gned. Our argument n the th rd sect on s that d ff cu t es n arrang ng market transact ons n the way spec f ed n the dea -type papaya-buy ng con-

196 tract tend to cause the add t on of h erarch ca e ements to the contract. These may be d v ded broad y nto (1) d ff cu t es of pred ct on of spec f cat ons the c ent w want to make of a contractor s performances (e.g., one cannot descr be what weapon one wants manufactured unt one has done the deve opment work, yet one wants to start too ng up for manufactur ng ear y so as to reduce the t me between deve opment and operat ona ava ab ty; Marschak, B ennan, and Summers 1967, 49–139; see a so Stockf sch 1973); (2) c ent or contractor uncerta nty about the costs of carry ng out the performances, resu t ng n a w sh to make strateg c read ustments e ther n the performances or n the compensat on dur ng contract performance; and (3) nab ty to measure c ear y the performances to be demanded or the cond t ons determ n ng compensat on (e.g., the nseparab ty of eng neer ng m stakes and construct on neff c ency n cost overruns n ma or construct on pro ects means that eng neer ng and construct on performances cannot be measured separate y, even though they are often n separate contracts. The "markets and h erarch es" terature assumes that the usua opt ma way to arrange work s through a market. Hence, what has to be exp a ned s "market fa ure." S nce most work by nd v dua s s not arranged d rect y through a market, there s a good dea to be exp a ned. The exp anat ons of market fa ure most often have to do w th fa ures of nformat on (uncerta nty), monopo st c advantages of e ther the buyer or se er (sma numbers n a g ven market), or d ff cu ty n measur ng outputs or commod t es ("teamwork," etc.). Hence they pred ct that uncerta nty, sma numbers barga n ng, and teamwork shou d produce non-market arrangements such as vert ca ntegrat on or h erarchy. Our genera argument takes the form of f nd ng s tuat ons n wh ch work s arranged through contracts (and hence through markets) that are character zed by uncerta nty, sma numbers barga n ng, and teamwork. The rece ved theory pred cts h erarchy, but we observe contracts. We w show that h erarchy can be arranged through contracts, hence "through the market." Sect on 3, then, sets a ser es of funct ona requ rements that var ous contract dev ces w be ca ed on to so ve. The genera argument of the terature s that h erarchy s a genera -purpose structure for fu f ng these funct ons, for ad ust ng performances to an uncerta n future f ow of events. But s nce we w observe these funct ons be ng requ red n parts of the economy that are n fact arranged through contracts, there must be contractua funct ona subst tutes for h erarchy, perhaps ord nar y of ower eff c ency but ca ed forth by the spec a c rcumstances of part cu ar ndustr es.

197 If h erarch es ord nar y do some th ngs better than contracts, we can ook for those contractua dev ces that s mu ate the operat on of h erarch es to do those th ngs. Sect on 4 be ow decomposes the funct ons served by h erarch es n ad ust ng to a chang ng wor d, n order to spec fy what contracts have to do to s mu ate h erarch es. We w argue f rst that contracts often spec fy author ty systems, n the sense of spec fy ng how one w recogn ze a commun cat on requ r ng a change of performance as b nd ng. In part cu ar, f a c ent spec f es a change n contractua cond t ons n m dstream, t w ord nar y have a so e ther to assume the r sks th s exposes the contractor to or to compensate the contractor for assum ng the r sk. In short, to spec fy the r ght to change the contract, one must ord nar y spec fy many aspects of the consequences of ssu ng an author tat ve commun cat on. To back up th s author ty, contracts often spec fy an ncent ve system, n the sense th s has w th n the f rms: a method of ty ng rewards to performances that does not g ve a the revenue due to better performance to the emp oyee. In order to ach eve max mum f ex b ty, the c ent shou d reserve the r ght to change the ncent ve system as cond t ons change. W th n a f rm, for examp e, one may get a ower comm ss on for se ng popu ar goods, bonuses for mov ng th ngs that turn out to be turkeys. S m ar y, a contract may spec fy how the compensat on scheme may be changed. To contro uncerta nt es n costs and pr ces, contracts often prov de an adm n stered pr c ng system. Th s may n some cases mean an author tat ve way to determ ne the market pr ce; the "adm n stered" aspect n such a case s then on y that the part es to the contract are n the future requ red to accept the market pr ce so determ ned, whether they want to or not. But the dev ces may be much more comp ex than th s, as for examp e n the Eng sh system of b dd ng on b s of quant t es n construct on pro ects. The quant ty surveyor descr bes the pro ect as a st of quant t es of var ous k nds of work (such as cub c feet of concrete foundat on to be poured). The contractor b ds a pr ce for each quant ty n the st, wh ch then spec f es a un que pr c ng system for change orders for that part cu ar contract: they are descr bed n the same way by the quant ty surveyor, as a st of quant t es, and pr ced accord ng to the pr ces or g na y b d by the contractor. A h erarchy prov des for reso ut on of conf cts w th n the f rm, w thout rout ne appea s to a court. S m ar y, contracts often prov de a system for reso v ng d sputes, somet mes w th severa ayers. Eng sh construct on contracts, for examp e, often prov de that a named eng neer (presumed to have nonpart san profess ona standards) w reso ve a d s-

198 putes between the c ent and contractor for the nter m so work can proceed, but that the eng neer s dec s on can be appea ed to a named structure of b nd ng arb trat on. These structures, nterna to the contract, are sub ect to the usua appea s to the courts (as are dec s ons n h erarch es n f rms), but they are ntended to serve the funct ons of keep ng d sputes from harden ng nto expens ve and d srupt ve ega batt es. F na y, h erarch es estab sh standard operat ng procedures for organ zat ons—to secure eff c ency and d spatch n process ng organ zat ona matters, to ensure that unobservab e qua t es n the output w be regu ar y ach eved by qua ty contro , to prepare for unusua dangerous events such as f res, and so on. S m ar y, contracts for arge pro ects often nc ude a schedu e (whose ega status s n genera precar ous), orders for components of nuc ear p ants often spec fy a ser es of qua ty contro documents and X-ray f ms to be supp ed for each we d, and mar ne nsurance contracts mp c t y spec fy standards of read ness for emergenc es ("seaworth ness"; see He mer 1985a). A structure w th eg t mate author ty; w th a man pu ab e ncent ve system; w th a method for ad ust ng costs, quant t es, and pr ces; w th a structure for d spute reso ut on; and w th a set of standard operat ng procedures ooks very much ke a h erarchy, very tt e ke a compet t ve market. Yet a these features of h erarchy are rout ne y obta ned by contracts between f rms n some sector of the economy. Sect on 5 be ow asks what t means for the theory of the market that one common use of the market s to set up o nt adm n strat ve structures between a c ent and a contractor, rather than to trade spec f ed performances for spec f ed compensat on. That s, the mode of the market n Coase, W amson, Teece, and others s not a descr pt on of the construct on contract ng market, the market for weapons R & D, the market for the serv ces of franch sed automob e dea ers (see Coase 1937; Dah and L ndb om [1953] 1976; W amson 1975; Teece 1976; and L ndb om 1977); nstead t s an abstract on from the econom cs textbook, n wh ch the adm n strat ve ngenu ty embedded n the contents of contracts has d sappeared. But when h erarchy s def ned by contrast to th s dea zed vers on of the market, ts features are ndef n te. A common danger of the dea -type method used n th s terature s that one of the types, usua y the most nterest ng one, s def ned res dua y, by contrast w th an empty dea -type nto wh ch few emp r ca observat ons fa . Th s means not on y that ntermed ate cases are m sana yzed, but that even the po es of contract and h erarchy are poor y def ned.

199

An Extended Definition of Hierarchy We w descr be be ow how a ser es of funct ons that are not nc uded n the s mp f ed dea -type market contract can be ach eved through mod f ed contracts. We need to show that they are n fact rout ne y ach eved by those structures we usua y ca h erarch es. It w therefore be usefu to show how h erarch es themse ves, created n the usua way by charter ng corporat ons and by dec s ons of boards of d rectors, carry out the funct ons that can be done by contracts. We w try to show how corporat ons create a structure w th a work ng author ty system, and an effect ve ncent ve system w th observat ons of nd v dua s performances and rewards for d fferent a performance, accepted operat ng procedures, eg t mate d spute reso ut on structures, and nterna pr c ng systems that contr bute to organ zat ona rat ona ty. F rst we w try to spec fy the e ements out of wh ch these funct ons are constructed w th n f rms, the forma e ements of h erarch es. Then we w try to descr be br ef y how the structure so constructed dea s w th uncerta nt es of spec f cat ons, uncerta nt es of costs, and unobservab ty of performances. Our ma n purpose here s to cast our trad t ona know edge of organ zat ons nto a form that descr bes ts re at on to the contractua mater a s we w go through.

Formal Elements of Hierarchies We w argue that most of the h erarch ca ntra-f rm structure we need to ana yze s made up of f ve e ements: (1) abor contracts y e d ng sub ect on to author ty systems by emp oyees; (2) f duc ary re at ons, espec a y nvo v ng boards of d rectors, entrust ng them w th w de d scret on as representat ves or trustees of the stockho ders and others; (3) the ega persona ty aspect of organ zat ons that def nes them as or g ns of dec s ons, a ow ng them to set up performance measurements and wage, sa ary, pens on, career deve opment, or other ncent ve systems, to enter nto contracts to mot vate other f rms by appropr ate ncent ve systems, and to change these by organ zat ona dec s ons; (4) the governance of the act v t es set up under these structures by product on programs or standard operat ng procedures; and (5) meet ngs (w th agenda contro ed by the h erarchy or by standard operat ng procedures) for nterna d spute reso ut on. A norma h erarchy, then, cons sts of abor contracts, f duc ary re at ons, the exerc sed r ght to measure and reward performances, standard operat ng procedures, and dec s on-mak ng and d spute reso v ng meet ngs.

200 Labor Contracts . S mon (1957b), n a fundamenta paper on the theory of the emp oyment re at on, showed how the centra phenomena of author ty can be generated n exchange re at ons. [4] Br ef y, the argument s th s. The centra phenomena of author ty are re ated to a "zone of nd fference" of the subord nate (the term was nvented by Chester I. Barnard). By a zone of nd fference we mean a set of act v t es, a of wh ch t wou d be rat ona for the worker to exchange at h s or her wage rate. The h gher the wage rate as compared to a ternat ves ava ab e to the worker, the more costs n terms of unp easant or dangerous obs the worker wou d be w ng to do to reta n the advantage of the wages. The h gher the wage rate n a g ven ob as compared to the compet t ve wage rate, the arger n genera w be the zone of nd fference of the worker. Converse y, as the compet t ve wage rate goes down, for examp e by unemp oyment dur ng a recess on, the s ze of the zone of nd fference at a g ven wage rate goes up. Author ty s strengthened by pay ng h gh wages and by bad t mes outs de the f rm. The zone of nd fference s a so ncreased f the tastes of the workers do not make a ess des rab e act v ty much more cost y to them than another act v ty. The more equa d fferent act v t es are n the m nds of the workers, the more f ex b ty author t es have. One h res as trave ng sa espeop e those who do not m nd trave ng, and one tr es to so arrange work act v t es that one part cu ar cho ce of act v t es does not resu t n the emp oyment of the worker s over at h gh wages. Under what cond t ons w an emp oyer prefer to buy a zone of nd fference rather than part cu ar act v t es (e.g., by subcontract ng)? If the emp oyer knew t wanted part cu ar act v t es n advance, t cou d buy them on the abor market at the compet t ve wage rate or set up an ncent ve system n wh ch the powerfu mot ves of, say, emp oy ng sons and overs at h gh wage rates cou d be mob zed. And c ear y th s s often done. Rather than run automob e sa es bureaucrat ca y, car manufacturers a ow fam es or sma corporat ons to own the ob of se ng cars through the franch se system. Th s a ows car dea ers to use the r bus ness power to reward the r ch dren and re at ves, produc ng a sma ar stocracy w th nher tab e status n the core of the most modern bureaucracy. Presumab y, car manufacturers do th s for rat ona reasons, s nce they know perfect y we how to run bureaucrac es of emp oyed peop e, w th nepot sm ru es, order y careers, and a the rest. The centra quest on of rat ona ty from the po nt of v ew of an emp oyer s the d str but on of uncerta nty. An exchange partner s ke y to want author ty when t does not know wh ch part cu ar act v ty t s ke y

201 to want but s reasonab y sure t w want some act v ty. Under these cond t ons t s rat ona to become an emp oyer, to buy a zone of nd fference rather than a set of part cu ar act v t es. Prov ded one s cont nuous y n a g ven ne of work, f one s faced w th uncerta nt es of spec f cat ons, uncerta nt es of costs, or nab ty to d v de up the work so that t forms decoup ed packages, one w prefer to buy a zone of nd fference rather than spec f c performances. But the s tuat ons n wh ch we w f nd the uncerta nt es and the coup ed act v t es above and yet a so f nd contract ng are those n wh ch e ther one s not cont nuous y n a g ven ne of work (e.g., when one s a c ent for a bu d ng or an o ref nery, one s not cont nuous y n the construct on bus ness) or when one s not competent to run such a bus ness (as n weapons deve opment or buy ng a software system). Under those cond t ons one wants a zone of nd fference but a so wants to contract for the work. For our purposes, however, t s enough that author ty systems can be constructed f one can buy zones of nd fference, and the abor contract, comb ned w th carefu d fferent at on of fam y from work, w usua y do the ob. Organ zat ons that have tasks requ r ng h erarchy w tend to have emp oyees. F duc ary Re at ons . H stor ca y, the aw of corporat ons deve oped out of the aw of trusts and the aw of agency rather than out of that of master-servant re at ons. That s, from a ega standpo nt, the centra e ement of a corporat on s not that t emp oys peop e but that others entrust t w th the r money. Boards of d rectors are a so emp oyees of the corporat on, but they are ma n y, ega y, trustees. And operat ng execut ves of a corporat on are a so emp oyees of the corporat on, but ma n y they are, ega y, agents of the board. Stockho ders, ke m nor ch dren for whom a trust s created, cannot ega y exerc se the r ownersh p of f rm goods by d spos ng of them or us ng them. On y the board, ke the trustee, has operat ng contro (e.g., "possess on") over the resources of the f rm. L kew se, an off cer of the corporat on s ega y bound to exerc se h s or her agency n good fa th and not to exceed the author ty granted, but h s or her acts ( f ega w th n the aw of agency) comm t the corporat on and ts resources. What the stockho ders buy n h r ng a board, and what the board buys n h r ng off cers, s not so much a zone of nd fference as a prom se of respons b ty, a prom se to carry out a f duc ary ro e as a trustee or as an agent. The cap ta market s shot through w th these f duc ary exchanges;

202 n fact, the ponderous conservat sm of bankers and stockbrokers and corporat on awyers s more an eth ca mperat ve of f duc ary exchanges than a character type of the very r ch. Inher tance aw prov ded for trusts because the he r was often not trusted (e ther by the court or by the person who d ed) to adm n ster the estate rat ona y. Ch dren were a ways presumed ncompetent. So the bas c purpose of the trust was to prov de w sdom n the management of an estate when w sdom was presumed ack ng. The cho ce of a trustee genera y requ red not on y w sdom but e ther d s nterestedness or dent f cat on w th the nterests of the he r. The f rst of these a ms of the aws of trust, w sdom, s ref ected n the ega prov s on that trustees (and often agents) must act as "a reasonab e man" wou d act n the c rcumstances. The norms do not spec fy what spec f c act ons the trustee or agent shou d carry out. Nor s t any excuse n the aw that the act on taken s not forb dden n the document creat ng the trust or the agency; f that act on was not one that a reasonab e man wou d have taken under the c rcumstances, the trustee or agent can be he d ab e. It s a defense for the agent, but not for the trustee, that he or she was spec f ca y nstructed to carry out the unreasonab e act ons. The second of these a ms, d s nterestedness, s ref ected n "conf ct of nterest" norms, wh ch try to ensure that the trust or agency s used n the nterest of the benef c ary or pr nc pa rather than n the nterest of the trustee or agent. Ident f cat on w th the nterests of the benef c ary or pr nc pa s often ach eved by norms of e ect on and d sm ssa , as n the case of boards of d rectors, or precar ous or d scret onary tenure, as n the case of cab net off cers, off cers of corporat ons, or commerc a agents. The prob em w th both of these normat ve prov s ons s that they are eas y corrupt b e, and the aw s a very neff c ent way to prevent corrupt on. Hence, the norms of w se and d s nterested serv ce are often supported by structura dev ces prov d ng for cont nua rev ew of the r performance. The most common such dev ces are co eg a ty (a comm ttee as trustee, e.g., a board of d rectors), pub c ty (e.g., annua reports w th spec f ed contents), and aud t ng. In the r fu deve opment, then, the norms of trust or agency prov de for s x features (as de from the obv ous one that the trustee shou d be we pa d for the serv ce): (1) reasonab eness or w sdom as the performance standard, (2) d s nterestedness, (3) e ect on or precar ous tenure, (4) co eg a ty or comm ttee dec s on mak ng, (5) pub c ty, and (6) aud t. In genera , arge transfers of cap ta to new enterpr ses take p ace n the context of exchange re at ons w th a these features. These forma prov s ons are ord nar y supported by ex-

203 tens ve norms of gent eman y bus ness behav or and mutua re at ons among the peop e nvo ved as trustees and centra agents of the corporat on that are we descr bed by the Span sh phrase of the group, hombres de conf anza —men of (that s, n whom we have) conf dence. Status Systems and Performance Incent ves . If a centra d st nct on between the t mes a c ent uses contracts and the t mes t bu ds a department to do the work s the cont nuousness of that ne of work, we shou d expect that a ch ef feature of the h erarch ca ncent ve system wou d be cont nu ty of the exchanges between emp oyee and emp oyer. When an exchange re at onsh p s cont nuous, such as bureaucrat c emp oyment, marr age, or the exchanges among subs d ar es of the same corporat on, current exchange s mod f ed by the expectat on of future exchange. That s, n the future there w be a f ow of va uab e goods f the exchange re at onsh p s not destroyed by the current exchange. The ru e of caveat emptor s not appropr ate because the person who comm ts fraud or renders a poor performance ga ns on y a sma amount n the current exchange but sacr f ces the who e (d scounted) va ue of the f ow of future exchanges (Becker and St g er 1974). Norms about "fa r" exchanges thus serve the nterests of both part es, n that whatever d sadvantage they may susta n from unfa r norms of fa rness, those norms protect them from the destruct on of the re at onsh p and hence protect the cap ta zed va ue of future exchanges. Further, a reputat on for fa rness may advantage one or more of the part es, so that, for examp e, an emp oyer may keep on h gh sen or ty workers who no onger pu the r oad because f r ng them wou d underm ne prof tab e exchanges w th younger workers who hope someday to have sen or ty themse ves (Becker and St g er 1974; Cap ow 1964). The f rst requ rement of such a system s that the future exchanges shou d n fact be secure, so that the future advantages for wh ch one sacr f ces h s or her current nterest w n fact be there. The bas c phenomenon n an effect ve comm tment s that each party be sat sf ed that at a re evant po nts n the future the va ue sacr f ced by the partner by stopp ng the exchange w be greater than the advantage that partner can get by stopp ng unexpected y. G v ng hostages does th s, for examp e, because the hostage s more va uab e to the person g v ng the hostage than to the one rece v ng t; thus the atter can sacr f ce the hostage f the re at onsh p s broken. Ev dence of g v ng the partner a monopo y pos t on by eav ng the market (e.g., by stopp ng f rt ng upon marr age) shows that one ntends to sacr f ce the who e va ue of the re at onsh p n case of rupture, not ust the d fference between the va ue of th s stab e exchange

204 and the va ue of we -exp ored a ternat ves. Pass ng contro to the partner—for examp e, by a supp er se ng out and becom ng a subs d ary to ts customer (see the ana ys s of F sher Body and Genera Motors n K e n, Crawford, and A ch an 1978, 308–310)—guarantees the re at onsh p by transferr ng contro over the fu f ment of the se ng partner s ob gat ons to the parent. An mportant concrete man festat on of comm tment norms s norms of sen or ty. If the contract of comm tment were estab shed n a t me ess nstant, on mode of a marr age that starts at the wedd ng and asts "unt death do us part," the r sk ar ses that on y one of the part es w turn out to be tru y comm tted. A one-s ded ho d ng of hostages, a one-s ded monopo y, a one-s ded transfer of contro over one s resources, a ead to s avery. One genera so ut on to the r sk of one-s ded comm tment s to move the comm tment by sma stages, so that at each po nt ( f a goes we ) the partners w ose more by bugg ng out than s be ng r sked by ncreased comm tment. And n the emp oyment re at on and ts ana ogues, the growth of sen or ty ru es, of secur ty of tenure n an agr cu tura tenancy, or of tenure n a un vers ty s suff c ent y v ta to the nature of the re at onsh p that ts absence s a s gn of ruth ess exp o tat on. If a comm tment s effect ve, t estab shes an area of b atera monopo y for the two part es. The sk ed autoworker has sen or ty on y n Packard Motors, and the workers who know Packard s product on process are the workers w th sen or ty. W th n the range between the pr ce of abor at wh ch Packard s w ng to go out of bus ness and that at wh ch the workers w qu t n sp te of the r sen or ty r ghts, any pr ce s poss b e. A b atera monopo y does not have a un que equ br um rate of exchange, and the range of poss b e rates s set by the degree of comm tment. The reason a of th s s mportant to the ncent ve system of cont nuous h erarch es s that many of the ncent ves used, espec a y at the ower manager a eve s, are status system rewards. The va ue of a promot on s not ma n y th s month s ncreased paycheck, but the expectat on that the rate of exchange n th s part cu ar exchange has permanent y a tered for the better. Thus, the reward for a worker mprov ng h s or her human cap ta s a comm tment at a new rate of exchange. Un ess the who e system s organ zed as a permanent status system, the nvestment (and the cont nuous d sc p ned pursu t of ob ect ves spec f ed by the super or) does not seem so worthwh e (Becker 1960). Status systems w th promot ons as ncent ves are a so more appropr ate for execut ve and manager a obs, n wh ch one wants to reward d scret on whose resu ts are on y observab e over qu te a ong per od (Jacques [1956] 1972, 22–42). It wou d be qu te hard to dev se a p ece-rate system for obs whose span of d scret on was a month or s x months.

205 Cont nuous exchange between the f rm and ts emp oyees, then, makes t poss b e to construct a more effect ve career ncent ve system. Max Weber s spec f cat on of career ncent ves as a centra def n ng feature of bureaucracy was shaped by h s rea zat on that such an ncent ve system produces the h ghest eve of d sc p ne for comp ex nterdependent tasks. Thus we w expect h erarch es to have a permanent y organ zed status system n wh ch r ghts are attached to pos t ons rather than to performances, w th nd rect attachment of rewards to performances com ng from nk ng promot ons to udgments of competence (but see Co e and S ege 1979). Standard Operat ng Procedures . Cont nu ty of work n the same ne makes standard operat ng procedures product ve. Much of the transact on cost of contract ng for spec f ed work s actua y the neff c ency of prototype product on. If exper enced a rp ane manufacturers can produce the second a rp ane of a g ven mode for about 80 percent of the cost of the f rst, the fourth for about 64 percent of the cost of the f rst, and so on, then contract ng one at a t me w th un que spec f cat ons rather than buy ng off the she f can be extreme y expens ve. The v rtue of standard operat ng procedures and of product on programs s not that they are standard, but that they embody organ zat ona earn ng (New-house 1982; see a so Scherer 1964, 120–121). One of the ma n th ngs a h erarchy uses ts contro over a zone of nd fference for s to make mprovements between the prototype and the second, between the second and the fourth, between the fourth and the e ghth, and so on. At the beg nn ng of a product on run there s not much one can do that s as neff c ent as buy ng the same act v t es today that one bought yesterday. Even after the product on process s runn ng smooth y, we can pro ect from exper ence n produc ng software that about 40 percent as much aga n as the nvestment to that po nt w st be spent on mprovements of the sort that programmers ca "ma ntenance." In product on processes these mprovements range from s mp e gs or patterns to the tra ned eff c ency of an exper enced mach n st s movements (Burawoy 1979, 46–73; see above, Chapter 2). In cost account ng the categor es of the system evo ve so that they can be eff c ent y aggregated for a re evant cost ana yses for dec d ng about products, or about execut ves, or about nvestments to mprove a g ven process (see Chapter 3). Product on- ne ma ntenance programs are mproved by better est mates of when parts shou d be prevent ve y rep aced, by deve opment of better d agnost c tests of what a s a mach ne, and by rep ac ng hard-to-repa r mot ve systems w th easy ones. In many ways, the centra th ng

206 one sets up f rms to do s to deve op standard operat ng procedures, and then to mprove them ncrementa y. Meet ngs . Perhaps the centra us on of the nov ce organ zat ona ana yst s that the organ zat on chart s a p cture of the author ty system. In the upper part of the corporate h erarchy the re at ons between ranks are more ke that between courts of or g na ur sd ct on and the appea s courts over them than ke that between pr vates and sergeants. The h gher ranks organ ze meet ngs n wh ch prob ems and d sputes of the ower ranks are roned out. The organ zat on chart serves as a const tut on that spec f es who w organ ze comm ttee meet ngs, not who w make dec s ons. Further, for the most mportant dec s ons there w be a sequence of meet ngs, or at east a sequence of rev ews and n t a s cu m nat ng n a meet ng. Standard operat ng procedures usua y spec fy the route of the paper on wh ch the dec s on contents are recorded, and th s n turn spec f es the sequence of meet ngs or rev ews that make dec s ons; n pract ce, a mportant dec s ons appear as author tat ve after a meet ng on them has been he d. Tyranny n such organ zat ons s most often organ zed as t was n e ghteenth-century France, where the k ng expressed d ssat sfact on w th a ser es of outcomes by nvent ng a new court, one more under h s thumb, to treat matters of roya nterest (ma n y taxat on matters; see St nchcombe 1982, 90–92). Execut ves nst tute tyrann es by contro ng meet ngs. Us ng the h erarchy as a const tut ng mechan sm for meet ngs creates an ag e system for reso v ng d sputes—though perhaps not one we wou d be happy to entrust the adm n strat on of ust ce to. The ag ty s part cu ar y mportant n secur ng dec s on on th ngs that have to be dec ded n a hurry. If d sputes are appea ed up the h erarchy, of course, the meet ng of the board of d rectors has the f na say. H erarch es are se dom para yzed by nterna d sputes, though as the ate-e ghteenth-century French roya adm n strat on shows, th s can happen.

Hierarchy and Uncertainty The comb nat on of abor market contracts nvo v ng a zone of nd fference w th career ncent ves n a more or ess stab e status system makes t poss b e to embark on a ne of act v ty w thout know ng prec se y what one s about. W th n a w de range, the emp oyees w obey, and a career execut ve who has done h s best on a cance ed mode s happy to be promoted to ass stant product on manager for the successfu mode , f perhaps a b t p qued not to be product on manager of h s own mode . Thus

207 the progress ve concret zat on of spec f cat ons, wh ch we w f nd be ow s created by contractua contort ons, s a norma h erarch ca course of events. D spute reso ut on mach nery produces the eg t mat on of the dec s on on the mode , and the d sappo nted execut ve was probab y cowed n a meet ng he attended. S m ar y, cost uncerta nt es do not requ re the payment of a r sk prem um to the department that may overrun ts budget. The corporate funds that bear the cost uncerta nt es are the same ones that obta n the returns. The ncent ves to manage costs for the f rm are as good as can be arranged, for the f rm co ects a the benef t of a the cost sav ngs. F na y, the capac ty to reorgan ze nformat on f ows, author ty, and measurement of outcomes so as to adm n ster nterdependent act v t es together reduces the prob ems of unobservab e performances. The career ncent ve system can re y on the overa udgment of the super or on the qua ty of work, rather than observ ng separate performances. In sp te of the tendency of super ors to g ve a the r subord nates equa rat ngs, and n sp te of mutua recr m nat ons between nterdependent peop e about who caused the d ff cu t es, one can keep an ncent ve system go ng and more or ess correct ts fau ts. One m ght st have to agree w th Ecc es astes, that "under the sun,. . . . the race s not to the sw ft, nor the batt e to the strong, nor yet bread to the w se, but t me and chance happeneth to them a ." But t me and chance can be adapted to n organ zat ona performance measurement better than they can n the r g d t es of norma contracts.

Prediction of Performance Requirements and Performance Measurement The centra reason for wr t ng adm n strat ve prov s ons nto contracts—prov s ons that contractua st pu at ons may be changed by spec f ed methods— s that the future s uncerta n. When that uncerta nty nvo ves so many cont ngenc es that t s too expens ve to g ve a terat ons of contract performances for a of them n advance, some mechan sm for change or adaptat on needs to be bu t nto the contract. For conven ence, we can treat these uncerta nt es under three broad categor es: (1) uncerta nt es of the c ent about what t w want, or uncerta nt es of spec f cat ons; (2) cost uncerta nt es, due to contractor techn ca or cost uncerta nt es, to c ent gnorance, or to commerc a or ega uncerta nt es n the c ent-contractor re at onsh p; and (3) prob ems of observab ty of contractor defau ts, so that w thout cont nuous ntervent on the c ent does not know whether the contract performances have been de vered.

208 If the c ent can ant c pate that t m ght change ts purposes, t w want to prov de for chang ng the performances. If the actua costs of performances d ffer from the est mates made at the t me of s gn ng the contract, the rat ona ty for the c ent of comp et ng the contract may change f the c ent has to pay those costs; on the other hand, the contractor m ght not be w ng to s gn the contract w thout a proh b t ve r sk prem um f t s to assume a cost r sks. If the c ent ant c pates be ng uncerta n whether the contract has been fu f ed, t w want to be ab e to nst tute detect ve work or record keep ng to reassure tse f.

Uncertainties of Specifications C ents may ant c pate that they w want to change the r spec f cat ons of performances over the course of the contract for three ma n c asses of reasons. F rst, the contract tse f may nvo ve exp orat on of the poss b t es n the wor d; for examp e, n an R & D contract: one wants to buy research and deve opment work on a part cu ar a rp ane eng ne concept on y f such an eng ne cou d be bu t, as udged after some pre m nary workup s done (Marschak, B ennan, and Summers 1967, 63–90). Second, the c ent may ant c pate changes n the state of the wor d to wh ch t wants contractor performances to adapt; for examp e, a car manufacturer w want the dea ers to push mode s that are se ng s ow y so no money w be ost on the product on- ne too ng for that mode (Macau ay 1966, 12, 15, 18, 33, 46, 89, 167–168, 171). F na y, c ent preferences may change because of a change n the reg me of the c ent organ zat on or because of exper ence w th the performances or products of the contractor dur ng the contract. Somet mes such changes n preference can be ant c pated; for examp e, the Department of Defense knows that operat ona exper ence w suggest mod f cat ons of the eng neer ng spec f cat ons for weapons, but the exact mod f cat ons cannot be pred cted (Stockf sch 1973; Bendor 1985). The ant c pat on that one w know more about the poss b t es as the contract runs ts course s at the core of contracts for profess ona serv ces. When peop e contract for a phys c an s adv ce, they f nd out what med ca serv ces they want. The spec a nst tut ona protect ons of profess ona -c ent re at ons are des gned n part to protect buyers who cannot know what they want unt they f nd out what the wor d s ke, what d sease they have (Parsons 1939, 34–49). When profess ona contracts are s gned for arge-sca e pro ects, as when an o company contracts for an eng neer ng consu tant f rm to deve op spec f cat ons for an o p atform, the exp orat on of what s poss b e s part of the contract. Such contracts are very genera y negot ated

209 on some sort of cost-p us bas s—a most a sure s gn of "h erarchy" penetrat ng market re at onsh ps. In weapons deve opment, the R & D part of the contract s very often wr tten on a cost-p us bas s, wh e product on contracts after a weapon has been shown to be a rea poss b ty are much more ke y to be f xed-pr ce ncent ve-fee or f xed-pr ce contracts. [5] The cost p us bas s shows that what performances the government w requ re n an R & D contract depends on what eng neer ng poss b t es turn out to be v ab e. Eng neer ng theory s qu te uncerta n n pred ct ng the performance character st cs of a h gh-techno ogy system from an ear y paper p an. It s a so very uncerta n as a bas s for pred ct ng the costs of reach ng a g ven techn ca poss b ty unt that poss b ty has been spec f ed n deta (Marschak, B ennan, and Summers 1967). Even after a poss b ty has been deve oped and man fested n prototype product on, the c ent cannot a ways exp ore ts costs n the market w thout add t ona expense. A draw ng that acks many deta s of exact y how to produce the tem can be c ar f ed (w th n the same f rm that d d the deve opment) by conversat ons between the eng neers and a crew of sk ed workers. To trans ate th s conversat on nto deta ed shop draw ngs for the product on process for a comp ex weapons system or a computer so these can be produced e sewhere can cost m ons of do ars. [6] That s, to draw up the c ent s spec f cat ons n spec f c enough form that the c ent can buy from severa supp ers rather than from on y one deve oper costs a ot of profess ona work, because t s the R & D contractor, not the c ent, that knows what the c ent wants. Somet mes the uncerta nty about the poss b t es nvo ves determ n ng natura facts rather than exp or ng techn ca poss b t es. In the deve opment of the arge Ekof sk o f e d n the southern Norweg an North Sea, the f rst techn ca udgment was that the f e d probab y was "not commerc a " because the rock type was not permeab e enough; each we wou d dra n on y a sma area and wou d not pay for tse f. In the "contract" between the Norweg an state and the o compan es exp or ng the f e d, there was a m tat on on exp o t ng the f e d too fast—for examp e, not wast ng the gas by burn ng t off but wa t ng unt the gas p pe nes were bu t. But the on y way to get nformat on about the permeab ty n a reasonab e amount of t me was to exp o t a few we s "too fast," to est mate when the product on started to fa off. Uncerta nty about the state of the wor d a coup e of m es down, then, requ red that the exp o tat on rate cond t ons of the contract be suspended for a t me—unt the product on features of the rock were determ ned—and then re nstated. Th s suspens on and then reassert on of the exp o tat on

210 rate s a sh ft n spec f cat ons over t me due to exp orat on of product on poss b t es (Moe et a . 1980, pt. 2, 7, 11). H dden features of the wor d may be revea ed by contractua act v ty and may requ re chang ng spec f cat ons as the h dden becomes man fest. One very common form of c ent change n spec f cat ons n construct on s due to fa ure of the mag nat on. The c ass c examp e s the ght sw tch drawn beh nd the door—when the craftsman sees the ocat on he can te the sw tch wou d be better on the other s de of the door. A typ ca , more subt e examp e occurred n the bu d ng of ear y concrete o dr ng p atforms n Norway. In the ta concrete shafts of these r gs, a ot of p umb ng and e ectr ca work has to take p ace after the shaft s poured. If the shaft has supports for scaffo d ng poured nto the ns de wa , one can nsta and repa r equ pment by scaffo d ng supported by the wa s, rather than hav ng to bu d scaffo d ng from the bottom. But scaffo d ng does not appear on des gn draw ngs, so no one saw th s poss b ty unt a prototype ex sted. Contracts for the other r gs under construct on were changed to spec fy such supports. The genera po nt s that n some k nd of contracts t s contractua act v ty that d scovers the product on poss b t es. As those poss b t es are progress ve y revea ed, one wants to change the spec f cat on of performances n the contract accord ng y. To spec fy a poss b e states of the wor d wou d be expens ve and ot ose, s nce mechan sms to adapt performances as poss b t es are revea ed can be bu t nto the contract. The purposes of the c ent may change, not on y because t f nds out about the wor d, but a so because that wor d changes. In many contractua s tuat ons the c ent can ant c pate that some genera k nd of change n the wor d may change ts purposes, w thout want ng or be ng ab e to spec fy a poss b t es n advance. Perhaps the s mp est of these s a change n the overa vo ume of the market for the ne of goods a c ent produces. In contracts w th franch se car dea ers, for examp e, t may be reasonab e to spec fy a market share for a dea er but not an abso ute number of cars to be so d. In a recess on, there s no reason to expect the proport on of Ford Motor Company cars among a those so d to decrease, but every reason to expect the abso ute number to decrease. In deve op ng nforma standards of reasonab eness to def ne the operat ona mean ng of franch se contracts that a owed the manufacturer to requ re "sat sfactory" sa es, market share ca cu at ons of dea er performance requ rements became standard. These market share standards rep aced pract ces toward the beg nn ng of the Great Depress on n wh ch Ford cont nued to force a constant stream of unse ab e cars on the dea ers (Macau ay 1966, 13).

211 In h gh-techno ogy ndustr es, techn ca deve opments outs de the c ent-contractor re at onsh p often make spec f cat ons of performances archa c. Th s s most obv ous n m tary techno ogy, when techn ca deve opment s carr ed out by the potent a enemy. M tary supp y contracts change as the Sov et Un on deve ops defenses or offens ve weapons. But n add t on, for examp e, IBM has n the past fo owed the strategy for keep ng ts techn ca spec f cat ons secret for as ong as poss b e so as to preserve ts monopo y over sa e of per phera s for onger, wh e App e has fo owed the strategy of encourag ng others to deve op and se add-ons and software. Both c ear y expected techn ca deve opments e sewhere to change the r re at ons w th the r c ents. IBM thought that others wou d underm ne contracts for per phera s; App e thought that others cou d ncrease the va ue to the c ent of the r computer, and so ncrease future sa es (Toong and Gupta 1982). In both cases, the compan es cou d be expected to mod fy the r contracts w th supp ers so that the monopo y or deve opments wou d be encouraged, ead ng n turn to d fferent responses to techn ca deve opments among those supp ers. When an acc dent happens to a g ven sort of sh p, the c ass f cat on soc et es that cert fy the sh ps for nsurance compan es often requ re changes n the sh ps before they w recert fy them. Such soc et es rout ne y nvest gate acc dents w th a v ew to requ r ng prevent ve measures on other sh ps. Thus, the prov s on n mar ne nsurance contracts that sh ps must be c assed by a named c ass f cat on soc ety s an nd rect requ rement that the po cyho der respond to changes n know edge of the wor d, n part cu ar know edge of causes and acc dents as represented n c ass f cat on standards (He mer 1985, 63–67). Changes n the standards of regu atory author t es, such as the CAB requ rement for new nspect ons and new ma ntenance procedures for the eng ne supports of the DC-10 after the 1979 Ch cago d saster (Newhouse 1982, 98), are a so embedded n nsurance contracts n prov s ons stat ng that the po cy s not va d un ess the p anes sat sfy government regu at ons. Acc dents that often happen dur ng the course of construct on, such as cave- ns, co apse of heavy structures before comp et on, and weather or f re damage to part y comp eted structures, w cause c ents to want to change the act v t es requ red of contractors n order to ach eve the performances spec f ed n the contract. The reason construct on has a much h gher acc dent rate than the operat ons phase of the same bu d ngs s that the construct on process s not carefu y des gned for safety, and there are stages when the ncomp ete structure s weaker, more exposed, more unstab e, than the comp ete structure w be. Theoret ca y, th s r sk s assumed by the contractor or the contrac-

212 tor s bond ng company, un ess otherw se spec f ca y prov ded n the contract. In actua ty, however, the schedu e, the degree of f nanc a stab ty of the contractor and hence ts f nanc a capac ty to carry out the performances, and somet mes even the techn ca poss b ty of construct on p ans are genera y affected by severe acc dents. In such a s tuat on var ous standard contractua c auses are trad t ona y generous y nterpreted by the courts, creat ng a requ rement that both part es adapt to the s tuat on. The requ rement for such adaptat ons are therefore genera y mp ed n the contracts by the trad t ona ega nterpretat ons. Uncerta nty of the aw tse f, espec a y uncerta nty of substant ve regu at ons such as env ronmenta standards, create changes n c ent purposes. For examp e, near y a year of work by an eng neer ng crew was requ red to mod fy the des gn of the Statf ord B p atform n the Norweg an North Sea to conform to new, str cter safety standards (St nchcombe 1985e, 1985c). The extra eng neer ng work thus enta ed nvo ved a change n the contract between the o compan es ho d ng the concess on n the f e d and the eng neer ng contractor. Env ronmenta and safety regu at ons v rtua y a ways change dur ng the construct on of a nuc ear power p ant n the Un ted States, requ r ng changes n the performances demanded of both the eng neer ng contractor and the fabr cat on and construct on contractors (Cohen 1979, 76–77, 90–95). In the case of software deve opment, where t s very d ff cu t to spec fy n advance what the c ent wants, adaptat on of the software to c ent needs depends on p ac ng the user-or ented members of the programm ng staff (the "arch tects") n author ty on the pro ect team, so that they gu de the mp ementat on programmers. When th s author ty structure n the contractor (the "system producer") fa s, t takes more t me to repa r the software system to sat sfy users than t wou d have taken to des gn t correct y n the f rst p ace (Brooks 1975, 47–50). The c ent s preference s not rea y chang ng; rather, what the c ent rea y wants departs over t me from what the contractor s nterpret ng as the c ent s spec f cat ons—un ess arch tects have author ty. F na y, we can recur to the contract for profess ona serv ces n wh ch c ent spec f cat ons—for weapons, for examp e—are deve oped by profess ona s. Profess ona "ma pract ce" or ncompetence then produces spec f cat ons that do not n fact ref ect c ent preferences. The av on cs for the F111 f ghter a rp ane (the F111 was a so known as the TFX; av on cs are the p ane s nav gat ona and a m ng e ectron c systems) ended up be ng much more expens ve and much ess re ab e than prev ous systems, though they d d do more whenever they worked. The Department of Defense rev sed spec f cat ons n m dproduct on to get a cheaper and

213 more re ab e system. The bad performance was due to a m staken set of eng neer ng pred ct ons about what was poss b e n a g ven "state of the art." Contractua act v ty can change spec f cat ons of performances by show ng that some performances are not among the product on poss b t es, as we by f nd ng some that are (Cou am 1977, 127–132). The spec f cat ons changes we have d scussed so far bas ca y ref ect changes n know edge of the wor d or changes n the wor d. But c ent preferences can change n a more d rect way. A s mp e examp e s a change n reg me n the c ent organ zat on. When Kennedy was e ected and McNamara became secretary of defense, the m tary ph osophy of the government changed. For examp e, the ro e of a rcraft n support of ground troops ncreased n mportance compared to the nuc ear bomb ng m ss on, and the ob ect ve of common weapons for severa serv ces was ntroduced. An attempt was therefore made to mod fy the TFX a rp ane, whose range and speed (and therefore we ght) were adapted to strateg c reta at on, for carr er operat ons and support of ground troops. The management of the actua des gn was de egated to the a r force. The fact that the mod f cat on was a fa ure enab es us to observe the conf cts n ob ect ves between the new top reg me and the a r force, for the a r force preference for fast, ong-range p anes warred w th the new preferences of the pres dent and the secretary. In the ong run, the a r force won over McNamara, and the p ane was not usefu for the m ss ons of the other serv ces (Cou am 1977, 237–336). A second source of d rect preference change s c ent exper ence w th the product. Operat ona test ng of weapons a most nvar ab y shows that the m tary d d not want what they asked for n the eng neer ng spec f cat ons, but nstead wanted a mod f cat on of t (Stockf sch 1973). And n order to se a sat sfactory software system, one needs to adapt t n the des gn stage so that c ents can mod fy t eas y when they earn from exper ence what they rea y want to use t for (Brooks 1975, 117–118). Even those mod f cat ons of software that are ca ed "ma ntenance," wh ch are done by the supp er and are nc uded n the contract of sa e, are rea y changes n c ent spec f cat ons due to c ent exper ence. They come about because the test prob ems that software des gners use to debug software do not adequate y represent the prob ems the c ents w use the system for. So c ent exper ence f nds the new "bugs," that s, funct ons for the programs wh ch the or g na program des gn does not a ow for but wh ch can reasonab y be construed as a eg t mate c ent expectat on for system performance. For a heav y used software system, such postsa e redes gn or "ma ntenance" may eas y amount to 40 percent of the tota deve opment cost (Brooks 1975, 121).

214 A th rd d rect source of changes n c ent preferences der ves from the fact that most c ents are organ zat ons. When a centra author ty n an organ zat on de egates dec s on-mak ng power (or revokes a prev ous de egat on), t often changes the operat ng organ zat ona preference funct on. The a r force pred ect on for heavy, ong-range, fast a rp anes poor y adapted for ground support or carr er operat on meant that the de egat on to the a r force n the TFX-F111 deve opment resu ted n a f ghter adopted on y by the a r force (Cou am 1977, 237–336). A common sh ft n preferences by de egat on s due to profess ona perfect on sm. Many profess ona s have a strong preference for techn ca y e egant so ut ons, wh ch often overr des costbenef t cons derat ons that are more mportant n the centers of u t mate power n the c ent organ zat on. Organ zat ons often go to great engths to put "more reasonab e" profess ona s n charge of des gn un ts (K dder 1981; 119, 142), wh ch shows they fear that nterna de egat on may change organ zat ona preferences. For th s reason, contracts norma y adm n stered by a profess ona department may prov de for per od c rev ew (w th poss b e reor entat on away from techn ca perfect on sm) by h gher author t es or operat ona arms of the c ent organ zat on. R & D contracts and eng neer ng contracts are espec a y ke y to prov de such rev ew po nts. Many c ent organ zat ons determ ne the r concrete preferences by a genera preference for prof ts. Sh ft ng prof tab t es therefore change the c ent s preferences among concrete a ternat ves. For examp e, car manufacturers too up each year for a arge number of mode changes. Some of those mode s se we , but some move s ow y. S nce too ng for s ow-mov ng mode s s a sunk cost and not usefu for other purposes, the add t on to prof ts made by se ng a ess popu ar mode car s greater than that made by se ng a hot-mode car. But f franch se dea ers contro nvestments of se ng efforts, car manufacturers wou d ke to change performances demanded of franch se dea ers to get them to move the s ow mode s. The purpose of th s extended st of examp es s to show that there are arge parts of the economy n wh ch the bas c assumpt on of the abor contract—that the emp oyer may change ts m nd about what t wants the worker to do (S mon 1957b)—ho ds between organ zat ons. [7] Corporate and governmenta c ents change the r m nds about what they want from corporate supp ers and contractors. They know ahead of t me that they may want to change the r m nds w thout destroy ng the contractua re at onsh p, ust as emp oyers do not want to h re new workers on the open abor market for each change of task. Of course, the court w demand that the contract be wr tten n such a way that the performances requ red at any part cu ar t me—or the opt ons, such as contract can-

215 ce at on, that can be used to reward or pun sh conform ty w th changed requ rements—be def n te. Both the norma econom c mode s of a market transact on and the ega mode of a contract tend to obscure the degree to wh ch arge numbers of contracts are (rea st ca y, though not ega y) agreements to de ver an ndef n te good or serv ce for an ndef n te pr ce. A system for adapt ng to c ents sh ft ng purposes as they d scover product on poss b t es, determ ne how the wor d has changed, or change the r preferences can render the performances requ red by the contract "p ecew se def n te," y e d ng an unfo d ng def n teness to be enforced by the court. And as c ents w have been mak ng those p ecew se respec f cat ons n the ght of costs or of forgone a ternat ves as spec f ed n part by pr c ng mechan sms bu t nto the contracts, the contractors compensat on w kew se have been rendered p ecew se def n te. The assumpt ons of our s mp est mode s of market transact ons, that c ents know what they want and w cont nue to want the same th ng throughout the transact on, do not therefore f t a arge number of actua market transact ons. Econom sts who ana yze weapons research and deve opment throw away the Econom cs 100 assumpt ons a most w thout not c ng (Marschak, B ennan, and Summers 1967; a so Scherer 1964). S m ar y, pract ca sem nars for contracts spec a sts n the construct on ndustry d scuss the techn ques for wr t ng the prov s ons dea ng w th "change orders" so that work can go forward on the change wh e compensat on for t s n d spute, w th hard y a g ance at the ega theory of def n te performances and cons derat ons. Our purpose here has been to focus susta ned attent on on s tuat ons n wh ch rat ona c ents w want to prov de contractua y for be ng ab e to change the r m nds, w thout want ng to create a who e h erarchy of wh ch they are the bosses.

Cost Uncertainties In many of the same ndustr es n wh ch we observe prov s ons for c ents to change the r m nds we a so observe contractor ncapac ty to pro ect costs accurate y. Scherer (1964) shows that when defense contracts are not renegot ated partway through, the standard dev at on of the percentage cost overrun or underrun s about 10 percent. That s, as de from any b as nduced n the mean cost est mate by strateg c barga n ng ncent ves (wh ch b as n d fferent ways w th d fferent types of contracts), sheer var at on among contract outcomes causes a part cu ar est mate to be w th n 10 percent of the mean percentage overrun or underrun for that type of contract on y about two-th rds of the t me; the other th rd of the t me t s more than 10 percent from the mean (see Scherer 1964; 192, 195–196, and the graphs 196–199). Defense contractors s mp y do not know how

216 much a weapon w cost. The standard dev at on of costs as compared to est mates s arge compared to the average percentage prof t on the contracts. Systemat c data from other areas s rare, because cost data are not rout ne y pub c. A Swed sh study of bu d ng construct on suggested a range from 10 to 27 percent for work ng t me spent n d fferent pro ects on coffee breaks, wa k ng to and from the ocker rooms, and other nonwork act v t es (Kre ner 1976, 97). C ear y the contractors cou d not pred ct th s var at on very we , or they wou d have taken prevent ve measures. S nce abor t me s a arge share of construct on costs, cost uncerta nty for construct on contractors wou d seem to be n the same range as for weapons contractors. Thus, even when performance spec f cat ons are def n te, the appropr ate compensat on s very genera y m sest mated. S nce often a s ng e pro ect s a arge share of the tota sa es of a contractor, such cost uncerta nty s a ser ous r sk. Th s r sk may be aggravated by pena t es for not meet ng a schedu e. In one case of construct ng m ss e s os w th such pena ty c auses, contractors apparent y added about 50 percent r sk prem um to the cost the government had est mated for the work. The government sh fted the r sk to tse f w th a cost-p us-fee contract, and bu t the s os more cheap y. [8] Because the c ent pays for the cost r sk, t may be to everyone s benef t to p an to adapt to the r sk, to agree to work together to m n m ze t, rather than mere y sh ft t between part es. As background to the dev ces usefu to such adaptat on (dev ces short of h erarchy), we can spec fy some of the causes of cost uncerta nt es. The f rst source of contractor cost uncerta nty s uncerta n techn ca nformat on. An R&D contract wou d have no research e ement n t f the techn ca so ut on were obv ous and a that had to be spec f ed were the deta s, deta s that cou d be descr bed (see Marschak, B ennan, and Summers 1967; on c v an R&D pro ects, see Scherer 1964, 178). The part of software deve opment that can be accurate y costed, for examp e—the wr t ng of nes of code—accounts for on y about a s xth of the deve opment cost. The h gh rema n ng costs resu t because d saggregat ng a user s funct ona requ rement nto an mp ementat on strategy ("system arch tecture") requ res a good dea of expens ve t me, and then the mp ementat on never works because t has bugs. The naccuracy of the est mate of debugg ng t me s so great, Brooks says, that a software pro ect usua y spends ha f ts debugg ng t me w th debugg ng "90 percent comp ete" (Brooks 1975, 154–155). R&D and software deve opment have n common that performance cr ter a for the product set by the c ent can-

217 not be ach eved n a techn ca y rout ne way. On y rout ne techno ogy can be accurate y costed. But any arge un que pro ect can meet techn ca surpr ses. The so samp es on the bas s of wh ch foundat ons for a bu d ng were costed may not have been a good enough samp e, and much extra foundat on work may have to be done (Kre ner 1976, 95–97). An o p atform base may s nk mproper y when be ng sett ed on the bottom and have to be d scarded ( tse f not a tr v a ob) (Moe et a . 1980, pt. 2, 113–114). A new, ghter mater a for the huge fans that make bypass et eng nes pract ca may break when a ch cken corpse s thrown at a prototype fan at a h gh speed to s mu ate the eng ne swa ow ng a duck n f ght; the t tan um that rep aces the new mater a may requ re new we ght ad ustments throughout the p ane (Newhouse 1982). In add t on to techn ca cost uncerta nt es, est mators manua s n construct on warn the est mator to take account of the sk ed abor market n the area (Page and Nat on 1976). [9] If there are many more obs than workers, the foreman s author ty s underm ned; f there are more workers than obs, both the sk and the d sc p ne of work crews can be ncreased. But how t ght the abor market w be at the t me a pro ect s done cannot genera y be pred cted accurate y when the pro ect s est mated. T ghtness w th n the f rm has the oppos te effect n the defense ndustry. If a defense contractor has many emp oyees but few pos t ons on pro ects because of bad uck at the Pentagon, the pro ects t does have w have h gh off c a overheads nf ated by sa ar es of peop e work ng on new b ds, as we as h gh concea ed overheads as pro ects are oaded up w th extra workers. Scherer (1964, 183, 187) shows that ean eff c ent weapons pro ects are done by contractors w th a ot of work, fat neff c ent ones by contractors try ng to reta n va uab e personne for better t mes. Of course, th s strategy s on y poss b e because many defense contracts are cost-p us-fee contracts. The cost of arge, spec a zed mach nes such as a rp anes, sh ps, heavy presses, or dr ng r gs tends to dec ne rap d y w th exper ence w th a part cu ar mode . The ru e of thumb among a rp ane manufacturers s that as one doub es the number of mach nes produced one cuts the cost of each add t ona one by 20–25 percent (Newhouse 1982, 19–21; Scherer 1964, 120–121; see a so above, Chapter 5). Th s means that the average cost of product on s as unpred ctab e as the s ze of the market. The s ze of the market for cap ta goods s very unpred ctab e, h gh y respons ve to the bus ness cyc e, and strong y re ated to the prosper ty of part cu ar c ents. Contractors cannot pred ct the r costs for arge cap ta mach nes because they do not know how far down the earn ng curve

218 they w get. It s n part for th s reason that a rp ane manufacturers w not des gn and produce a rp anes un ess they have a arge advance order book for them. Contractors very often depend on subcontractors who are themse ves exposed to cost uncerta nt es. When a arge number of arge subcontracts are et on a f xed-pr ce bas s, at east one subcontractor on the pro ect w ke y get the ob because t made the b ggest est mat ng m stake. S nce such a arge subcontract may be a arge share of the subcontractor s turnover, bankruptcy s not out of the quest on. Ro s Royce went bankrupt (to be saved by nat ona zat on) because of overruns n deve op ng the eng nes for Lockheed s w de-body a r ner (Newhouse 1982, 62–72); f Ro s Royce had not been rescued, Lockheed wou d have gone bankrupt as we . S m ar y, the computer deve opment crew featured n Tracy K dder s The Sou of a New Mach ne was substant a y de ayed because the ch p manufacturer started to go bankrupt (K dder 1981, 231–232, 268). (In th s case, s nce the contracts for de very of computers were not yet s gned, the ch p supp er was not techn ca y a subcontractor.) Contractors can m sest mate the r costs a so because the r h stor ca cost data are bad. For many arge contracts, the contractor s est mated cost s synthes zed from est mates of what quant t es of var ous e ementary act ons need to be carr ed out (e.g., how many yards of excavat on), mu t p ed by a cost per un t der ved n part from organ zat ona h stor ca cost data. But f the categor es of e ementary act ons are not f ne enough, the f n shed work may be costed w th data that nc ude rough work, the c ear ng of rregu ar terra n for a p pe ne may be costed w th data co ected from work done on ess broken terra n, or the ob of ocat ng survey respondents who are drug add cts may be costed w th data on ocat ng respondents who are phys c ans. The cost per un t may be m sest mated from the h stor ca data. Further, the mu t p er may be out of date ow ng, for examp e, to nf at on, techn ca change n the act v ty, or a dec ne n the qua ty contro standards of a cruc a sem f n shed goods supp er. Such sources of naccuracy p ague a cost account ng, but they are more ser ous and harder to detect when product on s not cont nuous. For one th ng, s nce contractors often have to bu d up and tear down the r organ zat ons rap d y, a constant overhead cost such as a competent cost-account ng department s often an expens ve uxury (St nchcombe 1959). Even when the contractor knows the costs (or even to the degree t knows them), the c ent may be exc uded from that know edge. The cost of b dd ng on a bu d ng construct on ob runs from 3 to 5 percent of the va ue of the ob, even after the spec f cat ons are comp ete y drawn up.

219 Contractors are therefore not w ng to b d f there are many ser ous b dders, because b dd ng on many obs they do not get ncreases the r overhead and can make them noncompet t ve n the ong run. Ord nar y, then, a c ent gets a samp e of three or so from a rough y norma d str but on w th a standard dev at on of 10 percent (Scherer 1964, 192, 195–196, and the graphs 196–199). [10] Because the c ent often does not know what accounts for the d fferences among contractors (Kre ner 1976, 80–81), t may not be ab e to obta n good nformat on from the market about costs even nsofar as the contractors know them, nor can t ega y penetrate the contractor for d rect cost nformat on. The government somet mes tr es to penetrate defense contractor costs. Even when they ask for separate pr c ng of components of a weapons deve opment contract and devote government eng neer and accountant hours to t, they cannot a ways get cost nformat on to a ow them to choose among a ternat ve strateg es (Scherer 1964, 204–207). In add t on, often contractors know of techn ca poss b t es that they do not revea because they do not effect ve y have a profess ona (f duc ary) contract w th the r c ents. [11] I once knew an operat ons research consu tant who had d scovered that the ma n schedu ng dec s on task a sk ed execut ve d d cou d be reproduced by a transformat on of var ab es and a s mp e near equat on. By concea ng that equat on, he ra sed the pr ce of h s adv ce on the schedu ng prob em. A p aster ng contractor w rare y te a potent a c ent how much drywa wou d probab y cost. Techn ca nformat on that wou d a ow c ents to cut contractor prof ts s hard to come by. The trad t ona theory n th s f e d says that when c ents are confronted w th uncerta nt es that may mp nge on de very dates, on supp er so vency, on qua ty contro prob ems as contractors desperate y make up for cost ng m stakes, they w tend to ncorporate the uncerta nty w th n the r organ zat on, so they can contro t as far as poss b e and pred ct when they cannot contro t (J. Thompson 1967, 36, 40–42). But we have observed that n much of the economy they bu d whatever adaptat ons they do make nto contractua re at onsh ps. They do th s even when t puts contractors n a pos t on of nformat on advantage that may be (and often s) used to the c ents d sadvantage—though the theory argues that the chance of exp o tat on by the contractor s a pr nc pa cond t on favor ng vert ca ntegrat on. [12]

Observability of Performance When two act v t es depend on each other (are "c ose y coup ed"), they shou d be adm n stered together under the same respons b ty, and on y when act v t es are decoup ed shou d they be d v ded between d fferent

220 f rms or be done by d fferent departments w th n a f rm. The more two act v t es are nterdependent, the more nformat on has to f ow between them, the more comb ned author ty s necessary for mak ng ad ustments n both, and the more d ff cu t t w be to a ocate respons b ty among act v t es for costs, de ays, nfer or qua ty, or other d ff cu t es (A ch an and Demsetz 1972). Th s decoup ng pr nc p e or team product on pr nc p e means n part cu ar that h gh y nterdependent act v t es shou d not be separated by contract ng parts of them, for that separates nformat on f ows, author ty, and measurement of outcomes between two or more f rms. A der vat on from th s pr nc p e s that act v t es that have to be performed n ser es shou d on y be separated among d fferent f rms f one s not n a hurry, so that de ays (to g ve a breath ng space), nspect ons, and nventor es of part y f n shed work between stages can buffer nterdependenc es (J. Thompson 1967). Yet we observe that weapons deve opment contracts are near y a ways rushed to stay ahead of the Russ ans; that de ays n construct on adm n stered by contracts are often the pr mary determ nant of net present va ue of arge cap ta pro ects, so such pro ects are rat ona y rushed (Moe et a . 1980, pt. 1, 111, 241); or that a r ne a rframes stand wa t ng for the de very of eng nes from a separate eng ne supp er at thousands of do ars a day ost per frame n nterest a one (Newhouse 1982, 166). S m ar y, when the system be ng produced s h gh y techn ca y nterdependent, the qua ty contro determ n ng the re ab ty of components determ nes a so the re ab ty of the system (Green and Bourne 1972). Th s means that the qua ty of work on a subcontracted component determ nes the effect veness of another: fau ty de very of hot or coo a r through a bu d ng s heat ng system eads to comp a nts about the fact that the w ndows do not open, and unre ab e av on cs causes a f ghter-bomber to f y very fast n the wrong d rect on. To get the contractor to take respons b ty for the performance of the who e system, a c ent often has to a ow the contractor author ty over the components (Scherer 1964, 110–112)—author ty that s often adm n stered and set up through contracts. [13] The centra d ff cu ty for contractua arrangements that such techn ca nterdependenc es among components create s that the performances n two or more separate contracts are nterm xed n such a way that the qua ty, t me ness, or cost of the separate performances are not separate y measurab e. Th s n turn means that t s d ff cu t to estab sh wh ch contractor was n defau t when the o nt performance was nadequate. Severa other common cond t ons have the same effect on perfor-

221 mance observab ty. When performances are d ff cu t to observe yet h erarchy s a so d ff cu t, spec a prov s ons may have to be made n the contract to render the performances contro ab e— wh ch means to s mu ate a h erarchy by contractua means. The most obv ous prob ems of observab ty come when the performance s tera y h dden, as n the ength and compos t on of p es for a bu d ng, the second coat of pa nt on a three-coat pa nt ob, the mean t me to fa ure of a computer or copy mach ne ust be ng bought, or, n nternat ona trade, the f sca and monetary po cy of the government that backs the currency n wh ch the contract s to be pa d. [14] In the nature of the case a contract for profess ona serv ces poses great d ff cu t es n performance measurement. One cannot eva uate the qua ty of a serv ce accurate y f one cou d not have prov ded t onese f. One may know, for examp e, that the rate of ceasarean sect ons n the Un ted States s much h gher than t s n other countr es where the materna and feta -neonate death rates are ower (e.g., the Nether ands or the Scand nav an countr es), mean ng that the procedure must be excess ve y prescr bed. Yet an nd v dua mother s not n a pos t on to eva uate the profess ona adv ce she rece ves n her part cu ar case. The extant econom c theory pred cts that the phys c an w prescr be the procedure that br ngs the h ghest fee; s nce such ncent ves cou d be corrected, one hopes t s th s rather than ncompetence that accounts for the overprescr pt on. [15] Except n very gross cases, defau ts n performance of profess ona contracts are d ff cu t to prove, and a ternat ve contro mechan sms are wr tten nto the norms of profess ona pract ce rather than nto the contract. In many contractua systems for arge-sca e pro ects, the c ent bu ds a arge profess ona system w th n ts own organ zat on to superv se the adm n strat on of the contract. Scherer co ected rank data from expert udges separate y on the performance both of the government team superv s ng a set of defense contracts and of the contractor. The performances were h gh y corre ated, nd cat ng that c ent performance and contractor performance are too nterm xed to be separate y observed. [16] In the Norweg an North Sea, the rat os of eng neers n the c ent pro ect organ zat on to eng neers n the eng neer ng contractor s pro ect organ zat on have apparent y ranged between 1:3 and 1:7. Under such cond t ons there s a arge f ow of nformat on, consu tat on, approva s and d sapprova s, and the ke from the c ent s eng neer ng crew to the eng neer s crew; th s means that at each step of the way the c ent s and the eng neer s performance are deep y nterm xed, and every step on the way to an eng neer ng contractor s defau t has been approved or caused

222 by a c ent d rect ve. When th ngs go wrong, when the pro ect s de ayed or becomes more expens ve than p anned, there s a f ow of recr m nat ons across the boundary that s near y as arge as the f ow of nformat on. In genera , such deep ntervent on makes the contractor s respons b ty for defau ts very d ff cu t to prove n court. We see the same process n reverse when we observe that defense contractors on f xed-pr ce contracts tend to refuse to ntroduce eng neer ng changes, even though they w supposed y be re mbursed for the costs of such changes, because they cannot recoup the fu cost of the d srupt on of product on (Scherer 1964, 180, 237–238). When the c ent ntervenes w th a change order, t s not poss b e to d scr m nate exact y the costs attr butab e to c ent ntervent on from the costs norma y due to act v t es for wh ch the contractor s respons b e. The genera po nt here s that when e ements of h erarchy are bu t nto the contract, by h gh Defense Department contro of contractors, by superv s on of the deta s of eng neer ng work by a arge c ent pro ect crew, or even by an " so ated" eng neer ng change n a f xed pr ce contract, the separab ty of contractor performance s underm ned. Thus, h erarch ca e ements bu t nto the contract tend to underm ne the ega ob gat on of the contractor for the performances spec f ed and to requ re further re ance on h erarch ca dev ces to ad ust to the consequences of poor observab ty of performances. F na y, we shou d note that when the contractor s n defau t, many contracts prov de that the c ent (and somet mes a so the surety bond ng company) can resume author ty over the performances and arrange to have them done by other contractors or by ts own work force. The costs of such subst tute performance are an e ement n determ n ng the ega va uat on of the contractor s defau t. That s, substant ve h erarch ca d scret on and act v ty superv sed by the c ent s nvo ved n determ n ng the damages of the defau t.

Uncertainty and the Requirements for Hierarchy Future contractua performances under certa nty g ve no d ff cu ty for e ther econom c theory or the ega theory of contract. But the manufacture of arge mach nes such as a r ners or computer systems, the product on of advanced weapons systems, and the bu d ng of arge cap ta construct on pro ects a frequent y take p ace under cond t ons of ex ante sh ft ng and uncerta n c ent spec f cat ons—ma or cost uncerta nt es that t s rat ona to adapt to rather than to nsure—and ex post uncerta nty about whether the contractor has ndeed done the performances. A of these uncerta nt es generate pressure to create structures connect-

223 ng the c ent w th the contractor (and the contractor w th any subcontractors) such that ad ust ve moves can be made author tat ve. That s, such uncerta nt es create the requ rement that h erarch ca e ements be bu t nto the contractua system.

Elements of Hierarchy in Contract Contents In the ast sect on we showed that the funct ons for wh ch h erarch es are set up are often necessary to accomp sh the c ent s purposes n contractua re at onsh ps. We have a so seen, n the preced ng sect on, how h erarch es n fact fu f those funct ons. The sett ng can prov de some gu dance n mod fy ng our Econom cs 100 or Bus ness Law 100 mode of the papaya-buy ng contract to take account of the comp ex ty we f nd n construct on, weapons, a rp ane, or computer system contracts. We start by tak ng from our mode of h erarchy the e ements that we expect to be present when peop e have to so ve h erarch ca prob ems by contractua means. "H erarch ca e ements" n contracts can be descr bed as cons st ng of f ve structures: (1) command structures and author ty systems; (2) ncent ve systems, support ng author ty systems and a so gu d ng the use of a contractor s d scret on by a structure of d fferent a rewards part y so ated from the market; (3) standard operat ng procedures, that descr be rout nes that nvo ve act ons by both contractors and c ents; (4) d spute reso ut on procedures, part y so ated from the court system and from the market; and (5) pr c ng of var at ons n performances part y so ated from the market, nc ud ng espec a y pr c ng based on contractor costs. C ear y a structure w th re at ons of command, an ncent ve system part y so ated from the market, standard operat ng procedures, nterna d spute reso ut on structures, and subun ts whose "pr ce" s determ ned ma n y by costs s qu te near to what we have descr bed above as a typ ca "h erarchy."

Command Structures and Authority Systems The na ve not on of an author ty system s one nvo v ng "command": a superv sor se ect ng an act on for a subord nate to carry out. Above the eve of foreman or secretary, author ty systems rare y depend on command, though an author tat ve descr pt on of an end to be ach eved by an nd v dua s perhaps not so uncommon. At m dd e management eve s and above, rev sed m nutes of comm ttee meet ngs, a str ng of n t a s approv ng a proposed budget a ocat on, standard operat ng procedures adopted by the re evant author t es and wr tten nto the organ zat ona

224 regu at ons, or the quest on of a super or, "And what do you th nk you (or we) ought to do next?" w th ts mp c t approva of whatever comes next, are more common ways of exerc s ng author ty. Author ty systems are systems by wh ch f ows of nformat on are cert f ed as eg t mate or author tat ve, so that the r sk of be ng wrong s removed from the person who acts n accordance w th the nformat on and s a d nstead on the eg t mators of the commun cat on (Barnard 1946). What we have to ook for n contracts, then, are systems for cert fy ng a g ven commun cat on (other than the contract tse f) as author tat ve, and therefore as red str but ng the r sks of be ng wrong. An e egant examp e occurred w th the ntroduct on of X-ray and son c nspect on of stee products. These nspect ons made poss b e a much more accurate determ nat on of whether a g ven p ece of stee actua y sat sf ed the standards spec f ed n the contract, not ust on the surface but a the way through. S nce a most a the ncreased accuracy was n detect ng heretofore undetectab e fau ts, the new dev ces enab ed c ents to ra se the de facto standards for stee . The stee producers argued that the spec f cat ons for stee n the contracts extant at the t me rea y ncorporated the cert fy ng procedures that were ava ab e when the contracts were s gned; the c ents argued that h dden defects v o ated the spec f cat ons as much as v s b e ones, and that nspect on procedures not spec f ca y descr bed n the contract were mp c t y eft to the d scret on of the c ent. (The dec s ons apparent y went aga nst the stee producers n most countr es, and now nspect on procedures are spec f ed n new contracts.) The ega ssue, however, was c ear y, By what standards s the qua ty contro cert f cat on to be taken as eg t mate? Such qua ty contro systems are common features of contracts and const tute a c ear examp e of an author ty system n Barnard s sense of mark ng spec f c commun cat ons as author tat ve. It s obv ous that they d str bute the r sks of m stakes, s nce the stee producers were c ear y c a m ng the r ght to the r trad t ona proport on of m stakes, on the bas s of wh ch they had made the r b ds. An approva s systems that s more c ear y superv sory s that descr bed above for defense contracts and arge o nsta at ons, n wh ch a arge staff of c ent eng neers rev ews and approves p ans drawn up by R&D contractors or eng neer ng consu tants. The arger the c ent s eng neer ng organ zat on that rev ews the contract s eng neer ng work, the f ner the net of approva s can be. Th s makes the sequence of steps that have to be approved more dense, wh ch means that sketches, drafts, pre m nary spec f cat ons and draw ngs, f na spec f cat ons and draw ngs, tender documents, techn ca eva uat ons of the b dders rep es, techn ca changes

225 proposed after contract start, and costs for a of these must a be approved by the c ent. As de from the extra work nvo ved n creat ng a document for approva for each ntermed ate dec s on, such a f ne net of approva s stra ns out any d scret on or or g na ty the contractor s eng neers m ght have been aff cted w th. Such a system makes the c ent eng neer nto a superv sor w th restr ct ons on h s or her ora and v sua contact w th the work, thus structur ng the ncompetence of the superv s on and ncreas ng ts bureaucrat c comp ex ty (see the hear ngs on th s prob em n m ss e programs c ted n Sherer 1964, 375–376). A arge number of nspect on or "comm ss on ng" prov s ons have the same genera effect. In genera , the nspect on structure set up by the c ent s not spec f ed n trad t ona construct on contracts but s assumed accord ng to ndustry pract ce. In arge cap ta mach nery construct on, such as ref ner es or power generat ng stat ons, the contract often spec f es a separate comm ss on ng phase, dur ng wh ch author ty s genera y reconcentrated n the c ent s hands and the eng neers or pro ect managers from the contractor are put at the d sposa of the c ent s superv sor of comm ss on ng to a d n the test ng of a equ pment. In sh pbu d ng, some part of the nspect on and test ng s carr ed out by the c ass f cat on soc et es that cert fy the sh p to the mar ne nsurance compan es. S nce c ents do not want sh ps they cannot nsure, and s nce produc ng an nsurab e sh p s part of the contractua respons b ty of the sh pyard, th s step serves a so as an nspect on on beha f of the c ent; ndeed, t s pa d for by the c ent rather than by the nsurance company (He mer 1985a). More d rect prov s on for command structures are often wr tten nto contracts. For nstance, a standard form for subcontracts by a pr me eng neer ng contractor conta ns the anguage:

Phys ca Exped ng

When (company name) deems adv sab e h s order sha be sub ec o phys ca exped ng by (company name) represen a ves who sha be gran ed access o any and a par s o he se er s or sub-supp er s p an nvo ved n he manu ac ure or process ng o h s order

No actua capac ty for the exped ter to order peop e around s spec f ca y prov ded, but author ty s c ear y be ng a ocated. The prov s ons about change orders n construct on and arge eng neer ng pro ect contracts qu te often conta n anguage to the effect that the contractor s to accept the orders of a spec f ed person ( n Eng and th s person, typ ca y ca ed "the Eng neer," s emp oyed by the c ent

226 and s named ear y n the contract) on a change orders, even when the compensat on for the change has not yet been agreed. The command author ty of the Eng neer over contractor personne s thus estab shed by des gnat ng the work to be a change order and not part of the or g na contract. The or g na contract, however, spec f es th s ntermed ate h erarch ca phase, unt the amendment of the or g na contract by the change order can be agreed on (or arb trated). Macau ay (1966) reports that n the bad o d days, the automob e manufacturer s "reg ona representat ve" (the member of the manufacturer s market ng staff who dea t d rect y w th the franch sed car dea ers) n effect had power to set quotas for sa es, for car nventory, for parts orders, for sa es and d sp ay space, and for other features of the dea er s bus ness conduct not spec f ed n the contract. These came to be po nts of d spute ma n y when they ncreased costs and nvestments for the dea er and ncreased gross vo ume for the manufacturer but cou d decrease prof t rates or ncrease r sks for the dea ers. The contract was wr tten to prov de that the manufacturer cou d cance the franch se un ess the dea er s performance was sat sfactory to the manufacturer; but th s meant that, ack ng any other source of udgments of sat sfactory performance, the reg ona representat ve cou d g ve orders w th the sanct on of the r sk of cance at on. Aga n the author ty structure s not spec f ed n the contract, but the sanct ons n the hands of the reg ona representat ves were, and the pract ce of accept ng the r udgments created the command author ty n pract ce.

Incentive Systems By an ncent ve system we mean a way of measur ng or otherw se observ ng eve s of performance of a contractor or of a contractor subun t and a ocat ng d fferent a compensat on based on the eve of performance, w thout further recourse d rect y to the market. An ncent ve system, then, s an enc ave n the market w th n wh ch spec a ru es of reward and pun shment app y. Of course, the broad eve s of compensat on tend to be set w th an eye to the market, and a compensat on system must produce a correspondence between output and costs that s v ab e n the market. But, for examp e, the p ece rate systems n most Amer can factor es pay ess for each add t ona p ece (above the quota that mere y earns the guaranteed da y m n mum wage) than they do for each p ece nc uded n the m n mum, wh ch reverses the pattern to be expected on the bas s of marg na revenue product v ty. The o der "Stakhanov te" Sov et ncent ve systems were c oser to the neoc ass ca dea . [17] Hence p ece rates are ncent ves n a ru e-governed enc ave, rather than mere y market ref ect ons of d fferent a mach n st ta ent.

227 Actua y, by th s def n t on a norma f xed-pr ce forward sa e contract creates an ncent ve system. For the term of the contract, the contractor can se the c ent a g ven quant ty at the f xed pr ce, even f t manages to reduce costs dur ng the course of the contract. Thus the contractor co ects the fu va ue of a cost reduct ons, and none of that add t on to contractor net revenue s d ss pated by compet t ve react ons to these new cost poss b t es. Th s super or ncent ve effect of f xed-pr ce contracts s consc ous y used by the Department of Defense and some arge corporate c ents as the bas s for a po cy preferr ng such contracts. The average prof t as a percentage of tota sa es on f xed-pr ce defense contracts runs about three t mes as much as that on var ous types of costp us contracts (Scherer 1964, 159–60). Yet the Department of Defense be eves t gets cheaper weapons from f xed-pr ce contracts; the r super or ty as ncent ve (and compet t ve) systems thus apparent y overr des the h gher rewards to the contractors. Once the bas c dev ce of bu d ng an ncent ve system nto the contract s recogn zed, t can be extended to a w de var ety of substant ve measures of performance. In sharecropp ng and putt ng-out p ecework systems, the ncent ve s based on the quant ty of (acceptab e) output. The d ff cu t es of such an ncent ve system n a modern economy have been most fu y stud ed for the Sov et Un on (Gran ck 1976). In the agency re at ons much ana yzed n the modern econom c terature, f nanc a outputs prov de the measures, wh ch one wou d not want to do f the workers cou d eat part of the output before sa e (Ross 1973; Becker and St g er 1974). Many construct on contracts prov de for pena t es ( qu dated damages) for de ays beyond an agreed-upon comp et on date, and often rewards for ear y comp et on as we . Franch se contracts for sa es out ets genera y n effect set up an ncent ve system to mot vate sa es efforts and sa es-enhanc ng nvestments by the franch se ho der w thout recourse to compet t ve b dd ng for the franch se. The Department of Defense has exper mented w th mak ng fees on R&D contracts depend on eng neer ng performance cr ter a as measured on the resu t ng weapon. They have most y d scarded the system because contractors successfu y argued that f the government had d sapproved ntermed ate proposa s that wou d have resu ted, f fo owed, n a h gher compensat on, then the government owed the compensat on; thus, the ncent ve system was ncompat b e w th the degree of deta ed author ty over the deve opment that the government wanted to reta n (see Scherer 1964, 179, quot ng U.S. Defense Department, "Incent ve Contract ng Gu de," 43). More genera y, the government s cost-p us- ncent ve-fee type of contract tr es to create an ncent ve system partway between that of a f xed-pr ce contract and a cost-p us-f xed-fee contract.

228 As one wou d expect, t appears that when the r sk s h gher, the contractors force the government to assume greater proport ons of the cost overruns (and correspond ng y, the contractors g ve up a arger part of the cost underruns, wh ch wou d be prof ts under a f xed-pr ce contract); that s, as one moves toward h gher eve s of nherent r sks and cost uncerta nty, one moves toward more h erarch ca ncent ve systems ( ower contractor shares n the cost r sks) (Scherer 1964, 308).

Standard Operating Procedures The st of documents requ red by the operator of an o f e d before t w pay a requ s t on (St nchcombe 1985e) s a contractua pract ce that constra ns everyone e se to produce the appropr ate documents; t prov des an order y way of proceed ng n mak ng purchases, a "standard operat ng procedure." But the compan es operat ng n the standard way produc ng the standard documents are the consu t ng eng neer ng f rms, the manufacturers and assemb ers of mach nery, the construct on contractors, and so on. It s the power mp c t n the r ght to award contracts that enab es the o company to set up a standard operat ng procedure for a comp ex of f rms nvo ved n supp y ng goods and serv ces to them. The purpose of requ r ng a these documents s on y rare y that the nformat on s needed for some dec s on by the c ent. Instead t s to make sure, as far as one can w th documents, that the contractor has fo owed rout nes spec f ed by the c ent n mak ng purchases, creat ng a tra of paper for aud t ng f necessary, prov d ng documentat on n case the regu atory author t es quest on a dec s on, and so forth. The nformat on, then, s most y not n the contents of the documents but n the r ex stence, and what that ex stence commun cates s that the contractor has n fact fo owed standard operat ng procedures spec f ed by the c ent. Tenders often requ re that the b dders on a arge pro ect prov de PERT d agrams or other forma schedu es w th est mated e apsed t me unt var ous pro ect "m estones" are reached. Genera y assoc ated reports on the ach evement of the m estones are requ red as we , so that c ents can est mate pro ect comp et on de ays as ear y as poss b e. Th s means n fact that c ents w usua y ntervene to speed up pro ects that fa beh nd. The ntervent ons often have the ega effect of transferr ng part of the r sk for de ays to the c ent, wh ch s cons dered to have taken respons b ty for meet ng the schedu e by tak ng t (part y) from the contractor s contro . The ega d sadvantage of nterven ng s often suffered for the sake of estab sh ng a standard operat ng procedure to govern the schedu e. Somet mes the standard rout nes are estab shed by trad t ona prac-

229 t ce. For examp e, the mar ne nsurance contract st conta ns a "sue and abor" c ause, wh ch was much more mportant before rad o commun cat ons. It requ red the po cyho der and ts agents ( n effect, the capta n of the sh p) to take whatever ega , phys ca , and commerc a measures were necessary after a covered acc dent to m n m ze the tota oss. C a ms aga nst the party ab e for the damage had to be f ed, repa rs essent a for the safety of the sh p had to be made, per shab e cargoes had to be so d, and so on. The costs were to be borne by the nsurer. Such prov s ons prov ded rout nes for transferr ng the capta n s agency re at on to the new pr nc pa , wh ch now wou d suffer the oss. The substant ve contents of the ob gat ons were def ned by the s tuat on (poss b e sources of oss), by the precedents n mar ne nsurance ad ud cat on, and by trad t ona requ rements for good udgment on the part of capta ns (He mer 1985a, 255–260).

Dispute Resolution The same "Eng neer" n Eng sh construct on contracts who s supposed to adm n ster d sputed change orders s a so genera y charged w th author tat ve ( nter m) nterpretat on of the contract n case of a d spute between the c ent and the contractor. The deo ogy here s that the Eng neer s an ndependent profess ona who, though emp oyed by the c ent, w g ve an ob ect ve nterpretat on of the contract. In Eng and tse f, th s usua y works to reso ve the d spute; when the contract anguage s exported abroad, however, t tends to resu t n an nter m c ent d ctatorsh p through the Eng neer s agency, because profess ona nst tut ons protect ng the Eng neer s autonomy are not we deve oped e sewhere. Many such contracts a so spec fy arb trat on n case e ther the c ent or the contractor d sagrees w th the Eng neer s dec s on. That s, a h erarchy of nom na y ndependent appea s s spec f ed w th n the contract, before the contract s appea ed to the courts. Under pressure from threatened eg s at on to protect franch sed car dea ers, car manufacturers deve oped a h erarchy of hear ngs and appea s, programs to he p agg ng dea ers, and vo untary compensat on for dea ers who were cance ed (usua y by arrang ng for a new dea er to buy them out), before app y ng the market sanct on of ook ng for an a ternat ve dea er. S nce the eg s at on was n fact neffect ve n protect ng dea ers, these procedures were protect ons for the dea er from the manufacturer s market udgment that t cou d do better e sewhere, rather than an actua form of precourt d spute reso ut on (Macau ay 1966). If these dev ces app ed to un on zed workers, they wou d be ca ed "gr evance procedures" and they wou d be taken as a c ear nd cat on of the deve opment of n-

230 dustr a ust ce w th bureaucrat zat on (Se zn ck, Nonet, and Vo mer 1969; see a so W amson 1975, 30, 73–80). There are a great many nforma equ va ents of these forma hear ngs nvo v ng a contractor and a c ent. Many tems on the agenda of a week y or month y meet ng between c ent personne and contractor personne are d sputes between the two organ zat ons: techn ca d sputes; d sputes over what costs w be a owed; d sputes over schedu es, qua ty of workmansh p, d srupt on of the c ent s bus ness, where to put the f d rt; or d sputes over a c ent s emp oyee who kes "to go over and ratt e the contractor s cage a b t." Often the purpose of such meet ngs s descr bed as "to c ar fy" the po nts of the agenda, wh ch avo ds nam ng them as d sputes.

Nonmarket Pricing Perhaps the surest nd cator that a c ent ntends to change a contract dur ng ts fe s pr c ng based on contractor cost. Th s ord nar y nvo ves comp ex procedures that supposed y ensure that costs are reasonab e, such as the st of procurement documents ment oned above or the comp ex cost-account ng pract ces requ red of defense contractors. The genera dea of such contracts s that the reward for the organ zat on s competence—the fee or prof t—shou d be separated from the costs nc dent on c ent change n spec f cat ons, c ent adaptat on to cost uncerta nt es, or c ent contro to secure fa thfu ach evement of unobservab es. Hosp ta serv ces, eng neer ng consu tant serv ces, R&D work, change orders n construct on contracts, "sue and abor" work of a sh p capta n n mar ne nsurance—a are usua y pr ced on a cost bas s. And cost pr c ng usua y means h erarchy. Cost pr c ng, though, n genera means comp ete y eav ng the market, and so g v ng up compet t ve contro s on pr c ng. When the techno ogy of a ne of work s not chang ng too rap d y, the nst tut on of a "quant ty surveyor" system, as used n construct on n Eng and, becomes poss b e. The quant ty surveyor takes the arch tectura and eng neer ng draw ngs of a pro ect and trans ates them nto quant t es of techn ca y d st nct act v t es. Thus for examp e, we d ng p pes together f fty feet n the a r s techn ca y d fferent from we d ng them on the ground and w have a d fferent pr ce. The contractor then b ds on the b of quant t es, g v ng a pr ce to a un t of we d ng f fty feet up, another pr ce to a un t of we d ng on the ground, and so on. Th s pr c ng system y e ds the contractor s b d on the contract. For the contract tse f, th s pract ce has the advantage of stat ng prec se y what s nc uded n the contract (the b of quant t es s taken as f na ), of avo d ng many contractor m stakes n

231 read ng draw ngs (mean ng that the b ds are ser ous), and of sav ng contractors est mat ng work so that more contractors are w ng to b d. But the contract usua y spec f es that a reasonab e amount of work under change orders w be performed w th the same pr c ng system. The quant ty surveyor takes the quant t es off the draw ngs for the change and determ nes the pr ce. Th s s not cost pr c ng post hoc, but pr c ng of work outs de the contract by a schedu e of pr ces per un t bu t nto the contract. But presumab y the pr ces per un t ref ect the contractor s est mate of ts m n mum costs p us a reasonab e prof t for the k nd of work spec f ed, under compet t ve pressure. Further, th s process produces a f xedpr ce ncent ve system dur ng the work on the change order, even w th the contractua prov s on of h erarch ca ntervent on by the c ent. It n effect uses market processes (and some very expens ve and t me-consum ng work by a quant ty surveyor) to get a market est mate of costs rather than a cost-account ng est mate of costs.

Contract Simulations of Hierarchies The ana ys s of typ ca contract contents n severa f e ds shows that when the extant theory pred cts vert ca ntegrat on but we do not f nd vert ca ntegrat on, we f nd nstead contractua prov s ons that may be expected to produce the effects of h erarch es. These prov s ons prov de for c ents to eg t mate commun cat ons that w be taken as author tat ve by emp oyees of the contractor; they arrange ncent ve systems by ru e part y outs de the market; they estab sh standard operat ng procedures that nvo ve regu ar f ow of nformat on between the contractor and the c ent and govern the procedures of the contractor as we as ts ob ect ves; they estab sh ru e-governed systems for mak ng fa r dec s ons n case of d sputes between the contractor and the c ent w thout recourse e ther to the courts or to c ent cho ce of another contractor; they pr ce var at ons n performance by nonmarket ru es, espec a y by costs or by quant ty measurements of outputs. These prov s ons often have a negat ve effect on the trad t ona ob gat ons under the contracts. The d ff cu ty when the Department of Defense sets up an ncent ve system to get the contractor to do what they ater dec de they do not want t to do, resu t ng n governmenta ab ty for the ncent ve even though the contractor does not sat sfy the performance requ s tes for t, s typ ca of the ega d ff cu t es one gets nto n wr t ng contracts w th h erarch ca e ements. The genera po nt s that one cannot requ re the contractor to do someth ng e se than s spec f ed n the contract w thout underm n ng requ rements e sewhere n the contract or accru ng ab t es that one wou d otherw se not have had. Wh e

232 much contractua ngenu ty has gone nto bu d ng h erarch ca e ements nto contracts n these f e ds such that the ega ob gat ons of both part es are c ear, ngenu ty does not a ways sat sfy the udge.

Theoretical Conclusion Our ma n descr pt ve conc us ons are embedded n the preced ng sect on. Br ef y they are that f ve e ements that make up h erarch es are often found n contractua anguage and that these e ements ach eve purposes of dea ng w th uncerta nt es that rat ona c ents w often want to dea w th. The fundamenta cond t ons that ca forth such contractua e ements are those that ord nar y produce h erarch es n the ord nary sense of the term. In a sense, the theoret ca conc us on s that the aw and the economy can arrange between f rms n the market everyth ng they can arrange n the abor market. When stated th s way, the ma n propos t on s unsurpr s ng and a pr or qu te ke y. It does not fo ow that there s no use form ng a concept of h erarchy w th f ve e ements (or some other number), an dea -type concept whose contrast ng concept s an dea -typ ca market transact on. Such a strategy w be usefu on y f n fact the var ous e ements cohere emp r ca y. S nce n our case they cohere ( f at a ) n a market sub ect to compet t ve evo ut onary pressures (Ne son and W nter 1982), we have to argue that the var ous e ements of h erarchy are funct ona y re ated. We have occas ona y noted above var ous emp r ca corre at ons n h erarch ca features of contracts: for examp e, the usua nonmarket pr c ng of change orders n construct on coheres w th arrangements for the Eng neer s h erarch ca command of those changes. We cou d go through the other n ne re at onsh ps between the f ve h erarch ca e ements we have found n contracts to show the r emp r ca re at onsh ps as we . To some degree our ana ys s of ex stent corporate h erarch es argues such an emp r ca corre at on. When one s ser ous about sett ng up a h erarchy for the ong run, one takes out a corporate charter, h res peop e, sets up an author ty system and an ncent ve system, reso ves d sputes nterna y, and the ke. The concret zat on of a f ve features n the norma corporate h erarchy therefore argues n favor of the emp r ca un ty of the concept of h erarchy. The funct ona un ty of the features of h erarchy s mp c t y argued above n the sect on ent t ed "Pred ct on of Performance Requ rements and Performance Measurement." The argument takes the form that a

233 f ve features of h erarch es (author ty systems, ncent ve systems, standard operat ng procedures, d spute reso ut on procedures, and nonmarket nterna pr c ng) are usefu when a c ent may want to change spec f cat ons, when a contractor or a c ent cannot pred ct costs very we , or when performances are not eas y separate y measured. The end resu t of th s argument, f t s accepted, s actua y a cont nuous d mens on between markets (or papaya-buy ng s mp e contracts) and h erarch es. For examp e, the work on a arge cap ta construct on pro ect done h erarch ca y under change orders s norma y about 20 percent. One cou d th nk of such a construct on contract, then, as norma y about 20 percent of the way between a pure market sa es arrangement and an ntraf rm h erarch ca arrangement. Perhaps the norma R&D contract for weapons s 80 percent of the way toward mak ng the contractor s eng neer ng department nto an arsena deve opment staff—a department of government (Stockf sch 1973; a so see note 6). There s noth ng wrong w th descr b ng such a d mens on, a ong wh ch dec s on systems range, by ts two ends, contract and h erarchy. It s mp f es theoret ca d scuss on. The cont nuous nature of the var ab e from " ns de" to "outs de" the f rm does, however, make t hard to do emp r ca work w th the d chotomy. When, for examp e, Teece g ves stat st cs on what percentage of the d str butors, p pe nes, or o we s are owned by the ref ners and conc udes that the o ndustry s not very vert ca y ntegrated (Teece 1976, 36, 63–64, 96), one wants to te h m to come n out of the ra n. In the ght of th s d scuss on, the W amson hypothes s (W amson 1975) can be restated n cont nuous form. [18] Rough y t s that the greater the poss b ty of chang ng c ent spec f cat ons, the more the cost uncerta nty, and the more d ff cu t t s to measure performances, the h gher the transact on costs of obta n ng performances n the nterf rm market rather than the abor market w be. Whenever these costs are not pa d for by the compensat ng advantages of contract ng (e.g., ow ng to be ng a construct on c ent for on y a short t me, or to d ff cu t es of R&D performance n government arsena s), we w expect to f nd h erarch ca adm n strat on n the everyday sense of the term. We mere y add to th s some spec f cat on of what contracts w tend to ook ke when the compensat ng advantages of contract ng are h gh. Another way to ook at these same theoret ca conc us ons s to ask, What e se bes des ega y b nd ng performance demands w we expect to f nd n contracts? Or, Why e se are th ngs put n contracts bes des to estab sh the r ght to damages f spec f c performances are not carr ed out? The bas c answer we have g ven s that the add t ons to contracts serve as the regu at ons of a forma organ zat on. A corporat on has a ot

234 of author tat ve "const tut ona " wr t ng that s not n the charter or other ega documents and that has the force of aw on y as ong as the rea aw s not ca ed n. Ru e books, memos w th enough n t a s, and m nutes of the meet ngs between the product on programm ng department and the shop super ntendent serve as such ega y unenforceab e forma arrangements n h erarch es. In a contractua re at on there s no other p ace to put the st of documents to be subm tted w th a request for re mbursement, or to spec fy who s n charge dur ng a d spute over pr c ng a change order, or to arrange other matters for wh ch one needs a h erarchy, except n the contract. The norma expectat on s that the c ent and the contractor shou d agree, forma y and n wr t ng, on such matters of o nt adm n strat ve structure but that t wou d ord nar y not be worth anyone s wh e to go to court over t. We wou d not be surpr sed to f nd ha fway through the contract performance a new ru e from the c ent s account ng off ce demand ng a new document before re mbursement (for examp e, a cert f cat on of the exchange rate at wh ch currency was bought to pay the subcontractor), and one wou d not expect the contractor to sue for damages over such an ncrease n performances demanded. Putt ng these prov s ons n the contract has the v rtue of exp c tness more than the advantage of ega b nd ngness. What usua y has to be made b nd ng under such c rcumstances s the overa contro of the ncent ve system by one of the part es. The near y h erarch ca factua arrangements descr bed by Macau ay (1966) for franch sed car dea ers were a ref ect on of the prov s on that cance at on was reserved to the manufacturer and that the manufacturer a owed a eve of prof t to the franch se ho der that made t worthwh e for the car dea er to to erate be ng pushed around. That s, the funct on of the ega y precar ous f ow of nstruct ons generated by h erarch ca structures bu t nto contracts s to set up a forma organ zat on—a h erarchy—that ncorporates e ements of the c ent organ zat on and of the contractor organ zat on nto a new un ty, under c rcumstances n wh ch the trad t ona theory n th s f e d wou d pred ct vert ca ntegrat on. When two powerfu nte ectua trad t ons have bu t powerfu but dea zed concepts, such as "market transact on" n econom cs or "contract" n aw, nto powerfu nte ectua systems, the concepts tend to shape percept on. Everyth ng appears as e ther the dea zed vers on or someth ng very d fferent—"h erarchy." It s a great advance to conce ve that d chotomy as a var ab e to be exp a ned, as Coase and h s successors have done. But th ngs n the wor d are hard y ever d chotom es, and they

235 are espec a y un ke y to be d chotom es when the economy sets a arge number of nte gent men and women ook ng for the ntermed ate ground. Our argument has been that n many sectors of the economy, such nte gent search for contractua means of creat ng sma h erarch es has been go ng on for generat ons. From the po nt of v ew of th s book, the centra troub e w th the econom c mode of a pr ce-quant ty transact on, and w th the ega mode of an enforceab e contract, s that by exc ud ng the uncerta nty of the future, co ect ng news about what that future has brought n, and co ect ng nformat on and mak ng dec s ons to a ow qu ck and rat ona response to that news, econom sts and awyers have e m nated the organ zat ona aspects of the networks of contracts. We have, however, agreed w th the markets-and-h erarch es trad t on that a h erarchy formed out of networks of abor contracts s a part cu ar y easy way to set up nformat on-co ect on and dec s on-mak ng systems; that comb nat on forms a const tut ona mechan sm by wh ch many such nformat on-co ect on subsystems can be created and ntegrated. But other types of const tut ons can organ ze and eg t mate the creat on of nformat on-co ect on and dec s on-mak ng structures. To f gure out what they do, we f rst ooked br ef y (aga n) at what h erarch es n fact ook ke, how they are des gned so that the r structure can be changed w thout chang ng who s n charge. Th s short account s very much a compressed vers on of what we saw Du Pont do ng nterna y when they ntroduced the r mu t d v s ona structure (see Chapter 4); t states the assumpt ons about what h erarch es are that are mp c t n the processes that h erarch ca Du Pont used to change how t responded to the news, to change ts system of mak ng dec s ons. These const tut ons genera y spec fy how the abor contract s concrete y used to create a system of author ty; how peop e at the top can be trusted w th other peop e s money; how an ncent ve system d fferent ated from the market can be set up to mot vate emp oyee performance, career deve opment, or subcontractor performance; how rout nes and standard operat ng procedures can be created and, more mportant, mproved; and how d sputes can be reso ved nterna y w thout the expense, de ay, and uncerta nty of the aw. Such a const tut ona system, n the markets-and-h erarch es trad t on, s what enab es h erarch es to outperform markets under cond t ons of uncerta n spec f cat ons or costs, of sma numbers n the market or f rst-mover advantages, or of other sources of "market fa ure." But our assent to the core of the markets and h erarch es trad t on, comb ned w th the observat on that networks of contracts between f rms

236 are the core organ z ng dev ces n many of the most uncerta n parts of the economy, suggests that we have to ook for how s m ar const tut ona systems are set up w th other types of contracts than those of the abor market. We f nd those const tut ona dev ces wr tten nto the contracts between f rms, where they are as ega y precar ous as they are n the aw of the abor contract or the aw of f nanc a f duc ar es that are the ega bas s of the const tut ons of ord nary h erarch es w th n f rms. Such organ zat ona dev ces, such pseudoh erarch es, do not need the certa nt es of contract aw to make them run, any more than su ts about the r ghts of emp oyees versus emp oyers are the norma way to make a corporate h erarchy run. The uncerta nt es of research and deve opment n the weapons bus ness, or of contract construct on, produce the prob ems of nformat on process ng that we have been ana yz ng throughout the book. When n Chapter 2, for examp e, we were pred ct ng that the construct on ndustry shou d have a h gh sk eve because of ts work- eve uncerta nty, we d d not need to not ce that those sk s were t ed together by contracts, rather than by a h erarchy, nto a s te organ zat on. Here we do not need to not ce that, when contracts create a const tut on n wh ch a comp ex adm n strat ve task of coord nat ng many d fferent tasks at a construct on s te can go on, the adm n strat ve task s substant a y s mp f ed by the sk ed workers be ng ab e to take up a ot of the uncerta nty, wh ch therefore does not have to be adm n stered at the h gher eve . The genera zat ons about the growth of nformat on-process ng structures toward sources of news about uncerta nt es that matter to the work do not depend on the work a be ng executed by a s ng e ega ent ty, "an organ zat on" as we wou d usua y formu ate t. Prov ded that n sett ng up the mu t organ zat ona system that does the coord nated work the network of contracts forms a const tut on that can create status systems, standard operat ng procedures, nterna d spute-reso ut on systems, and the ke, the network of organ zat ons can, n the re evant respects, act ke a h erarchy. Contracts as they actua y ex st n much of the economy are not to be exp a ned by a branch of econom c theory, nor by a branch of ega theory, but by a branch of organ zat on theory.

7— Segmentation of the Labor Market and Information on the Skill of Workers The Fundamental Uncertainty of the Labor Contract No f rm that h res a new person knows qu te what t s gett ng, or n part cu ar what t w get on the average over the e ght or so years t can expect to emp oy the new person. Even wh e the person s emp oyed the f rm does not know exact y what t has got. Let us start w th the measurement of performance after the worker gets the ob, s nce f that s mposs b e to obta n, pred ct ons of that performance at the t me of h r ng cannot he p but be yet more nexact (March and March 1978). We w ana yze f rst performances of ath etes n profess ona ath et cs, for stat st ca measurement of performance s sure y better deve oped there than n pract ca y any other area (Kahn and Sherer 1986). Measurement s done by outs ders rather than by management, so that managers need not dec de on a cost bas s how carefu y to measure performance. Ath etes on baseba , basketba , footba , or soccer teams work under standard cond t ons, cond t ons very near y dent ca to those of peop e who p ay the same pos t ons on other teams. The ob ect ves on a teams are dent ca : to score n a we def ned set of ways n a standard number of games aga nst a set of teams that are near y dent ca n compos t on. The performance appropr ate for d fferent pos t ons s a so measured n standard ways, so that n baseba , for examp e, batt ng averages, earned run averages, or f e d ng averages span performances appropr ate ( n d fferent m xtures) to outf e ders, p tchers, and nf e ders. The d ff cu t es w th even th s h gh y deve oped measurement system for the work of ath etes are we known. The f rst, most obv ous d ff cu ty s that the measurements are stat st ca , and (g ven that ha f of a the teams ose and the other ha f w n, except n games ke soccer where t es are common) they are stat st ca under cond t ons where certa n k nds of uncerta nty are max m zed as far as the oppos ng teams and the overa

241 managers of the compet t ve system can manage (for a descr pt on of the nforma management of such a system of compet t on, Ba nese cockf ght ng, see Geertz 1973a). Th s means that every measurement s on y as stat st ca y re ab e as the samp e s ze perm tted by a season of games (or the season so far) makes t. For examp e, scor ng s a very unre ab e measure n soccer, because the expected number of scores per person n a season s very ow. P tchers who p tch few games, or pass ng by backs other than the quarterback, present comparab e prob ems. But these are on y extreme forms of the common prob em of random measurement error n a performance stat st cs. In add t on, d fferent sports depend to d fferent degrees on teamwork, and d fferent pos t ons or ro es n any one sport matter more or ess to the performance of others. Thus, nd v dua performance stat st cs cannot adequate y summar ze performance, because the qua t es ref ected n one person s scor ng (or other performance) nc ude qua t es of teammates. Th s fact has shown up h stor ca y n Amer can sports n the s ower rac a ntegrat on of the pos t ons of p tcher and catcher n baseba or of quarterback n footba , where the mportance of the pos t on s contr but on to teamwork makes measurement of performance d ff cu t. Some measures of performance (e.g. the number of doub e p ays n baseba ) are strong y affected by nseparab ty of teamwork performances. F na y, some star performers are very popu ar w th the fans, wh e some are a good dea ess popu ar. Except for the factor of race (where t s fa r y c ear that n basketba a wh te star s more va uab e to a team than a b ack star of equa performance, because there are more wh te fans than b ack fans, and wh tes prefer a star of the r own race; see Kahn and Sherer 1986), the ab ty to get favorab e not ce n the news and br ng fans to the stad um s not eas y measurab e. So even n ath et cs, under the best of cond t ons, emp oyers do not know qu te what va ue they are buy ng for what they pay a worker. Further, the very exce ence of the measurement demonstrates a second po nt, that the performance of ath etes over the course of the r p ay ng fe and the ength of that p ay ng fe are not we pred cted by scouts or by recru tment contract rank ngs. Sports recru ters f nd out more sure y than other personne departments when they have made a m stake n udgment ( n fact, the m stake rate of personne departments may be the worst-measured performance character st c of the bus ness wor d, and sure y s n the c v serv ce). But of course, many sports have h gh n ury rates as we . The ong-run performance of an Amer can footba runn ng back, even w th a ong ser es of knee operat ons, s not very we pred cted by whether the knees he d up through co ege.

242 Th s d ff cu ty of pred ct ng an ath ete s performance at the t me of recru tment occurs even though the s tuat on n wh ch the measurement of the ta ents and tra n ng of the recru t takes p ace— name y, co ege or amateur sports or farm c ubs—s mu ates the actua work s tuat on fa r y we . Th s d ffers from the s tuat on encountered by a recru ter of, say, eng neers, who cannot exam ne many br dges bu t by a budd ng c v eng neer, not even br dges bu t n a farm c ub that s not n the same eague as a ma or c v eng neer ng f rm. Consequent y the contract of an ath ete n any g ven year of h s or her career depends nt mate y on current measurements of product v ty, rather than on cert f cates cons st ng, perhaps, of batt ng averages from h s or her co ege or farm c ub days. The bonuses for s gn ng are of course a part a est mate of how much of the ath ete s tota va ue can be re ab y pred cted at recru tment t me. In other parts of the economy, the abor contract has sen or ty prov s ons aga nst ayoffs and for mov ng up to better obs by bump ng ess sen or workers or for tenure after a s x-year tr a per od, types of ev dence of emp oyer and emp oyee unw ngness to make cont nuat on and reward depend on current performance. These structures of tenure are w ped out n profess ona ath et cs by the poss b ty of good measurement of current product v ty. If sen or ty or tenure n the abor contract nd cates poor measurement of current product v ty, the emp oyer—and even more, perhaps, the emp oyee—must be uncerta n about what the emp oyer has n fact bought. Th s uncerta nty s ower n ath et cs than n other sorts of work. Part of the reason the poss b ty of measurement and pred ct on s so much worse n most obs than n ath et cs s that the emp oyer outs de ath et cs w not be p ay ng the same game as now n the future: the market w change, the techno ogy w be mproved, a d fferent sort of teamwork w be needed because d fferent groups w be work ng together, and so on. The art f c a r g d ty of the ru es of sports makes t poss b e to pred ct what one w want n e ght years n a th rd baseman. In e ght years n the automob e ndustry, Genera Motors may be compet ng w th the Japanese on qua ty rather than w th Ford on sty e and pr ce, and so may be want ng workers who w suggest how to avo d m stakes rather than workers who can keep the ne go ng whether the carburetor gasket f ts or not. An emp oyer outs de ath et cs wants to recru t now someone who w be good at do ng whatever the f rm dec des to do n e ght years, rather than someone who w st be good at what a th rd baseman has a ways done. Part of the reason the abor market as a who e s ess good than ath et cs at pred ct ng product v ty s that schoo s are not n genera good

243 s mu at ons of work fe; the measurements at career beg nn ngs are of what makes potent a recru ts good at schoo work, w th an overemphas s on nte gence and an underemphas s on qua t es of character. Insofar as schoo recommendat ons do comment on qua t es of character, we wou d a be better off to gnore them, because ord nar y the teachers who wr te the etters have very tt e opportun ty to observe those parts of fe n wh ch such qua t es p ay a great ro e. Part of the reason the abor market as a who e s worse than ath et cs at measur ng product v ty and reward ng t s that emp oyees w not stand for rewards based on performance un ess they get a ot out of t. One can h re profess ona ath etes even though one s go ng to f re them at rough y th rty-f ve years o d on the bas s of dec n ng performance stat st cs, to w thho d sen or ty pr v eges, to make the r future ncome uncerta n depend ng on such th ngs as the success or fa ure of knee operat ons, on y because one s w ng to pay a ot for the years they work. Not know ng what one s buy ng, or even what one has bought, makes the abor contract a perenn a nformat on prob em for management. It m ght be mag ned that at east the worker knows what he or she produces, but th s a so s not true. What workers conce ve as va uab e to the f rm s strong y determ ned by the obs they occupy. Manua workers typ ca y be eve that what rea y ho ds a f rm together s phys ca product on; c er ca workers be eve that f the f ow of paper does not go r ght, everyth ng w come apart; managers be eve that the bott eneck factor n company success s management; f nanc a off c a s of the company be eve that the proper structur ng of debt and equ t es s the fundamenta determ nant of the m xture of r sks and prof ts of stockho ders and debt ho ders, and so determ nes whether or not the f rm w cont nue to ex st. Thus, at east some of them must overest mate the r product v ty for the f rm. And emp oyees very genera y have v rtua y no dea what the r ta ents wou d br ng on the open abor market.

Institutional Substitutes for Measurement of Productivity Mut p e personne nformat on systems n organ zat ons and n the abor market are d rected at f ng n the gaps n th s p cture of poor measurement and poor pred ct on of the product v ty of abor. These systems of nformat on nc ude cert f cates from the educat ona system and somet mes from the profess ona system as we (e.g., spec a ty boards n med c ne); un on membersh p n the crafts; promot on dec s ons n nterna

244 abor markets; nforma respect n the work group; whatever exper ence f rms have about whether men or women work out better under the sa ary, promot on, and ncent ve systems they typ ca y set up for c er ca workers; ob eva uat on schemes; recommendat ons by workers n the p ant about new h res (a remarkab e proport on of workers are recru ted th s way); footnote c tat ons that d str bute honor n the sc ent f c commun ty and so nd rect y determ ne appo ntments, promot ons, and sa ar es n un vers t es; and so on. The nformat on that eads emp oyers to prefer one k nd of workers to another (e.g., that eads arge manufactur ng corporat ons to prefer peop e a ready emp oyed there but that eads un vers t es be ow the very f rst rank to prefer not to h re the r own graduates as ass stant professors) means that d fferent segments of the abor force w have d fferent opportun t es: d fferent ncomes, d fferent eve s of emp oyment secur ty, d fferent chances of promot on, and the ke. Consequent y, abor market segmentat on s fundamenta y a so ut on to a prob em of the nformat on structure about peop e and the r performance that confronts organ zat ons and emp oyees. But s nce peop e s rewards depend on these f ows of nformat on, they have an nterest n hav ng the f ows conta n nformat on that says they are better than others, whether that nformat on s trustworthy or not (Edwards, Re ch, and Gordon 1975). The abor market part of the prob em of organ zat ons s shot through w th agency prob ems, prob ems of the two part es to a contract hav ng d fferent amounts and k nds of nformat on and want ng to use those nformat on d fferent a s to the r own advantage. The examp e of an agency prob em we gave n Chapter 1—whether a secondary schoo teacher s cert f cate of aud ov sua competence was nformat on about whether that teacher s ke y to se ect good nstruct ona f ms—demonstrates the deep prob em of gett ng unb ased nformat on about worker product v t es. And t s th s mposs b ty of ga n ng good nformat on about emp oyee performance that creates many opportun t es for emp oyees to monopo ze opportun t es, by monopo z ng the chance to generate good nformat on about the r performance. We f nd re at ve y tt e monopo zat on n profess ona ath et cs (for nstance, the res dua rac sm n ath et c teams h r ng and promot on s not at a comparab e to the rac sm or sex sm that st ex sts for obs pay ng comparab e sa ar es n corporat ons or un vers t es), because n sp te of the caut ons we have out ned above, one can measure the product v ty of ath etes very we ndeed. The purpose of th s chapter s to exam ne the nterp ay of nformat on on performance, agency prob ems b as ng the nformat on, and monopo zat on of opportun t es by segments of the abor market.

245

Types of Information about Work Performance An organ zat on needs nformat on about the mmed ate past work performance of a worker so as to a ocate rewards fa r y, and t needs nformat on about future work performance n dec s ons about h r ng, promot on, and retent on. In order to be used as a bas s for rewards, nformat on needs to be, to some degree, "aud tab e," so that peop e do not s mp y get to choose the r own reward eve s. In order to be usefu for pred ct ng future performance, an emp oyer needs nformat on about response to a range of cond t ons that fa r y represent the expected future of the work the person w do, and aga n the nformat on needs to be aud tab e, so that workers do not un atera y dec de on the r retent on and promot on and potent a recru ts do not determ ne who shou d be h red by prov d ng fa se nformat on. The genera prob em w th nformat on systems about emp oyees s that the more accurate and w de-rang ng a person s observat on of the worker s, the more ke y t s that the person w be a fe ow worker (or a fe ow student). Th s br ngs up two prob ems: the fe ow worker may not be competent to render a udgment that wou d be accepted by a reward-g v ng or h r ng execut ve; and the fe ow worker may have h s or her own fate nt mate y ntertw ned w th the nformat on g ven, and so be mot vated to w thho d or d stort t. Informa pressures among fe ow workers tend to d scourage g v ng accurate nformat on to management, espec a y f that nformat on s negat ve (see, e.g., Ha e 1984, 121–122, 138–144, 161–162). Informat on from a compet tor for the same promot on m ght, however, be shaded n a negat ve d rect on for reasons of rat ona nterest, wh e nformat on from a fe ow worker m ght be negat ve y shaded for vengeance f the worker had been too attent ve to the eva uator s w fe. Informat on from h erarch ca super ors tends to be based on ess deta ed observat on but to have d fferent, genera y sma er, agency prob ems (but see the ana ys s be ow of why un on f rms are more product ve, for contrary ev dence). The nformat on that a super or g ves about the worker s performance g ves the super or part a contro over a reward for that worker. The super or, then, may trade the benef t of g v ng more favorab e nformat on to super ors for cooperat on. Wh e the organ zat on genera y benef ts, n the short run, from the cooperat on, the degraded nformat on about the worker s a cost that must be pa d n return. Of course, the cooperat on requ red of a superv sor does not a ways contr bute to organ zat ona ob ect ves, as sexua harassment cases demonstrate (cf. Char ton 1983).

246 Informat on s therefore often based on measures of performance that are eas er to aud t, such as un ts produced, hours recorded by a t me c ock, sa es made, or (h gher up the h erarchy) costs reduced, prof ts ncreased, de very t mes reduced, scrap f gures down, and the ke. There are four broad types of d ff cu t es w th such measurements. The f rst s that they too can be man pu ated by nterested workers on the scene; for examp e, barga n ng between workers and f rst- ne superv sors about how much t me shou d be a owed aga nst a p ece rate for down t me of a mach ne s common n mach ne shops, because management cannot get accurate measurements of that down t me from anyone but the workers and superv sors nvo ved (Burawoy 1979). The second d ff cu ty w th forma performance measurement s that the measures used may not be good est mates of true performance. In shoe factor es n the Sov et Un on where performance was measured by numbers of shoes produced under cond t ons of eather shortage, for examp e, factor es produced on y shoes of the sma est s ze (Gran ck 1954). When respons b ty n dea ng w th extraord nary c rcumstances s a core funct on of a ob, rout ne measurements of product v ty tend to be worst exact y when the workers are do ng the extraord nary parts of the r obs the best. For examp e, more so d ers under one s command are k ed wh e one s successfu y attack ng or defend ng, and product v ty of a chem ca p ant s owest when the workers are putt ng out a f re. The th rd d ff cu ty s that such aud tab e measures tend to r g d fy the measurement of performance, so that techn ca change or reorgan zat on underm nes the measurement system. W th r g d performance measurement, the nterests of workers embedded n the measurement system tend to ead to sabotage of techn ca and organ zat ona change. Fourth, such spec a zed measurement s genera y expens ve. One st needs the pyram d of superv s on even f product v ty s measured separate y. One needs cooperat on and the transm ss on of nforma sk s w th n the work group (and one has to pay the costs of cooperat on) to dea w th the ord nary cont ngenc es of work, even f peer udgments are not used n the personne process (cf. B au, [1955] 1963). Because, for examp e, one cannot subst tute the records of product v ty, produced by an nspect on department, for superv s on by the foreman of a mach ne shop, measurement of performance by nspectors s an extra expense. Spec a performance measurement therefore tends to be narrow y focused on a few aspects of performance and to be done at a h gh eve of abstract on, s nce t wou d be too expens ve f t were not rad ca y s mp f ed. For h r ng and promot on dec s ons one needs to measure performances that w pred ct what k nd of worker a person w be n the fu-

247 ture. Even f one or the other of the performance measurement types out ned above has been carr ed out before the h r ng or promot on, pred ct ve nformat on has four d ff cu t es add t ona to those assoc ated w th the measurement of current and past performance. F rst, the performance that the h rer or promoter tr es to measure may not pred ct the sorts of performances that w be requ red n the future ob. Ord nar y the person do ng the h r ng cannot pred ct exact y what the performance requ rements w be for the ob be ng f ed, because of techn ca and organ zat ona change or because personne off cers do not know much about the ob themse ves. A cruc a factor n the pred ct on may be how we a person s accepted by the future fe ow workers and superv sors who can teach h m or her to do t we ; personne peop e can usua y pred ct th s on y by us ng crude categor es such as ethn c ty and sex, and those are not very good pred ctors. A second add t ona troub e s that the performances measured prev ous y may be systemat ca y d fferent from those requ red n the ob a person s be ng recru ted for or promoted to. Often much of the nformat on for recru tment s about schoo ng, and the performances measured there (typ ca y tests n courses or genera ab ty tests), wh e fa r y we aud ted, are not good samp es of the tasks requ red for success n obs. Work done n a ower pos t on s often not representat ve of what the worker w need to do n a h gher pos t on (part cu ar y n terms of "respons b ty," management of peop e, and management of nformat on f ows at a h gher eve of abstract on, a of wh ch tend to be more mportant n h gher pos t ons), so that exce ent performance ower down may not pred ct good performance h gher up (th s uncerta nty can be part y contro ed by ett ng peop e who m ght be promoted p nch h t for super ors when those are s ck or on vacat on). A th rd broad add t ona d ff cu ty w th recu tment nformat on s that peop e prov d ng the nformat on often have no comm tment to the organ zat on they are supp y ng nformat on to, so that etters of recommendat on a rank peop e as among the top 10 percent. Th s s one reason why the more aud tab e parts of peop e s prev ous exper ence (grades rather than recommendat ons, and grades n ca cu us rather than grades n soc o ogy) tend to overemphas zed. A fourth broad d ff cu ty w th recru tment and promot on nformat on s ack of comparab ty. It s we known that t s harder to get good grades at Swarthmore or Reed or some other ead ng p aces than at ess d st ngu shed co eges and un vers t es, but there s no genera formu a to trans ate grades from one schoo nto the r equ va ents at another. S m ar y, w th n a company some superv sors w be we known to g ve gen-

248 era y very h gh rat ngs to the r subord nates, others to be somewhat tougher, but no genera trans at on to a standard degree of softness or toughness s ke y to ex st. Even w thout such var at ons n the standards of the udges, there wou d be d fferences n what s be ng measured. An eng neer who has been do ng an exce ent ob se ng a rp anes produced by a part cu ar company does not show the same v rtues n h s or her work as an eng neer who has been do ng an exce ent ob des gn ng eff c ent heat transfer from the exhaust to the ncom ng a r and fue of the eng nes of those a rp anes. Even f both workers have subst tuted successfu y for the r super ors, the r super ors have d fferent obs as we . Thus, pred ct on of future performance s even more d ff cu t than measurement of past and current performance.

A General Theory of Certification By a "cert f cate" we w mean a summary of nformat on about some performances of a worker or a cand date for a ob that has been sub ected to some sort of spec a procedures that are thought to make t more trustworthy (the genera approach of th s sect on s taken from He mer 1984; see a so Shap ro 1987). The f rst th ng we have to do, then, s to d st ngu sh d fferent cert f cates accord ng to the degree of forma zat on of both the samp ng of performances and the protect on of the trustworth ness of the udgments. At the extreme ow end of forma ty s the "cert f cate" of sen or ty n a part cu ar ob. The un verse samp ed n the know edge that a person has he d a ob for some t me s the correct un verse for assess ng performance n that ob ( nsofar as the ob s not pred cted to change much), but the samp ng process te s us on y that no negat ve nformat on has come to ght that was suff c ent y ser ous to f re the worker. No part cu ar efforts have ord nar y been made to check and protect the trustworth ness of the nformat on, to make sure that everyone who shou d be f red has been f red. At the h gh end of the var ab e of forma ty s the degree (w th a transcr pt) from a part cu ar co ege or other schoo , wh ch ord nar y spec f es n what courses performance was samp ed and cert f es that the udgments were made by peop e re at ve y un nterested n prov d ng spur ous nformat on. One can be reasonab y conf dent that much of the nformat on was der ved from penc and paper tests or from wr t ng ass gnments, and that the sub ects of those tests wou d be recogn zab e by someone who had taken a s m ar course at another schoo . Cert f cates can be produced as an offshoot of a the types of nforma-

249 t on systems descr bed above, by fe ow workers or bod es of peers, by super ors g v ng an overa udgment, by systemat c performance measurement systems of a more or ess numer ca and aud tab e sort. Those prov d ng the nformat on may have samp ed performances h gh y pred ct ve or ess pred ct ve of ob performance, may have more or ess comm tment to prov d ng the h r ng or promot ng organ zat on w th accurate nformat on, and may be prov d ng nformat on n a common metr c (such as Graduate Record Exam nat on scores) or n noncomparab e form. The centra soc o og ca po nt of a th s s that a nformat on that s taken nto account n recru tment, retent on, or promot on tends to need eg t mat on. The ma n reasons for th s are pract ca : performance measurement requ res observat on of a worker or recru t n performances that are re evant to the work he or she s to do; udgment of that nformat on requ res that the udge too be competent n the same performances; and transm ss on of the nformat on requ res honesty and d scret on on the part of the transm tter. Second, t s not on y performances n schoo that become cert f cates. Sen or ty s a most a ways a part of the forma record, and s rough y of equa mportance to educat on n determ n ng ob eve , ob secur ty, and pay. Sen or ty s a cruc a cert f cate of qua f cat on, part cu ar y for the ob a worker now ho ds, because what t acks n systemat c measurement and trustworth ness of the udgments t makes up n re evance. The performances of wh ch t s a sort of record are those that are cruc a for do ng the ob. If abor market segmentat on s due to d fferences n var ous popu at ons opportun t es to g ve acceptab e cert f cates of the r competence n a ob, then a b g source of segmentat on s the tendency of most obs to be awarded next year to the person who ho ds them th s year. Sen or ty s a so one of the components n promot on systems or " nterna abor markets" that character ze arge bureaucrat c f rms, the var ous eve s of government (espec a y n educat ona , hea th, and m tary branches), and certa n arge nonprof t organ zat ons such as pr vate un vers t es. S nce n genera not a workers get the same number and s ze of promot ons, and s nce educat ona qua f cat ons determ ne wh ch promot on adder d fferent peop e are on w th n the same organ zat on, sen or ty n th s case s more comp ex than n the case of a ob qua fy ng a worker for that same ob. Most promot ons go to whatever groups are now emp oyed n arge bureaucrat c organ zat ons, espec a y to those who entered from co ege. We have suggested n Chapter 2 some of what sen or ty n pos t ons to wh ch one s recru ted from h gher educat on may be cert fy ng.

250 The other ma n source of cert f cat on n nterna abor markets, aga n espec a y n those career adders that start w th recu tment from h gher educat on, s the forma op n on of super ors. As one gets nearer to the top of the h erarchy, more of such op n ons are formed n comm ttee rather than by nd v dua super ors, wh ch means that to some degree the top ranks of the " nterna abor markets" of arge organ zat ons must be treated as a subtype of craft and profess ona cert f cat on. At the top peop e are cert f ed by a body of peers. The th rd ma n type of cert f cat on s based on observat on of work by peop e a ready cert f ed as sk ed n that work, and t s protected from the most extreme forms of nterpersona ogro ng by a "comm ttee of peers"–type structure. As Am ta Etz on has po nted out (1975), comm ttees of peers are actua y made up of one s super ors, not one s peers. (The or g na not on presumab y was that the peers are peers of one another, ke the House of Lords, not peers of the cand date.) What the comm ttee of peers structure means s that no one person s a part cu ar h erarch ca super or who has the cert f cat on w th n h s or her own power. Instead, for examp e, the assoc ate and fu professors of a department meet as a comm ttee to recommend to the dean whether a g ven ass stant professor shou d be promoted, or the body of craftsmen meet to cert fy someone as a ourneyman. Interna abor markets n corporat ons and arge bureaucrac es tend to be more craft ke toward the top than they are n the m dd e, because as one advances nf uence over promot on chances beg ns to pass away from a s ng e h erarch ca super or toward a group, the "management comm ttee" or, n the extreme, the board of d rectors. The bas c dea of peer cert f cat on systems s that the true requ rements of a ro e can be taught and udged on y by those who can do the ro e themse ves, who have teach ng and superv sory respons b ty for the cand date for the ro e, and who have mu t p e nforma contacts w th the cand date. Qu te genera y a strong sen or ty e ement s bu t nto these cert f cat ons, as n the f xed term of apprent cesh p n many trad t ona crafts or the f xed term of nternsh p or res dency of phys c an tra nees. The expectat on s often that anyone adm tted to the peer tute ary per od can earn to perform the ro es nvo ved, so years spent n the apprent cesh p are n effect the on y measures of competence n the cert f cate. Qu te often, however, there are add t ona exam nat ons or performances (e.g., the trad t ona "masterwork" of some art st c crafts) or add t ona exam nat on of the work (e.g., the assessment of pub shed work by a comm ttee on a promot on n a un vers ty). So wh e ass stant professors rare y get promoted w thout a respectab e f ve years sen or ty n

251 the ob, they may not get promoted even w th the sen or ty because the r work s not udged to be up to the standard of the profess on. We w expect the features of the cert f cat on system for part cu ar pos t ons to vary w th the character of the nformat on prob ems nvo ved n those pos t ons. In genera , the more the cruc a uncerta nt es of the system n wh ch a ro e s p aced are dea t w th n the ro e tse f (as, for examp e, med ca d agnos s n hosp ta s s dea t w th by phys c ans, not managers), and the more the nformat on to dea w th those uncerta nt es tends to be concentrated at the owest eve , the more one has to re y on ro e ho ders (peers) for udgments. Thus n genera , the h gher the sk eve of a ro e, and the more mportant the uncerta nt es dea t w th by workers sk s n the organ zat on as a who e, the more the organ zat on must tap the nformat on he d by peers, that s, by the sen or peop e who perform the ro e themse ves and who work w th the cand date. When t s d ff cu t for peers to exp a n what t s they do and to udge whether someone e se s do ng t we , and when bureaucrat c super ors cannot measure output we because of poor observat on or ncommensurab ty of outcomes, as n secondary schoo teach ng or po ce patro work, then very genera cert f cates of appropr ate character and background for the work are comb ned w th sen or ty (for a genera descr pt on of po ce and teach ng promot on and sa ary systems, see Sp erman 1986, 54–65), but usua y w th re at ve y sma reward for sen or ty. Where the sk tse f s obscure, but the output of nterest to the organ zat on s read y measurab e and aud tab e, as among who esa e or rea estate sa espeop e or ath et c stars, cert f cates of competence and sen or ty usua y p ay a very sma ro e and rewards are based on measured performance. Where cruc a uncerta nt es have to be dea t w th by teamwork, mak ng nd v dua performance d ff cu t to measure, as n cont nuous process manufactur ng, sen or ty m xed w th h erarch ca super ors udgment tends to dom nate. These genera cons derat ons are comb ned n d fferent ways nto f ve ma n k nds of nst tut ona systems for cert fy ng peop e for obs n modern econom es, wh ch we w out ne n the next f ve sect ons; these systems are based on (1) cont nu ty or tenure n the present ob; (2) promot ons from w th n n arge organ zat ons; (3) craft and profess ona cert f cat on by peers; (4) recru tment of a person from w th n the fam y or of onese f as owner; and (5) the odd case of cert f cat on through work n a un on shop. After exam n ng these ma n nst tut ona systems for cert f cat on of competence, we w ook at how the opportun t es to prov de nformat on and to have t d ssem nated to the peop e do ng the

252 h r ng are d str buted. If a peop e needed to get a ob was to have the competences to do t, the abor market wou d be much more compet t ve than t now s, because competences can be earned n a sorts of ways. But one has to earn a competence n such a way that trustworthy nformat on about that competence can reach a potent a emp oyer at a t me when that emp oyer s mak ng a dec s on about whom to h re (for a fu er ana ys s, see He mer 1984). S nce one of the ma n forms of cert f cat on s be ng a ready n the ob, and f someone sat sfactory s n the ob an emp oyer does not make a dec s on, compet t on takes p ace ma n y on the rare occas ons of ob vacanc es. But even n the case of ob vacanc es, on y certa n types of nformat on are acceptab e cert f cates to a potent a emp oyer for pred ct ng a potent a recru t s future performance. And s nce the nformat on embedded n those cert f cates s generated n nst tut ons that produce the opportun ty for a person to perform, produce observat ons on that performance, and guarantee to some degree the ntegr ty and comparab ty of the report of those observat ons, not everyone cou d conce vab y prov de cert f cates of competence because not a have access to those nst tut ons. Th s s tuat on produces a strat f cat on of opportun t es n the abor market. In genera , those who cannot produce cert f cates of spec a competence occupy p aces n the "secondary" abor market, a market of open compet t on for bad obs.

The Great Segmenting Factor Is Who Holds the Job Now Around 1970 the status atta nment terature, the terature about who got what obs and how soc a or g ns nf uence what obs they got, assumed that everyone got the h ghest occupat ona status they cou d, so t assumed un form ty of the va ues of workers choos ng obs (though these authors never d scussed ust what they were assum ng). The "causes" n th s terature, then, were amounts of peop e s resources, espec a y human cap ta resources, such as educat on, and ord nary cap ta or wea th, wh ch fac tate or h nder that max m zat on (the c earest statement of th s theory I have seen s Ke ey 1972); n other words, the causes were not var at ons n sub ect ve processes (e.g., n how much peop e va ue prest ge versus money returns, or secur ty versus h gher average ncome), but rather var at ons n the s tuat ons that determ ne peop e s poss b t es. In the past decade, bes des prov d ng the ma n substant ve mater a that has been used n the deve opment of soc o og ca methodo ogy, the

253 status atta nment terature has prov ded an env ronment n wh ch var at ons n abor market demands by f rms have become pr nc pa var ab es for exp a n ng the d str but on of abor market outcomes. And the new var ab es are prov ded by nst tut ona features of the emp oyers . Overa , the centra nst tut ona feature of emp oyers s that, by and arge, they w emp oy the same workers n the same obs th s year as they d d ast year. For examp e, the average ob occup ed around 1970 by a ma e worker under f fty-f ve fa s vacant n Yokohama about once every e ght years, ow ng to e ther promot on w th n the f rm or movement between f rms; n Detro t such a ob wou d fa vacant about once every four years (Co e and S ege 1979, 70–71). So n Japan, about seven-e ghts of the obs occup ed ast year wou d st be occup ed th s year by the same man; n the Un ted States, about three-quarters wou d be. In 1970, the typ ca ma e worker between th rty-f ve and f fty-f ve n the Detro t study (Co e and S ege 1979), eft the f rm he worked for on the average of every fourteen to twenty years; n Yokohama n 1971, the typ ca ma e worker eft the f rm on the average of every th rty-three years. If these men d d not move among f rms, then chances are they wou d change obs w th n the f rm on y once n s xteen to twenty years n Detro t, once n twenty years n Yokohama. On average, then, f a ob s occup ed by someone over th rty-f ve, nobody s go ng to move nto t any more often than about once every ten to f fteen years (rough y ha f the obs be ng vacated by promot on, the other ha f by the worker eav ng the ob), n e ther Japan or the Un ted States. Jobs occup ed by young workers are more ke y than those occup ed by o der workers to fa vacant because of both k nds of mob ty. Jobs occup ed by workers n arge organ zat ons are more ke y than those n sma organ zat ons to fa vacant due to promot on , wh e those n sma organ zat ons are more ke y to fa vacant because of mob ty among f rms . But the overwhe m ng nst tut ona fact about the abor market s that the boundary around the worker- ob pa r s the most mpermeab e of abor market boundar es—among markets, perhaps on y the hous ng market for owner-occup ed hous ng has a ower rate of movement (the hous ng market as a who e, at east n the Un ted States, has very rough y the same rate of movement of peop e as does the abor market). H gh-sen or ty workers n m dd e age move out of the r obs about as often as homeowners move out of s ng e-fam y houses that they own. Th s means that the overa structure of the abor market ref ects the mob ty patterns of peop e now work ng wh ch took p ace up to forty years ago when they were young. The tendency of peop e to reta n the r

254 obs means that ho d ng a ob now does not represent an equ br um created n the t me ess nstant of econom c theory. About one-s xth of a awyers n the Un ted States are women, even though about one-th rd of a aw schoo graduates n 1987 were women; past patterns of mob ty nto the ega profess on w be cast n stone unt the t ght person- ob t es between ma es and obs as awyers are d sso ved by ret rement or death (cf. L eberson and Fugu tt 1967 on the comparab e prob em for the races; Cap ette 1982). A sorts of f rms have " nterna abor markets" n the sense that the peop e who w occupy the obs next year are most y n the f rm a ready. What th s comes down to s that the f ex b ty of the abor market n terms of mob ty s ma n y f ex b ty over what to do w th the new cohort—peop e between twenty and th rty who are sett ng nto the obs that they w most y then ho d (except for promot ons, f they are n the b g f rm–government sector). How th s group sett es n w heav y determ ne what the abor market ooks ke th rty to forty years from now. If 33 percent of aw schoo graduates, but on y 10 percent of eng neer ng schoo graduates, are women, and f entry rates n those f e ds are about the same between 1986 and 1990, then n 2020–2025 the new recru ts w be between s xty and s xty-f ve years o d. A tt e over 33 percent of the awyers and a tt e over 10 percent of the eng neers aged s xty to s xty-f ve n 2020 (a few more of the men w have d ed than the women) w be women. To put rea numbers on th s, n 1980 about 65,000 eng neers out of 1,401,000, or 4.6 percent, and about 74,000 awyers and udges out of 530,000, or 14.0 percent, were women (Stat st ca Abstract of the Un ted States, 1984 , 416). In 1981, about 7,700 out of 75,000 eng neer ng graduates were women, or about 10.3 percent (D gest of Educat on Stat st cs 1983–84 , 120); for aw the f gures are 11,768 out of 36,331, or about 32.4 percent ( b d., 126). By and arge unt 2020, the fo ks now n the profess on w shape the sex compos t on of the ega and eng neer ng e tes. At around that t me present aw and eng neer ng graduates w have moved nto the e te to rough y doub e the respect ve percentages of women. So the fact that fem n sm has had a ot ess effect on eng neer ng tra n ng than on ega tra n ng recent y w st be a fact set n stone n the e te of the eng neer ng and the aw e te abor force n 2020. Th s means, however, that when the abor market aggregates do move because of some outs de cause, such as a decrease n the number of e ementary schoo –aged ch dren, th s movement has enormous mp cat ons for the on y peop e who are mob e, name y the peop e ust enter ng the abor market. For nstance, the tota number of teachers d d not

255 change much between 1975 and 1981, as a percentage of what t was n 1975 (see b d., 49; D gest of Educat on Stat st cs 1976 , 53). But because a most a the teachers a ready n serv ce wou d ho d onto the r obs, the ad ustment of young peop e was tremendous. Dur ng that same per od, the number of teachers graduat ng went down from 167,000 (D gest 1976 , 118) to 108,000 (D gest 1983–84 , 120), or by over a th rd. In the same years, the number of eng neer ng graduates went up by over 50 percent, though sure y the tota number of eng neers rose by on y 3–5 percent or so; s m ar y, bus ness adm n strat on graduates ncreased by over 75 percent, and computer sc ence graduates more than tr p ed n ust s x years (D gest 1976 , 119; D gest 1983–84 , 120). These ncreases d d not occur because th ngs were rea y boom ng n th s per od, by factors of 1.5 to 6, n management, eng neer ng, and computer work; rather f the dog of abor demand moves ts h ndquarters a tt e, the ta of demand for new graduates has to wag a ot. The tota s ze of the abor force has grown some ate y: from about 95 m on to about 114 m on, or by about a f fth, n the e ght years from 1975 to 1983 (Stat st ca Abstract, 1985 , 390). But obs d d not grow that much, so unemp oyment went from about 8 m on to about 11 m on ( b d., 390, 406). The unemp oyment rate , then, was about steady, at about 6 to 10 percent. But young peop e s xteen to n neteen had a rate of unemp oyment that var ed from 16 percent to 23 percent, and for peop e twenty to twenty-four the rate var ed from 10 percent to 16 percent (Stat st ca Abstract, 1987 , 390); that s, the tota range of var at on for teenagers was about tw ce that of the abor force as a who e. Rough y speak ng, f unemp oyment overa goes up 1 percent, the unemp oyment rate for peop e out of schoo between s xteen and n neteen goes up 2 percent, and that for peop e out of schoo between twenty and twenty-four goes up 1.5 percent. In other words, the f uctuat ons of tota emp oyment h t the new cohort harder. Th s tendency a so shows up n wages. In the b g f rm sector, about a m on un on zed workers (m sce aneous ourna st c sources, Spr ng 1985) have contracts that have separate sca es for recent y h red workers. Because the new cohort s so b g, f rms and other organ zat ons can get good young workers more cheap y than good o d workers. So compan es not want ng to f ght the r un ons agree w th the o der workers represented n un ons that for the f rst ten or so years on the ob new workers w get ower wages than the o d workers n the same obs, and they w progress up the sen or ty sca e at a certa n d stance beh nd the o der workers. So the ad ustment n the wage rates respond ng to the ncrease n the tota supp y of workers s be ng d sproport onate y borne by young

256 workers. (My adv ce to the reader s to be born n 1933, when the b rth cohort was the sma est re at ve to the tota abor force; one w have opportun t es far beyond what one s mer t wou d otherw se earn.) In add t on, the new cohorts take up the uncerta nty d sproport onate y. New bus nesses are a ot more ke y to fa than o d ones, espec a y n new ndustr es. For nstance, the death rate of f rms n the h gh-tech e ectron cs ndustry s a ot h gher than n the o d- ne ndustry of newspapers (Freeman, Carro , and Hannan 1983). Newspapers by and arge emp oy o d peop e, e ectron cs f rms young peop e. So when f rms fa , t s more often young peop e who have to scramb e. If emp oyers ooked on y at comparat ve wage rates, they wou d n genera prefer to ay off workers now ho d ng the obs they have and to h re nstead teenagers at the m n mum wage or, by mov ng the factory, to h re Koreans or Ta wanese. But c ear y most emp oyers prefer the peop e they have now to the peop e they m ght get by ay ng a the o d workers off and start ng anew n the cheapest poss b e abor market. Emp oyers be eve they know someth ng va uab e about the workers who now occupy obs n the r organ zat ons that they do not know about other workers. Let us br ef y specu ate about what an emp oyer m ght know about workers w th sen or ty. F rst, f rst- ne superv sors wou d presumab y comp a n about a worker s m sbehav or f that behav or were cr t ca to product on. The s ence of f rst- ne superv sors that s nd cated by sen or ty, perhaps supp emented by a vague nvest gat on of whether the worker has "kept h s or her nose c ean," then, means that there s noth ng wrong w th a worker s performance n any cruc a respect. Second, those c ose to product on, at east, know that a worker has to earn a ot nforma y from other workers about the qu rks of the mach nes or procedures or c ents and how to dea w th them, and sen or ty n a part cu ar ob usua y measures at east a m n ma f t nto the product on work group. That nforma earn ng a so prov des workers w th c ues to at east the m n ma requ rements of teamwork that make a part cu ar operat on go. Product on nes or other "unsk ed" systems run a good dea better than eng neers a one can guarantee because workers usua y cooperate to remedy defects n the r mach nery by baby ng t (Kusterer 1978), by he p ng one another dea w th techn ca nterdependenc es, where troub e for one creates troub es for others, and so forth. The workers often know the r oca part of the product on or serv ce process better than the r superv sors cou d teach them. Sen or ty n the ob cert f es that workers know the techn ca and soc a competences that are earned n the work group.

257 Th rd, some of the rout nes that workers are requ red to use on the r obs are used very often, and some of those may create troub es f done bad y. For nstance, te ers count money many t mes a day, and t s mportant that they shou d get th s task r ght (Kusterer 1978, 79–80). Whatever those frequent y used rout nes are, t s exact y those rout nes wh ch workers present y n the ob are h gh y pract ced at. Thus, they may be more product ve than workers w th the same reperto re of sk s but d fferent amounts of pract ce n the sk s most used n the r obs, because the others have pract ced on other obs. Th s s ess true of course, the more rap d y the task compos t on of a part cu ar ob changes (for nstance, t s ess true of ma ntenance e ectr c ans whose tasks change very frequent y than of construct on e ectr c ans who may work for days on a s ng e task). But n many organ zat ons the task compos t on of part cu ar ro es s qu te stab e, and n such organ zat ons the peop e n the ob are cert f ed to be better f tted for those obs by pract ce than subst tutes w th the same reperto res of sk s. Fourth, workers who ho d the ob now and have he d t for some t me show, by cont nu ng to ho d t, that the ncent ve system for that ob s adequate to keep them work ng at t. One of the best pred ctors of be ng w ng to do a ob s hav ng been w ng to do t for some t me. Sen or ty n a ob s thus a measure of character and mot vat on, wh ch s un que y a measure of how the workers w ke y respond to the ncent ves that n fact they w be confronted w th n the ob. New h res have much h gher turnover than o d ags. There are many rat ona reasons for an emp oyer to prefer the workers t has, because the r ho d ng the obs they do s nformat on about what sorts of obs they can ho d. Many of the costs of turnover are of course the sheer bureaucrat c troub e of h r ng peop e or n the expense of tra n ng them. But many of the costs of turnover come from the fact that the nformat on emp oyers have about workers and the r f t to the work system, to the rout nes and ncent ves of the work ro e, s worse n cond t ons of h gh turnover. What makes these facts nst tut ona y cruc a s that peop e who cou d ust as we have been h red n the f rst p ace but were not can never show that they ho d the ob, w th a that that mp es about whether they w have the ob next year. The argest cause of abor market r g d ty s the preference of emp oyers to st ck w th the ev s they know rather than f y to others they know not of. Of course, the who e s tuat on wou d be most unstab e f t were not for the fact that workers a so prefer emp oyers they know they can sat sfy, prefer nforma systems on the shop f oor to keep product on go ng that

258 they have earned to manage, prefer work that they have thorough y mastered so they can, when they choose, perform t sem consc ous y, and of course, prefer work where the ncent ves seem to them adequate to make t a worthwh e. The fact that workers have he d obs for a wh e shows them, too, that these are obs they can ho d. Sen or ty on the ob, then, s moderate y re ab e nformat on about whether peop e can do the ob, and perhaps the best ava ab e under cond t ons (wh ch are norma ) n wh ch the management tse f cou d not do the ob very we . The stor es of ear y un on organ z ng days of managers stuff ng tar paper nto the furnaces to make smoke, to te str k ng workers that they were manag ng f ne, ustrates the d ff cu ty of managers or staff workers earn ng the thousands of deta s nvo ved n how the ne rea y runs. Be ng an a rp ane eng neer does not make one a f ghter p ot, nor does manag ng a product on ne make one a sem sk ed operat ve. Yet management s n a better s tuat on to know how to do a sem sk ed product on ob than a random worker p cked from the street, so the workers w th sen or ty w exceed such rep acement workers even more than they exceed managers. Thus, the centra cert f cate n the abor market s the mp c t cert f cate of a ready ho d ng the ob, a cert f cate that s espec a y va uab e for cert fy ng one for that part cu ar ob.

Segmentation by Internal Labor Markets: Promotions Go to Those Now Employed by Big Firms and Government The second b g nst tut ona fact on cert f cat on s that b g f rms and government organ zat ons have mob ty w th n them among pos t ons, presumab y most y promot ons, wh e sma f rms n the c ass ca sma f rm compet t ve sector have a most no promot ons. Rough y three-e ghths of a movements among obs n both Japan (a ow abor mob ty soc ety) and the Un ted States (a h gh abor mob ty soc ety) are between pos t ons w th n f rms (Co e and S ege 1979, 70–71), f ve-e ghths between f rms. These w th n-f rm movements represent " nterna abor markets" n the strong sense that even peop e who d sso ve the person- ob t e tend to move w th n the boundar es of the f rm. Such mob ty structures are character st c of bureaucrat c organ zat ons n the eng neer ng-based ndustr es (manufactur ng of mach nes and genera y meta s fabr cat on, chem ca s, o , e ectr c ty generat on and d str but on, te ephone and te egraph serv ces), n f nance, and n government (St nchcombe 1979b, 231–233). The effect of the s ze of the

259 Tab e 3 Average Years Be ween n ra rm Job Changes by S ze o F rm or De ro and Yokohama

S ze

De ro (1970)

Yokohama (1971)

1–9

50

100

10–99

17

50

100–999

11

25

1000+

8

12

Source Ca cu a ed rom Co e and S ege 1979 80–81

f rm on the promot on rate s ustrated n Tab e 3, wh ch shows that promot ons are much more frequent n arge f rms than sma ones. Th s s often ca ed the creat on by arge f rms of an " nterna abor market," but t s cruc a to reca that even peop e who do not change obs are n the abor market: they se the r abor under cond t ons n wh ch the emp oyer cou d h re someone e se, and t s ust as much an nterna abor market n the sma f rm sector when very few change e ther f rms or obs. The po nt here s not that a young woman who gets a nonc er ca ob n a b g f rm w be promoted more s ow y than the men; rather very probab y a forty-year-o d woman does not now work for a arge f rm or the government n a ob that eads to promot on, and so she s a ot ess ke y to get a promot on over the next twenty-f ve years than s a man of forty. S nce there are very few promot ons n the sma f rm sector emp oy ng a ot of women and b acks, and s nce promot ons w go to peop e emp oyed n f rms where promot ons take p ace, t s a ready a foregone conc us on that even w th perfect y fa r po c es w th n f rms on promot ons, most promot ons over the next coup e of decades w go to wh te ma es. (It s mportant to note that the d fference between arge and sma f rms n promot on rates may be ower for women; n arge f rms women are d sproport onate y n dead-end obs.) But what s cruc a from our po nt of v ew s that the nformat on used to pred ct performance n the new ob cons sts n arge measure of observat ons of performances n the o d ob. In many obs covered by a co ect ve contract, that performance s cert f ed by sen or ty a one. As we have argued above, th s s not an nformat on-free cert f cat on, and a person promoted to a h ghery e d ng mach ne n a mach ne shop because he or she has worked another mach ne for years s n fact a good bet. Furthermore, obs are ord nar y connected to other obs n such a way that exper ence n one s thought to be re evant for exper ence n the

260 other; there are "career adders." Th s s c earest where the h gher ob s rea y the same as the ower ob, as n the case of fu professors and ass stant professors. Where the h gher ob s as f rstne superv sor of the work of peop e do ng the same ob, t s genera y conceded that exper ence n the superv sed ob s re evant. Eng neer ng superv sors are recru ted from among eng neers, software deve opment superv sors from among programmers, f oorwa kers from among sa espeop e, foremen n the crafts from craftsmen, and so on. Th s means that many promot ons w be restr cted not on y to peop e w th n the arger f rms and governments, but a so to peop e on the f rst rungs of career adders. Where the nformat on s udged by a comm ttee of peers (aga n, th s means a comm ttee of super ors) n an nterna abor market, the work of the cand date s to some degree observab e by that comm ttee. It s more observab e n un vers t es when the s gn f cant work s pub shed wr t ng; teach ng work s not usua y very observab e n Amer can un vers t es. But n bus nesses, the h gher one goes n the organ zat on, the more everyday work s done n comm ttee and by ong te ephone conversat ons among execut ves n d fferent departments, wh ch means that a var ety of peop e who m ght promote a cand date for h gher management have had dea ngs w th the cand date. Thus, a cand date s work s more observab e by the promot on comm ttees toward the top of the manager a h erarchy than t s for f rst- ne superv sors toward the bottom. Promot on systems, or nterna abor markets n the strong sense, make much use of the same sen or ty nformat on that s used n reta n ng the same peop e n the same obs. But n add t on, f nformat on about ob performance s ava ab e to super ors by d rect observat on, and f the pos t on be ng f ed s of cons derab e mportance to the product v ty of the organ zat on as a who e, performance eva uat ons are used as a supp ement to sen or ty nformat on (Rosenbaum 1981). Th s n turn means that the cert f cat on of ob performance for h gher and more d ff cu t obs tends to be udged by a "comm ttee of peers," though n most pyram da organ zat ons the h erarch ca super or ty of the peers s obv ous. At any rate, the dynam cs of recru tment to h gher management and to superv sory pos t ons n h gh-status staff departments tends toward the profess ona mode , to be d scussed n the next sect on. Genera y n systems of promot on from w th n, promot ons to ower pos t ons are based pr mar y on sen or ty n the next ower ob, promot ons to m dd e pos t ons are based pr mar y on eva uat ons by h erarch ca super ors, and promot ons to top pos t ons are more s m ar to profess ona or craft recru tment based on eva uat ons by a comm ttee of peers.

261

Worker-Controlled Recruitment in Professional and Craft Occupations The th rd b g nst tut ona fact n ob cert f cat on s that some emp oyers h re cert f ed sk ed or profess ona workers from other emp oyers n the same ne of work, produc ng a pattern of mob ty n wh ch peop e get the same sorts of obs they had before when they move to a new emp oyer. In th s case, the centra boundar es n the abor market are around occupat ons as we as around f rms (though somet mes rather than around f rms, as n h gh y unstab e ndustr es such as bu d ng construct on). Occupat ona assoc at ons such as craft trade un ons or profess ona schoo s govern mob ty nto the occupat on. The fundamenta genera zat on about entry nto craft and profess ona obs (and, as we have argued above, nto h gher management and top staff obs n bureaucrat c organ zat ons) s the more democrat c contro over recru tment nto a set of obs s —that s, the more entry and tra n ng are contro ed by workers n that set of obs—the fewer women, b acks, Mex can-Amer cans, or mm grant workers are emp oyed n the group (He mer 1986b, 766; see a so Cohn 1985; G enn and Fe dbert 1977). There are three ma n p aces n the abor market where "co-optat on" s a common method of recru tment—that s, where members of a body of workers he p tra n, se ect, and f nd obs for young potent a members of the body. These are (1) the profess ons, espec a y the h ghest-status profess ons n a g ven area of work—for examp e, professors n prest ge co eges and un vers t es; awyers, but not para ega s; dent sts, but not denta hyg en sts; phys c ans, but not nurses or ab ass stants n a hosp ta (but chem sts who work for chem ca or o compan es who do the same sort of aboratory work as ab ass stants n hosp ta s are nc uded here, because wh e ab ass stants n hosp ta s are recru ted and tra ned by patho og sts, profess ona chem sts do recru t the r own fo owers [Schroeder and F n ay 1986]); (2) craftsmen (or craftspeop e, f the genera zat on I am argu ng for were not true), such as craftsmen n the construct on trades; mach n sts of var ous k nds (e.g., too and d e makers, pattern makers); mo ders n foundr es; pressmen and typographers n pr nt ng, at east before computer typesett ng; cooks and chefs n fancy restaurants; ong-hau truck dr vers dr v ng b g r gs; and r verboat p ots n Mark Twa n s day—such workers are usua y recru ted, tra ned, and cert f ed by those who a ready occupy pos t ons as craftsmen; and (3) top management, execut ves h gh enough that they are ke y to be on the board of d rectors as we as to ho d an execut ve pos t on. In each of these cases an o der trad t on of organ z ng recru tment

262 app es, n wh ch the tra n ng occurred on y when someone who knew the trade a owed a nov ce to come nto h s off ce or shop and earn that trade (Jackson 1984, pass m; Carr-Saunders and W son 1933; Lockwood 1958). Now recru tment n these crafts or profess ons s often by schoo s or by adm ss on to pract ce n a hosp ta or some other organ zed and quas -bureaucrat c structure, but there s st "peer group" contro over who gets n. A few other work sett ngs are marked by recru tment w th a peer group character, but these do not have the same abor-market-w de "occupat ona " form: n m n ng, new underground m ners have c ass ca y been recru ted by the work ng m ners and so have had ethn c and sex cont nu ty w th the m ners, wh e aboveground workers are h red by more bureaucrat c methods and very often are of a d fferent ethn c ty than the underground m ners. For examp e, n Ar zona around the turn of the twent eth century the aboveground workers at a m ne wou d be Mex cans, the be owground workers f rst Corn sh t n m ners, then ater Ang os more genera y (Boswe and Bush 1984, 141–142; see a so Boswe 1986). Wh e professors n h gh-prest ge p aces have a ot of contro over recru tment, n un or co eges or n what are now ca ed "state un vers t es" the adm n strat on has genera y more contro over recru tment. The pred ct on, then, s that aff rmat ve act on w th respect to race, gender, and ethn c ty w have gone further n un or co eges and state un vers t es than n prest ge un vers t es or tony bera arts co eges. How, then, does t come about that when workers contro recru tment to the r own ranks (when recru tment s "democrat c"), the recru tment pattern s so d scr m natory? It rem nds one of the observat on that to ma nta n the sub ugat on of a ma or ty of another ethn c ty stab y n the ong run, as s done n South Afr ca or Israe , there must be democracy w th n the ru ng ethn c group. On y f, say, or enta Jews have a way to express the r nd gnat on over the treatment they get from the Ashkenaz m, say, by support ng the L kud, can the more oppressed part of the Jews be prevented from o n ng the r nterests to those of the Arabs and underm n ng the so dar ty of the ru ng ethn c ty. S m ar y, the Boers n South Afr ca used an Eng sh-type par amentary system to become the ru ng ethn c ty, then crowded out the Eng sh through the democrat c system (democrat c w thout b acks, need ess to say) rather than a y ng w th, say, the co ored. Recru tment s one of the severa cases n wh ch democracy has undemocrat c effects, and we wou d ke to know why. Let us cons der professors a m ng to h re another professor n the r department. How often shou d one expect them to propose to the adm n strat on to h re the cheapest abor the un vers ty can get? The peop e

263 who do the h r ng are ke y to be e ther on y professors themse ves (usua y funct on ng n a recru tment comm ttee) or ( n the Un ted States) a professor who s a so a part-t me adm n strator who gets pract ca y no more money for be ng cha r of a department and who w go back to h s or her professor a ro e after a term as cha r. The recru ters, then, have no nterest n beat ng down the wages of the professorate, and qu te often have the reverse nterest. (Someone try ng to h re me to d me that he wanted to go to the adm n strat on propos ng the h ghest sa ary we cou d support w th arguments, because then we [ t wou d on y be "we" after I came, of course] cou d use that as a benchmark for what the others of comparab e mer t ought to be gett ng.) Now contrast th s s tuat on w th that of a dean of a un or co ege, say n Gary, Ind ana. He or she s ke y to be h r ng qu te a few teachers, and w soon earn that the co ege can h re Un vers ty of Ch cago graduate students to teach a course for a few thousand do ars; they w be pretty smart, qu te ke y fema e, but w accept ow wages part y because they are w ng to work part-t me so they can work on the r d ssertat on the rest of the t me. For the vocat ona (and h gh schoo makeup) c asses, the un or co ege needs to h re teachers fu t me, and f t s w ng to h re women t can get except ona y good peop e who thought they were go ng to be stuck a the r ves teach ng n a h gh schoo . The un or co ege can get women a ot cheaper than t can get men, for t wou d have to h re men away from a mach ne shop or an auto repa r shop or an eng neer ng ob. And of course, f a un or co ege h res b ack peop e, s nce by and arge they st cannot get rea y good obs, t can probab y get them cheaper, too. A dean s own sa ary w not go down n the future because he or she has recru ted a ot of cheaper abor; the dean s not h r ng h s or her own rep acement, ust subst tutes for more expens ve fu -t me co ege teachers w th advanced degrees. And econom z ng on facu ty sa ar es w sure y ncrease the chances that a dean w be promoted or w ncrease h s or her sa ary. So the fundamenta dynam c s set up. If an execut ve s future depends on m n m z ng the costs n a department or nst tut on, then he or she w tend to h re ess qua f ed workers, and w tend to prefer cheaper b acks, women, and mm grants, espec a y ega mm grants, f the organ zat on can use them. Patho og sts, even though they are profess ona s, are not h r ng the r future rep acement when they h re a ab ass stant. The r future rep acement w come out of a med ca schoo ( n 1981 that meant that three out of four of the rep acements w be ma e, Stat st ca Abstract, 1984 , 170), so a patho og st w not be beat ng down h s or her own wages by h r ng cheaper women ab techn c ans (w th four

264 years of tra n ng) nstead of more expens ve men chem sts (w th four years of tra n ng). On the other hand, n a chem ca company f execut ves h re chem sts to work n the ab they w very ke y be h r ng the r own rep acement, and the new h res wages w short y be compet ng w th the r own (Schroeder and F n ay 1986; F tzg bbon 1988). If the h rers are members of a group that co-opts and tra ns ts own members, then f they beat down the wages of peop e com ng n, they beat down the r own wages as we . If h rers are execut ves who h re a ot of peop e who are not go ng to be promoted nto the r own pos t ons, then f they beat down the wages of the peop e they h re, they themse ves w get promoted faster. One of the eas est ways to beat down the wages of new h res s to h re d sproport onate y from among women and m nor t es. We thus pred ct that the more democrat c recru tment s, the fewer women and m nor t es w be recru ted; the more recru tment s n the hands of bureaucrat c or entrepreneur a off c a s, the more women and b acks shou d be recru ted. But for those m nor t es and women who do get h red by a body of fe ow workers, the h rers have an nterest n see ng that they get the same pr v eges and are not used by the organ zat on to beat down the genera eve of wages. So those few women and m nor t es h red by a body of peers are ess ke y to be exp o ted (He mer 1986b; Shack-Marquez and Berg 1982). Gary Becker ([1957] 1971) argues that those who h re w th the ob ect ve of m n m z ng costs w h re more b acks and women, on the assumpt on that t s entrepreneurs (who w never be rep aced by the peop e they h re) who do the h r ng; and n the compet t ve ndustr es, where Becker f nds most recru tment of women and b acks, entrepreneurs more often do n fact do the h r ng. The pressure for execut ves to h re women and m nor t es s ntens f ed f there s a ot of compet t ve pressure on the organ zat on. In a re at ve y abor ntens ve sma f rm sector such as appare or contract construct on, a f rm may be n one of two bas c s tuat ons. The f rst s where the cost of abor s standard zed throughout a compet t ve market by craft un ons, so that a emp oyers are faced w th the same wage rate for sk ed workers (somet mes there s a nonun on sector that s ess eff c ent, qu te often w th sem sk ed workers that need a ot more superv s on; that market s tuat on w be ana yzed be ow). In such cond t ons the emp oyers do not m nd too much turn ng over the recru tment of workers to craft trade un ons, because then they get the same sort of workers as the r compet tors for the same wages. The second cond t on s that the compet t on over abor costs re gns unrestra ned, and emp oyers have to get the cheapest abor they can n order to stay compet t ve. In

265 Tab e 4 Compe

ve and Recru men Cond ons De erm n ng he Sex Race and C zensh p Compos on o he Amer can Labor Force



Un on or Pro ess ona Recru men



Compe

ve Pressure Sma F rm

Yes

No

Wh e ma e perhaps excep or some segrega ed cra s (e g s eep ng car por ers) cons ruc on mach n ng

B ack ema e Puer o R can Mex canAmer can ega s ore gn born ex es appare ood many serv ce ndus r es

Wh e ma e un vers y acu es phys c ans n hosp a se ngs op managemen o arge rms

Fa r y wh e ma e o en n ndus r a un ons au o chem ca s o e ec r ca equ pmen

Lower Pressure O gopo y

compet t ve cond t ons w th no sh e d ng monopo es n the abor market, a rap d sw tch to women, b acks, mm grants, and ega s tends to take p ace (G. Becker [1957] 1971; Cohn 1985). So we get the four cond t ons descr bed n Tab e 4.

Family Recruitment in Small Firm Sectors A b g nst tut ona fact n some ndustr es s that emp oyers h re themse ves and parts of the r fam es to do some of the work of the f rm. Th s s most character st c of sma -f rm sectors of the economy, espec a y n farm ng and the serv ce ndustr es. Peop e who have careers as owner of a f rm are not governed by the same sort of re at on between the supp y of abor and the demand for abor that character ze other emp oyment re at onsh ps, and these sectors tend to "oversupp y" abor dur ng downturns of the bus ness cyc e. Peop e do not seem to be very good pred ctors of the r own work performance as entrepreneurs, as s ev denced by the arge fa ure rate of sma bus ness enterpr ses. Rough y ha f of a new starts of sma bus nesses n the urban economy go out of bus ness n the f rst coup e of years, nd cat ng that the entrepreneur d d not pred ct very we h s or her own ab ty to make the enterpr se go (or d d not pred ct the other resources needed very we , wh ch amounts to the same th ng). In add t on, the sma bus nesses n the s ng e-fam y res dent a and subcontract ng parts of the construct on ndustry tend to be destroyed on a

266 arge sca e n bus ness recess ons and depress ons, because construct on act v ty s very sens t ve to the bus ness cyc e. The same s true of arch tectura f rms (B au 1984, 115–132). Phys c ans, awyers, dent sts, and a few other k nds of profess ona s are ord nar y successfu se ng the r own abor, but they have some government he p n cert fy ng the r ab t es n that work and n support of the r monopo y. For reasons that have noth ng to do w th entrepreneur a ta ent, about ha f of a sma farm ng bus nesses d sappear n each generat on n modern countr es whose urban economy s deve op ng fast enough to absorb them. But a stud es of, for examp e, adopt on of nnovat ons by farmers show arge d fferences n product v ty. Corporate farms and partnersh ps get rough y 50 percent more sa es per acre than nd v dua fam y farms (computed from Stat st ca Abstract, 1987 , tab es 1102–1103), wh ch probab y nd cates poor udgment by many fam y farmers about whether they shou d be entrepreneurs n that bus ness. Thus, wh e overa the pos t on of the owner of a bus ness h r ng h mse f or herse f s a pr v eged one n the abor market, peop e do not show a great dea of ab ty n udg ng or pred ct ng the r own work performance. The cert f cate that a person gets from hav ng owned a f rm s to a cons derab e degree a cert f cate of foo hard ness.

Union Membership As a Certificate of Productivity Un on workers are about 10 to 30 percent more eff c ent than other workers (Freeman and Medoff 1984) and so do not run nto hard res stance n barga n ng up the r wages unt they exceed 30 percent above nonun on workers. In January 1979, the med an ma e un on member made about 28 percent more than the med an ma e nonun on worker; un on zed fema es made about 50 percent more, but there were very few of them (Stat st ca Abstract, 1984 , 435). Un on zed workers are more often b ue-co ar workers n manufactur ng or transportat on (41.4 percent of ma e b ue-co ar workers were un on members, 29.1 percent of fema e; f gures n th s paragraph are from Stat st ca Abstract, 1984 , 435) as opposed to (a) serv ce workers (ma es 24.7 percent, fema es 10.8 percent); (b) c er ca and k ndred workers (ma es 32.9 percent [many be ng post off ce etter carr ers], fema es 12.2 percent); (c) profess ona , techn ca , and k ndred workers (ma es 19.4 percent, fema es 26.5 percent [most of the un on zed profess ona workers are teachers, and teach ng s a heav y fema e profess on, wh ch s why more fema e profess ona s are organ zed than ma e ones]); (d) man-

267 agers and adm n strators (ma es 8.4 percent, fema es 5.3 percent), and (e) sa espeop e (ma es 3.3 percent, fema es 5.1 percent). In genera , un on zed workers work n ndustr es where there s qu te a b t of compet t on, though not as much perhaps as n a ot of sma f rm compet t ve manufactur ng. Freeman and Medoff (1984) show that un on zat on of the work force on y pushes down prof ts n the ess compet t ve sector. That s, on y when there are excess prof ts n an ndustry, because of o gopo st c advantages, do un ons manage to get part of what wou d otherw se be prof ts. In compet t ve ndustr es, un on and nonun on f rms have about the same prof t rates. So the quest on s, how do un on emp oyers stay n bus ness f they have to pay rough y 30 percent more for the r workers? The bas c answer seems to be that un on zed workers work harder or w th more sk than nonun on workers. Apparent y they get more work out per day than workers n nonun on shops or on nonun on crews. There are a number of d fferent poss b e exp anat ons for th s. In some craft trades ke the pr nt ng ndustry, peop e ev dent y start n the nonun on f rms (Jackson 1984, 263). After they get thorough y tra ned, so that they can work faster and more accurate y than somebody st n tra n ng or w th on y a few years exper ence, they get themse ves obs n the un on zed sector and o n the un on. When nonun on f rms that pay about 30 percent ess than un on f rms se n bas ca y the same market (though there are some k nds of pr nt ng that unt recent y were done on y on a un on bas s), the on y way the un on f rms can compete s to produce the same products us ng about 70 percent of the worker hours that the nonun on f rms use. A s m ar th ng seems to happen n the construct on ndustry, though there the market s perhaps more segregated. In metropo tan areas, most b g obs (most commerc a and ndustr a construct on, and a most a government construct on) are run on a un on bas s; repa r obs on houses n most trades are run on a nonun on bas s; and the bu d ng of houses s run both ways. In the compet t ve branches, espec a y hous ng construct on, where un on and nonun on over ap, the un on construct on f rms have to get by wh e pay ng 30 percent more than other f rms. Un on f rms usua y do not get obs for wh ch the c ent pays by the hour, because t seems to a noncraft c ent so much more expens ve to h re un on abor, and they have no way of know ng th s abor s more eff c ent. Perhaps they have a so heard about un on featherbedd ng, and so mag ng that un on work w be more expens ve because of work ru es. A most a repa r work s done by nonun on f rms, because t s a most a ways done on an hour y bas s. But where f rms b d for the obs, the un on f rms can do a r ght. Th s

268 means that the un on f rms must be about 30 percent more eff c ent than the nonun on f rms n construct on. Here aga n, I suppose the greater eff c ency of un on workers ref ects the nonun on sector serv ng as an apprent cesh p system for the un on sector. In factory work we f nd the same genera p cture, that un on f rms pay ng more than the nonun on f rms are compet t ve, though they can t go too h gh above the nonun on wage and st make t. Here severa d fferent th ngs seem to be go ng on. The f rst s mere y that a un on contract st mu ates management to nst tute a speedup that they cou d have done w th nonun on workers but somehow have not got around to. So part of the exp anat on s that un on f rms have more eff c ent managements, managements that dr ve the workers harder because they have to se at the same pr ce as the r compet t on wh e pay ng h gher wages. Incent ve systems and ndustr a eng neers are often part of the system ntroduced to speed up the work after the company goes un on. Second, because the system of ndustr a re at ons that s nst tuted under a un on contract seems fa rer to the workers, they may be w ng to work harder and take orders more cheerfu y. Gr evances go nto the gr evance system rather than caus ng su en workers to nst tute a secret s owdown because they th nk they are not be ng treated fa r y. Th rd, t may be that un on f rms can be more se ect ve n h r ng the r workers than nonun on f rms can. Wh e most personne recru tment processes cannot d st ngu sh a good worker from a bad one, there may be some added fuss ness n un on f rms about who s h red, and s nce the f rm s pay ng more than other f rms, t can get the better workers. For examp e, t seems that a ot of b ue-co ar obs are f ed through the recommendat ons of the f rms own workers (Granovetter 1973, 1974). It may be that n un on f rms managers that keep the r ear to the ground and hear about good prospects among fr ends and re at ves of the r good workers can, because they are pay ng h gher wages, genera y get those fr ends and re at ves. So un on f rms recru t from hard-work ng fam es and hard-work ng fr endsh p c rc es, by fo ow ng out the recommendat ons of the r good workers and not sten ng to the r sh rkers. The overa resu t, however, s that one way or another work ng for a un on f rm cert f es that one can ho d a ob n a un on f rm, wh ch n turn means that one can ho d a ob n wh ch the who e system (whatever ts or g n) makes one cons derab y more eff c ent than workers n nonun on f rms. Thus, the fact that un on f rms tend to offer the r un on obs next year to the r own workers, so that they too have the arge ma or ty of the r obs f ed by the same person who had the ob ast year, resu ts n the cert f cate of greater product v ty be ng perpetuated n the same so-

269 c a groups who were more un on zed ast year. As the above data show, the more un on zed groups are d sproport onate y the ma e work ng c ass. Let us p ck up the thread of the argument aga n. Cert f cates that g ve emp oyers reason to pay above the go ng rate for unsk ed abor are generated n f ve ma n ways: through ho d ng the ob now (sen or ty); through data co ect on w th n arge f rms and governments, ead ng to promot ons; through cert f cat on of profess ona s and craftsmen by bod es of peers; through cert f cat on by own ng a bus ness that one s a manager; and through occupy ng a ob n a un on shop. Each of these cond t ons carr es d fferent amounts and k nds of nformat on. Depend ng on what nformat on s most prob emat c for part cu ar ro es n the organ zat on, we wou d expect more than one cert f cat on system to be n p ace. For examp e, we have argued that n management the system for cert fy ng competence for top ro es s d fferent from that for cert fy ng competence for ro es at the m dd e and bottom. The centra overa d ff cu ty to wh ch these var ous systems of cert f cat on respond s that the measurement, and more espec a y the pred ct on, of the qua ty of performance s prob emat c. About a we know about why un on members are more product ve s that they work n more product ve organ zat ons; but from a brute econom c po nt of v ew, th s means that the economy shou d (and does) pay un on members more. Wh e the preference for un on workers s ord nar y mposed on f rms from the outs de and aga nst the r w s, that preference apparent y a ows them to pay the r workers more. S m ar y, much of the nformat on that comes from the fact that a g ven worker now occup es a g ven ob s nchoate, and has been hard to ana yze systemat ca y here. But s nce so many emp oyers use that nformat on to dec de whom to emp oy next year, t serves as a cert f cate n the abor market. Sen or ty cr t er a are espec a y prom nent when the qua t es that const tute qua f cat on for the ob are a so nchoate, as n teach ng and po ce work. A such cert f cates, generated n d fferent ways, and measur ng d fferent k nds of qua f cat on for work, tend to preserve pos t ons of pr v ege n the abor market.

The Secondary Labor Market These b g nst tut ona facts (stab ty n obs, nterna promot ons n arge f rms, occupat ona cont nu ty n some occupat ons, fam y recru tment n some sectors, and the assoc at on of un on zat on w th product v ty) taken together d st ngu sh, f rst of a , between peop e ns de the boundary mark ng off the emp oyed from those not emp oyed. Peop e

270 now emp oyed have a much better chance of be ng emp oyed next year than those not now emp oyed. But a second d st nct on has preoccup ed recent abor market soc o ogy. Th s s the d st nct on between those ns de any boundary that protects pr v ege among those emp oyed (such as be ng emp oyed by a arge f rm that g ves promot ons or h gher pay, be ng qua f ed as a craftsman or profess ona , own ng a f rm that g ves the owners emp oyment, be ng emp oyed n a un on zed shop or crew), genera y ca ed the "pr mary abor market," and those emp oyed but not protected by such a boundary, genera y ca ed the "secondary abor market." Rough y speak ng, those outs de the c rc es of pr v ege amount to about 25 percent of those emp oyed. About one out of four obs (and about one out of four obho ders) have h gh rates of nterf rm mob ty and h gh rate of mob ty nto the status of unemp oyment, and have ow wages, ow un on zat on, and other features show ng ack of pr v ege. The peop e who occupy these unpr v eged obs n most modern econom es are d sproport onate y ethn c m nor t es, women reenter ng the abor market n m dd e age, teenagers, and peop e recent y unemp oyed (Doer nger and P ore 1971). In underdeve oped econom es they are very often recent m grants from rura areas v ng n fave as or b donv es and work ng on the fr nges of the urban serv ce sector. From the po nt of v ew of th s chapter, what s d st nct ve about such popu at ons s that they have very tt e opportun ty to prov de cert f cat on that w sat sfy an emp oyer w th a good ob on offer that they can be pred cted to do we n that ob. The ma n k nd of cert f cat on that sat sf es emp oyers s that one a ready occup es such a ob—and that s the hardest k nd of cert f cat on for members of the secondary abor market to g ve. Those sorts of forma zed nformat on that ust fy promot on n nterna abor markets are even more naccess b e to members of the secondary abor market, and the bod es of peers that govern se f-recru ted status groups have no reason to regard such peop e as potent a peers, and much nterest n keep ng out ow-wage abor. Such ack of cert f cates no doubt ref ects n part a true ack of competence, but th s does not answer the soc o og ca quest on of how such ack of cert f ab e competence (and perhaps of true competence) s systemat ca y d str buted. After a , peop e who end up w th the good obs were a so born ncompetent.

Conclusion A most a modern organ zat ons must face a great dea of uncerta nty n the r pred ct ons about worker product v ty over the per ods n wh ch

271 those workers w be emp oyed. There s a great dea of uncerta nty even when those workers are a ready emp oyed, because a the forms of uncerta nty ana yzed e sewhere n the book make t uncerta n exact y what one w be ask ng workers to do, exact y what ncent ves they w get f they do that, and exact y what the econom c or other consequences for the organ zat on w be f they do t. Organ zat ons have even more uncerta nty when they are recru t ng new workers. Th s uncerta nty matters more for some organ zat ons than for others. Genera y speak ng, f an organ zat on s confronted w th uncerta nty that requ res sk ed var at on of rout nes at the worker eve (as, for an extreme examp e, s requ red n bas c sc ent f c research), then the prob em of p ck ng workers who can dea successfu y w th the uncerta nty s cruc a to organ zat ona success. In those cases w expect organ zat ons to deve op personne systems that make heavy use of nformat on from fe ow workers (peer rev ew systems), w th a the prob ems of mot vated d stort on of personne nformat on that th s nvo ves. Such peer systems are ke y to app y on y to the profess ona or craft subsect on of the organ zat on; thus, for examp e, the bus ness off ce of a un vers ty w probab y use tt e peer rev ew. Organ zat ons that requ re a great dea of teamwork, and n wh ch superv sors contr bute strong y to teamwork, w be expected to use a good dea of h erarch ca nformat on from super ors and super ors of super ors n pred ct ng workers future performance. The dea -typ ca " nterna abor market," n wh ch promot ons are governed by h erarch ca super ors, shou d be a norma outcome of uncerta nt es that revo ve ma n y about the amount of teamwork. Most work that s ne ther too d ff cu t nor too nvo ved n teamwork but that s poor y measured by re ab e manager a nformat on, such as teach ng or po ce work, shou d resu t n cert f cat on systems organ zed around sen or ty. The ma n cert f cate nd cat ng that person can manage the ob (whatever the ob rea y nvo ves) s that he or she has been manag ng t a ready. In part cu ar, the h gher apparent product v ty of un on f rms, wh e ke y as myster ous to un on f rm managers as t s to econom sts and soc o og sts, s apparent y, whatever ts cause, adequate y preserved by keep ng un on workers on the r obs. Work that can be eas y assessed by aud tab e measures n an env ronment that s suff c ent y stab e that the work s repet t ve and requ res the same ta ents (and the same measures of ta ent) shou d be pa d accord ng to ts marg na product v ty, ust as econom sts assum ng perfect nformat on pred ct. Trave ng sa espeop e, ath etes, and ettuce p ckers seem to be the ma n examp es. The genera resu t, then, s that nformat on about worker perfor-

272 mance s ord nar y very bad, and pred ct ons of future worker performance even worse. Peop e apparent y cannot pred ct the r own future performance as workers n bus nesses they start for themse ves we enough to keep from go ng bust; and they can presumab y be expected to do worse at pred ct ng the performance of others. That s why the abor market nst tut ons n most parts of the economy ook so tt e ke the nst tut ons of comm ss ons on sa es, or superstar contracts n ath et cs, or subcontract ng by p ece of ettuce p cked, wh ch wou d be the dea -typ ca outcome f emp oyers actua y measured marg na product v ty n any d rect way. What one needs to know for ana ys s of nst tut ons co ect ng and us ng nformat on on workers s what nformat on about the future performance of the work s conta ned n the ma n k nds of cert f cates emp oyers use. We have tr ed to ana yze what sort of nformat on sen or ty, educat on, cert f ed or forma zed performance measurements from prev ous obs, product on outcome measurements, and the ke conta n, so as to g ve a funct ona ana ys s of abor market nst tut ons. Th s funct ona ana ys s g ves someth ng of a Pang oss an v ew of abor market d scr m nat on: the nst tut ons that resu t n rac a , ethn c, and gender d scr m nat on are rea y a for the best n th s best of a poss b e wor ds. Of course, they are not best for a purposes, and espec a y not for the purposes of bu d ng a fa r soc ety. The po t ca funct on of th s ana ys s, then, s to try to suggest why d scr m natory abor market nst tut ons are so res stant to change, and to prov de a bas s for propos ng ways to change them. The purposes these nst tut ons best su t vary w th the type of nformat on system, and we have used th s fact to exp a n where n the economy the d fferent types of system are used. Know ng why nst tut ons w th a d scr m natory effect are found where they are may be po t ca y usefu n underm n ng them. The sc ent f c funct on of th s ana ys s s to exp a n why some k nds of abor market segmentat on are found n some parts of the economy, d fferent k nds n other parts. G ven that exp anatory purpose, t s not surpr s ng that we do not succeed very we n exp a n ng why n a parts of the economy, the genera tendency s for a such nst tut ons to d scr m nate aga nst women, aga nst rac a y oppressed groups, aga nst teenagers, and aga nst recent mm grants or guest workers. The connect on between nformat on, udgment, and reward s centra to the structure of status systems (St nchcombe 1964, 25–40). Whenever nformat on about work performance s centra to the overa funct on ng of an organ zat on, status systems w be shaped by the requ rements of that nformat on system. In co eges and secondary schoo s, the nforma-

273 t on requ rements for the grad ng of students are eas y sat sf ed, and the preservat on of the reward va ue of the resu t ng cert f cates s of greater funct ona mportance. For examp e, the proport on of students f unk ng f rst-year chem stry var es hard y at a w th the average SAT scores of enter ng freshmen, because the reward va ue of a grade n chem stry s pr mar y the re at ve rank ng t prov des a student as compared w th other premed ca students, not the abso ute grade eve . Converse y, re ab e performance of ma ntenance tasks n comp ex, and frequent y mod f ed, mach nery w be hard to measure and pred ct, but cruc a for a chem ca p ant. The status system of ma ntenance peop e n such p ants w therefore be shaped by the prob em of co ect ng mean ngfu nformat on on craft workers performance. Thus, the requ rements of grad ng chem stry students and qua fy ng ma ntenance workers are d fferent, even though they both depend u t mate y on chem ca know edge. Hence, we f nd grad ng by standard zed tests by chem sts who themse ves were promoted by comm ttees of peers n co eges, and no grad ng, but nstead apprent cesh p and craft contro , for ma ntenance workers. We therefore pred ct d fferent k nds of strat f cat on systems n organ zat ons w th d fferent nformat on requ rements, and d fferent status systems n d fferent parts of the organ zat on depend ng on the part cu ar nformat on requ red. We w expect promot on n un vers t es to depend not on h erarch ca nformat on from the bus ness off ces, but on facu ty comm ttees of peers, w th the accompany ng d fference n the mot vat on of h gh-status peop e n the bus ness off ces to h re women and m nor t es, those n the facu ty to h re wh te ma es and other h gh-pr ced abor. But we w expect the very top ranks of both to have the same ow proport ons of women and m nor t es, because as one approaches the top of the bus ness off ce h erarchy peop e are recru ted more by comm ttees of peers, rather than recommended for promot on by the r own super ors— ust as n the case of facu ty. The var at ons between and w th n organ zat ons n the strat f cat on system are ke y to be exp a ned by var at ons n the nformat on needed about peop e for d fferent product ve tasks.

274

8— Class Consciousness and Organizational Sociology: E. P. Thompson Applied to Contemporary Class Consciousness Introduction By c ass consc ousness we mean the tendency of peop e to th nk of the r pos t on n the arger soc ety n terms of the r pos t on n an emp oy ng organ zat on. Workers are c ass consc ous when they th nk of the r gr evances at work and the r nterests n po t cs as both der ved from the r emp oyment re at on to part cu ar organ zat ons. The centra ro e of organ zat ona pos t on n c ass consc ousness means that such consc ousness s a eg t mate sub ect of organ zat ona soc o ogy. But the fact that c ass consc ousness s a concept on of one s re at on to the arger soc ety means that t a so fa s w th n po t ca soc o ogy. Chapter 7 was about how emp oyers f nd out about workers and how that shapes the arge structure of the abor market. Here the cruc a po nt s that the who e ob ect of an ncent ve system s to commun cate to workers where the r nterests e. On y f workers nterests can be connected to some sort of performance measurement w the ncent ve system mot vate attent veness, mprovement of rout nes, phys ca effort, concern for prof ts, and the earn ng and teach ng of sk s. The nformat on conveyed to workers about where the r nterest es makes the system run, or makes t fa . S nce performance measurement s prob emat c, as we showed n Chapter 7, what the ncent ve system commun cates to the worker about h s or her nterest s sub ect to cu tura nterpretat on. A part cu ar k nd of nterpretat on w th arge soc a and po t ca consequences s c ass consc ousness. Our f rst task n th s chapter s to def ne c ass consc ousness. To do so we reana yze E. P. Thompson s great work on the deve opment of c ass consc ousness n Eng and, supp emented by some of Dav d Lockwood s observat ons on the mportance of emp oyers creat on of categor es of workers who have dent ca abor contracts. The resu t s a def n t on that

275 takes nto account the fact that c ass consc ousness s a pro ect on of organ zat ona pos t on on the nterpretat on of the arger econom c and po t ca order of soc ety. Thus t nvo ves both organ zat ona and po t ca consc ousness. The prob em we mmed ate y confront n der v ng a theory from th s def n t on s that peop e s re at on to a arger soc ety s determ ned by that soc ety as we as by the r organ zat ona pos t on. There are enormous var at ons n the re at on of c ass consc ousness to po t cs n var ous soc et es. Our second task n th s chapter s therefore to descr be that var at on, and to so ate the ma n var ab es pred ct ng var at ons n the type of c ass consc ousness soc et es show. The two categor es of forces we so ate are po t ca and organ zat ona , correspond ng to the two components of the concept of c ass consc ousness. The po t ca var at on s the way work ng-c ass c t zensh p comes to be estab shed, w th the three ma n subvar ants be ng def n t on of workers c t zensh p by sov et-sty e revo ut ons (e.g., Russ a, Ch na), mob zat on of d senfranch sed workers n bera soc et es (e.g., Eng and, Scand nav a), and enfranch sement w th tt e worker mob zat on (e.g., the Un ted States, Japan). Before the ndustr a revo ut on v rtua y no emp oyed workers were c t zens; after that revo ut on they were genera y c t zens as def ned by one of these three subvar ant def n t ons. The organ zat ona var at on s the growth of emp oyment of arge numbers of workers w th dent ca abor contracts, such as craftsmen a h red at an occupat ona wage rate, or sem sk ed workers h red as assemb y- ne tenders at a wage determ ned by a category n a bureaucrat c wage system. (The mportance of categor ca determ nat on of the abor contract was argued n Lockwood 1958.) The propos t on s that c ass consc ousness s max m zed when workers have had to be mob zed to ach eve enfranch sement and when arge groups of them are emp oyed at un form wage rates to do s m ar work. Before the ndustr a revo ut on few workers were emp oyed as members of categor es of workers hav ng dent ca abor contracts; after the ndustr a revo ut on many were e ther art sans or factory sem sk ed workers. The organ zat ona factor—that there are standard contracts for groups of workers so that the worker enters a predef ned structura "s ot" when tak ng a ob— s the ma n way c ass consc ousness s connected to ndustr a zat on. Factor es and sk ed trades n manufactur ng c t es produce groups of workers who f s ots n a prev ous y organ zed product on system on standard terms. Soc et es bas ca y do not have work ng-c ass consc ousness before ndustr a zat on. But the prob em of negot at ng the pos t on of groups of s ot ho ders

276 s def ned d fferent y n d fferent po t ca systems. Work ng-c ass consc ousness s the resu t of the nterp ay of the organ zat ona process of putt ng workers nto s ots on the one hand, and the po t ca def n t on of what r ghts and bert es such groups of s ot ho ders have on the other. These po t ca def n t ons vary among the ma n types of po t ca systems. In the th rd sect on we take up E. P. Thompson s great ana ys s (1963) of the nterp ay between occupat on and factory emp oyment n Eng and on the one hand and Eng sh po t ca cu ture on the other. I n neteenth-century Eng and, workers had to mob ze to ga n the franch se. It was a bera soc ety n wh ch workers had many po t ca r ghts (perhaps the most mportant ones were " mmun t es" rather than r ghts, such as freedom from arb trary search and se zure, a certa n freedom of speech, etc.) and n wh ch there was po t ca compet t on n Par ament. Thus, Eng and was the f rst country to produce a "modern" work ng-c ass consc ousness n a bera soc ety. Thompson s strategy s to show how the cu tura themes of the Eng sh poor, espec a y the po t ca themes, were used by Eng sh workers from about 1790 to about 1830 to nterpret the mean ng of occupat ona , ndustr a , and po t ca oppress on. Th s then forms the bas s for understand ng how work ng-c ass organ zat ons such as trade un ons came nto ex stence, and why they formed po t ca a ances on the eft of the par amentary system. We w then app y Thompson s cu tura strategy (as we have extracted t), comb ned w th Lockwood s ana ys s of the soc a and econom c bas s of groups of s ot ho ders, to ana yze var at ons n c ass consc ousness n d fferent parts of modern bera econom es. F na y, we app y th s genera theory to exp a n why the serv ce sector of modern econom es s n genera poor y un on zed, and why so many workers n th s sector do not vote for work ngc ass part es. The two bas c var ab es we dent fy are the tendency of serv ce workers to have nd v dua zed abor contracts, and the tendency for the mean ng of the r work to be def ned n terms of conservat ve symbo s. Ind v dua zed abor contracts make the r membersh p n categor es w th a common wage rate and common cond t ons of work prob emat c. The da y use of conservat ve symbo s n work prov des a framework for nterpret ng gr evances and opportun t es n non-c ass-consc ous terms.

Unity in Diversity: Why Are Societies with Factories So Much Alike? The fundamenta causa forces assoc ated w th ndustr a zat on that br ng about work ng-c ass consc ousness are, then, the organ zat ona one of

277 bureaucrat zat on of the wage re at onsh p and the po t ca one of ega zat on of the suffrage re at onsh p. F rst I w say what these are, and why they are assoc ated w th the reorgan zat on of product on n the course of ndustr a zat on. Th s part of the ana ys s ust f es treat ng c ass consc ousness n a chapter n a book on organ zat ona soc o ogy.

Bureaucratization of the Wage Contract The fundamenta nnovat on of ndustr a zat on, part cu ar y n ts factory form but a so n ts art sana form, s that a arge number of workers are h red by one emp oyer to do essent a y the same work n the same p ace, and are tra ned to do that work n essent a y the same way (Lockwood 1958). Th s means that w th n the f rm t seems unfa r to make nd v dua and d fferent arrangements w th each worker, and the soc a bas s of the nd v dua zat on of contracts between workers and emp oyers d sappears. The emp oyer then makes the contracts of emp oyment for categor es of workers un form and sub ects the workers to un form standards of work (e.g., un form output cr ter a); the category of workers becomes a "co ect ve worker." Consequent y, the abor market wage rate becomes a soc a and psycho og ca rea ty to the workers, because t s an exp c t part of the ncent ve system. Th s s n essence a co ect ve contract before co ect ve barga n ng. C ass consc ousness as nf uence on the wage rate s co ect ve worker part c pat on n the co ect ve contract that a ready ex sts . The organ zat on of trade un ons and other dev ces for worker representat on s, n some sense, mere y recogn t on n workp ace po t cs of a cond t on that s created by the nature of factory product on: name y, that there w be a co ect ve contract for categor es of workers. In art sana or craft-type product on, the un form ty of workers s produced by a abor market–w de tra n ng and exper ence system, one n Western European and North Amer can c t es descended from med eva urban gu d structures (see esp. Sewe 1986, 1980; Hobsbawm 1984, 214–251). Here deas of the fa rness of wage rates spread beyond spec f c emp oyers to the oca abor market as a who e; n the end there s a percept on that a mach n sts n a part cu ar p ace, for examp e B rm ngham, ought to make the same wage. Genera y the trade un on structures to respond to th s sort of market abor categor zat on form ear er, and are stronger, than those represent ng factory workers. The cruc a feature n the causat on of trade un ons s thus not rea y "bureaucrat zat on" of the abor contract by a factory management. Instead t s that there s a soc a y estab shed un form ty of workers, together w th a un form ty of factua wage rates for that un form c ass of

278 workers e ther w th n the factory or n a oca abor market. It s un versa sm of the abor market re at on for a category of workers, rather than the bureaucrat zat on emphas zed by Lockwood (1958), that s cruc a . Industr a zat on great y ncreases the proport on both of urban craft workers n the soc ety and of factory operat ves; thus t ays the bas s for c ass consc ousness.

Legalization of the Suffrage Relationship The workers po t ca s tuat on descr bed by E. P. Thompson for the Eng and of 1790–1830 s one w th d fferent suffrage re at ons n d fferent par amentary d str cts (see, e.g., the ana ys s of act ve worker part c pat on n Westm nster: 451–471; a page numbers c ted n the fo ow ng w th no further spec f cat on are to Thompson 1963). Such a system has many d st nct types of oca po t ca arrangements, each w th ts own determ nat on of who part c pates n po t cs and how nf uence s arranged. It s a much-reduced form of the near y nf n te po t ca var ety of oca po t ca re at ons under feuda sm. A feuda government n feuda sm s heyday was a ramshack e centra government bu t on an astound ng var ety of oca po t ca systems. As the centra governments ga ned more power, there were a sorts of demands, both by the k ngs and by the popu ace, to make the system of oca po t ca organ zat on more regu ar and un form. For nstance, there are demands by the nob e house of the Estates Genera e ected ust before the French Revo ut on to award nob ty on y for mer t, to make commerc a taxes regu ar and un form, to regu ar ze the court system and make ust ce more ega zed, and so on (We tman 1968, 269–441; Markoff 1975). Hav ng such nat ona un form ty mposed on oca governments throughout the country, and a un form set of cr ter a for who gets to vote and who can ho d off ce and what each off c a can do and the ke, means un form ty n oca po t ca nf uence w be nat ona y ega zed. The ntroduct on of such un form ty s of course fu of conf ct, because somebody oses every t me the oca po t ca const tut on s changed. Lega un form ty of suffrage s an mportant po t ca precond t on for work ng-c ass consc ousness, s nce t makes the po t ca demands of d fferent groups of workers n d fferent parts of the country un form. The ega un form ty of a the oca po t ca systems produced by a grow ng power of the nat ona government usua y d v des the upper c asses, as the reform movement n Eng and that eventuated n the Reform of 1832 d d—some parts of the upper c ass ost some po t ca pr v eges (e.g., oca ar stocrats m ght ose the r "rotten boroughs"), wh e other parts ga ned. The un form ty presents the opportun ty for work ng-c ass mo-

279 b zat on to nf uence the nat ona structure of suffrage, rather than on y to nf uence the po t cs of part cu ar par amentary d str cts. The d v s on w th n the upper c asses dur ng the reform movement may create opportun t es to form a ances between the workers and the former y d spr v eged bourgeo s e. When such an ntroduct on of nat ona un form ty takes p ace n an o garch ca or d ctator a context, one of the th ngs that usua y gets regu ar zed and ega zed s the underpr v eged status of workers. Some workers who were prev ous y enfranch sed are ke y to ose po t ca nf uence, and at any rate the ega category of "workers who are d senfranch sed" s created on a nat ona sca e. But that makes t obv ous what a worker has to do f he or she wants to have more work ng-c ass nf uence on e ther nat ona or oca po t cs—to change the nat ona cond t ons of suffrage so that the poor can vote, ho d off ce, and so on. Oppress ve ega zed un form ty s a bas s of work ng-c ass un ty. When democrat c suffrage had a ready been ntroduced nto many oca systems before the work ng c ass grew, as n the Un ted States, then there was very tt e pressure to form a nat onw de work ng-c ass movement around the quest on of suffrage. When trade un ons formed ater, they tended to be ess po t ca than they were n Eng and and most of Europe. Thus, t s nat onw de mob zat on for nat onw de ega zat on of work ng-c ass suffrage that max m zes c ass consc ousness.

Urbanization Th rd n mportance to bureaucrat zat on of the abor re at on and ega zat on of suffrage as causes tend ng to connect ndustr a zat on and ndustr a sm to work ng-c ass consc ousness s ndustr a sm s urban z ng effect. In pre ndustr a Europe, even n countr es ke Eng and where there was an agr cu tura pro etar at, there was not c ass consc ousness among workers n agr cu ture. A b g reason for th s was that agr cu tura workers ved n agr cu tura v ages that were c ass heterogeneous. When ndustry came, espec a y factory ndustry, workers moved nto sma ndustr a towns and c t es. The quest on of suffrage for the work ng c ass has thus a ways been a quest on of suffrage for the urban poor, and has been separate from the quest on of suffrage for sma farmers or agr cu tura aborers.

E. P. Thompson's Conception of Working-Class Consciousness The ma n th ng that d st ngu shes the soc o og ca approach to ndustr a sm, and to the process by wh ch soc et es became character zed by mod-

280 ern, eff c ent manufactur ng and mass d str but on, s ts nterest n c ass consc ousness. Soc o og sts genera y character ze soc et es as hav ng more or ess ndustr a sm, and more or ess c ass consc ousness. That s, soc o og sts do not n the f rst nstance ask the quest ons, how much does an nd v dua person fee ke he or she s a worker, ob ect f h s or her daughter marr es someone other than a worker, fee that on y workers po t ca and econom c nterests shou d be defended, and so on. These are quest ons about nd v dua dent ty as workers, wh ch the soc o og st genera y tr es to answer after he or she has character zed the soc ety n wh ch the nd v dua ves. Thus, when we want to know whether, for examp e, a b ack worker n the Un ted States th nks of h mse f or herse f f rst and foremost as a worker or as a b ack person, a soc o og st attacks th s quest on by ask ng f rst, How far s the soc a , econom c, and po t ca fe of b ack peop e n the Un ted States organ zed around the r status as workers, rather than around the r status as b ack peop e? After hav ng answered that quest on n suff c ent deta so as to understand what s go ng on at the soc eta eve , he or she tr es to exp ore how th s soc a state enters nto a person s consc ousness, nto the se f-def n t on and the def n t on of h s or her nd v dua soc a , po t ca , and econom c nterests, g ven how the soc ety s organ zed nto c asses and nto ethn c groups. E. P. Thompson asserts that the Eng sh work ng c ass had been "made" by the 1830s. What spec f ca y d d he mean by that? F rst, a broad character zat on (see a so Sabe 1982, 127– 193): he meant that one can wr te sens b y about Chart sm, a Br t sh movement favor ng ncreased democracy n the Br t sh government that deve oped after the per od Thompson covers, by ask ng how the work ng c ass responded to the movement. What one w be ta k ng about s the work ng c ass as an organ zed po t ca and soc a force w th a sense of ts own dent ty, a more or ess re ab e fo ow ng, and a v s on of tse f and of ts h stor ca nterests. If, however, one wr tes about the react on n Br ta n to the French Revo ut on at the beg nn ng of the per od Thompson treats (about 1790), one must wr te nstead about th s and that democrat c tendency, w th an occas ona art san, an occas ona trade un on– ke group of workers, adher ng to the movement. There wou d be no c ear tendency for a sorts of workers to react a ke to the r common econom c and po t ca nterests, wh ch m ght have ed them to favor the rad ca tendenc es of the French Revo ut on. So one cou d not wr te a sens b e h story of the democrat c tendenc es n Eng and (or Great Br ta n) re ated to the French Revo ut on from the 1790s to 1815, say, n terms of how the work ng c ass reacted to the Jacob n tendency.

281 We w say that a soc ety s character zed by a h gh degree of c ass consc ousness when the fo ow ng two tendenc es are strong. 1. Many members of the abor force def ne the r ma n econom c nterests n terms of the wage rate of some group of workers, want that wage rate to be h gher, and o n and support some sort of trade un on to ra se that wage rate. The more they th nk of the r own wage eve as be ng set by the genera wage eve of a workers, the more th s s "c ass consc ousness" as opposed to " ob r ghts" consc ousness, but we want to nc ude both broad (c assw de) and narrow ( ob r ghts) def n t ons of c ass nterest here. C ass-consc ous workers enter nto conf ct over the sett ng of genera wage rates w th the emp oyers who pay those rates. Such work ng-c ass consc ousness s n contrast to workers be ng nterested pr mar y n the pr ces of commod t es they produce (as sma farmers or peasants are n many agr cu tura systems, and as some art sans were n pre ndustr a Europe) or n the s ze of government handouts (as most of the poor of Rome are reputed to have been) or pens ons (as a good many Northern C v War veterans were; see Skocpo and Or off 1983) or soc a nsurance benef t eve s (as peop e over s xty-f ve are n the Un ted States today; see St nchcombe 1985d). 2. Po t ca nterests of the poor n soc ety (often n many d fferent concrete po t ca organ zat ons) are organ zed separate y from the r ch, creat ng a workers movement that (a) favors the nterests of workers n wages, and n trade un on organ zat ons that defend wages: that s, defends workers " ndustr a " r ghts and nterests; (b) favors the extens on (or defense) of the franch se n po t cs to a workers, or at east a ma e workers, at east for the urban part of the economy: that s, dent f es "democracy" w th the po t ca nterests of urban workers and defends democracy so conce ved; and (c) un f es a arge part of the wage workers n the soc ety n defense of the above two-pronged program (econom c and po t ca ), as aga nst one or more part es of the r ch and powerfu that oppose the power of trade un ons, favor the reduct on of wages, and favor restr ct on of the franch se to "the r ch, the w se, and the we born." Thompson s say ng more spec f ca y that on y a sma part of the work ng popu at on n 1790, f rst, dent f ed ts ma n econom c nterests w th the wage eve of a group of workers, and w th the correspond ng trade un ons that defended that wage eve , and, second, dent f ed ts po t ca nterests w th an ncrease n the po t ca power of urban wage workers, espec a y by advocat ng extens on to them of suffrage and the r ghts of po t ca organ zat on. There were "the abor ng c asses," but not "a work ng c ass," n 1790; by 1830 there was a substant a work ng c ass nterested n wage eve s, trade un ons, and the extens on of suffrage to urban ma e workers.

282 Such a def n t on of work ng-c ass nterests prov des " nformat on" to workers n part cu ar organ zat ons about how to nterpret the organ zat ona ncent ve system. It te s them to ook at t as a co ect ve contract rather than as a reward for nd v dua product v ty. It te s them not on y that those w th opposed nterests n the co ect ve wage eve a so oppose the enfranch sement of workers and the po t ca eg t macy of trade un ons, but that the r part cu ar emp oyer s a po t ca opponent as we . It te s them that the r nd v dua troub es at work are gr evances to be processed n the oca co ect ve barga n ng system n the workp ace, or n co ect ve movements n the arger soc ety. In short, t concret zes the uncerta n future of the nd v dua abor contract by p ac ng t n the context of co ect ve nst tut ons of c ass. Throughout h s book Thompson s try ng to show the mmense var ety of thought and exper ence of work ng peop e and the r eaders (many of the eaders were, of course, workers, though many were not) n wh ch these centra character st cs of modern Br t sh h story were formed. One of the hardest th ngs to keep n m nd when read ng or wr t ng h story s that no one n 1790 knew how t was a go ng to come out by 1830. For one th ng, n 1790 very few peop e worked for wages, and when they d d t was most y n the countrys de where the soc a and po t ca cond t ons were very d fferent from those n the c ty of London or the ater ndustr a d str cts of the north of Eng and. For another, the who e dea of a "wage rate" had no ega ex stence, nor d d t ex st much n soc a pract ce or even as an nte ectua construct. R cardo, the f rst great theor st on the quest on of how wages n genera came to have the eve they do, was st a re at ve y unknown academ c scr bb er even by the end of Thompson s per od (wh ch ends about when R cardo d ed). In 1790 near y everybody thought that what a worker got pa d was the resu t of a barga n that he or she as an nd v dua came to w th h s or her emp oyer. An nd v dua worker who d d not ke th s wage rate cou d go work for h mse f or herse f (wh ch d d not seem so unrea st c n those days). Econom c nterests n 1790 d d not natura y d v de themse ves nto the nterests of workers n h gher wages and the nterests of emp oyers n ower wages. The dea of a co ect ve barga n n wh ch workers had some power over wages was an abom nat on. Peop e d d not know that they wou d be th nk ng n terms of co ect ve wage rates by 1830. S m ar y, there was very tt e not on genera y n 1790 of what anyth ng that m ght be ca ed "democracy" was a about, and certa n y very tt e tendency to dent fy t w th po t ca nterests of a body of work ng peop e. Work ng peop e d d not vote, and the dea that they wou d know how to run th ngs was as fore gn as the not on that students m ght be ab e to run a un vers ty wou d be n the 1980s. Probab y most workers n

283 1790 had very tt e not on that po t cs affected them, had no trusted representat ves of the r nterest to formu ate po t ca ob ect ves for them, and n other ways were psycho og ca y and organ zat ona y "nonc t zens." Not conce v ng that they were members of a po t ca body, workers genera y had no more po t ca op n ons about what ought to be done n Eng and than we have about what ought to be done n, say, Japan, where we are nonc t zens. By ow work ng-c ass consc ousness at the soc eta eve n 1790, we do not mean that fo ks d d not have econom c gr evances about the r s tuat on or that they d d not fee a enated from the po t ca system because the r gr evances were not so ved. We mean nstead that there were few t es between econom c gr evances and co ect ve act on that ra ses wages n the economy, and a so that there were few t es between workers po t ca gr evances about the r own power essness and any co ect ve movement that favored democracy and def ned t as the part c pat on of work ng peop e n the shap ng of po t ca dec s ons. What Thompson s document ng s the hundreds of ways that ord nary peop e cou d come to th nk about the r gr evances and the exper ences of the r econom c and po t ca ves, and the ways that the r eaders (some of them from other c asses) cou d formu ate the r nterests n the econom c and po t ca systems and get a fo ow ng. For examp e, ord nary peop e had no neat p cture of a gradua deve opment from a state of worker econom c and po t ca apathy to a state of co ect ve nterest n the wage barga n and of co ect ve organ zat on nto a c ass party n favor of democracy (mean ng un versa urban ma e suffrage). Instead there was much work ng-c ass enthus asm for a var ety of movements, nc ud ng Owen te soc a sm (formu ated by an emp oyer), wh ch den ed the c ass conf ct by try ng to organ ze work w thout wage re at ons; Method sm, wh ch def ned many character st cs of work ng-c ass cu ture as s ns and nterpreted the bad ot of workers as the resu t of the r s nfu ves; Jacob n sm, whose adherents thought the democracy created by the French Revo ut on was a ot better than the par amentar sm of the r ch that governed Eng and; the Church and K ng mobs that went after Jacob ns; craft trade un ons whose eaders wanted to d vorce craft quest ons from po t cs; and craft un on sts who thought revo ut onary po t ca act on and a sort of re at ve y nnocent terror sm were necessary. The po nt s that h story as t s ved s a ot more comp cated than h story as one m ght reconstruct t, f one knew n advance that n 1830 there wou d a work ng c ass organ zed to some degree nto trade un ons and ready to funct on together n a prodemocrat c po t ca movement. What the ast chapter of Thompson (711–896) s ma n y about, then,

284 s the great var ety of thoughts that st had nf uence among workers after 1830. These deas were the egacy of what we m ght, ook ng at t retrospect ve y, ca "the strugg e to def ne Eng sh workers as a c ass." Here Thompson def nes the cr ter a he has used to sort through th s enormous var ety of po t ca thought of the 1820s and ear y 1830s and to say that (ba anc ng as one has to to respect the ntegr ty of the thought of the peop e actua y funct on ng on the scene) the Eng sh work ng c ass actua y ex sted by 1830 n a h stor ca y s gn f cant sense. The appropr ate cr ter a for Thompson to use n dent fy ng quotat ons from the documents as man fest ng the ex stence of a c ass that s a ready made are (1a) that the nterests of the workers are def ned pr mar y n terms of wages; (1b) that the nterests of workers are def ned as be ng n oppos t on to those of emp oyers; (1c) that workers are shown as be ng nterested n bu d ng trade un ons, n defend ng trade un ons, n co ect ve barga n ng about wages and work ng cond t ons, and so on; (2a) that peop e connected to workers because they are eaders of trade un ons, or e ected off c a s appea ng to a work ng-c ass const tuency, or eaders of r ots or other co ect ve man festat ons composed of workers, are n favor of democracy n the sense of extend ng the suffrage; (2b) that those same peop e defend po t ca y the r ghts of trade un ons and of other workers organ zat ons; and (2c) that those same peop e conce ve of the opponents of democrat zat on ( .e., the r own po t ca opponents) n c ass terms, as propert ed and r ch, as nterested n ower ng wages, as un nterested n the we fare of the work ng peop e. More concrete y, we w be nterested throughout the m dd e sect ons of th s chapter n s x features of deo og ca thought that we, fo ow ng Thompson, have dent f ed as be ng nd cators of grow ng c ass consc ousness. To supp y catchwords, they are (1a) wage nterest, (1b) econom c conf ct over wage eve s, (1c) trade un on organ zat on, (2a) favor of democracy (un versa suffrage) n po t cs, (2b) defense of c ass organ zat ons (trade un ons) n po t cs, and (2c) a c ass conf ct def n t on of po t ca conf ct. The fundamenta observat on we start off w th s that there was very tt e c ass organ zat on or c ass consc ousness n the sense that Thompson def ned t n e ghteenth-century Europe, and that by the m dd e of the twent eth century t was pract ca y un versa n the advanced bera soc et es (Western Europe, the Un ted States, Canada, Austra a, and New Zea and). So we start w th the soc o og ca observat on that n fact c ass consc ousness, n the sense we (fo ow ng Thompson) have def ned t , s a feature of ndustr a zed soc et es; hence, t h stor ca y came about dur ng the n neteenth century n a the soc et es then undergo ng ndustr a zat on. The bas c po nt of the Thompson book s tt e has n fact

285 been sa d when that has been sa d, because we st have to exp a n how workers created the structures and consc ousness that are the features of ndustr a zed soc et es. An add t ona reason for want ng to understand the connect on of factory and craft organ zat on to c ass consc ousness s that at the present t me, a ot of s gns po nt to an eros on of c ass consc ousness, n the sense def ned above. Th s eros on seems somehow to be connected to the growth of the serv ce sector and of bureaucrat c management, nvo v ng the growth of m dd e-c ass emp oyment.

Cross-national Variation in Class Consciousness The argument above s that n many ways the centra transformat on of soc a structure that was ntroduced w th ndustr a sm s the po t ca and econom c organ zat on of the work ng poor and the r ncorporat on as c t zens, as def ned by the nst tut ons bu t up w th ndustr a sm n the econom c and po t ca order of soc ety as a who e. Th s was a bas c transformat on n how the poor of the soc ety came to th nk of who they were, of what cou d be done about the r econom c s tuat on, of what the r po t ca nterests were, and of what they wanted the state to do about the r cond t on. It s a most true to say that a soc et es that are ndustr a zed have trade un ons n one form or another, wh e very few underdeve oped countr es do; t s a so a most true to say that a ndustr a zed soc et es mob ze the workers nto po t cs under a system of un versa suffrage and have part es that at east c a m to represent the nterests of the workers n oppos t on to other part es or po t ca tendenc es that oppose those nterests. Most underdeve oped countr es bas ca y have ne ther trade un ons nor worker po t ca part es (Inke es and Sm th 1974). But there are two ma or var at ons from the p cture g ven by Thompson, that are represented by the Sov et Un on on the one s de and by Japan on the other. The core of th s compar son comes out of Barr ngton Moore s great book Soc a Or g ns of D ctatorsh p and Democracy (1966) and, more spec f ca y on the p ace of abor re at ons, from Re nhard Bend x s Work and Author ty n Industry (1956; see a so Wa der 1986; Burawoy 1985, 156–208).

Class Consciousness in Soviet Societies In the USSR, organ zat ons that or g nated as work ng-c ass part es and trade un ons became the po t ca ru ers, and w ped out the represen-

286 tat ves of the upper c asses. Then "reasons of state" became the respons b ty of the ru ng work ng-c ass party and of the trade un on apparatus, and they devoted themse ves at east as much to preserv ng ru e, to bu d ng m tary m ght, and to ncreas ng product v ty as to represent ng the workers nterests wh ch they had or g na y represented. A new m dd e c ass n the economy and a m dd e and upper c ass n the state grew up w th n the she of organ zat ons represent ng the deo ogy and po t ca power of the work ng c ass. One resu t of th s evo ut on s that the upper c ass n the USSR does not ve near y as we as the upper c ass of most cap ta st countr es of the same eve of deve opment (e.g., Ita y, Spa n, Venezue a, Uruguay, S ngapore, Hong Kong, and the o states of the Arab an pen nsu a are perhaps at about at the same eve of deve opment as the USSR). On y a very upper-c ass person n Moscow has a three-bedroom apartment, for examp e, and the upper-c ass dachas outs de Moscow that one hears about n ant -Sov et t rades are the sort of cab n that a th rd of the popu at on of Norway have access to, a s ght y crude vers on of a sma one-fam y house. But the Sov et upper c asses have more po t ca power than the upper c asses of these ntermed ate cap ta st countr es, because the tsar st nob e and bourgeo s c asses and the r po t ca representat ves have been destroyed. From a cu tura po nt of v ew, the powerfu n the Eastern b oc are an uncha enged po t ca ru ng c ass w th a very strong cu tura "surv va " of ega tar an sm n econom c pr v eges. But that ru ng apparatus s a so connected w th the work ng c ass n a very d fferent way than Western ru ng apparatuses are (Wa der 1986). In part cu ar, the ru ng group has deep penetrat on nto the Commun st party and the trade un ons. The comm tment of the trade un ons and of the Commun st party to product v ty has meant, for examp e, that what we ca " ndustr a eng neer ng" ( ncent ve systems and mach ne speedups) s more common, s defended by the trade un ons aga nst worker attacks, and resu ts n at east "temporary" workers putt ng out more effort than Amer can workers do. Harast (1978) g ves an account of a mach ne shop that s a ot ke the one stud ed by Burawoy (1979) n the Un ted States. The f nd ngs nd cate that there s cruder exp o tat on of these workers n Hungary, sponsored n many ways by the un on, than n the core of monopo y cap ta sm n Ch cago. Burawoy (1985, 156–208) argues that the core exper enced and sk ed workers n sov et factor es he p management respond to the great uncerta nt es of supp y (cf. Gran ck 1954, 1967) and so are not treated as ruth ess y as the temporary workers descr bed by Harast (see a so Wa der 1986, 28–84).

287 The genera p cture n Eastern Europe and Ch na nc udes work ng-c ass revo ut on, and then the ncorporat on of "reasons of state," nc ud ng espec a y advances n econom c product v ty, nto the agenda of the work ng-c ass movement. Thus the c ass consc ousness of sov et soc et es s channe ed and contro ed by organ zat ons w th a strong nterest n the success of factor es, so that ts consequences are much muted as compared to the c ass consc ousness descr bed by E. P. Thompson.

Class Consciousness in Corporatist Capitalism When ndustr a zat on takes p ace under a strong r ght-w ng "corporat st" reg me that suppresses work ng-c ass organ zat on as much as they can throughout the process, work ng-c ass po t ca representat on can be ach eved e ther by revo ut on, ead ng genera y to the creat on of sov et soc et es, or by other means. Japan from about 1890 to 1945, Germany from 1933 to 1945, and to some degree Ita y, Spa n, and perhaps Braz at var ous t mes are examp es of conservat ve corporat sm, w th m ted and sponsored work ng-c ass representat on n conservat ve reg mes. The resu t ng nst tut ona order after the corporat st reg me n turn oses power ( s overthrown) depends a ot on how democracy and the freedom to organ ze trade un ons are subsequent y ntroduced. The bas c genera zat on s that f modern forms of c ass consc ousness are ntroduced nto such a system w thout be ng extracted from a re uctant r ght-w ng government by a thorough y mob zed work ng-c ass movement (workers were not mob zed to ntroduce suffrage and trade un ons nto Japan n 1945–1946, for examp e), then a country deve ops nst tut ons of work ng-c ass consc ousness ke trade un ons and eft-or ented part es n a bas ca y conservat ve env ronment. Worker organ zat on for conf ct w th the upper c asses s much ess sa ent n Japan and s m ar countr es than n countr es ke Eng and and the Un ted States, where a bera reg me pres ded over ndustr a zat on. The Un ted States s perhaps ha fway between Br ta n and Japan n th s regard, w th the quest on of un versa suffrage not hav ng been so ved n the Un ted States through a mob zat on for conf ct w th the upper c ass but mob zat on nonethe ess hav ng been necessary to get the r ght to organ ze trade un ons and to force co ect ve barga n ng on emp oyers. There s therefore a Labour party w th a soc a st deo ogy n Great Br ta n, a Democrat c party w thout much soc a st h story n the Un ted States, and apparent y permanent-m nor ty eft po t ca part es n Japan. In Great Br ta n, trade un ons have a fundamenta y ant management at-

288 t tude on product v ty quest ons; n the Un ted States, the r att tude s susp c ous but product v ty dea s are made n co ect ve contracts; and n Japan, workers part c pate act ve y n ncreas ng product v ty. A ong the d mens on of mass veness of work ng-c ass mob zat on, rang ng from Japan to Russ a, France s probab y further to the eft than Great Br ta n; certa n y t has more rad ca - eft po t ca part es represent ng the work ng c ass, and more cantankerous, f not more powerfu , un ons than Great Br ta n (Ga e 1983; T y 1986; Hobsbawm 1984). I wou d p ace most other European countr es between France and Great Br ta n (cf., e.g., Esp ng-Andersen 1985 and Stephens 1986 on the Scand nav an countr es; Kocka 1986 on Germany; and, for a genera survey, Sturmtha 1953); a have more rad ca work ng-c ass movements than the Un ted States. Espec a y n the po t ca part of our def n t on of c ass consc ousness, Europe s genera y more eft than the Un ted States.

The Culture in Which Class Consciousness Grew The comparat ve p cture sketched above suggests that, w th some m nor var at on, the or g ns of trade un ons and work ng-c ass part es n most bera soc et es took a form s m ar to that descr bed by Thompson for Eng and. In th s sect on we out ne how Thompson ana yzed the or g ns of c ass consc ousness n Eng and, to see how the stat c comparat ve soc o ogy out ned above appears n a h stor ca ana ys s. The f rst few chapters of Thompson are meant to g ve the f avor of the cu tura presuppos t ons of var ous sorts that w become re evant to the work ng-c ass def n t on of what s about to happen to workers w th ndustr a zat on. The dea of cu ture that s man fested here s n some ways s m ar to that advocated by the French anthropo og st C aude Lév -Strauss, common y ca ed "structura sm." Lév -Strauss argues that a cu ture cons sts not so much of a set of be efs as of a set of a ternat ves for nterpret ng th ngs .

"Members Unlimited" The contrast n Thompson s chapter 1 (17–25) s between the corporate organ zat on of po t ca thought and work ng-c ass rad ca sm. The corporate dea s, for examp e, that our nterests as art sans n the shoemak ng trade are the nterests of the corporat on (the gu d) of shoemaker masters and ourneymen; th s s n contrast w th the not on that everyone ( nc ud ng shoemakers—on shoemaker rad ca sm, see Hobsbawm and Scott 1980) m ght be nterested n, and better off, mp ement ng rad ca

289 deas. The contrast here s between def n ng nterests by the ega and econom c nterests of c osed corporat ons rather than by un versa and democrat c ega and econom c nterests of "the peop e." Thus the contrast s between a c osed art san gu d and the London Correspond ng Soc ety, a rad ca democrat c group from around the beg nn ng of the n neteenth century that started ts ru es w th the words "That the Number of our Members be un m ted." The po nt here s not that the not on of un m ted so dar t es—what we wou d ca "un versa sm" n soc o og ca argon—was common among the work ng c ass. Thompson knows that the London Correspond ng Soc ety was an e te group, w th some two thousand members n 1793; he s therefore not descr b ng the natura cond t on of the poor n the 1790s when he presents the London Correspond ng Soc ety s dea of a common nterest n rad ca not ons. He s ust demonstrat ng that the cu tura baggage of the London work ng c ass at the t me nc uded an a ternat ve to the c osed occupat ona corporat on—name y, the London Correspond ng Soc ety, wh ch had n common w th the eventua shape of work ng-c ass consc ousness that t extended so dar ty beyond the craft. One a ternat ve for nterpret ng workers exper ence dur ng ndustr a zat on, then, was n terms of the pr v eges of c osed corporat ons, wh e another a ternat ve was n terms of an un m ted number of members of the common peop e.

"Christians and Satan" The second and th rd chapters of Thompson s work (26–76) show that work ng-c ass cu ture was character zed not on y by the heavy-handed protestant stra ns that wou d be man fested n Method sm, but by a hearty, fun- ov ng, dr nk ng, braw ng cu ture as we . Th s theme says that a har ot sn t rea y a bad, she part y represents good d rty fun; and that dr nk ng and cock f ght ng may be ob ected to by stuffy parsons, but a rea man kes a tt e amusement. Aga n, the po nt s not that there were not p enty of sober, God-fear ng workers n the 1790s. Both s des of the contrast between bawdy a e dr nkers and sober evange ca s def ned themse ves by contrast w th each other, and so each "understood" the way of fe of the contrast ng group to some degree. That contrast prov ded two ways of def n ng what t meant to be a worker n 1790s, and t wou d ater turn up n a Method st-versus-c ass-so dar ty way of def n ng who the peop e were that a good worker wou d dent fy w th. Aga n, part of the po nt s that the nascent work ng c ass cou d understand th ngs e ther way: that, for nstance, the poor cou d be bad because they were s nners or good because they were sober, and even

290 a worker who had a good dea of the "bad" n h m wou d understand the force of the argument from s n. The good Method st kew se understood to some degree the defens ve argument that a man wants a b t of fun after work, after a .

"The Free-Born Englishman" Chapter 4 (77–101) shows that patr ot sm and respect for author ty was contrasted w th a pos t ve aff rmat on of the va ue of be ng et a one and hearty ob ect on to an off c a or other author ty that oversteps what s r ghtfu y due h m (an off c a was of course somet mes a queen, but was usua y a man, n Eng and). Though of a ater era, the ne from G bert and Su van s P nafore s nd cat ve: "For a Br t sh tar [sa or] s a soar ng sou /As free as a mounta n b rd." The song was conv nc ng even n a soc ety where n fact a Capta n ran the sh p w th an ron hand: for he on y d d so n a context where work ng sa ors had the r ght to have a b t of fun when n port or to be Method sts. The bas c contrast here s that of reverence for the crown and church as respons b e for the nat on and ts sou versus an aff rmat on of

reedom rom abso u sm ( he cons u ona monarchy) reedom rom arb rary arres r a by ury equa y be ore he aw he reedom o he home rom arb rary en rance and search some m ed ber y o hough o speech and o consc ence he v car ous par c pa on n ber y (or n s semb ance) a orded by he r gh o par amen ary oppos on and by e ec ons and e ec on umu s (a hough he peop e had no vo e hey had he r gh o parade huzza and eer on he hus ngs) as we as reedom o rave rade and se one s own abour Nor were any o hese reedoms ns gn can aken oge her hey bo h embody and re ec a mora consensus n wh ch au hor y a mes shared and o wh ch a a mes was bound o ake accoun (79)

The fundamenta dea s "You can t push me around." Aga n, the po nt s not that a poor men, n every s tuat on, wou d th nk automat ca y about the proper m ts of author ty, about be ng free-born Eng shmen (women were most y not thought of as free born, though they were n fact often defended by the m ts of author ty nvo ved n the ega vers on of the concept). In fact, there s a contrast between the not on that the upper c asses know what they are do ng and shou d be et do t, w th coerc on to back them up f necessary, and the dea "They can t push me around." As the re at ons of author ty at work

291 and n po t cs changed over the forty years between 1790 and 1830, workers cou d nterpret those changes n terms of the contrast. Conservat ve workers wou d recogn ze what rad ca s were ta k ng about when they used the phrase free-born Eng shmen , and the rad ca s wou d recogn ze the not on that soc ety wou d come apart f the ru ers were not respected.

"Planting the Liberty Tree" The ast chapter of Thompson s background portra t (102–185) s about ate nnovat ons n th s genera work ng-c ass trad t on, e ements of what we common y ca "the En ghtenment." Here the cu tura contrast s between the c ear ght of reason and the obscurant sm of trad t on. The centra f gure n a th s, accord ng to Thompson, was Tom Pa ne, or g na y a corset maker but ater a sort of profess ona revo ut onary pub c st n the Amer can, French, and now Eng sh rad ca movements. The bas c theme of the aff rmat on of reason s that anybody can read (or cou d f they were taught, and they ought to be taught), and anybody can reason. The En ghtenment ana ys s of soc a fe s that when they do reason, peop e w conc ude that the system s r d cu ous when t tr es by appea s to trad t on to prevent the conc us ons from be ng taken nto account. Thompson s carefu to po nt out that th s rad ca sm s not c ass consc ous by h s def n t on; rather, t s what we wou d ca "popu st." In the En ghtenment pos t on, t s not spec f ca y c ass-consc ous thought that s super or (that s, the Marx st pos t on), but genera y rad ca , "reason ng," thought of a k nds. The En ghtenment pos t on defends what a Marx st nowadays wou d ca "petty bourgeo s rad ca sm." The a ternat ve v ew s that the nst tut ons we ve under have a remarkab e w sdom of the r own. If pressed, some peop e cou d g ve a rat ona zat on of how that w dsom came to be n nst tut ons, ones often theo og ca or ar stocrat c (w sdom s n the b ood of the ru ers). But the contrast ng not on to the En ghtenment ana ys s s, rough y, that "the heart hath reasons that reason knows not of," w thout exp c t theo og ca reason ng. To stoy s short story ([1886] 1982) about the s mp e herm ts on an s and taught the Lord s Prayer by a pass ng b shop, who forget the words and so wa k across the water to ask the b shop to teach them aga n, s a perfect express on of th s att tude. It s not scr pture that counts, but the good trad t ona hearts of the s mp e poor. In Thompson s background chapters we have a descr pt on of the cu tura mater a s out of wh ch, so to speak, h story—spec f ca y workers and the r eaders—wou d form an nterpretat on of work ng-c ass nterests. The " eft" po es of these contrasts—(1a) extens ve so dar ty,

292 (2a) hearty and somet mes rowdy ma e conv v a ty, (3a) a "you can t push me around" att tude, and (4a) secu ar reason—are subthemes of trad t ona Eng sh work ng-c ass cu ture; workers wou d v ew the events of 1790–1830 through these themes, eventua y to arr ve at the def n t on of the r po t ca and econom c nterests n c ass terms. Converse y, the "r ght" po es—(1b) narrow corporat sm, (2b) pur tan ca re ect on of the se f (and of the worthwh eness of the se f s body part cu ar y), (3b) the v ew that the author t es have the nterests of the who e soc ety to ook after and "know better than us," and (4b) the ph osphy of "the heart hath reasons that reason knows not of," w th ts respect for trad t on—form the cu tura bas s of nterpretat ons w th n the work ng c ass that h nder the format on of c ass consc ousness. The d st nct veness of Thompson s approach to th s quest on s see ng the who e process not as the nev tab e unfo d ng of the "correct" nterpretat on of work ng-c ass nterests, but as the nte ectua strugg es of a bunch of workers to f gure out what was go ng on, n terms of the cu tura her tage that they had. The cu tura her tage s not , n Thompson, a "set of pre ud ces" that predeterm ne how that bunch of workers w see the Eng sh po t ca repress ons dur ng the Napo eon c wars, the econom c exp o tat on of the ear y ndustr a factor es, and the reorgan zat on of nonfactory work by the expans on of the market for sk ed workers. Instead the cu ture s a set of nte ectua too s and mora a ternat ves that can be used to def ne the mean ng of events, and to def ne w th respect to those events who s on our s de, who s on the other s de. In Thompson, cu tura causat on operates ke a person s vocabu ary: vocabu ary does not predeterm ne what that person s go ng to say, but prov des a more or ess m ted set of a ternat ves n wh ch to frame what he or she wants to say. One shou d th nk of Thompson n the background chapters as g v ng the cu tura and mora equ va ent of a d ct onary rather than a sort of "proto-work ng-c ass doctr ne."

The Cultural Perception of Exploitation, Oppression, and the Wage Bargain Th s cu tura background means that the facts of po t ca oppress on and of that spec a sort of exp o tat on character st c of the ntroduct on of ndustr a sm are not suff c ent to exp a n the shape of work ng-c ass so dar ty that eventua y deve oped n Eng and. The react on to those facts s, n Thompson s v ew, a creat ve response by the work ng c ass, us ng both the vocabu ary of mora and soc a udgment that was part of the r her tage and the further mora vocabu ary that wou d deve op dur ng the per od.

293 After an extens ve quotat on of a cotton sp nner s ana ys s of what has been happen ng n Manchester, Thompson summar zes what was seen as "exp o tat on" n the econom c re at ons n the new factor es:

he r se o a mas er-c ass w hou rad ona au hor y or ob ga ons he grow ng d s ance be ween mas er and man he ransparency o he exp o a on a he source o he r new wea h and power he oss o s a us and above a o ndependence or he worker h s reduc on o o a dependence on he mas er s ns rumen s o produc on he par a y o he aw he d srup on o he rad ona am y economy he d sc p ne mono ony hours and cond ons o work oss o e sure and amen es he reduc on o he man o he s a us o an " ns rumen " (202–203)

Thus, n Thompson s chapter on exp o tat on (189–212) there are two conc us ons. The f rst s a stat st ca and mater a one, that probab y between 1830 and 1845 or so workers average rea wages ncreased, but that that ncrease was not very substant a , and at any rate came after about forty years of war and other sorts of cr s s n wh ch rea wages probab y rema ned stab e. What "stab ty" means n th s context s not that everybody stayed the same, but that an equa number got better off as got worse off . So, to use an examp e of Thompson s, wh e the coa m ner under forty who, work ng a ot harder w th the new methods of extract ng coa , was earn ng and consum ng more, a coa m ner over forty who was unab e to work because of an n ury, whose r sk was ncreased by work ng too fast, or who m ssed be ng n ured but was s cker than a worker of h s age wou d have been n o den days because there was more dust n the m nes, m ght have been worse off (211). What stab ty means for the f rst forty or so years of what we norma y ca the ndustr a revo ut on s that rough y an equa number of workers exper enced " mm serat on" as exper enced "progress." The second po nt s that a ot of what workers comp a ned about d d not have to do w th exp o tat on, so much as w th hum at on. The mpos t on of a factory d sc p ne that kept workers from send ng the "boy" out for a pot of a e n the m dd e of the morn ng s not so much a decrease n the standard of v ng as an nterference w th workers ves n a way that s none of the bosses bus ness. The sp nn ng m owner d d not cons der h mse f rea y to be " n the same trade" as the sp nner runn ng the enny. Th s meant t was not be evab e e ther for the emp oyer or for the workers to present quest ons of econom c we fare n a corporat st so dary fash on.

294 Thus, even f workers ves turn out on average to nvo ve about the same amount of tea and sugar after ndustr a zat on as before, f they th nk that they are ent re y dependent for the r way of fe on someone who does not g ve a damn about the r we fare, they fee they are at r sk. They fee they have to use the methods of conf ct and po t cs to protect themse ves, rather than trust ng to the good w of the upper c asses. When peop e have to put the r who e ves nto the hands of a system, they want to assure themse ves that that system s "trustworthy." Thompson s genera po nt s that whatever the stat st ca f ow of mprovements and degradat ons, workers d d not th nk they were dea ng w th a trustworthy system, a system n wh ch the ru ers cared about the r fates, respected the r act on to defend those fates, and regarded them as persons w th nherent va ue to be respected and protected. Part of Thompson s po nt s about the creat v ty of the eaders, who had these mater a s to work w th and used them to def ne the p ace of workers n the arger po t ca order. But part of h s po nt s that s nce these mater a s were part of the work ng-c ass cu ture, the eaders creat v ty was not mere y e te hermet c specu at on; t was a so a bas s for d scourse w th the mass of the workers. Thompson, n h s ast chapter on c ass consc ousness (711–832), pays a ot of attent on to the d scourse of severa eaders of the trades un ons of London, us ng the r organ zat ona connect on to show that qu te a ot of workers be eved that these eaders were ta k ng about the po t ca and econom c cond t ons of workers ves. But for most of the per od that Thompson s ta k ng about, such organ zat ona connect ons were suppressed by the Br t sh government. How, then, s Thompson go ng to make a case that h s sp nner n Manchester s ta k ng to the workers, s engaged n c ass d scourse rather than n rad ca onan sm (199–203)? He does th s n part by show ng the cont nu ty n the themes between what the sp nner says n 1818 n Manchester and what var ous contrast ng v ews regarded as good and worthwh e work ng-c ass fe a generat on ear er, n the 1790s. It s because the 1818 sp nner s d scourse has themes from the cu ture of the 1790s poor that we can conc ude t s part of a d a ogue between the sp nner and the rest of the work ng c ass. Both are concerned w th what makes a respectab e worker and what makes a trustworthy system to entrust a worker s fate to. The ma n cu tura themes n Thompson s work are those that app y to the work ng c ass as a who e (or even Eng and as a who e); they are not d st nct ve to part cu ar f rms or trades. To exp a n var at ons n c ass consc ousness w th n the work ng c ass of modern soc et es, we need to use the same strategy that we (and espec a y Thompson) have used to

295 exp a n var at ons n c ass consc ousness over t me n Eng and, but now to construct exp anat ons at the f rm and occupat ona eve s. We want to turn var ab es that we have descr bed across soc et es, and over h stor ca t me n Eng and, nto var ab es that can exp a n why hosp ta s have ess c ass consc ousness than chem ca factor es, but chem ca factor es ess than automob e factor es. That exp anatory prob em occup es the rest of th s chapter.

Constitutionalism in Modern Organizations Organ zat ons that c ass fy emp oyees nto categor es and treat them a ke produce c ass consc ousness not on y because they create co ect ve barga ns by the r very structure, but a so because they often do not ve by those co ect ve barga ns. Once categor es have been created that are supposed to be treated a ke, workers get nd gnant when dent ca y p aced workers are not treated a ke. Arb trary author ty threatens the eg t macy of structures that are supposed to be fa r more than t does the eg t macy of structures that are supposed to be a network of nd v dua barga ns. But a such structures have been h stor ca y created n abor markets n wh ch the norma structure was a matter of nd v dua contracts, each d fferent, and n wh ch arb trary author ty was the ord nary course of events. Consequent y, the growth of "const tut ona sm" n organ zat ons, w th m tat ons mposed on management so that they are ob ged to fo ow the r own ru es (the ru es they are requ red to create by the r product ve process), reduces conf ct between management and c ass-consc ous workers (for a s m ar argument, see the contrast between Br t sh and Amer can mach ne shops n Burawoy 1985). We can d st ngu sh four broad types of const tut ona sm n modern ndustr a organ zat ons, wh ch we ca (1) due process, (2) ob eva uat on, (3) co ect ve barga n ng, and (4) good sportsmansh p n the ncent ve games.

Due Process Se zn ck, Nonet, and Vo mer (1969) argue that bureaucrat c organ zat ons n modern soc ety ( .e., exact y those organ zat ons hav ng abor contracts that are standard for members of categor es) tend to deve op gr evance procedures that m t the author ty of superv sors and management. These gr evance procedures prov de an opportun ty to accuse management of not fo ow ng the r own ru es prov d ng for equa treatment of peop e w th s m ar obs. The management as a who e has an nterest n fa r treatment, treatment accord ng to un versa st c personne po cy, whether or not they have agreed to those standards w th un ons.

296 Such structures of gr evance process ng tend to deve op workers r ghts to br ng w tnesses on the r own beha f, to know the nature of the charges aga nst them, to appea , not to have the sanct ons aga nst them come nto effect unt the case s reso ved, and so on. By equa z ng the status and powers of workers and management before the gr evance comm ttee, the organ zat on ncreases ts capac ty to use ts author ty n the way t has spec f ed n the ru es govern ng categor es of workers (and of course decreases ts capac ty to use arb trary author ty). Such structures st create c asses and c ass consc ousness, but they are ess ke y to create c ass host ty and consc ousness of c ass oppress on.

Job Evaluation A cruc a part of the structure of un form contracts for emp oyees of a g ven category s the re at on among the categor es. Consequent y, bureaucrat c organ zat ons are more ke y than those w th more nd v dua zed contracts to deve op ob eva uat on schemes that spec fy the re at ons among the contracts that app y to var ous categor es of workers. If these re at ons among categor es are deve oped so as to seem fa r to the workers, they tend to reduce the sort of ob-consc ous categor ca host ty due to one category of workers not be ng treated proper y as compared w th some other category of workers. Many such schemes are ntroduced w th extens ve consu tat on of workers about what makes one ob more worthwh e than another, and so worthy of h gher pay (So tan 1987, 158–171). If management agrees to m t tse f so that t does not reward one category n the system w thout correspond ng rewards for the others, thus keep ng the re at ons among categor es that seem fa r to the workers, they aga n reduce the host ty n c ass consc ousness. Of course, such a system tends to bu d nto tse f d scr m nat on aga nst women, when there s a consensus between management and the better-organ zed workers that categor es of workers who are most y women ought to get ower wages. In the ong run th s tendency may underm ne the so dar ty between ma e and fema e workers.

Collective Bargaining Co ect ve barga n ng s rare y ntroduced by management n order to pac fy and mprove the product v ty of ts workers. However, a ne of argument n econom cs has urged that co ect ve barga n ng prov des the advantages of const tut ona sm, and so mproves work d sc p ne. Schumpeter carr ed the argument furthest, stat ng that on y a soc a st government m ght be ab e to prov de enough work d sc p ne to save cap ta sm (1942, 211). The argument was preva ent enough n the group

297 of abor econom sts at the Un vers ty of Ca forn a at Berke ey to cause them to start a ser es of stud es n the ate fort es and f ft es ent t ed "Industr a Peace Under Co ect ve Barga n ng." Recent demonstrat ons by Freeman and Medoff (1984) that un on workers genera y have h gher product v ty (10 to 30 percent h gher) than nonun on workers n the same ndustry, mak ng the r wages h gher even n very compet t ve markets, support th s argument. Even when management recogn zes ts nterest n fo ow ng the ru es mp c t n a modern f rm structure that has dent ca abor contracts w th arge groups of emp oyees, un ons may st be necessary to d sc p ne management to fo ow the r own nterest (R. Freeman 1976). Just as the Eng sh and Amer can ega systems are eg t m zed by the fact that both the p a nt ff (or prosecutor) and the defendant have awyers, so the const tut on of the modern bureaucrat c enterpr se s eg t m zed by categor es of emp oyees hav ng the r un ons. Managements do not have enough good sense to keep the r own structures n good const tut ona work ng order, to eg t mate the r own author ty; they need un ons to g ve the workers vo ce, ust as soc ety needs oppos ng attorneys to g ve defendants vo ce.

Good Sportsmanship M chae Burawoy (1979) observed that a arge bureaucrat c f rm w th ncent ve pay for mach n sts had changed ts po cy over the years so that t p ayed the game more fa r y now. In an ear er study of the same f rm, the nvest gator found that when product v ty went up so that many workers were earn ng ncent ve prem ums, the base requ rement before prem ums were earned was boosted. But when Burawoy was there, peop e were perm tted to keep the r w nn ngs w thout chang ng the base product v ty. S nce Burawoy observed that the workers were treat ng the ncent ve system as a game, the appropr ate mage for descr b ng th s new behav or of management of not chang ng the ru es every t me the workers started to w n s "good sportsmansh p." Burawoy argues that th s good sportsmansh p by management tended to cause workers to focus attent on on the r persona performance and the s ze of the r prem ums, rather than on the bas c terms of the game. But the bas c exchange of money for effort was bu t nto the terms of the game. Hence, management s pract ce of good sportsmansh p tended to create deo og ca hegemony, a systemat c ack of attent on by workers to the nherent y conf ctua sett ng of the terms of the abor contract for the category of mach n st. Thus t reduced c ass consc ousness.

298 The overa pred ct on, then, s that the more const tut ona the procedure of arge bureaucrat c organ zat ons s (the more t s character zed by due process, consensua ob eva uat on, co ect ve barga n ng, and good sportsmansh p by management n the ncent ve games t p ays w th workers), the ess host ty there w be n the c ass consc ousness that such structures nev tab y tend to create.

Debureaucratization, or Individualizing the Labor Contract The structures for reduc ng c ass consc ousness a eave the bas c categor ca abor contract ntact. An a ternat ve strategy s to nd v dua ze the abor contract as far as poss b e. For examp e, a common pract ce n sett ng compensat on eve s n Amer can un vers ty abor contracts s as fo ows. The un vers ty centra adm n strat on computes the overa ncrease n abor cost t can afford the next year, then d saggregates th s nto percentage goa s for d fferent deans and heads of serv ce d v s ons, usua y reserv ng a cont ngency fund to dea w th spec a cases. Deans and heads of d v s ons n turn g ve percentage goa s to the r department cha rs, usua y sav ng some a owance for spec a cont ngenc es. Department cha rs then d v de th s amount among the facu ty, a m ng to meet potent a outs de offers, to reward sen or ty, and to reward other spec a mer ts. Though the average ncrease s set by the centra adm n strat on, that average s nd v dua zed n a set of comp ex and part y secret po t ca processes that g ve everyone the mpress on that they have a spec a abor contract, one that recogn zes the r spec a mer ts or the r most recent outs de offers or the fact that they had bad uck w th ourna referees the ast coup e of years. Thus, the categor ca udgment of compensat on eve s for facu ty as a who e does not appear c ear y n the process of negot at on between an nd v dua facu ty member and h s or her department cha r or dean. There seem to be three ma n types of structures that work to nd v dua ze the abor contract (or to nd v dua ze the apparent abor contract, nsofar as there s a concea ed categor ca abor contract beh nd the nd v dua ones): promot ons and careers as pr mary rewards, ncreas ng the comp ex ty of the category system, and nd v dua y negot ated sa ar es.

Promotions and Careers For promot on systems to underm ne peop e s sense of hav ng the r econom c fate determ ned by the fate of the category n wh ch they fa , promot ons have to be reasonab y frequent. Two forces tend to ncrease

299 promot on rates: a h gh rate of growth of the overa enterpr se, creat ng vacanc es at the top that can be f ed from w th n (St nchcombe 1974, 123–150), and a h gh proport on of tota obs above the eve at wh ch recru tment takes p ace (B auner 1964, 131–132, 148–154). In comp ex status systems w th many peop e severa steps above the recru tment eve , each ret rement creates many promot ons (Wh te 1970). It has been a eged that part of the purpose of creat ng the comp ex strat f cat on system w th the manua work force of the stee ndustry was to create promot ons that cou d underm ne stee worker c ass so dar ty (Stone 1975). More br ef y, categor es of obs can be un form wh e rates of promot on between categor es are nd v dua zed, and th s underm nes the c ass consc ousness–creat ng effect of categor es of dent ca abor contracts.

Complexity of the Category System The comp ex ty of the category system n wh ch workers are c ass f ed has an effect n add t on to that of creat ng poss b e promot ons. The s ze of the group n wh ch abso ute y c ear so dar ty s created s n genera nverse y proport ona to the number of categor es. There may be, for examp e, aboratory techn c ans, nurses, occupat ona therap sts, soc a workers, and bookkeep ng superv sors w th about the same eve of pay and cond t ons of work n a hosp ta , and each ob type may have severa eve s of sen or ty or spec a zat on. But the heterogene ty of the category system may mean that a part cu ar Laboratory Techn c an II does not know anyone e se n prec se y the same category as she or he s. The equ va ent Soc a Worker (Psych atr c) IV, Occupat ona Therap st Superv sor, and Surg ca Nurse may never recogn ze the r mp c t membersh p n a category w th near y dent ca abor contracts to Laboratory Techn c an II.

Individually Negotiated Salaries Somet mes category systems have some bas s of mer t, such as, for professors, the ength of one s v ta or the s ze of outs de offers, that can ust fy separate treatment of category members for some part of the r status prerogat ves. A professors may have tenure, but on y a few of them have sa ar es n the top 10 percent of professors sa ar es. It s most often the sa ary that s detached from the rest of the category system and made part of a more nd v dua zed barga n, so we have named the strategy for ts ma n man festat on. Bonuses, comm ss ons, stock opt ons, easy work oads, and other perks may a so be nd v dua zed by app y ng eg t mate measures of mer t, hence reduc ng c ass consc ousness n the re evant

300 group of workers. In th s case, wh e the abor contract s tru y nd v dua zed, many subparts of t are treated n a categor ca fash on. Two other forces that can reduce c ass consc ousness are to recru t workers from ess consc ous abor market segments, and to subcontract part of the work to f rms that operate n d st nct abor market segments.

Labor Market Segmentation and Class Consciousness When there are arge categor es of ow-sk ed workers w thout much chance of promot on n an enterpr se, t may be poss b e to f those categor es w th abor force segments who are, for other reasons, ower n c ass consc ousness. In recent years th s has meant espec a y women and teenagers (both have often been exc uded from un ons, a fact that tends to reduce the r c ass consc ousness). Even when arge groups of c er ca workers are emp oyed under very standard zed cond t ons, they are re at ve y un ke y to form un ons ( n the Un ted States) f they are fema e. S m ar y, the operat ves n text e m s or c garette factor es are often Southerners or women or both, and have been ess m tant than one wou d expect g ven the r work cond t ons and wages (B auner 1964, 70–73). In many parts of the serv ce sector, ow-sk obs have been f ed by women or teenagers, wh ch he ps to exp a n the ower degree of c ass consc ousness n fast-food restaurants or among grocery c erks. Ana ys s of the s ze of th s effect s very d ff cu t, because part of the reason abor markets are segmented n the f rst p ace s so that women and teenagers can be recru ted to pos t ons n wh ch the genera cond t ons of the ob (precar ous tenure, ow sk , few career prospects) d scourage c ass consc ousness. They are segmented n add t on so that such ow-wage abor can be kept out of obs whose cond t ons are favorab e because of strong trade un on contro . These processes do not un form y work very we , as s ev denced by re at ve y h gh eve s of organ zat on among grocery c erks n some parts of the country and n some cha ns (Wa sh 1988).

Subcontracting and Class Consciousness The fundamenta context of bureaucrat c un form ty outs de the sk ed trades s the f rm. Workers do ng dent ca work n d fferent f rms do not necessar y see themse ves as be ng members of the same category, espec a y f they are n d fferent oca abor markets. Consequent y, one cons derat on for an assemb y f rm n the "buy or make" dec s on (the dec s on on whether the f rm shou d manufacture parts tse f or subcontract to supp ers) s whether there are oca abor markets n wh ch the category of workers requ red to make the part get ower wages than that

301 category does w th n the f rm. For examp e, f mach n sts n rura M ch gan are emp oyed by Genera Motors, they w probab y have to be pa d wages comparab e to F nt mach n sts. If, however, they are emp oyed by a oca y owned mach ne shop w th good re at ons to the oca vocat ona teachers, they can probab y be pa d wages comparab e to auto mechan cs n rura garages. And f (as s usua y the case) auto mechan cs n rura garages earn ess than UAW mach n sts n F nt, a subcontractor can manufacture the part much more cheap y than Genera Motors can, f ts eff c ency s equa . Both groups of mach n sts may separate y be qu te c ass consc ous, but that c ass consc ousness s ess ke y to resu t n equa wages for the parts manufacturer and the assemb y f rm (Sabe 1982, 31–77). In fact, parts supp ers to the automob e ndustry do seem to have ower wages than automob e manufacturers.

Low Unionization of the Modern Service Sector: Theory To dea w th the genera not on that the movement toward a serv ce economy has caused some of the dec ne n c ass consc ousness, we have a prob em of def n ng the dependent var ab e. In E. P. Thompson there s no quest on what work ng-c ass consc ousness s: the workers he quotes are a c ear y workers, and the un ons found n the ndustr es he stud es are c ear y un ons. But s, for nstance, the Amer can Med ca Assoc at on, a "work ng c ass"–consc ous organ zat on, mere y one made up of qu te r ch workers and consequent y a ot more conservat ve n econom c and po t ca ph osophy than, say, the Un ted Auto Workers? The Amer can Bar Assoc at on s very we organ zed, and often they take n t at ves aga nst reforms that wou d g ve ega serv ces more cheap y, such as so-ca ed no-fau t d vorce—that s, they defend the econom c nterests of ega art sans n much the same way stock ngers and others n ear y-n neteenth-century Eng and d d. For these profess ona assoc at ons, the quest on s whether they are un ons or not, and whether the group consc ousness they show shou d be ca ed a narrow form of c ass consc ousness—c ass consc ousness w thout a ances w th other work ng-c ass organ zat ons. An exce ent examp e of the prob em can be seen n the fact that w th n the census occupat ona category of "Profess ona , Techn ca , and Re ated Workers," women are more ke y to be ong to trade un ons than men are. Th s s because women are overrepresented n the teach ng profess on, one of the few profess ona occupat ons that are h gh y un on zed. F fteen or twenty years ago teachers were hard y un on zed at a ,

302 because the Nat ona Educat ona Assoc at on nc uded pr nc pa s and super ntendents (superv sors supposed y cannot be members of the r workers un ons n the Un ted States) and d d not engage n co ect ve barga n ng. The aw d d not at that t me prov de for representat on e ect ons n pub c schoo s. We w take as a rough nd cator of a group s c ass consc ousness whether or not ts organ zat ons app y for the ega r ght to be co ect ve barga n ng agents. Th s cr ter on can be defended by the theory of c ass consc ousness above, and d v des occupat ona organ zat ons n a way that agrees w th my ntu t ons. Severa sectors of the serv ce ndustr es, however, have cons derab e c ass consc ousness even n the Thompson sense. Some of the most powerfu ear y un ons were formed by h ghstatus ra road workers, who were qu te we organ zed by the 1880s. The ra roads were the f rst arge-sca e modern bureaucrat c ndustry to be a most comp ete y un on zed n the Un ted States. They were pretty we organ zed by the F rst Wor d War (L cht 1983). Stee - (after the break ng of the Homestead str ke n the 1890s) and autoworkers were not rea y organ zed unt the m d th rt es. Neverthe ess, the overa p cture s one of ow c ass consc ousness n the serv ce sector.

Service-Sector Class Consciousness: Demography The Demography of Class Contact A bas c genera zat on from po t ca soc o ogy (L pset, Lazarsfe d et a . 1954) s that the more one assoc ates w th workers, the more eft one s op n ons and vot ng w be and the more ke y one s to be drawn nto those act v t es that are connected to work ng-c ass m eux. If a person s a ow y wage worker but s work ng as a servant n an upper-c ass househo d n an upper-c ass sect on of town, he or she w be very un ke y to vote Democrat c n the Un ted States or Labour n Eng and, very un ke y to earn to p ay darts n Eng and or poo we n the Un ted States, and very un ke y to o n a un on or to part c pate n a work ng-c ass demonstrat on. Converse y, a grocer w th a store n a work ng-c ass d str ct whose conversat ons w th c ents are a w th workers, who ves over the store among a -work ng-c ass ne ghbors, and who shares the ethn c ty and re g on of the ne ghborhood and so meets work ng-c ass peop e n h s or her vo untary assoc at ona fe, s very ke y to vote Democrat c or Labour or to come out on demonstrat ons on beha f of work ng-c ass gr evances or even to make contr but ons to the workers str ke funds when the commun ty sp ts apart dur ng a str ke.

303 Many serv ce obs so ate the person from soc a contact w th other workers and requ re ntense contact w th c ents. Broad y speak ng, c ents are a cross-sect on of the c ass system, so contact w th c ents tends to move peop e n upper-c ass profess ons (e.g., the c ergy) n a eftward d rect on n econom c deo ogy (though n a more fundamenta st d rect on n theo ogy), peop e n owerc ass occupat ons n a r ghtward d rect on. The broad effect on the peop e who m ght otherw se o n work ng-c ass organ zat ons n the serv ce sector s to decrease work ng-c ass consc ousness (Hughes 1958). Th s genera zat on depends a ot on the compos t on of the c ente e. For examp e, U.S. senators are very upper-c ass fo ks w th a c ente e that s very near y an exact cross-sect on of the popu at on, on whom they depend and w th whom they soc a ze a ot because they depend on them. It s a so perhaps the upper-c ass group most even y sp t between Democrats and Repub cans, and t has more proun on members than pract ca y any other upper-c ass profess on. Some other spec a zed groups tend to be near ha f eft, ha f r ght as we , such as abor awyers, for obv ous reasons. Those awyers who spec a ze n creat ng new f nanc a nstruments or adv s ng corporat ons on corporate reorgan zat ons tend to be qu te r ght-w ng, wh e cr m na and d vorce awyers or po t ca awyers (e.g., d str ct attorneys) tend to be much farther eft. In med c ne, the phys c ans who ta k to the r c ents (psych atr sts, ped atr c ans, genera pract t oners) tend to be farther eft than those whose pat ents are most y unconsc ous (surgeons, anesthes o og sts) or not present (patho og sts). Genera y speak ng, the teamsters, ra road eng neers, p ots, and so on n the serv ce ndustr es have tt e contact w th c ents and are qu te c ass consc ous, wh e ma e sa espeop e are very ke y to ta k to the r c ent and very un ke y to be n un ons. The more the demography of the workp ace nc udes c ents, and the more ntense the nteract on of serv ce workers w th c ents s, the more those who m ght otherw se be d sposed to o n un ons or to vote eft w nstead be nonun on and vote to the r ght.

Gender Demography in Service Industries Women are d sproport onate y recru ted nto the serv ce sector, espec a y nto ts ower-status pos t ons. Even when women are factory workers, they are harder to organ ze nto un ons than men ( n 1980, 41.4 percent of ma e b ue-co ar workers were un on members, 29.1 percent of fema es; n that same year, 24.7 percent of ma e serv ce workers were un on members, 10.8 percent of women [Stat st ca Abstract of the Un ted

304 States, 1984 , 441]). In many ways, the ma n k nd of "peasant workers," n Sabe s terms (1982, 101–109), are women. He def nes peasant workers as those whose ma n fe s e sewhere than at the r work, who are w ng to work hard for poor pay n nsecure obs because they are mak ng more than they cou d make where the r fe s. If more women than men are, ke peasant workers, temporary obho ders enter ng the abor market from unorgan zed sectors (e.g., the househo d or the farm sectors), th s shou d depress the r c ass consc ousness. In genera , women rece ve more ntense and more mora st c soc a zat on than men; and, n genera , modern mora ty s conservat ve. In add t on, women tend to bear the respons b ty n the home for matters of taste, that s, for so manag ng the home that the soc a status d sp ayed s as h gh as the fam y can afford (Bourd eu 1984). If anyone s ts on the front porch n an undersh rt and sp ts on the steps, t s probab y the man; f n rura areas anyone s go ng to eave o d mach nes out n the yard and g ve the mpress on of a unkyard off ce, t w probab y be the man. So the genera zat on that we w come to be ow, that peop e more respons b e for matters of taste tend to be more conservat ve, app es more often to women workers than to men. Except when the r ght-w ng cand date s be gerent and enthus ast c about state v o ence (e.g., consp cuous y n favor of cap ta pun shment), women n most countr es vote more to the r ght. In most countr es, women o n un ons ess than men as we , and are ess c ass consc ous n op n on surveys. Women tend to work w th other women, a fact that can re nforce them n the r conservat sm, s nce the r co-workers are, stat st ca y speak ng, ke y to be conservat ve too. Of course, when the co-workers are eft-w ng, organ zed, c ass-consc ous women, the reverse ho ds.

Service-Sector Class Consciousness: Selling Status Symbols Qu te often the work of the serv ce sector s produc ng or se ng symbo s, and qu te often those symbo s have a conservat ve t nge. Peop e tend to come to be eve and va ue anyth ng they say, so f the r work requ res them to produce symbo s w th a conservat ve t nge, they tend to become conservat ve.

High Status Fronts to Sell Status Symbols One k nd of symbo zat on that peop e take up n many serv ce ro es conveys a sense of the k nd of person they themse ves are. Some of those symbo s represent hum at on (more on th s be ow), such as wa tress

305 un forms. But qu te a ot of the symbo s fa se y nd cate that one s onese f upper c ass. For women, the symbo s of be ng of a h gher c ass than the ob wou d nd cate are ( n 1987) sk rts and dresses, med um makeup, and dyed ha r. If a work sett ng tends to nduce women to adopt such symbo s of off c a pub c woman ness, t w tend to conservat ze them as we . For examp e, one wou d expect beaut c ans to be more conservat ve than other craftswomen. W th men, the comparab e groom ng symbo s today are frequent shav ng (or frequent psuedoshav ng, produc ng the sharp ne around the beard to show that the man has not ust et t grow), neckt es w th the assoc ated dress sh rts, po sh on the shoes, and, at the extreme, "bus ness" su ts. The more a ro e requ res peop e to dress ke bus nessmen, regard ess of the content and status of the ro e, the more ke y t s that the workers w not have c ass consc ousness.

Conservatizing Deference Some serv ce-sector ro es requ re symbo z ng deference, espec a y toward c ents. As Robert McKenz e and A an S ver showed so effect ve y for Eng and n Ange s n Marb e (1968), a deferent a att tude s c ose y assoc ated w th vot ng Tory. Ro es that have a ot of deference bu t nto them, such as barbers or sa esmen (or occas ona y sa eswomen) of expens ve th ngs such as haute couture and BMWs, shou d have ess work ng-c ass consc ousness than d me store c erks or checkout personne n a supermarket; wa ters shou d be ess c ass consc ous than servers n cafeter as; peop e who work n rac ng stab es shou d be ess d sposed to o n un ons than other agr cu tura pro etar ans; and so on. A ot of the "emot on work" that Ar e Hochsch d ta ks about n The Managed Heart (1983) comes down to show ng deference to the c ent. Such emot on work n a ro e shou d n genera decrease work ng c ass consc ousness.

Taste As a Conservatizing Force P erre Bourd eu (1984) g ves a pretty comp ete out ne for France of what sorts of th ngs are apprec ated by the re at ve y unschoo ed r ch, and what sorts are apprec ated by the rather we off but h gh y educated upper-m dd e c asses. By and arge, th ngs more apprec ated by the upper c asses of e ther k nd are the same sorts of th ngs that women apprec ate more than men. For examp e, we -to-do h gh y educated peop e are ke y to apprec ate c ass ca mus c as we as modern popu ar mus c, to take the r mus c s tt ng down n a concert rather than danc ng at a n ght c ub, to prefer The Art of Fugue or The We -tempered C av er to "The B ue Danube" or the Hungar an Rhapsod es, to ke Da or Rousseau

306 more than Raphae , to th nk of themse ves as "art st c" rather than "consc ent ous," to th nk that even eng neers ought to have a b t of human t es tra n ng n co ege, and so on. The same contrasts wou d be true of women as compared to men. Qu te a ot of the h gh-marg n trade n the serv ce sector comes from produc ng and se ng goods that appea to nouveaux r ches, who have not yet had the r tastes perfected by severa generat ons of fam y w th good human t es c asses, or se ng those goods that show more "o d money" tastes. So we wou d expect peop e n the serv ce sector se ng such th ngs as go f or tenn s essons, nter or decorat on, haute couture, s mm ng sa ons, and so on, to be re at ve y conservat ve. I suppose that n genera French teachers n the Un ted States wou d be more conservat ve than Span sh teachers for the same reason, even though there s probab y more reputab e eft-w ng symbo sm n French cu ture than n Span sh; n the Un ted States one by and arge earns French n order to be fash onab e, Span sh to be a soc a worker. Converse y, the fo ks who make and se beer (at east n countr es where the work ng-c ass dr nk s beer) are ke y to be eft sts, and owners of pubs or beer ha s are ke y to be more eftw ng than owners of w ne shops or cockta bars. Beer advert s ng symbo zes ma e work ng-c ass v rtues, and beer ha s appea to ma e work ng-c ass c ente es rather than to m dd e-c ass coup es (LaBonte 1987). The argument wou d then be that the parts of the serv ce sector d sp ay ng re at ve y tt e c ass consc ousness are those that se h gh-marg n goods to the r ch, perhaps espec a y those that se ma n y to the r ch w th sma er educat ona cap ta . But s nce those h gh marg ns n part ref ect more serv ce-worker abor, the h gh-marg n trade shou d have a arger conservat z ng effect than the th n-marg n trade of goods a med ma n y at the work ng c ass and poor.

Service-Sector Class Consciousness: The Small Firm Effect The th rd b g force produc ng ess c ass consc ousness n the serv ce sector s that superv s on of serv ce-sector workers s d fferent from that of workers n the manufactur ng sector.

Short Hierarchies Most of the serv ce sector, except transportat on and ut t es, s run on a sma -f rm bas s, that s, the prof t taker h mse f or herse f, or the h gh execut ve who represents the prof t taker, s on y a few steps away from

307 the worker n the h erarchy, and often works a ongs de the worker. Th s means severa th ngs. F rst, the super or s much more ke y to superv se by po nt ng out the goa s he or she s try ng to ach eve, rather than by "Braverman z ng" or rout n z ng the ob. Second, someone w th a comm tment to prof t mak ng s nearby n terms of soc a d stance: a worker can ta k to the head of a restaurant or the owner of the McDona d s franch se n a way that s mposs b e w th the ch ef execut ve off cer of Genera Motors. Th rd, the mob ty chances of a worker are not def ned by sen or ty, but are of a cap ta st character; thus, the same ob ect ve mob ty rates operat ng n the serv ce sector as n manufactur ng wou d probab y have a conservat z ng effect n sma f rms, because there the mob ty s more often nto the cap ta st ro e of owner rather than nto the sa ar ed ro e of foreman. Amb t ous serv ce-sector workers are probab y more conservat zed by the r amb t on than members of the manufactur ng or other arge-f rm sectors.

Fewer Categorical Labor Contracts Sma er f rms are genera y assoc ated w th not estab sh ng categor es of workers and attach ng the same pay to a members of the same category and the same sen or ty. Pay s a barga n w th nd v dua s, not w th a category of peop e, a fact that reduces c ass consc ousness, as argued above (Lockwood 1958).

Piecework Compensation Much of the superv s on n the serv ce sector s on a sort of "p ecework" system, where c ents dec s ons make up the p eces that the workers are pa d by. Comm ss ons are common n sa es work, t ps common for wa ters, ha rdressers, and tax dr vers, and n add t on gett ng part of the tota take rather than a wage s common for tax dr vers. What s be ng e c ted n these schemes s not "product v ty" n some obv ous sense, but the capac ty to f nd and nf uence c ents to take the serv ce, buy the good, or g ve the t p. Such ncent ve systems are genera y nd v dua z ng, mak ng the workers th nk the r wage depends on the r own efforts rather than on a genera wage rate (Burawoy 1979; Browne 1973). But the k nds of p ecework found n the serv ce ndustr es often make the worker dependent on f gur ng out what the c ent wou d ke; th s ntens f es the effects of ro e requ rements that press workers to assoc ate w th and be fr end y to upper-c ass c ents or to show deference. The more a ro e requ res the worker to f gure out the w shes of the c ent, the more the conservat ve character of the average c ent (as compared w th the average worker) penetrates the worker s consc ousness.

308

Unionized Environments and Class Consciousness In some parts of the serv ce ndustry, however, the overa re at on between a corporat on and the emp oyees s so shot through w th un on e ements that even peop e who wou d otherw se not be very c ass consc ous end up n un ons and somet mes ta k ke c ass-consc ous workers. A ove y case s a r ne stewardesses: they are fema e, wear the med um amount of makeup that ref ects the r h gher c ass of c ente e, the r ob s to be deferent a and to f gure out how to sat sfy upper-c ass and upper-m dd e-c ass c ents, and unt recent y they were more or ess exp c t y requ red to be "peasant workers," to eave the bus ness when they got too o d. One wou d thus not expect them to very c ass consc ous. But everybody e se n the ndustry, even the execut ves n charge of the sma work group they are n (the p ots), are a un on zed. Compared to anyone e se n wa tress ro es w th an upper-c ass c ente e, then, f ght attendants are pretty c ass consc ous.

Conclusion The organ zat ona facts that create soc a c asses, and so create the bas s of c ass consc ousness, cannot be understood w thout understand ng the po t ca cu ture n wh ch c ass re at ons are understood by the workers. We have tr ed to bu d a theory of what c ass re at ons ook ke n modern soc et es by comb n ng organ zat ona soc o ogy w th po t ca h story, tak ng nsp rat on from E. P. Thompson. We have perhaps made E. P. Thompson nto more of a theor st than he wou d be comfortab e w th (he s supposed to be eve that "theory s ust az ness n the arch ves"), but the resu t s someth ng that can be used to understand abor re at ons of modern corporat ons, who create c asses by creat ng ob categor es, and to understand why many modern serv ce ndustr es do not create much work ng-c ass consc ousness. The fundamenta argument of th s chapter s that s nce c ass consc ousness s the pro ect on of organ zat ona statuses onto a arger po t ca and econom c canvas, the cu tura too s workers have to make these pro ect ons go as far to exp a n c ass consc ousness as does the organ zat ona process that creates groups w th dent ca abor contracts. Those too s are part of genera work ng-c ass cu ture, as our ana ys s of Thompson shows; but they are a so shaped by the sorts of structures that ex st w th n the organ zat on, such as gr evance procedures, ob eva uat on schemes, or nd v duat on of the abor contract, to def ne the mean ng of the categor es nvo ved n abor contracts. They are a so shaped by cu -

309 tura def n t ons erected by the work tse f, espec a y n the serv ce ndustr es, where a good part of what a worker se s s a set of symbo s of upper-c ass status. But those cu tura def n t ons cannot redef ne everyth ng. Or better, they can redef ne everyth ng, but not to be anyth ng they p ease. The serv ce sector s ess c ass consc ous n arge measure because t does not n fact put as many workers nto arge groups w th categor ca abor contracts, as we as because w th n those categor es workers have to present themse ves as f they were upper c ass n order to se d amond r ngs. Craftsmen and craftswomen, and the sem profess ona s who make up the craft trade-un on structure of the a r nes ndustry, serve an upperc ass c ente e ust as ewe ers do, but they do so n a structure that treats them as equ va ent members of a category earn ng a standard wage rate. So wh e we w not expect, say, stewardesses or p ots to be as m tant as m neworkers, we w expect them to be a good dea more c ass consc ous than workers n reta ewe ry. The pro ect on of organ zat ona status on the arger soc ety depends n part on the symbo s that accomp sh th s pro ect on; but t depends a so on the type of organ zat ona status system n wh ch t s embedded. The argument of th s chapter, then, has been that peop e get much of the nformat on about the r nterests from the nformat on n the ncent ve system of the organ zat on they work for. In part cu ar, they do not earn that they are part of a abor market, and thus that the r wages w be competed down to the owest eve the r emp oyer can pay, from the obscure operat ons of that market tse f. Instead they earn t from the categor es n the ncent ve system of the r emp oyer. When un form category systems of emp oyer organ zat ons are pro ected onto the arger canvas of the po t ca system, they produce the raw mater a of c ass consc ousness, of trade un on organ zat on, and of eft ega tar an po t cs. But the very fact that the way the ncent ve system of the f rm or of the oca sk ed trade appears to peop e produces the bas c v ew they have of the r econom c and po t ca nterests means that var ous features of that oca scene can affect what they see to be the r nterests. Further, t means that the arger processes of worker enfranch sement and trade un on organ zat on can affect what workers see when they ook at the oca ncent ve system. For most of th s book we have treated the uncerta nt es that organ zat ona systems dea w th as hard facts. C ear y they often must be hard facts, or ob ect ve y measured features of markets wou d not have observab e effects on sk d str but on or on d v s ona zat on, and turn ng nvent ons nto nnovat ons wou d be ust another ob peop e do, nstead of a ma or force for adm n strat ve changes. But

310 equa y c ear y, t does not do to be dogmat c about how far facts are ob ect ve; some facts are more ob ect ve than others. How far an nd v dua worker s nterest s actua y co ect ve s not mmed ate y g ven to workers n the abor market, as t s g ven that the customer w have the bu d ng bu t the way he, she, or t wants t rather than however s conven ent for the construct on contractor. So the degree to wh ch a abor market ooks to the worker n t as though t ought to be un on zed s not as ob ect ve a fact about that abor market as s the degree to wh ch the construct on market ooks ke a set of contracts for un que bu d ngs, des gned to customer spec f cat ons. C ass consc ousness thus ustrates a more genera fact about modern soc ety. Whatever the cu tura apparatus that nterprets facts n some sector of the economy, very genera y the facts themse ves were generated by the nformat on-process ng systems of organ zat ons. The d scovery process n the aw s most y app ed to get access to organ zat ona records to use as ev dence n court, though the government may take a very d fferent v ew from that of IBM about what those facts mean (F shman 1981, 380–411). The facts used by the tax system are generated by account ng and payro systems n organ zat ons, though genera y ne ther the organ zat on nor the emp oyee s favorab y d sposed to the uses made of those facts. The facts generated by the reg strar of the un vers ty, a comp at on of udgments of professors, are centra to some of the cert f cates used n the abor market, though many professors are vague y uncomfortab e that the r udgment of a student s understand ng of Shakespeare s worth money n the abor market. Thus, t s not unusua that eft-w ng part es and trade un ons make use of the facts generated by the categor es of the ncent ve systems of organ zat ons. They make use of them part y because they can do no other. It takes a good dea of argument to get professors to see that the core determ nant of the r sa ary ra se s what percentage everybody s ncrease has to add up to, as dec ded by the un vers ty adm n strat on, wh e t takes very tt e argument for an autoworker to see the same th ng—and a that s mp y because the f rst percentage s nd v dua zed before the professor s note on next year s sa ary comes (and t comes after the schoo years ends, mak ng compar son d ff cu t), wh e the second s the percentage that appears n the autoworker s pay packet, dent ca to the percentage n other workers packets. No doubt the emp oyers n automob e manufactur ng wou d rather g ve the same mpress on of nd v duat on that un vers t es g ve, but they wou d have d ff cu ty runn ng an assemb y ne that way. Regard ess of the source of the organ zat ona d fference n nd v duat on, professors

311 are hard to organ ze for co ect ve barga n ng, autoworkers are easy, and that hard fact for trade un ons s created by the hard facts of the d fferent ncent ve systems. But trade un ons make d fferent use of such facts n the Sov et Un on than n Great Br ta n or the Un ted States, and d fferent uses st n Japan. The same k nds of workers are d sproport onate y act ve n trade un ons n a four soc et es, presumab y ow ng to deep facts about the ncent ve systems appropr ate for d fferent sorts of product on. But these s m ar t es are over a d by d fferences n the arge structure for nterpret ng the r mean ng n the po t ca cu ture as a who e. Such facts about var at ons n nterpretat on as those e aborated n th s chapter are what g ve the m ts to the "techno-Marx sm" of the f rst seven chapters.

312

9— University Administration of Research Space and Teaching Loads: Managers Who Do Not Know What Their Workers Are Doing What Business Universities Are In My purpose n th s f rst sect on s to spec fy why the uncerta nty of sc ent f c advance s centra to the work of un vers t es, by try ng to descr be what bus ness un vers t es are n. Th s w then serve as a background for exp or ng why adm n strat on of un vers ty space and teach ng oads s as t s and, n turn, for say ng why what ooks ke a pretty haphazard system s actua y more or ess rat ona . Most research space s a ocated as f to sovere gn states—the departments and research centers. Teach ng oads tend to be adm n stered by departments, w th re at ve y ntense superv s on by deans but tt e adm n strat on from h gher up. I w argue that the very unpred ctab ty of research, and hence of research space and facu ty t me needs, s the m croscop c exp anat on of why the centra space adm n strat on of a un vers ty s so r g d, and why the adm n strat on of teach ng oads var es so much across departments. The f rst prob em s to exp a n why organ zat ons that make most of the r money from teach ng shou d emphas ze research so much. My argument compares what un vers t es do to what banks do; that s, both banks and un vers t es are f duc ary nst tut ons. F duc ary nst tut ons are essent a to create an exchange system that trades and poo s d fferent k nds of uncerta nt es. What a modern ead ng un vers ty n the Un ted States or Eng and buys s d fferent from what t se s. It buys professors ach evement n a very uncerta n wor d of advances n sc ence and scho arsh p, and t se s prest ge educat on. For the most part, the prest ge educat on s n the we -estab shed consensus part of the var ous d sc p nes; thus peop e n, say, chem stry, w th many d fferent approaches to what s an nterest ng research quest on, w be w ng to use the same e ementary chem stry book. Further, they w teach the standard use of the standard aboratory equ pment n the teach ng aborator es, a though the equ pment n the r research aborator es s purpose-bu t and un que.

313 So a un vers ty buys n the h gh-r sk market for sc ent f c advance n chem stry, and se s most y n the market for cert f ab y exce ent and standard educat on n the profess on of chem stry. (In the ast four decades, the NSF and NIH have bought research serv ces from the phys ca and b o og ca sc ences departments of ead ng un vers t es on a sca e never seen before 1945—so th s argument app es n pure form on y up to 1945.) The professor s h red to do research, wh ch s a d fferent sort of work from what the student (or the goverment) pays tu t on for: teach ng. The un vers ty buys research on the abor market on the genera cr ter on that someone who knows chem stry we enough to advance t w , on average, be ab e to prov de re ab y up-to-date teach ng of chem stry s consensus. The h gh-r sk market for sc ent f c advances, then, prom ses ess r sk that students w get out-of-date or wrong teach ng. The po nt s that students wou d be unab e to h re n the h gh-r sk market for sc ent f c advances themse ves because assess ng the r sks of be ng wrong about sc ent f c advances requ res eva uat on by peers who understand the parad gms and the r uncerta nt es and the deta s of aboratory procedure. A student s not n a pos t on to udge whether a person has rea y advanced know edge or s ust some sort of crank. The corre at on between students eva uat on of nd v dua professors and the professors research product v ty s re at ve y ow (I wou d guess t at around .2, based on the d fference n teach ng eva uat ons n the un vers t es versus the state co eges n Ca forn a many years ago). But the tu t on that a un vers ty can get away w th charg ng for a prest ge educat on (whether t s charged to the state or to the nd v dua student) s very h gh y corre ated w th the average sc ent f c and scho ar y product v ty of the facu ty. In some sense the students (or the r parents, or the r state eg s ature), then, pay the un vers ty to h re n a market for sc ent f c and scho ar y competence that s so r sky they cannot make good dec s ons n t for themse ves. They pay the un vers ty to prov de them w th a poo ed r sk funct on that s a re ab e cert f cat on of competence and up-to-dateness. A though parents cou d not choose the facu ty at MIT themse ves, they are conf dent that a cert f cate that one s a competent eng neer (or sc ent st) g ven by MIT as an nst tut on w be va uab e n the abor market, because t w ref ect rea earn ng.

The Analogy of Universities to Banks In the same genera way, banks buy future ncome streams n bus nesses or home mortgages that have much arger f uctuat ons and r sks than most savers are w ng to bear. By poo ng those r sks and ncome streams, by keep ng reserves, by ma nta n ng FDIC nsurance, and the ke, the

314 banks can se savers a very d fferent tempora pattern of ncome w th a very d fferent r sk funct on than s found n the future ncome streams they buy from the bus nesses they are end ng to. One s check ng account or sav ngs account s qu te qu d, ava ab e on short not ce at a h gh degree of certa nty; the bank, however, s earn ng the nterest on t w th both h gher r sk and arger nvestments that are qu te qu d. The bank wou d ose a ot of money by ns st ng that a factory bu t w th the r oan had to be so d off to sat sfy savers demands. In both banks and un vers t es, we have to trust the nst tut ons to run for us the r sks we are not competent to run (or do not want to run) ourse ves, so that we can buy a re ab y qu d degree n eng neer ng that can be cashed n for a ob at Du Pont (even though MIT has bought n a very uncerta n market of recent advances n quantum mechan cs that we cannot understand), or so that we can re ab y co ect our sav ngs account w th nterest when we want to make a down payment on a house (even though the nterest was earned by an nvestment that comes to matur ty at a d fferent t me and w th more r sk than we were w ng to bear, maybe f nanc ng the R&D at Du Pont). And t s because we have to trust them to act for us n markets that we can t ourse ves manage that I ca them both "f duc ary" nst tut ons.

Institutional Risk Management in the Anglo-Saxon versus the Latin World Many Lat n Amer can un vers t es, nstead of buy ng the fu t me of a ead ng sc ent st as shown by recent d scover es and mak ng h m or her teach rough y ha f t me, h re someone part t me who earns a rea v ng by part-t me work e sewhere and who can be cert f ed to know the consensua content of the course to be taught. That s, the abor that the un vers ty buys s a most exact y the same as the abor that the student buys: competent teach ng of a course. The prob em s the same as w th a bank that ends out on y to peop e or enterpr ses w th very tt e r sk and h gh y qu d resources: the c ent can get no ncreased returns ow ng to the bank s or un vers ty s nte gent part c pat on n h gh y r sky nvestments n cap ta equ pment or new know edge w th a h gh average rate of return. H gh y qu d and safe nvestments n sav ngs accounts can have h gher rates of return because banks poo r sks for the r savers. L kew se, h gh y competent and standard teach ng can be obta ned w thout much r sk by ett ng the un vers ty nvest n precar ous research ventures on the bas s of ts spec a zed know edge of where the var ous f e ds of research are probab y go ng.

315 Because Lat n un vers t es h re as f they were students, over t me the teach ng n eng neer ng schoo s and account ng schoo s does not change as rap d y w th advances n know edge n the average Lat n Amer can or Span sh un vers ty as t does n the average North Amer can or Br t sh un vers ty. Th s means n turn that nvestments n human cap ta n Lat n Amer ca and Spa n g ve ower rates of return than n North Amer ca and Eng and. More or ess the same th ng s true n nternat ona compar sons of nat ona cap ta markets. The k nd of certa nty that savers need s not produced by ega certa nty. Instead t s produced by hav ng a comp ex set of nsurance ke nst tut ons. L fe nsurance, for examp e, g ves savers r sks of be ng ab e to co ect when they d e that are d fferent from what wou d be obta ned f they had nstead bought shares n the Prudent a Bu d ng; the chances banks g ve of peop e be ng ab e to w thdraw the r sav ngs to make a down payment on a house are kew se d fferent from what those peop e wou d have had f they had owned a p ece of a mortgage on Houston rea estate. W th f duc ary nst tut ons, nvestments can be made even f the nvestments do not prov de the r sks and the overt me pattern of sav ngs and w thdrawa s that savers demand. Lega certa nty w not do that, because even f a saver s p ece of a Houston mortgage s ega y certa n, he or she cannot co ect on t un ess a buyer can be found n Houston, and nowadays (1989) a buyer cannot be found n Houston. My untutored mpress on s that banks n Lat n Amer ca are not as good at poo ng h gher r sks so as to g ve savers the r requ red qu d ty wh e nvest ng n ongterm econom c deve opment. Lat n banks tend to want to end on mortgages n rea estate rather than on corporate bonds.

What Is the Uncertainty of Scientific Research? Betting Where New Knowledge Is to Be Found Betting The f rst po nt one has to get stra ght n order to address the soc o ogy of sc ence s that the strategy of act ve y pursu ng new know edge s a d fferent process from that of exam n ng purported new know edge the extent to wh ch t s know edge. That s, do ng sc ence s not ust fo ow ng some ru es of ep stemo ogy. What a sc ent st has to do s take the present state of know edge, as he or she be eves t to be, and make nte gent

316 guesses about what e se (or what d fferent) s ke y to be the case, and further guesses about the method by wh ch the ke hood that the guess s r ght can be ncreased or decreased. The second ha f of th s guess ng or bett ng process s the strateg c part of ep stemo ogy. Ep stemo ogy s the study of var ous methods of ncreas ng or decreas ng the probab ty that var ous assert ons about the wor d are true. There are ots of deep ph osoph ca prob ems about how th s can be done, what makes them work, and the ke, wh ch const tute the ma n body of what s usua y ca ed the ph osophy of sc ence. But a sc ent st s posed w th the prob em of how, as a pract ca matter, best to affect the sub ect ve probab t es w th wh ch the body of sc ent sts w ho d a be ef that he or she proposes to nvest gate. It s, n other words, a matter of emp r ca or theoret ca strategy , a game n wh ch the ru es are n some sense ep stemo og ca standards, but the conduct of nvest gat on s a matter of choos ng the east-cost, h ghest-return method of beat ng that game. So ep stemo ogy bears rough y the same re at on to the prob em of "estab sh ng" new know edge as the ru es of footba , soccer, or horse rac ng bear to w nn ng n those games. Obv ous y, a person can be a very good udge of the ru es for a horse race w thout be ng a very fast horse or even a champ onsh p ockey. L kew se, a person can be a good ep stemo og st w thout be ng much good at chang ng the probab t es of var ous be efs that sc ent sts ho d. But converse y, an ent ty can be a very fast horse w thout be ng much of a udge of horse races, and be ng a successfu sc ent st does not much pred ct nte gence n the bus ness of ep stemo ogy or ph osophy of sc ence. Beh nd the dec s on to seek new know edge n a g ven p ace es more than a strategy for affect ng the probab t es w th wh ch sc ent sts ho d a be ef. Even more mportant s dec d ng what set of be efs t wou d be usefu to try to affect, wh ch nvo ves a sorts of udgments of probab t es that there s someth ng wrong w th the who e apparatus—theoret ca , observat ona , parad gmat c, etc.—on wh ch our present udgments of the probab t es of var ous be efs n the sc ent f c system rest. There are bets n theoret ca work as we . For examp e, peop e who ought to know say that when E nste n s spec a theory of re at v ty was deve oped and pub shed, there were resu ts by a famous exper menta st wh ch ooked uncha engeab e that showed that E nste n was wrong. On y many years ater d d someone determ ne that there had been a eak n the exper menta apparatus. E nste n was engaged n ook ng for a way to ntegrate the theor es of Lorenz and Maxwe and s mp y gnored the contrad ctory resu ts, hop ng, I presume, that there wou d turn out to be

317 a "spec a " exp anat on of them. There d d. The po nt here s that E nste n s bet was that a un f ed treatment of two theor es, both of wh ch had been emp r ca y product ve, was ke y to be truer than e ther a one, and ke y to generate pred ct ons that wou d te someth ng about the wor d that we had not been n a s tuat on to th nk of before. Of course, there m ght not have been a eak n the apparatus, and E nste n m ght have been qu te wrong. In such a s tuat on, where one set of nd cat ons, such as the exper ments, po nts one way, another set, such as more or ess conv nc ng theoret ca arguments, the other way, a sc ent st must p ace bets as seems w se. He or she has to be w ng to be wrong. A wonderfu examp e s the fact that two d fferent phys c sts have got the Nobe Pr ze for the d scovery of transuran um e ements (Zuckerman 1977, 212-213). In the Nobe address for the f rst d scovery ( n 1938; the other award was n 1951 for research carr ed out n 1940), Ferm (who pract ca y a phys c sts wou d agree deserved the Nobe for someth ng ) po nted out that there seemed to be someth ng wrong w th the standard nterpretat on of the resu ts of the exper ments that "estab shed" the ex stence of transuran um e ements, because Hahn and Strassmann had found bar um among the products of those exper ments. Ferm s exper ments nvo ved bombard ng uran um w th neutrons, and the actua resu t was f ss on of uran um, not fus on of the neutrons w th uran um atoms. But regard ess of the sc ent f c facts, wh ch of course I am not competent to comment on, Ferm had c ear y been mak ng bets (and gett ng them accepted n the sc ent f c commun ty, up to and nc ud ng the Nobe comm ttee) on the nterpretat on of h s exper menta resu ts, and he had been wrong—and he knew t (another measure of h s be ng a great phys c st) at the t me he accepted the pr ze for t. What had made the facts of the exper ments nterest ng was the r theoret ca nterpretat on, and that, as t turned out, was wrong. Another nterpretat on, however, wh ch perhaps unfortunate y had a ot of pract ca app cat ons, was a so nterest ng, and t ed to the famous exper ments show ng a nuc ear cha n react on n a basement at the Un vers ty of Ch cago dur ng Wor d War II—a so conducted by Ferm . In some ways, what d st ngu shes great sc ent sts s that they make a ot of bets n uncerta n c rcumstances (the r "or g na ty") and a ot of those bets turn out we (the r "competence"). But f even a great sc ent st makes a ot of bets n uncerta n c rcumstances, he or she w ose a ot. S m ar y, even f a bettor s so good at udg ng horsef esh (or udg ng ockeys) that he or she has a w nn ng record at the racetrack, f the bettor bets a ot he or she w ose a ot as we as w nn ng a ot. Ferm was not so hum ated at hav ng been wrong that he avo ded mak ng the pr ze

318 speech, or defens ve enough to try to concea that he was gett ng the pr ze for hav ng "estab shed" someth ng that was apparent y not true.

New Knowledge It s, however, not on y the change n the probab ty of a g ven b t of know edge, mmed ate y mp cated n an exper ment or other observat on or mmed ate y nvo ved as a emma of a new theoret ca der vat on, that determ nes whether sc ent sts w regard what has been estab shed as "new know edge." The overa eva uat on of sc ent sts about how new, and how much, know edge there s n a g ven b t of sc ent f c work s how much that b t of sc ent f c work changes the probab t es they themse ves use n bett ng on where to f nd new know edge. For nstance, a new f nd ng n parapsycho ogy, even f be evab e (and th s s tse f a prob em, regard ess of how far the parapsycho ogy work fo ows the ru es of eg t mate ep stemo ogy), w hard y affect the bets of pract c ng psycho og sts about where new know edge s to be found, because the mechan sms, theor es, exper menta methods, and the ke that are "usefu " n parapsycho ogy have noth ng much to do w th the mechan sms, theor es, and exper menta methods that are usefu n psycho ogy as a who e. In contrast, the fact that one cou d bu d T nkertoy- ke mode s of DNA to exp a n features of the X-ray d ffract on pattern, wh ch are compat b e w th prev ous know edge about the d stances between atoms and other ons and wh ch mmed ate y suggest how gene reproduct on n DNA and prote n product on n the ce m ght proceed, mmed ate y and dramat ca y changed the bets v rtua y a mo ecu ar b o og sts made about where new know edge was to be found (see Watson [1968] 1981). The b o og ca mechan sms were those wh ch genet c b o og sts and ce phys o og sts had been ook ng for, the atom c bond ng theor es used n bu d ng the mode s were so d theor es about the structures of mo ecu es, and the X-ray d ffract on method was and s a re ab e dev ce for gett ng certa n parameters of the structure of matter, and had won the Nobe Pr ze for peop e before. As a sub ect ve est mate, about a th rd of the papers n the Sc ent f c Amer can nowadays are about research n mo ecu ar b o ogy or ce phys o ogy that depends heav y on the Watson-Cr ck Doub e He x research; these papers represent, that s, research that has been fo ow ng out bets about where new know edge s to be found wh ch wou d be very d fferent f we d d not know what we do about the structure of DNA. Part of what determ nes how much a p ece of sc ent f c work affects the structure of probab t es n other peop e s m nds, on the bas s of wh ch they p ace bets represented by the r own sc ent f c work, s of course ts

319 ep stemo og ca va ue. A so d exper ment that cou d hard y have come out the way t d d un ess the theory t supports were true w affect the way peop e bet about re ated matters much more than wou d exper menta resu ts that cou d be due to a most anyth ng, among other th ngs the theory n quest on (see St nchcombe 1968, 17–28). But a ot w depend on how the resu ts re ate to the ongo ng programs of work of the sc ent sts who read them. For examp e, a resu t show ng that soc o ogy s samp ng des gns probab y cause us gross y to m sest mate the regress on coeff c ents that re ate sons occupat ons to fathers occupat ons wou d be much more va uab e know edge than a s m ar demonstrat on that we are gross y m sest mat ng coeff c ents re at ng a enat on from soc ety to a most anyth ng e se. Many more soc o og sts work on the regress on coeff c ents re ated to soc a mob ty than on those re ated to a enat on. Th s s because the soc a mob ty coeff c ents p ay a centra ro e n our theor es of soc a zat on for soc a ro es, n our theor es of rebe on n the po t ca system or c ass consc ousness, n our not ons of what s d st nct ve about ndustr a soc ety as compared w th peasant or pr m t ve soc et es. Regress on coeff c ents re at ng to a enat on are of nterest on y to a few soc a psycho og sts (e.g., n soc o ogy, Me v n Seeman) and to a few po t ca sc ent sts who work w th the M ch gan e ect on stud es (e.g., Ph p Converse). So when a soc o og st demonstrates someth ng about the b as of our samp ng methods w th respect to a enat on (St nchcombe 1979a; St nchcombe, Jones, and Sheats ey 1981), the work of most soc o og sts w rema n unaffected. The same s true of another b g b as, that househo ds of peop e who ve a one are substant a y undersamp ed, whereas peop e who ve n b g fam es are very substant a y undersamp ed, n most noncensus stud es—but when was the ast t me an average soc o og st was fasc nated by an art c e n wh ch the centra po nt had to do w th v ng a one or w th v ng n a arge househo d? The structure of theory s espec a y mportant for how w de the mp cat ons of a b t of sc ent f c work w be. The abso ute centra ty of the DNA work s perhaps the best examp e of th s po nt. Genet cs depends on the study of how genes rep cate. The centra way genet c nformat on s transferred to ce phys o ogy s by prote n synthes s, so the re at on between DNA and prote n synthes s, c ar f ed by know ng how DNA s bu t, s centra to ce phys o ogy. Know ng how prote ns themse ves are bu t a ows one to use var ous b o og ca methods to nvest gate the structure of part cu ar prote ns, and suggests as we how one m ght synthes ze prote ns re ated to a part cu ar prote n (see Latour 1980, an ana ys s of work that depends on app cat ons of these f nd ngs about DNA struc-

320 ture). S m ar y, quantum mechan cs was centra to phys cs n the 1920s and 1930s because the re at onsh ps between energy and matter were theoret ca y nvo ved n most branches of phys cs, not to ment on n many branches of chem stry, and quantum mechan cs suggested new d rect ons of research on a those matters. For our purposes, what th s means s that the amount of uncerta nty about what one s go ng to do next year n sc ent f c work s very great. On y f many sc ent sts wou d bet d fferent y than a g ven sc ent st w t be worthwh e for that sc ent st to do work on a top c. On y f he or she has a d fferent est mate from the rs about how the work w come out shou d he or she do the work. Consequent y, t s the areas of max mum uncerta nty that are most attract ve to sc ent sts, espec a y when they th nk they can bet more accurate y than other peop e about how the exper ment w come out. And as a resu t, un vers ty research cons sts of a sh ft ng set of "pro ects," wh ch are qu te d fferent th s year from ast, and w be qu te d fferent yet aga n next year. Th s a so means that the nformat on about what wou d be strateg c for a g ven sc ent st to do next year—what w get through a set of NSF referees, for examp e—w be concentrated among peop e n the sma commun ty of sc ent sts who themse ves know the probab t es attached to var ous be efs re ated to the proposed research, n other words, who know what wou d be va uab e know edge f t were found. It s such "peers," usua y a group of fewer than one hundred act ve researchers n a g ven f e d, who know whether a part cu ar pro ect s rea y d rected at a prob em whose so ut on w matter. Further, on y they w know whether the procedure proposed n the research s suff c ent y so d that other sc ent sts w be eve the resu ts. And on y they w know whether th s s the sort of work that a part cu ar sc ent st s ke y to be competent to do proper y, or whether t nvo ves know edge of a d fferent f e d as we as the f e d n wh ch the proposer has been tra ned. It s th s nformat ona mp cat on of the fact that t pays to do sc ent f c research on y on uncerta n sub ects that s centra to the prob em of the adm n strat on of sc ent f c work n genera , and to adm n strat on of space for research n part cu ar.

The General Character of University Research Space Problems Un vers ty research us ng aborator es, then, s usua y organ zed nto pro ects. Pro ects have a f n te fe and then are not repeated, though the fo ow ng pro ect may have s m ar enough character st cs so as to have

321 bas ca y the same space requ rements. Most research has to be done on y once (or at most a f n te number of t mes). Th s pro ect character of research means that the act v t es that occupy a g ven p ece of un vers ty space are a ways chang ng. The va ues of square feet requ red for, say, chem ca research that are entered n adm n strat ve documents are, n a sense, actuar a va ues: expected va ues of stat st ca processes that deve op over t me, w th a good dea of var at on over t me due to chance. That s, they represent the number of facu ty members and graduate students one expects to have n the chem stry department, the m x of sorts of research one can expect them to be nterested n, the uck those peop e can expect to have n gett ng the r research funded, and the amounts and k nds of space that a the resu t ng act v t es n pro ects can be expected to f t nto. The prob em of space p ann ng (as represented, for examp e, n the f gures of per person research space for d fferent d sc p nes n Bere ther and Sch nger 1968, 58–61) s therefore to est mate the expected va ue of the space requ rements of a future f ow of pro ects—the amount of space the un vers ty shou d be expected to prov de as nsurer that whatever the cont ngenc es, a researchers w at the appropr ate t me get what space they need to house the r research. The assumpt ons bu t nto that st of per person square foot requ rements n Bere ther and Sch nger s Un vers ty Space P ann ng (1968) are that the department w cont nue to have the research d st nct on that on average t has had, and so w be ab e to get grants of about the same quant ty as before; that t w have about the usua proport on of eaves, the usua rat o of graduate student pro ects to facu ty pro ects, and the usua exper ence of other personne cont ngenc es; that the m x of research-space- ntens veness w not change drast ca y, so that there w not a of a sudden be arge space users n departments that have not used much space n the past; and so on. Further, the way Bere ther and Sch nger s book s wr tten, t mp es that the research space requ rements est mated for the Un vers ty of I no s at Champa gn-Urbana do not ref ect happenstance d fferences from other un vers t es n the re at ve d st nct on of d fferent sc ence departments, or d fferences n emphas s on space-us ng versus space-sav ng sc ent f c top cs or modes of nvest gat on n part cu ar departments at Champa gn-Urbana. That s, attempts to g ve per person research space est mates n d fferent d sc p nes make an nsurance-type actuar a assumpt on that the un verse samp ed n the exper ence of space requ rements n the past w cont nue to obta n n the future. S nce one s pred ct ng the requ rements of a future m x of one-of-a-k nd pro ects, th s assumpt on s, of course, fa r y shaky.

322

Space Is Not Easily Transferred But the b g d fference between the adm n strat ve enterpr se of un vers ty space a ocat on and that of an nsurance company s that money to pay off nsurance c a ms s fung b e, eas y transferred from the one who pays prem ums to the one who has the acc dent. Space s not eas y moved to where t s needed, nor are un vers ty peop e eas y d v ded up so that one-th rd of the r research s n one bu d ng, one-s xth n a second, and the other one-ha f n st a th rd. The an ma s have to be near the mazes they are to run, and the cadavers and the r refr gerat on near the d ssect ng tab es. Furthermore, each researcher s n some sense h s or her own nsurer, because he or she pays a ot of the cost of not hav ng enough space to do the research. The un vers ty does not fa to get tenure because there was not enough space to do the research; the ass stant professor does. That means that each person wou d ke to have a reserve of space that can be used f needed; the department n turn wants to nsure ts facu ty members; the schoo wants to have a reserve of space so that departments n need can get t; and so on. Peop e do not fee they can trust peop e h gher up to take care of them f they do not take care of themse ves. So space temporar y empty but ( f a goes we ) to be f ed when the grant s approved tends to be kept n a qu d reserve near the bottom of the h erarchy. We have deve oped nst tut ons n the nsurance ndustry that most peop e can trust, though except n mar ne nsurance these nst tut ons are on y a century or two o d. Peop e be eve that when they have an acc dent the nsurance company w pay them (out of other peop e s prem ums), and n the meant me they w pay prem ums (to pay for those who have acc dents). The nsurance company makes money by co ect ng nterest on the reserves t keeps so that t can be trusted. Such nst tut ons of trust ex st on y at the owest eve s n a un vers ty (and they are precar ous even there). The resu t s that space resources are not as qu d as nsurance nst tut ons make money to be, because we trust nsurance compan es w th our f nanc a futures n ways we wou d not trust deans and provosts w th our research futures.

Information Problems in Managing Space Risks One of the bas c reasons we have d ff cu ty bu d ng research space f duc ary nst tut ons more s m ar to nsurance compan es (wh ch cou d take

323 from one researcher who s not much of a space-user th s year to supp y someone who s an ntens ve space user th s year, because we a know we w be taken care of when t comes our turn) s that there s a rad ca nformat on assymetry between the owest eve , wh ch does the research, and the h gher eve s, wh ch wou d have to manage the transfer of space. An nsurance company can be moderate y conf dent that c ents w not purpose y run the r cars nto trees n order to co ect the b ue-book va ue because there are other d sadvantages to be ng n acc dents, such as be ng w thout a car, gett ng hurt, and so on. There are prob ems of mora hazard n nsurance, but they are manageab e. But a professor m srepresent ng h s or her need for space n case he or she shou d need more (or because t wou d be n ce to have more) can be checked on y by someone who happens to know a ot about how many cadavers one has to keep refr gerated for a g ven k nd of t ssue samp ng, or how b g a mach ne has to be to crush tt e b ocks of hardened cement to test the strength of mater a s w th d fferent cur ng cond t ons (see Ha d 1952, 367), or whether an ndustr a eng neer rea y needs a separate tab e for h s or her PC or whether t can go on the desk nstead. The nformat on needed to compare the space needs of d fferent k nds of sc ent sts s concentrated n the (separate) owest un ts. Th s means that any poss b e conf dence that, say, the v ce pres dent for research or the Co ege of Arts and Sc ences dean can take care of a professor s research space needs re es on that v ce pres dent or dean be ng ab e to compare the p eas of t ssue samp ers aga nst those of mater a s sc ent sts, and to d st ngu sh the p eas of peop e who use a PC n p ace of a typewr ter and can put t on the r desks from those of peop e who use ded cated PCs w th necessary per phera s that cannot f t on the desk. So the " nsurer" at a h gher adm n strat ve eve n a un vers ty cannot co ect good enough nformat on to red str bute space n a r sky env ronment, because that nsurer cannot te at the t mes c a ms compete what the uck of the draw has n fact brought n n the way of space requ rements for the d fferent departments. The nsurer cannot te because the eve of deta of space needs and the nte gence to bet rea st ca y about space needs next year are ocated at the base of the organ zat on, and the peop e w th the nformat on have no nterest n spo ng the r own chances to nsure the r research space needs by commun cat ng exact y, n the va n hope that the adm n strat on w take care of them. Th s def c ency of adequate nformat on at h gher eve s a so means that peop e at ower eve s see h gher- eve attempts to manage space as an ntervent on by gnorant peop e nto matters they do not understand. Sc ent sts exper ence s that the r arguments, when they know them to

324 be tru y persuas ve, never persuade at the next h gher eve , wh e somet mes spec ous arguments of other co eagues or other departments do persuade. V ce pres dents for research see a ab empty and propose to rea ocate that space, w thout know ng that th s ab s mere y between pro ects and s (v rtua y) certa n to be f ed two months from now. Deans who propose that new peop e w have to f t the r research nto the space vacated by the person who ust ret red do not rea ze that they have ust cut out the three most d st ngu shed cand dates, who happen to be ntens ve space users, and have exc uded us from the ne of research that s where the d sc p ne s go ng.

Space Is Not Sovereign Territory in All Organizations The un vers ty pattern of departmenta sovere gnty contrasts w th, say, the re at ve fung b ty of she f space n a supermarket. Space throughout the store s more or ess s m ar n va ue for se ng a the d fferent var et es of goods; h gher management eve s can be trusted to g ve more space to a part cu ar ne of goods n a departmenta superv sor s ur sd ct on f the superv sor can show t w make more prof ts per foot of she f space, w th the nformat on on wh ch such est mates of prof ts are based hav ng a common form and a common metr c. Space n a supermarket s therefore much more rea ocatab e, much more adaptab e to sh ft ng buy ng hab ts and the chance var at ons of the c ente e of th s and that store, than s un vers ty space. As consumpt on of canned o ves goes down and consumpt on of refr gerated de catessen goods goes up, t s easy to rep ace a the scattered d sp ays of fancy o ves w th a short she f of severe cans, and to use the space saved for a de catessen counter (for a genera ana ys s of th s change n supermarkets, see Wa sh 1988, 185–220). The stab e stat st ca character of supermarket sa es make actuar a va ues good est mates of the va ue of a foot of space devoted to o ves, as compared w th a foot of refr gerated d sp ay of smoked sa mon. H gher managers can co ect accurate nformat on on feet of she f space w thout the cooperat on of ower managers or c erks. A the determ nants of fung b ty of space are h gher n supermarkets than n un vers t es.

How Universities Administer Space The resu t of these features of the nherent nature of un vers ty space adm n strat on s a sort of nested sovere gnty, mod f ed by rare nvas ons from h gher eve s w th spec a ust f cat on. An nd v dua has sover-

325 e gnty over the research space a ocated to h m or her, except on occas on of ret rement, eav ng the un vers ty, or nvas ons under spec a c rcumstances by the department. But nformat on on the underuse of space by an nd v dua w not ord nar y be forwarded by a department cha r to the dean s off ce, nor by the dean to the provost s off ce, because f anyone s go ng to get the advantage of nvad ng a facu ty member s h stor c space r ghts, t shou d be that department (or schoo ), not other departments (or other schoo s). On the occas on of new appo ntments or of ret rements, or of ma or po cy sh fts such as an attempt to bu d up e ectr ca eng neer ng or to start a program n cogn t ve sc ence, the quest on of departmenta space may be reopened at a h gher eve . And of course, a schoo can open the quest on of space when t has free space to offer— f " t" ( .e., one of ts donors) has ust bought a bu d ng or renovated aboratory space or f t has abo shed a department (or, ess ke y, abo shed a h gher adm n strat ve off ce). Then deans can make the negot at on w th ower eve s a s tuat on n wh ch everybody w ns. S m ar y, a schoo s ord nar y mmune from nvas on of ts turf by the v ce pres dent or provost eve , un ess there s a ma or red rect on of the un vers ty or un ess the h gher eve s respons b e for gett ng a ot of money for a new bu d ng. Space becomes fung b e at the next h gher eve on y when t s worth the wh e, at that h gher eve , to go to a ot of "po t ca " troub e to rea ocate th ngs, or when schoo s or departments are reasonab y sure of w nn ng by enter ng nto re at ons w th a h gher eve . What rea ocat on there s depends on the h gher eve hav ng nformat on (that t does not genera y use for rea ocat on) about the actua uses to wh ch space s be ng put, about space requ rements of d fferent sorts of act v t es, and the ke. Some of th s nformat on s the actuar a exper ence of the sort summar zed above, that for examp e, chem stry w usua y need four off ces worth of aboratory space for each emp oyee suff c ent y mportant to the department of chem stry as to occupy an off ce (ca cu ated rough y from Bere ther and Sch nger 1968, 82–83; chem stry departments of h gher eve s of d st nct on than that at the Un vers ty of I no s may need arger rat os). Some of the nformat on on wh ch h gher- eve nvas ons of prerogat ves are based s obta ned by keep ng track of what one earned n past renovat ons—keep ng the same person manag ng space renovat on n the dean s off ce for years, so that one can consu t h m (or, n the rare case of a bu d ngs expert who s fema e, her) when the t me comes for rea ocat ng space or for a ocat ng new space. The ass stant dean for space quest ons, un ke other ass stant deans, shou d not be recru ted

326 from the facu ty, because then v ta space nformat on w go back to the facu ty at the end of a professor s adm n strat ve ass gnment. The nformat on needed to nvade ower sovere gnt es cannot genera y be co ected by sett ng up a comm ttee of facu ty from var ous departments and te ng t, say, to come up w th ten thousand more feet for e ectr ca eng neer ng. Each comm ttee member w systemat ca y overest mate the probab t es of need ng a the current y underut zed space n h s or her department (for examp e, when a the grants app ed for for next year a come n at once). The facu ty representat ve from each department wou d be regarded as a tra tor f he or she d d not put the best face poss b e on that department s space needs, exact y because one cannot trust the h gher eve s to come up w th space when t s rea y needed n the future. The nformat on source for nvad ng the sovere gnty of a ower eve has to be kept autonomous from the ower eve , f rm y under the author ty of the h gher eve . The ass stant dean for space shou d have h s or her career n the adm n strat on, not n the facu ty. Space cannot n genera be adm n stered on a "federa " bas s, because the federated un ts cannot ord nar y be persuaded to g ve accurate nformat on to the author t es of the h gher eve s—because g v ng up space means tak ng a chance w th the future research of the department.

Toward a Philosophy of University Space Administration The argument above mp es that the defense of terr tory by ower un ts aga nst h gher ones n un vers ty fe s n part a rat ona response: t s the conservat sm of an nsurance company that res sts spend ng a ts reserves, because "t me and chance happeneth to them a ." Th s means that proposa s for mprov ng the space adm n strat ve system need to respond to the same needs that space sovere gnty w th occas ona nvas ons does; that a though space s very hard to make f ex b e, the nature of un vers ty research work requ res that t be f ex b e; that f ex b ty s hard to arrange too far from the eve where the f ow of chance var at ons n research space needs s known we enough to respond rat ona y to t. The fo ow ng e ements of a poss b e ph osophy of space management, then, are meant more to commun cate the nature of a strategy than to be "recommendat ons." The purpose of th s br ef out ne of a strateg c response by h gher adm n strat ons to the prob em of space s to ustrate the mp cat ons of th s ana ys s of the m crofoundat ons of an nst tut on that has not been very eg t mate, because ts deo ogy has not been exp c t y

327 formu ated. F rst, one has to d st ngu sh the sorts of nformat on that are ord nar y now kept at d fferent eve s n the adm n strat on, and how that nformat on re ates to the m crofoundat ons we have sketched above. The bas c m crofoundat on adv ce wou d be that these nformat on systems can be mproved on y f they respond to the nformat on about uncerta nt es at the bottom of the system, at the eve of the work ng sc ent st, n a way compat b e w th that sc ent sts nterests, so that accurate nformat on about the probab t es of space needs can become ava ab e to h gher eve s on the adm n strat on.

University Space Information Systems One prob em of a account ng systems s the cho ce of eve s of abstract on for d fferent purposes (for the prob em n manufactur ng, see Chapter 3 above). A un vers ty space adm n strat ve system genera y nvo ves (1) a very h gh y abstracted ("account ng-type") system, wh ch matches square feet of usab e space ("net square feet") w th budgetary un ts (schoo s and departments, nst tutes, centers, etc.); (2) an arch tectura system, wh ch nvo ves draw ngs and structura spec f cat ons; and (3) an nforma network of exper enced assoc ate deans and bu d ng and grounds personne , wh ch stores away n adm n strators m nds a we ter of spec f c facts about th s and that b t of space and th s and that department s needs. Much of th s network s nformat on s der ved from remember ng renovat on p ann ng for that space from the past. The fo ow ng suggest ons m ght mprove the capac ty of h gher adm n strators to respond to the uncerta nt es. The Square Feet Per Budgetary Un t System . Compar son of tab es n Northwestern Un vers ty s genera stat st ca report (Northwestern Un vers ty Off ce of the Provost 1986, 111–119, 127–134) makes c ear that grant do ars per fu -t me equ va ent (FTE) facu ty member s, other th ngs be ng equa , a good pred ctor of the rat o of research space to off ce space. For examp e, of the var ous departments that do chem stry of one sort or another, those w th more grants per facu ty member a so have more research space per off ce. Thus one wou d expect on a probab st c bas s that as departments change n re at ve uck at the sources of grant and contract money, they change n space needs as we . The poss b ty of arge f uctuat ons of a few grants suggests that the centra f e of space nformat on shou d have a spec a f e of the current space usage of a nd v dua grants above a certa n amount and, on b g grant getters grants, of changes n the grant do ars or the personne emp oyed. The tota s of grants assoc ated w th these arge grant getters (and arge users of space) shou d be cont nuous y updated, so that the un ver-

328 s ty or schoo can respond more centra y and qu ck y to opportun t es to rec a m space temporar y from a arge user down on h s or her uck, or to a ocate space qu ck y to deve op ng needs of someone whose research program has f na y taken off. The Arch tectura Draw ngs System . It s reasonab y d ff cu t for anyone w thout deta ed nst tut ona memory of the spaces nvo ved to remember the cruc a user-re evant d mens ons of the space users occupy. Consequent y, t s often not obv ous what s poss b e when t comes to mov ng peop e around. What one perhaps needs s a set of user-re evant character st cs obta ned from nterv ew ng peop e about the space they use, te ng what about the space s mportant to them. (I suppose the best sort of quest on to pose to work ng sc ent sts wou d be those that ask: "What wou d be the d ff cu t es of mov ng nto X ," where X s the name of a bu d ng or a d fferent w ng or f oor of the same bu d ng.) For some departments, f oor strength for heavy equ pment (or for heavy co ect ons of books or mater a s) wou d be a re evant d mens on, for others t wou d not. For most scho ars, ce ng he ght wou d not be a re evant var ab e, but for archeo og sts w th co ect ons to store t wou d be. S m ar y a r c rcu at on for an ma s, hoods to vent dangerous gasses (and a r ntakes that are not ust be ow such vents), so at on from v brat on, freedom from e ectromagnet c no se, ab ty to nsta mu t p e temporary te ephone nes for te ephone survey nterv ew ng, refr gerat on, hum d ty contro , and the ke are a character st cs of qua ty of space that are re evant to spec f c users. The dea wou d be, then, to attach to the f e of draw ngs a footnote for each room or other un t of space. The footnote wou d st mportant user character st cs of the space, e c ted by nterv ew ng the occupants or nterv ew ng the peop e who p anned the ast renovat on of that space. Mak ng Bu d ngs and Grounds and Renovat on Informat on Usefu for Exchange . The centra prob em w th the nformat on n the "o d boy network" of bu d ngs and grounds, assoc ate deans for adm n strat on, and assoc ate provosts s that t can n genera be eas y used on y at a h gher eve , one not nvo ved n mak ng the chancy f ow of future pro ects match the space ava ab e. Most such respons ve exchanges of space are made w th n departments or research centers, and f between departments, are eas est for deans to manage f the departments have come to an agreement beforehand. Part of the prob em of br ng ng the space know edge nto contact w th the know edge of research demands s that centra arenas of f ex b ty n research space ass gnments are the research centers rather than the departments. Prov d ng channe s between space

329 off cers n the deans off ces who dea w th departments and research centers cou d mprove those centers capac t es to negot ate space a ocat ons. Such channe s make the nformat on of such deans off c a s usefu n the p ace where the m crofoundat ons nformat on on research needs for space s best— n fac tat ng exchanges among departments and centers.

Interdepartmental Contracting Part of the troub e n gett ng departments to et go of space temporar y underut zed s that os ng space tends to be permanent, desp te vague prom ses from the h gher adm n strat on that f a coup e of rooms are g ven now to, say, the nterd sc p nary Amer can cu ture program, when the department needs a coup e of rooms for a new facu ty down the road a p ece, t w be taken care of. S nce those adm n strators have ne ther a budget of space to a ocate each year nor, n genera , the power to take space away from someone e se, the departments g v ng up space typ ca y do not be eve these prom ses. Arrangements cou d be made so that the department gett ng the space has t for a def n te term (depend ng on the expected durat on of the research program to wh ch the space w be devoted) and w g ve back f (a) the department now g v ng t up needs t at that t me and (b) the h gher adm n strat on does not prov de that department adequate a ternat ve space. Then one wou d subst tute for the weak guarantee by the dean a one a o nt guarantee between someone who has space and someone who a ocates space on beha f of the dean, mak ng the prom se more be evab e. The ob ect here s to do for the m n ature sovere gn ords n the un vers ty what term easeho ds of agr cu tura and d d for feuda sm: creat ng a v ab e exchange system w thout g v ng precar ous tenants permanent property r ghts.

Department and Center Architecture W th n departments or research nst tutes, the centra prob em s to rea ocate space over the short run. One year a g ven facu ty member s research emp oys f ve data entry c erks on souped-up PCs, and the next year on y two research ass stants do ng runs at the computer center; n the meant me, another facu ty member may go from tak ng samp es from monkey cadavers to runn ng exper ments w th ve monkeys, expand ng space needs great y. If there s some common research space that can be reapport oned, w th movab e part t ons, power sources every few feet, severa separab e ght ng contro s, and te ephone acks everywhere, and f the departmenta conference room can be chopped to d fferent s zes, somet mes crowded,

330 somet mes generous, depend ng on the overa stra n on the research space resources, then th s common space can serve as an nsurance reserve. (Supp y ng movab e dra ns, water sources, and exhaust hoods are prob ems that now seem techn ca y mposs b e; the nventor who so ves them w make phys ca sc ence aboratory space adm n strat on a ot eas er.) Such a reserve can pay th s year s c a ms on space from space used by another facu ty member or graduate student ast year, n return for a guarantee to the second that f he or she g ves up the space th s year, t w be ava ab e aga n when NIH comes to ts senses about support for that person s research. The genera adm n strat ve pr nc p e here s that the h gher the proport on of a departmenta space that s n qu d reserve at the departmenta eve , the ess nd v dua s have to nsure themse ves aga nst future expans on cont ngenc es by keep ng a hoard of space.

The Role of University-Level Research Policy The descr pt on above means that space po cy ooked at from h gher adm n strat ve centers s ke y to take the form of rare ep sodes, each un que, each f ed w th the concreteness of part cu ar bu d ngs, part cu ar space requ rements, part cu ar t es to strateg c dec s ons about h r ng new sc ent f c personne or start ng new schoo s and departments. These are ke y to nvo ve two ma n forms of strateg c po cy: fund-ra s ng pr or t es, and opportun st c ntervent ons n ower- eve space d sputes. Bere ther and Sch nger s Un vers ty Space P ann ng (1968) s d rected ma n y to the f rst of these quest ons: what k nd of space, and how much, can we bet we w need for a strateg c dec s on to expand, say, a chem stry department? Fund ra s ng at the Un vers ty of I no s s of course ma n y eg s at ve, so one can mag ne that a trans at on of chem stry needs nto square feet w tse f be enough to ra se the money: the pos t v st use of actuar a numbers of square feet per FTE sc ent st makes stra ghtforward sense to a eg s ature. In pr vate un vers t es, however, space needs, deve opment pr or t es, and fundab ty by pr vate donors and the federa government a have to be we ghed n the determ nat on of what sort of f nanc ng for bu d ng the adm n strat on shou d go after. Opportun st c ntervent ons tend to come when new space becomes ava ab e, when ma or expans ons or contract ons n departments are undertaken, when new, very d st ngu shed peop e are h red, or when a d st ngu shed person ret res or d es. It s character st c of these types of rea ocat ons that on y a few parts of the un vers ty are d rect y nvo ved (and on y a few p eces of space), that the nterested part es are ntense y concerned but hard y anyone e se s, that deta ed nvest gat on of the

331 space ava ab e and space needs s undertaken, and that therefore genera po cy cons derat ons are swamped by the concreteness of the nterests and nformat on on a part cu ar space prob em. Th s means n turn that genera po c es at the un vers ty eve are not much use. Wh e the re at ve pr or t es of sma parts of the un vers ty versus other sma parts are of course po cy ssues, and ought r ght y to nvo ve the h ghest genera po cy author t es of the un vers ty, genera un vers ty po c es are hard to br ng to bear on quest ons of such concreteness, where the constra nts are many and powerfu and the opt ons therefore very restr cted. The bas c thrust of the "recommendat ons" here for deve op ng the nst tut ons of space adm n strat on n un vers t es m ght be phrased as: bu d nst tut ons that fac tate exchange among the research entrepreneurs at the owest eve s. Th s w sound fam ar to anyone who has d pped nto econom cs after Adam Sm th, for t says essent a y that one has to bu d nst tut ons that fac tate the market, rather than nst tut ons that run th ngs from the center. The argument aga nst centra econom c p ann ng s that s nce the nformat on requ red to equate marg na costs to marg na revenue n thousands of nes of goods n thousands of factor es and farms and shops s w de y d str buted, bureaucrats cannot do t eff c ent y. S m ar y, n the a ocat on of un vers ty research space the know edge requ red to bet on the product v ty of that space n very d verse uses s not ava ab e to a space bureaucrat at the top of the system. Those to whom t s ava ab e, nsofar as t s ava ab e at a , are the d st ngu shed phys c sts or the d st ngu shed anthropo og sts at the base of another un vers ty. That means that even w th n the organ zat on, a more market ke structure of adm n strat on s rat ona . Even f a very bureaucrat c reg strar s off ce, count ng cred ts toward graduat on, s poss b e w thout substant a y underm n ng the modern zat on of the curr cu um, a very bureaucrat c research space adm n strat on s not rea y feas b e, because adm n ster ng research space sens b y requ res know edge of the nherent uncerta nty of bett ng on where new know edge s to be found.

The Teaching Productivity of Scholarly Reputations The d st ed and concentrated form of the nformat on from sc ent f c and scho ar y peers n other un vers t es s the sc ent f c or scho ar y reputat on. That reputat on nd cates that n genera , as far as those peers can te , a part cu ar person s one who can bet we about where new know edge s to be found, and so can make genera y good udgments about

332 such matters as what shou d go nto the curr cu um, whom the un vers ty shou d h re, what equ pment shou d be nsta ed n order to do reputab e work, and the ke. Students can be expected to know whether a teacher s commun cat ng we what t s that he or she s supposed to be teach ng; on y that person s peers from e sewhere can be expected to know whether the teacher s ke y to have good udgment about what to try to teach. Thus the pres dent, provost, or dean s not ke y to be a competent udge of the dec s ons that go nto dec d ng what the bas c course n phys ca chem stry ought to cons st of, how far the re at ve emphas s on d fferent top cs shou d be changed because of the new d rect ons that are deve op ng on the research front ers. The way the pres dent te s whether those dec s ons are ke y to be we s the same way that co eagues e sewhere te : by ook ng at whether the facu ty member who makes them (and h s or her co eagues n the department) seems to be good at bett ng on where new know edge s to be found.

The Rent of Reputations Whether an Amer can co ege or un vers ty can charge h gher tu t ons for ts teach ng does not depend much on what the students say about whether the facu ty are effect ve at commun cat ng what they are try ng to teach. The tu t on s heav y dependent nstead on the genera reputat on the co ege or un vers ty has for "qua f ed" teachers. What qua f cat on comes down to s the reputat on the facu ty has among other sc ent sts for good udgment about scho ar y matters and about research. Thus, we can use the tu t on of co eges and un vers t es (separate y for pr vate and state un vers t es) n the Un ted States as a measure of scho ar y d st nct on of the facu ty and not go far wrong. But th s means that the reputat on of the facu ty, espec a y among peers n other un vers t es, mu t p es what peop e are w ng to pay for the teach ng, much as a aw f rm be ng n a p ece of rea estate n the f nanc a d str ct mu t p es what peop e are w ng to pay for ega serv ces. The payment of h gh sa ar es to d st ngu shed professors, then, m ght be cons dered comparab e to advert s ng expenses; yet ust as advert s ng a aw f rm s address fa se y as n the f nanc a d str ct w not serve the purpose, so n a un vers ty on y outs de cert f cat on of the advert sement that the facu ty have good udgment makes t effect ve. Thus, the extra sa ary of a d st ngu shed professor s much more ke the extra rent one pays for an off ce next to the b g banks; the un vers ty, rather than h r ng a part cu ar scho ar s abor, s rent ng h s or her reputat on. Of course, f there s pos t ve ev dence that the d st ngu shed udg-

333 ment of the scho ar w not be app ed to the curr cu um, the effect s essened. Lead ng phys c sts at the Pr nceton Inst tute for Advanced Study do not have qu te the same effect on Pr nceton s tu t on as do ead ng phys c sts n the Pr nceton phys cs department, because t s known that the former do not concern themse ves much w th curr cu ar matters and nstruct on. St , the va ue of teach ng young peop e n a department s ncreased by the department s or schoo s genera reputat on, and so one can charge enough for professors serv ces to support them do ng ha f-t me research. A schoo w thout d st ngu shed phys c sts cannot charge enough (e ther to the state or to the students) to g ve the r young phys c sts fu sa ar es for do ng ha f-t me research, ha f-t me teach ng. The opportun ty for a young phys c st to nvest ha f t me n h s or her reputat on by do ng research, wh ch he or she can ater rent out at a h gher pr ce, s, from the po nt of v ew of the un vers ty, a rent pa d n advance for the future va ue of the reputat on. From the po nt of v ew of the young facu ty member, t s a rent pa d on h s or her "prom se." Peers, preferab y from e sewhere as we as from the un vers ty tse f, thus prov de udgments of prom se as we as reputat on. They are the on y ones who can supp y re ab e nformat on to make the bet of future product v ty that ust f es the nvestment of "rent n advance"—payment for ha f-t me research. Of course, w thout serfdom of facu ty members for the advance rent, the returns from that nvestment may accrue to another un vers ty. But n fact, most d st ngu shed sc ent sts rent the r ach eved reputat ons to the same un vers t es that nvested n the r ear y career, f the un vers ty reta ns them (see, e.g., Zuckerman 1977, 194–195). Joseph Ben-Dav d (1960, 1968–1969) argues that un vers t es w make such nvestments n young peop e on y when they are n strong compet t on for sc ent f c d st nct on w th other un vers t es n the same country. But th s means that ma nta n ng the va ue of the teach ng of a un vers ty or co ege depends more on ma nta n ng the reputat ona cap ta t s rent ng than on the abor expended on teach ng. Th s n turn means that the un vers ty ought perhaps to exp o t the d fferent a capac ty of d fferent facu ty members to ma nta n the r reputat ons, by g v ng the more ta ented (as measured by scho ar y reputat on) more t me for research and those who br ng n ess reputat ona cap ta heav er teach ng oads.

Getting the Work Done with Rented Reputations Th s overa s tuat on poses a d emma for un vers ty adm n strat on: how to co ect nformat on on the va ue of the research of var ous nd v dua

334 facu ty for the ong-run reputat ona hea th of the un vers ty, and how to ma nta n that hea th w th the money from h gh tu t ons from students or h gher budgets from state eg s atures. That d emma has at east two aspects: the determ nat on of teach ng oads for departments, and the determ nat on of teach ng oads for nd v dua s w th n departments. Broad y speak ng, the number of course enro ments per "regu ar" facu ty member, and very often the number of courses taught per facu ty member, s sma er n the aboratory sc ence departments n bera arts schoo s (and espec a y sma er n those departments whose members m ght teach n med ca schoo s) than n the human t es and soc a sc ences, and arger n bera arts, aw, and management schoo s than n med ca , denta , and eng neer ng schoo s. Most un vers ty departments n the Un ted States have forma y equa teach ng oads for a regu ar facu ty members, w th the oad ca cu ated n e ther courses or ecture courses ( .e., exc ud ng graduate sem nars) per year. Some departments n some un vers t es, however, count the teach ng of arts and sc ences graduate students not as a teach ng respons b ty but as a research respons b ty, and some do not count tutor a - ke superv s on of ndependent study or d ssertat ons as forma teach ng ( nforma y , of course, such nd v dua nstruct on s a ways cons dered rea teach ng). That s, un ess there s outs de pressure, as from a state eg s ature, for str ct account ng ru es, most un vers t es are fa r y ax about what s counted as a course and a ow departments a w de at tude n mak ng such def n t ons; under forma pressure, they may even nvoke d fferent nforma ru es to make the record ook r ght n d fferent departments. D fferent departments and d fferent schoo s are a so ke y to have d fferent forma standards of what s a norma teach ng oad. Presumab y, th s m ght ref ect the d fferent amount of pract ca d ff cu ty n teach ng d fferent sub ects; the sma student oads for e ementary anguage nstruct on, for examp e, no doubt part y ref ects the fact that the teacher has to hear the students rec te n order to correct the r pronunc at on. But teachers of e ementary anguage courses often have a very arge number of courses, are not pa d extra for the r reputat ons, and have very tt e t me eft over for research; nor, often, are they on the tenure track, the centra nd cat on that the un vers ty does not want to nvest n the r reputat ons. S m ar y, t s not usua y feas b e to have a very arge med ca schoo c ass gather on rounds to hear a d scuss on of the norma run of symptoms and d seases n a g ven spec a ty. The ma n po nt here, though, s that the nformat on systems that dec de such quest ons are heav y decentra zed to departments, and no doubt ref ect the "norma nvestment n prom s ng sc ent sts" that ead-

335 ng un vers t es make n the var ous d sc p nes. The reputat on for research product v ty that a chem st has to keep up w th s, of course, that of facu ty members at other ead ng schoo s, accomp shed n the amount of research t me those facu ty have eft over after do ng the r teach ng. Th s may be d fferent from what a mo ecu ar b o og st has to keep up w th, because many mo ecu ar b o og sts are n med ca schoo s, where teach ng oads are norma y ower. W th n departments there are a so var at ons n how teach ng oads are ad usted to the potent a payoff of nvestment n reputat ona cap ta . Some departments do th s exp c t y, by c ass fy ng peop e as "research act ve," w th a ower number of courses or a arger cred t for graduate teach ng, and nact ve, w th a arger number of courses or a arger proport on of undergraduate ecture courses. Somet mes the way one qua f es as act ve n research s s mp y to say that one has research under we gh; I was once n such a department, and severa of the facu ty d d not make that c a m, though some others who had not pub shed anyth ng n a ong t me d d. Somet mes, however, there s apparent y departmenta consensus on who are the act ve researchers, who are nact ve (by wh ch we mean that a strong cha r s dec s ons about the d str but on of d fferent a teach ng oads do not create nto erab e stra ns), so a forma departmenta dec s on a ocates ower teach ng to ead ng researchers, h gher teach ng to ess act ve researchers. Probab y th s s more common n sc ences w th a strong parad gm, such as phys cs or econom cs, than n areas w th a weak or mu t faceted set of cr ter a for research va ue, as s character st c of geography, soc o ogy, or management sc ence (cf. the c ass f cat on of sc ences n Wh t ey 1984, 158 and pass m). Somet mes n the phys ca and b o og ca sc ences qua f cat on as an act ve researcher depends on whether one has a grant that supports graduate students and postdoctora students. Very often facu ty members do not do the r own ab work, but nstead get grants, prov de and cr t c ze deas, and do much of the wr t ng up of resu ts. Somet mes research grants d rect y reduce teach ng oads by buy ng part of the facu ty member s t me; a comparab e arrangement s for facu ty to have o nt appo ntments w th a research nst tute, wh ch pays part of the r sa ary, preferab y by research grants. In both these cases some "outs de" body n contro of the money makes a udgment about the probab e va ue of the research and se ects some but not a of the facu ty to get ower teach ng oads by pay ng part of the r sa ar es spec f ca y for research. Such bod es use both the reputat ona nformat on that un vers ty processes use and peer udgment of the spec f c pro ect proposa subm tted. In some systems, part cu ar y those outs de the Un ted States, teach-

336 ng oads n un vers t es are determ ned d fferent y for d fferent ranks, and the ranks that facu ty members reach depend on how d st ngu shed the r research s. Professors (what Amer cans wou d ca "fu professors") n German- nsp red or Eng sh- nsp red systems are much rarer than n Amer can un vers t es, and not a scho ars who get tenure as teachers become professors. Professors often have d fferent teach ng oads, and at any rate have a d fferent m x of courses ( n part cu ar, more graduate nstruct on), than " ecturers," or whatever the ower ranks are ca ed. Named or endowed professorsh ps n Amer can un vers t es often carry reduced teach ng, somet mes by hav ng an endowed part of the sa ary n a research nst tute, and ke European professorsh ps are not expected to be ava ab e to a comers. The overa mp cat on of th s rough out ne s that a mu t p c ty of dev ces are used n an attempt to g ve facu ty members a "fa r" teach ng oad, so that they can bu d up the r own reputat ons and thereby that of the un vers ty, thus a so bu d ng up the tu t on rate the un vers ty can se ts teach ng serv ces for. It s character st c that teach ng oad dev ces are somet mes (a) excess ve y r g d, so that a facu ty members have the same teach ng oad regard ess of the return to nvestments n the r reputat ons (espec a y the same w th n s ng e departments or w th n un vers t es); (b) corrupt b e, because they use nformat on on y from the nterested part es or from oca departmenta co eagues; or (c) re ant on outs de peer udgment (as n research grants, amounts of pub cat on, d fferent a promot ons to professorsh ps n European systems or to endowed professorsh ps n the Un ted States). Whatever the forma dev ces are for mak ng the teach ng abor contract var ed, they are embedded n an nforma system that furthers the same end—of protect ng the reputat ona cap ta of the un vers ty through oose teach ng oads for h gh y reputab e researchers. In form ng a abor contract between a un vers ty and a part cu ar scho ar or sc ent st, there are a var ety of sutb e negot at ons about how much emphas s on research there s to be n the contract, and so mp c t y on how tt e teach ng abor s to be supp ed. When the nte ectua preparat on for one s teach ng w be the same as the preparat on for one s research (as nd cated by whether the cha r of the department asks what the cand date wou d ke to teach, w th few restra nts about d v s on between graduate, undergraduate, ecture, or sem nar, versus a c ear out ne of wh ch arge undergraduate courses the professor must take on; and nd cated as we by how few or many separate course preparat ons the cand date s expected to make), the proport on of the sa ary that s essent a y rent for a reputat on s mp c t y set h gh, the proport on that

337 s payment for teach ng abor, ow. Such "spec a pr v eges" of ow teach ng and h gh research have to be eg t mate w th n the department, but they do not ord nar y have to be defended n any arger adm n strat ve sett ng. There s genera y pressure at the owest adm n strat ve eve s of the system—w th n departments—to set the genera terms of the contract favorab y (that s, to make more of the contract count as rent on an ach eved reputat on or an advance on a future reputat on, as opposed to wages for teach ng) for the department as a who e, and to create one or another forma dev ce for d fferent at ng that genera m x of payment for teach ng versus research reputat on n the barga ns w th part cu ar facu ty members. The pressures for favorab e ( .e., research-or ented) barga ns are d fferent a y effect ve across the d sc p nes, genera y be ng more effect ve n the phys ca and b o og ca sc ences, part cu ar y n those attached to the med ca schoo , and ess so ( .e., more or ented toward teach ng) n d sc p nes at the human t es end. It s hard to make the d fferent a s among departments eg t mate. Most y they are made eg t mate by not be ng pub c. Soc o og sts usua y do not have much nformat on on the genera terms of the abor contract n chem stry—for examp e, how many courses chem sts have to teach, how graduate nstruct on s counted, whether they can buy t me off w th grants, and the ke. And of course converse y, chem sts know very tt e about what the abor contract of a d st ngu shed soc o og st genera y ooks ke.

Summary on the Academic Teaching Labor Contract To exp a n why the contract for teach ng abor n ead ng un vers t es rare y shows ev dence of any attempt by the emp oyer to get more teach ng for ess money, why sen or academ cs make much h gher sa ar es than beg nners for do ng the same teach ng abor for no more hours, why r cher and more prest g ous un vers t es pay ng h gher sa ar es requ re ess teach ng abor, we have to reconce ve what the abor contract s for. If we conce ve t to be a contract to rent a reputat on, so that the abor supp ed can be so d for h gher pr ces to students and grant g vers, many of these pecu ar t es are exp a ned. But the dependence of the va ue of teach ng work on the teacher s sc ent f c or scho ar y reputat on s tse f determ ned by the overa prob em of qua ty contro n advanced teach ng, g ven that the c ent who s go ng to pay a prem um for qua ty cannot genera y te whether t s, or w

338 be, h gh qua ty or not. S nce one cannot, however, trust academ cs to correct y assess the r own nte ectua va ue, because then the r sa ar es wou d depend on y on what they te the r emp oyer about how good they are. The ev dence such academ cs m ght prov de on the r own va ue wou d not be ke y to conv nce a parent that the bache or s degree s worth as much as a sma med umqua ty house. So one has to use nformat on com ng from a body of nformed udges who have no mmed ate nterest n the facu ty member s sa ary (or teach ng work), but who w know (as best t can be known) whether what the cand date s ke y to teach s ke y to be true. One has to have departmenta comm ttees and deans to trans ate that scho ar y reputat on nto promot on and tenure, but they cannot conven ent y generate the or g na nformat on that const tutes the true qua ty contro . St , before that reputat on can be so d n the form of h gher tu t on payments or arger subs d es per student from the state eg s ature, that reputat on needs to bu d up over t me, be genera zed across t me, and be genera zed to the un vers ty as a who e. That s why Amer can un vers t es have to nvest n the future research reputat on of young facu ty members, by pay ng them to do research whose va ue to the un vers ty w come ma n y when they are sen or scho ars who are known to be f rst-rate by a w de commun ty of scho ars outs de the un vers ty.

Conclusion The nformat on about what s a worthwh e use of un vers ty research space and un vers ty facu ty t me s generated f rst of a at the owest eve of the un vers ty, as a sc ent st or scho ar dec des what w probab y be of most use n the further scho ar y work of h s sma research group scattered a over the wor d. The un vers ty has to pay h gh wages for h m or her to produce know edge that w , on average, be read by th rty to one hundred peop e n other un vers t es, because on y those peop e w be ab e to prov de the nformat on needed for qua ty contro of the nte ectua content of teach ng. The un vers ty has a so to prov de the space for the research when the scho arsh p s space-us ng; the aborator es, when t s exper menta ; the brar es, when t s book-us ng; and the museums, when t s spec men-us ng. These two requ rements der ve from the bas c fact about advanced teach ng—that teach ng the best we know neff c ent y s much better than teach ng what we knew a generat on ago eff c ent y. The requ rements for research space and t me confront the un vers ty w th d fferent sorts of agency prob ems. The k nd of person who can

339 teach an up-to-date vers on of some spec a f e d may be most re ab y found by consu t ng the ead ng members of the sma research commun ty he or she s contr but ng to. Those peop e can be dent f ed by consu t ng the arger research commun ty (or subd sc p ne) n wh ch that sma commun ty s embedded. There are a so mu t p e measures of the mpact of a sc ent st that are more or ess mmune from gross agency prob ems, such as c tat ons n other sc ent f c work. By us ng sc ent f c or scho ar y reputat on as an ndex of facu ty nte ectua qua ty, a un vers ty can, n the ong run, make a case that the r advanced nstruct on s worth more than the advanced nstruct on of ess-research-or ented nst tut ons. Thus, a abor contract that ooks more ke a contract for the renta of a facu ty member s reputat on does, to a cons derab e degree, so ve the prob em of whom to h re, reta n, and promote, and of who shou d have author ty over the teach ng and research program of the un vers ty. But the demands on un vers ty space generated by the ser es of research pro ects that const tutes the nvestment n that reputat on s not so eas y sub ect to peer rev ew by outs de sources. The outs de sources who know pretty exact y whether an exper ment prov des ev dence for or aga nst a hypothes s do not know near y as we how much equ pment space, off ce space, and brary space the ne of research nvo ved shou d take. They do not need an accurate assessment of the space needs of another sc ent st e sewhere for the work they w sh to do; what they do need s an accurate assessment of the truth of that sc ent st s hypothes s for the r work. The agency prob em here s that the peop e w th n the un vers ty who have the best answer on the qua ty of a person s research and on h s or her space needs—name y, that very person— a so have an nterest n gett ng the owest teach ng oads and the ushest space a ocat on poss b e. One can use outs de nd cators of sc ent f c and scho ar y reputat on to udge research product v ty, and thus to prov de a so d bas s on wh ch to rent the reputat on and have that renta be a rat ona un vers ty nvestment; but one cannot use those same outs de "peers" to so ve the un vers ty space a ocat on prob em, because they do not have, and have no reason to g ve, accurate nformat on on the space needs. One may use genera scho ar y reputat on to we gh an nd v dua s p eas for more space, so that a person whose reputat on makes h m or her a s x-hundred-pound gor a gets space wh e a ghter-we ght scho ar does not; but the un vers ty needs too much know edge of the substance of space needs to a ocate space pure y on the bas s of reputat on. The bas c hypothes s of th s book s that adm n strat ons w grow n-

340 format on-process ng and dec s on-mak ng structures toward the sources of nformat on about the cruc a uncerta nt es of the r work. We can reformu ate our arguments here by say ng that an nformat on structure for the adm n strat on of space rea y has nowhere to grow. Those who have the best est mate of the actuar a va ue of the r space needs over the next few pro ects are the sc ent sts or scho ars nvo ved, and t s exact y they who can be expected to g ve on y overest mates. The best comprom se s to trust the nforma assessment of co eagues w th n the department, n a system of departmenta sovere gnty over research space. The baroque structures of the tenure and promot on rev ew process n ead ng un vers t es show that there are great prob ems n assess ng who w have the most va uab e reputat on to offer to the un vers ty for rent. But the way those structures grow out, toward co ect ng nformat on from the sma research commun ty n wh ch a sc ent st or scho ar s embedded, show that there s some so ut on to the agency prob em n ascerta n ng what reputat on a un vers ty shou d rent, at what pr ce, and w th what teach ng ob gat ons, to ma nta n that overa , nst tut ona reputat on. That overa reputat on ets the un vers ty se ts teach ng at much above what the cheapest teach ng abor cou d be h red for.

341

10— Conclusion Three Criteria for Theories of Organization Th s book has been gu ded by three ma n cr ter a for theor es of organ zat on: (1) they shou d exp a n how organ zat ons can be more rat ona than nd v dua s (though of course they are not a ways); (2) they shou d be soc a structura and thus ab e to exp a n, for examp e, why 10 percent of an organ zat on s parts can be absent on a g ven day and the organ zat on st work the same way; and (3) they shou d exp a n var ance deep ns de organ zat ons by var at ons n what the nd v dua parts have to do—and by extens on, they shou d exp a n var at ons among organ zat ons as we as why some parts of a s ng e organ zat on are so d fferent from others. After the genera argument nvo v ng these three cr ter a together s d scussed, each cr ter on w be presented n more deta w th ustrat ons from the argument. F na y, the mp cat ons for research strategy n organ zat ona soc o ogy w conc ude the book. The f rst cr ter on s that the theory shou d prov de the bas s for exp a n ng why soc et es that can organ ze arge-sca e comp ex organ zat ons hard y ever eave anyth ng mportant to nd v dua s. Our exp anat on s that organ zat ons ord nar y do a the th ngs we have been ana yz ng n th s book better than nd v dua s do: they teach the advances n sc ence better (Chapter 9); they produce both exp os ves and pa nts better (Chapter 4), dr for o a coup e of m es under the North Sea better (Chapter 3), and organ ze work better (better than we usua y manage n our househo ds) so that the peop e who do t have the appropr ate sk s (Chapter 7). The theory, then, has to exp a n why organ zat ons are more rat ona than nd v dua s. If, as we w argue be ow, the var ab es we want to use n the theory are descr pt ons of uncerta nt es and of the soc a structures of those parts, and the un ts of ana ys s are the parts of the organ zat on, the mech-

342 an sm that connects them s a postu ate of organ zat ona rat ona ty. In th s book, the postu ate of rat ona ty s what connects cond t ons n the env ronment to the structure of parts of organ zat ons. But the concrete substance of rat ona ty depends on what part of the tota uncerta nty a part cu ar part of the organ zat on has to dea w th. Thus, rather than postu at ng one rat ona ty for the who e organ zat on (such as "max m ze net present f nanc a va ue") and assum ng that the rest of the prob ems w be dea t w th by an opaque "product on funct on," we postu ate many d fferent rat ona t es depend ng on what uncerta nty has to be dea t w th to get a g ven subtask done. One of the many (and often the h erarch ca y dom nant) rat ona t es of an organ zat on s to max m ze net present va ue. But that cannot be done, for examp e, w thout prevent ng b owouts and hence w thout dea ng rat ona y w th the r sk of h gh-pressure gas com ng nto the dr ho e. Further, we do not connect the postu ate of rat ona ty to dec s ons (a mass of ethnography of dec s ons shows that they are not very rat ona ), but to soc a structures that gather nformat on and, more or ess rout ne y, make dec s ons. The postu ate of rat ona ty, then, s much more a gu d ng pr nc p e of growth of the structure than t s a not on that a s ng e ut ty s everywhere max m zed n each dec s on. And th s means that the tradeoffs "between rat ona t es" are ord nar y to be conce ved of as compet ng pr nc p es of growth of d fferent parts of the dec s on-mak ng system. For examp e, the structure for recru t ng, reta n ng, and promot ng facu ty n a un vers ty s organ zed to dea w th the uncerta nty of whether peop e w do good research n the future, and grows nto a baroque system of many eve s of comm ttees wr t ng many etters to the outs de "peers" to ask what dec s on the organ zat on shou d make. The structure for ass gn ng teach ng s or ented to a structure of requ rements for award ng degrees, for ass gn ng course grades, for sett ng forma teach ng oads and negot at ng except ons, often department by department, so as to get the teach ng ob done. They are nterdependent structures because the tu t on one can charge for the teach ng s determ ned by un vers ty reputat on for scho arsh p and because the research that bu ds the reputat on takes t me that m ght otherw se be used to teach. But the two structures do not grow at the same speeds and w th the same author ty n d fferent co eges and un vers t es, and the tradeoff between the va ues of teach ng and research n a part cu ar un vers ty s an outcome of the re at ve growth of two structures of rat ona ty. In some sense, th s mp c t cho ce s a ong-run cho ce of ut t es, and to some extent the tradeoff s determ ned by the h erarch ca y super or

343 eve s of the organ zat on. But the rat ona t es bu t nto the Amer can semester-hours course-cred t system that Thorste n Veb en so ked to make fun of are not the same as the rat ona t es bu t nto the use of c tat on counts to measure reputat on that o d-fash oned facu ty who th nk one ought to read a person s work to udge them (that nc udes me) ke to make fun of. As th s examp e shows, t matters a good dea whether the rat ona ty s ocated theoret ca y n the organ zat on as a who e or n ts subparts—a po nt we dea w th as the ast of our three cr ter a. If t s ocated n the subparts, then re at ve rates of growth of the subparts are the dynam c component n the cho ce of ut t es, and the ut ty funct on can be expected to change over t me. But the rat ona ty of dec s ons s a so determ ned by the soc a structure n wh ch those dec s ons are made, by how far that structure has grown to be ab e to br ng good nformat on about the k nd of uncerta nty that ma n y matters and to organ ze the dec s on process so that that nformat on can be used when t comes n. The prob em of whether one can trust the source of nformat on s pervas ve n organ zat ons, because the t me y nformat on about subparts of the uncerta nty tends to be concentrated among peop e whose spec a respons b ty t s to dea w th t—wh ch means that they need not report the best nformat on ava ab e f so do ng s to the r d sadvantage. So the prob em of the appropr ateness of the soc a structure n wh ch an uncerta nty s dea t w th has two ma n parts: whether the nformat on co ected s good, and whether the peop e who have the nformat on have ncent ves to use t for the benef t of the organ zat on or mot ves to m srepresent t so that they ook better. Performance measurement systems n part cu ar are shot through w th prob ems of correct y measur ng the env ronment confronted by the organ zat on n a part cu ar area of ts funct on ng so that they can eva uate performance re at ve to that (see, e.g., Ecc es and Wh te 1988, on the d ff cu ty of gett ng the appropr ate "market pr ce"—or "cost," wh ch n econom c theory s supposed to be the same—at wh ch d v s ons of a mu t d v s ona company se each other the r products). The nformat on system on a part cu ar area of the env ronment, pervaded by a part cu ar k nd of uncerta nty, "needs," then, to be constructed n such a way as to so ve the ma n d ff cu t es, substant ve and mot vat ona , for co ect ng nformat on on that sort of uncerta nty. The bas c assumpt on of our theor z ng to exp a n soc a structure var at ons w th n and among organ zat ons s that we can exp a n much of the var at on n d fferent parts of d fferent organ zat ons by those needs. We hope that th s approach can g ve us systemat c answers about why we fee , when we come to the end of O sen s ana ys s ( n March and O sen 1976)

344 of how a name ess un vers ty chose a successor to a name ess dean, that we have not on y a case study of a part cu ar un vers ty but a so a genera case of an organ zat on try ng to h re on the bas s of scho ar y d st nct on ( n sp te of h r ng a so for an adm n strat ve ob). We get that fee ng of genera ty, we argue, because we know we are not study ng a part cu ar dec s on that ust happened to come out the way t d d; rather, we are study ng a soc a structure shaped by the nature of the dec s on prob em, wh ch s shaped n turn by the k nd of prob em t s to get re ab e nformat on on scho ar y d st nct on. And th s br ngs us to our second cr ter on for theor es of organ zat on—that they exp a n var at ons n soc a structure rather than var at ons n the outcomes of part cu ar dec s ons. The b g d ff cu ty w th approach ng organ zat ona fe through dec s on mak ng, as n the S mon and March trad t on (S mon [1947] 1976; March and S mon 1958; Cohen, March, and O sen 1972), s that the fundamenta fact of the matter s def ned, n the f rst nstance, as what happens n a part cu ar dec s on. The aura of genera ty about the best work n that trad t on (e.g., March and O sen 1976) comes from trac ng what happened n a part cu ar case back to someth ng we recogn ze as genera . For examp e, the ease w th wh ch a search comm ttee for a new dean co ects nformat on about cand dates who have wr tten art c es n books ed ted by the former dean wou d c ear y be s m ar n the soc a structure of other un vers t es, as a professors wou d recogn ze. The reason econom sts have troub e ass m at ng the resu ts of such stud es nto the r theor es s exact y that the r dea of genera ty s not of a soc a structure bu t nto the organ zat on s mode of proceed ng, but of the human tendency to max m ze se f- nterest narrow y conce ved. Our bas c argument here s that the th ngs worth exp a n ng about dec s ons n organ zat ons are those that are due to cont nu ng features of soc a structure. That s, we be eve that one shou d fo ow one s ntu t on about what s exp cab e and genera n O sen s study of se ect ng a dean—wh ch s not that peop e make m stakes about scho ars reputat ons because they do not have t me and energy to get everyth ng r ght, but that un vers t es by the r nature co ect nformat on by the method of scho ars fo ow ng out the r own contacts to get good eva uat ons of the research that cand dates are do ng. But of course, to be sources of var ance n behav or between organ zat ons and the r parts, such features of the soc a structure have to be var ab es. Soc a structures of parts of organ zat ons have enough constancy to be worth exp a n ng, and enough var at on between organ zat ons and between the r parts to g ve someth ng to exp a n soc o og ca y. Our purpose n th s book has been to say what those soc a structura var ab es are, and why they are systemat c n many d fferent organ zat ons. Our

345 answer to the why quest on s that organ zat ons grow soc a structures to hand e dec s ons by creep ng toward nformat on about the ma n sources of uncerta nty that matter of the organ zat on. Of course, those sources of uncerta nty are d fferent for d fferent parts of the organ zat on. The b g troub e w th some vers ons of the cont ngency theory of organ zat ons (the c ass c source s J. Thompson 1967, but Thompson s a so nterested n nterna var at ons) s that t tr es to def ne one b g uncerta nty for the who e organ zat on. But f an organ zat on s dr ng for o n the North Sea, one part of the organ zat on faces the uncerta nty of whether h gh-pressure gas w b ow a m e of p pe and other equ pment out of the ho e; another part faces the uncerta nty of whether OPEC w fa apart (no onger uncerta n, though econom c theory taught that OPEC wou d come apart ong before any pract ca peop e cou d assume t actua y wou d), n wh ch case on y the r ch f e ds wou d pay; another part faces the uncerta nty of whether the mar ne nsurance ndustry w be ab e to put together a coa t on w ng and ab e to nsure the enormous nvestment n nsta at ons (He mer 1958b, c); another part faces the uncerta nty of whether the soc a st government w et them keep the prof ts the r contract w th that government g ves them; and so on. Th s s then not one b g uncerta nty, but ots of d fferent uncerta nt es, and one co ects the atest news about them n very d fferent ways. Consequent y, one has to bu d soc a structures to keep tabs on d fferent parts of the env ronment, to respond on the very d fferent t me sca es that these d fferent uncerta nt es are uncerta n on, to be competent to eva uate the r sks as the news comes n, and so on. The and so on s n the prev ous paragraphs are meant to suggest that an organ zat on has to be bu t to respond better to each of the var ous cont ngenc es t s faced w th than the average nd v dua cou d do, because t wou d be a very unusua nd v dua who cou d mon tor the nformat on about exact y when h gh-pressure gas m ght come nto a dr ho e and what coa t on of bureaucrats and po t c ans m ght conv nce the soc a st government to be a re ab e partner n ts contracts w th wor d cap ta st enterpr ses. The standard operat ng procedure of the var ous parts of the organ zat on, or the "program for dec s ons" n the S mon and March anguage (March and S mon 1958), s not on y a system for tak ng d scret on away from smart nd v dua s who m ght have guessed better n a part cu ar s tuat on; t s a so a system for co ect ng nformat on so that the ord nary peop e who are ke y to be at the po nt of dec s on at the cruc a t me are moderate y ke y to take the r ght dec s on both about b owout protect on and about what tack to take to defend o ndustry prof ts n Norweg an po t cs. The standard operat ng procedure that w do one

346 of these th ngs we s very un ke y to do the other we . A rubber stamp w th Bu sh t! on t (see Chapter 3) wou d destroy the de cate po t ca ba ance, wh e t s ke y to rem nd dr ers to pay attent on to what other dr ers say, rather than to the papers sent out by peop e onshore, f they want to stay a ve. Accord ng to our th rd cr ter on—that the un ts of ana ys s shou d be parts of organ zat ons—a theory that exp a ns why, say, the abor contract for professors n a un vers ty s so d fferent from the abor contract for eng neers n a manufactur ng f rm (see Chapter 9; a so Schroeder and F n ay 1986; He mer 1984) shou d a so exp a n why professors contracts are so d fferent from accountants contracts w th n the same un vers ty. A casua nspect on w show that ess than ha f of the peop e n a un vers ty have obs that depend on scho ar y competence and reputat on outs de the un vers ty. The fact that most organ zat ons have none, rather than ess than ha f, has to be exp a ned, to be sure; but so does the fact that the un vers t es most devoted to scho arsh p have most of the r peop e devoted to someth ng e se. A theory that s good for exp a n ng the d fferences between organ zat ons shou d be good for exp a n ng d fferences between parts of a g ven organ zat on. Th s cr ter on means that the theory deve oped by W amson (1975) to exp a n why some act v t es are organ zed h erarch ca y wh e others are organ zed n a market, for examp e, needs to be extended n two ways. In the f rst p ace, t shou d be ab e to pred ct when act v t es organ zed n the market tse f w have more h erarch ca e ements n the r contracts (as we tr ed to do n Chapter 6 and W amson has attempted n h s ater work [1985]). But second, t shou d be ab e to pred ct wh ch parts of an organ zat on w be organ zed more h erarch ca y, and wh ch more ke a market (as we fo owed Chand er do ng n Chapter 4, d scussed for un vers t es n Chapter 9, and s d scussed n re at on to the part cu ar prob em of nterna pr c ng n Ecc es and Wh te 1988). The cruc a po nts here are (1) that the transact on s too sma a un t to shape a soc a structure, wh e a ong ser es of transact ons reso ved w th s m ar sources of news about s m ar uncerta nt es can shape durab e parts of the soc a system, and (2) that the organ zat on s too b g a un t because the uncerta nt es t faces are too many and too var ous to have one dom nant outcome. At the oppos te end of the var at on n un ts of ana ys s (rang ng from transact ons to who e organ zat ons) s the theory deve oped n organ zat ona eco ogy ana yses of fa ure rates of who e organ zat ons (e.g., A dr ch 1979; Carro 1987). The theory re ates such rates to who e organ zat ons structura features and to features of the r env ronments. Natura y, th s m sses a ot of fa ures that d d not destroy the organ za-

347 t ons because n other respects the organ zat ons were successfu (NASA st ex sted after the Cha enger d saster). But more mportant from our po nt of v ew, t does not prov de the opportun ty to use nuanced measures of soc a structure—of whether, for examp e, the account ng department s so set up and so re ated to the rest of the organ zat on as to g ve pr or warn ng of fa ure of parts (see the d scuss on of Du Pont s use of account ng nd cators to ocate fa ng parts n Chapter 4), thus a ow ng structura changes to be made n the fa ng parts before the who e organ zat on s brought down. Everyone knows from exper ence w th the r own organ zat on that there are enormous var at ons from p ace to p ace n the organ zat on s soc a structure, n r ghts and ncent ves, n hours, n sty e of dress, n access to expense accounts, n autonomy, n respons b ty for cr s s management, n career prospects, and so on. If, as we have argued above, our centra var ab es ought to be soc a structura , then we are requ red to be ab e to ana yze these nterna var at ons. They appear to the newcomer as part of the natura order of th ngs; the newcomer to a un vers ty, for examp e, earns as a natura fact that professors are a owed to do pretty much as they p ease and be pr ma donnas, but that the rest of us are sub ect to the author ty of our h erarch ca super ors. But these d fferences pers st and are part of the natura order of th ngs for the next cohort of newcomers: they are part of the soc a structure.

Organizational Rationality Peop e who work for organ zat ons are he d to a h gher eve of respons b ty to be rat ona than are peop e answer ng on y to the r nd v dua des res need ng to be max m zed. Further, peop e who des gn nformat on and dec s on systems n organ zat ons are he d respons b e for both the r own rat ona ty and that of others who w be mak ng dec s ons n the system. Consequent y, the dea of ca cu at ng a the tradeoffs among a the costs and benef ts, the dea of co ect ng a the nformat on needed to make a rat ona dec s on ( f on y to cover one s ass), the dea of cons der ng benef ts some d stance n the future on rough y the same bas s as benef ts at the next moment—a are more near y ach eved n organ zat ons than n nd v dua ves. I take the "postu ate" that organ zat ons are more rat ona than nd v dua s as an observat on, and consequent y as someth ng we need to exp a n, as we as tak ng t as the bas s for the reason ng n th s book. The emp r ca assert ons of the arguments we have posed here can be turned on the r heads to exp a n the connect on between organ zat on

348 and rat ona zat on. We can say of the argument about d v s ona zat on n Chapter 4, for examp e, that the more we observe that organ zat ons are d v s ona zed when they are n mu t p e markets ( n the sense def ned n Chapter 4), the more support there s for the propos t on that organ zat ons are sources of rat ona ty. S m ar y, when organ zat ons use sen or ty as a cert f cate for se ect ng workers for moderate y stab e nterdependent product ve systems, but use more peer udgment of sk for cert f cates of workers ab ty to dea w th a var ab e f ow of cont ngenc es (as suggested n Chapter 6), we have further ev dence for organ zat ons as sources of rat ona ty. Th s book, however, has been wr tten the other way around, w th the rat ona ty of organ zat ons be ng postu ated. Note that th s s not a postu ate that organ zat ons do not make errors. Instead t s a postu ate that one can exp a n var at ons n the soc a structure of d fferent parts of organ zat ons on the presumpt on that the structure of the nformat on and dec s on system w grow toward those sources of uncerta nty whose rat ona management w matter most to the organ zat on. The evo ut onary pr nc p e here s that the organs of an organ zat on grow better adapted to the r env ronment over t me— n other words, that the rat ona ty of an organ zat on mproves (see Carro 1987 for ev dence that new organ zat ons fa more often than o der ones). At a part cu ar t me, for examp e, an organ zat on may be try ng to run the adm n strat on of a product nnovat on n the same way they run the product on of housepa nt. But th s w create such tens ons, such c ear rrat ona t es, that f the organ zat on keeps up ts comm tment to the nnovat on t w evo ve a t ghter ntegrat on of eng neer ng, market ng, and product on w th n the nnovat on s adm n strat on, w reward sk more strong y n the nnovat ng subpart of the organ zat on, w or ent ts market ng staff toward market ng add on nnovat ons through regu ar contacts w th former buyers rather than toward ow pr ce and steady supp es, and so on. S m ar y, f cost reduct on s a pr mary organ zat ona mperat ve, we w expect the cost account ng system to grow more rap d y n the d rect on of a more deta ed system of cost codes, of a more f ex b e system of ana ys s of costs, and of a system of f nanc a categor es that are more un que to the organ zat on rather than d ctated by the f nanc a and tax nst tut ons of the soc ety. In short, the postu ate s not that organ zat ons are rat ona , but rather that as organ zat ons become better organ zed, they w tend to mprove the r nformat on and dec s on systems and so become more rat ona . But our fundamenta postu ate here s that the core of organ zat ona rat ona ty s eff c ency n news process ng, not eff c ency n max m za-

349 t on. The reason dr ers are or ented ma n y to other dr ers s not that other dr ers have max m zat on schemes that come to better answers over the ong run, but that other dr ers, espec a y those n the same f e ds, have news about where a dr crew s ke y to get nto troub e. What makes nterna commun cat on such a h gh pr or ty s that on y those other dr ers w have news about what s two m es down the ho e, not that on y they w know what to do about a pocket of h gh-pressure gas or a stratum that tends to co apse. If they can f nd out where the h gh-pressure gas s ke y to be, dr ers sk s w a ow them to m n m ze r sks. And they get the r sk s from the r co eagues on y n the ong run: n the short run they get news from them. L kew se, the observat on that un vers t es wa t for outs de offers to f nd out how much the r facu ty members are worth (J. Thompson 1967) does not mean that the facu ty and adm n strat on of other un vers t es are so much smarter than our own. (For one th ng, th s cou d not be true for a un vers t es.) Rather, t means that facu ty and adm n strators at other un vers t es have no reason to overva ue the reputat ons they are buy ng, whereas the best experts on those reputat ons w th n the f rst un vers ty—name y, the owners of the reputat ons—want to exaggerate them. News about reputat ons of one s own facu ty has to be gathered outs de the un vers ty part y because outs de op n on s prec se y what reputat on cons sts of, but ma n y because the news from ns de s so eas y and rout ne y corrupted. Start ng w th the postu ate that rat ona ty for an organ zat on cons sts ma n y of react ng accurate y to recent uncorrupted news, then, s a d fferent theoret ca or entat on than mag n ng that the ma n source of rat ona ty s the ab ty to max m ze. For examp e, parts of one s own organ zat on are systemat ca y mot vated to prov de d storted news, as when one asks scho ars how h gh the r reputat on s n order to determ ne the r sa ary. The argument of th s book s that one shou d react to th s not on y by ask ng what coa t on n the organ zat on w contro what s actua y max m zed; one shou d a so ask how the organ zat on m ght grow ts nformat on and dec s on structures toward sources of uncorrupted news, so t can n the ong run max m ze what t set out to max m ze. One cou d, of course, use the same approach to exp a n the prob em of nd v dua rat ona ty—that nd v dua s grow competences and nformat on-co ect on strateg es wh ch enab e them to dea w th those uncerta nt es that are most mportant for max m z ng the r ut t es. But a who e fe s too comp cated to deve op such persona competences throughout the range of a mportant uncerta nt es. In contrast, the prob em of man-

350 ag ng the uncerta nt es of a department w th n an organ zat on s often suff c ent y m ted that spec a zed growth of soc a structura competence can pay off. Thus, w th n the department we w expect to f nd that nd v dua competences for co ect ng and react ng to spec a zed uncerta nt es ( .e., sk s) w grow (Chapter 2), and that spec a zed nformat on and dec s on systems n that department w deve op to detect, h re, mprove, and reta n those sk s (Chapter 7). We w a so expect spec a zed departmenta nformat on-co ect on systems to grow (Chapter 3) and author ty to ntegrate nformat on about a market w th d st nct ve uncerta nt es co ected by d fferent departments to be ntegrated by d v s ona organ zat on (Chapter 4). A th s comp ex ty n the nformat on and dec s on system, a the f es, the spec a zed computer programs, the cu tura e aborat on of enc osed departmenta systems, and the ke, s much more than cou d be crammed nto an nd v dua . And t s a about a part of the env ronment—how to dr for o at the bottom of the North Sea (St nchcombe 1985e), how to produce and market passenger a rp anes (Newhouse 1982)—that s much more spec a zed than the who e of an nd v dua s fe. If one mag nes that what peop e and organ zat ons are do ng to be rat ona s so v ng a max m zat on prob em w th g ven data, there s no reason to be eve that an organ zat on shou d do th s ob better than an nd v dua . If one mag nes nstead that rat ona ty s eff c ent process ng of news about uncerta nt es, t s obv ous that organ zat ons are ke y to do t better than nd v dua s, and that nd v dua s are ke y to do t better n the r organ zat ona ro es than n the r ves. Thus, the dea that organ zat ons are more rat ona than nd v dua s s nt mate y t ed up w th the concept on of rat ona ty as process ng and react ng to news about uncerta nt es.

What Makes the Dependent Variables Social Structural? The catch phrase we have used to denote the genera c ass of phenomena we are try ng to exp a n s " nformat on and dec s on system." When we descr bed the d fference between cost account ng systems that co ect data for use n mprov ng a product ve system and f nanc a account ng systems, we were not try ng to exp a n where a the numbers come from that enter nto a part cu ar est mate of how much a g ven mach ne w save, or how those are trans ated nto an "expected t me to payback" us ng assumpt ons of a stab e future market. Instead we were nterested

351 n the fact that the cost account ng system w be used, among other th ngs, for a ser es of ever-chang ng pro ects that nvo ve buy ng many d fferent mach nes and that need d fferent sorts of data, often un que to the p ant. F nanc a accounts have to answer the standard quest ons that tax author t es and cap ta markets rout ne y ask n a way that those markets can trust (e.g., n a way that s aud tab e). What s acc denta and cont ngent about cost account ng numbers s what happens n d stort ng the numbers about a part cu ar mach ne and reso v ng the res dua uncerta nt es n a part cu ar mach ne-buy ng dec s on; what s of the essence n account ng s the d fferent character st cs of the systems respons ve to the uncerta nt es of a ser es of cost-reduc ng "pro ects," as contrasted w th systems or ented to sat sfy ng the author t es and the cap ta market that they w get the r share of respons b y and honest y ca cu ated returns. We have n genera chosen to ana yze soc a structura systems rather than cu tura systems, the soc a re at ons across wh ch nformat on s transm tted and mod f ed rather than the category systems w th wh ch t s ana yzed. Gaye Tuchman (1978), for examp e, ana yzes the soc a structure of nformat on process ng n the news bus ness—the way story ass gnments create f ex b ty, say, or why the names of news sources are def ned as nd v dua newspeop e s pr vate property. But she a so ana yzes the cu ture that newspeop e use to def ne the nature of a story: news stor es may be spot news (a f re), soft news (the prob em of battered ch dren), cont nu ng news (a b s course through Congress), or "what a story" (the assass nat on of a pres dent). Tuchman shows that these categor es do not exact y character ze what s go ng on n the wor d or what the f na story w ook ke; nstead they ref ect the prob ems of schedu ng the work that the f ow of nformat on about a story creates, and n that sense are an examp e of the k nd of ana ys s the theory here wou d generate. The structura ana ys s shows why news organ zat ons wou d need a qu ck system of c ass fy ng stor es by what sorts of work-schedu ng prob ems they w create. The agency prob em of how to get newspeop e to deve op a network of news sources and the requ rement that peop e be reass gnab e to stor es outs de the r spec a ty, somet mes to become subord nate to ower-status peop e who were on the scene f rst, are th ngs of the sort we have been bu d ng theor es of. But the c ass f cat on scheme deve oped n the nforma cu ture of newspeop e s not the sort of var ab e that enters much nto our ana ys s here. The content of the cu ture used to def ne the uncerta nt es that, n our theory, create the structure, the content that sets the nformat on and dec s on structure nto appropr ate mot on, we have eft to one s de. Instead we want to know wh ch d fferent sorts of nformat on for mak ng

352 wh ch d fferent dec s ons the structure has to be ready for, not wh ch dec s ons t does n fact make when var ous uncerta nt es come n. The reason for eav ng the cu ture that s used to ana yze uncerta nt es out of our theory s that that cu ture gets ts stab ty on y from the structure of stab ty of the soc a organ zat on of wh ch t s a part. For examp e, t s on y because the dec s on about what goes nto the newspaper has to be made n t me to pr nt t before t s de vered that the news co ect on prob em s so pervaded by the prob em of tempora organ zat on of nformat on co ect on. Consequent y, the fact so ngen ous y demonstrated by Tuchman, that the concepts used to c ass fy news stor es have to do w th the tempora character st cs of the work that has to be done rather than anyth ng about the resu t ng story or the nature of the event, s from our po nt of v ew to be exp a ned by the soc a structure she a so descr bes. There s of course no contrad ct on n add ng a set of cu tura var ab es to be exp a ned by those soc a structura var ab es we have tr ed to exp a n here. Our on y po nt s that th s s a d fferent ob from the one we have tr ed to ana yze here. To put t another way, we do not ana yze why the structures we have out ned here somet mes make m stakes, why w th the best dr ng crew n the wor d somet mes one st s gets a b owout, why a f rm that stab zes abor re at ons by p ay ng the good-sportsmansh p ncent ve game w th ts mach n sts goes bankrupt even when the system works f ne, why no one was ready to cover John F. Kennedy s assass nat on (Tuchman 1978). These th ngs must be dea t w th n the c ass c stud es of dec s ons, such as A son s ana ys s of the Cuban m ss e cr s s (1971), because when one stud es a part cu ar dec s on one has to exp a n why t came out the way t d d. We have exp a ned not the outcome of part cu ar dec s ons but the structure because we d d not have a theory of error, a theory of why even when the structure was we des gned to dea w th the re evant uncerta nty the wor d st managed to foo t. The advantage of such a strategy s that the soc a structure stays n p ace to be stud ed. When O sen stud es the b ases of an academ c search process, as part of a d agnos s of why un vers t es are "organ zed anarch es" (March and O sen 1976, 82–139), the nterest s focused on the part cu ar causes that made the process dev ate from the dea set forth n the un vers ty s dec s on-mak ng procedure for facu ty appo ntments. But t s very unusua for an organ zat on to have the exact dea from wh ch th s part cu ar process dev ates; most organ zat ons do not co ect reputat ona nformat on from other organ zat ons for h r ng dec s ons. Wh e t s true that facu ty, for examp e, often do not show up for the meet ngs ca ed to take cruc a dec s ons, that s arge y because un vers t es do not

353 put them on the re evant comm ttees un ess they have a good dea of other research work, graduate student superv s on, and cosmopo tan respons b t es that c utter the r schedu es and take them out of town. Why shou d one have an dea that causes t to be mposs b e to ve up to t? Our exp anat on of why un vers t es are organ zed anarch es s g ven n Chapter 9; our argument wou d be on y that organ zed anarchy s a d agnos s of the state of the dependent var ab e: what k nd of nformat on and dec s on system s appropr ate for un vers t es. Br ef y organ zed anarch es are found n un vers t es (but much ess n teach ng co eges or secondary schoo s) because on y outs ders, "peers," can be trusted to g ve uncorrupted nformat on on the cruc a dec s ons of a research un vers ty and on the "reputat on" of peop e t m ght h re or promote. Its nformat on system, then, has to start w th a person s peers ns de the un vers ty, who w know how to co ect nformat on from the re evant peers outs de. S nce a un vers ty can charge a good dea more for ts serv ces f t has a good reputat on, t has to co ect ts nformat on through the peop e who carry the ma n work oad at the owest eve — ts facu ty. Th s nab ty to detach peop e from the r other ob gat ons to do cruc a adm n strat ve tasks means n turn that un vers t es must make the r dec s ons n comm ttees w th on y two-th rds of the r members present, who are at best devot ng on y 5 percent of the r work ng t me to co ect ng the cruc a nformat on for an opt ma dec s on, and who th nk the adm n strators respons b e for the dec s on are an rr tat ng and more or ess use ess excrescence on the body of scho ars. Other organ zat ons, the ones that can get fu -t me adm n strators to come to meet ngs for cruc a dec s ons, do not have to get the nformat on co ected at the owest eve by peop e of the h ghest reputat on as scho ars. Thus, the shape of the errors made n organ zed anarch es, the r genera nab ty to nvade the research space of departments and centers, and the comp cat ons of the r personne systems that bet on future research reputat ons n a w de var ety of f e ds are determ ned by the nature of un vers t es uncerta nt es and by the d ff cu t es of gett ng unb ased nformat on w th n the organ zat on on those uncerta nt es. These shape the k nd of soc a structures needed to process dec s ons, wh ch are more or ess necessar y organ zed anarch es. We have chosen here to theor ze about the soc a structure rather than about the dec s ons because w thout the soc a structura context the errors appear mere y as avo dab e m stakes, weaknesses of the human be ngs rather than "by-products of the best we can do under the c rcumstances." We have prov ded the bas s for ana yz ng the types of errors that may be d st nct ve of d fferent sorts of structures. Just as we expect dec s ons to be made n un vers t es w th on y two-th rds of the re evant

354 peop e at the meet ngs, so we w expect dr ers to pay too tt e attent on to how to save money on spare parts, marketers of software nnovat ons wr tten for a g ven nnovat ve computer to pay too tt e attent on to c ents who have bought a d fferent computer than the computer they are try ng to preserve the monopo y of, and the ke. S m ar y, we have (Chapter 5) g ven grounds to expect, for examp e, that the ways n wh ch the nformat on and dec s on system for product nnovat ons w d ffer from the arger organ zat on n wh ch t s embedded n s m ar ways n most organ zat ons. Further, we have g ven grounds to expect that when th s s not true, the nnovat ons have even ess ke hood of success than nnovat ons genera y do. Because the success of nnovat ons s very uncerta n n the best of c rcumstances, we w expect there to be a ot of errors n dec s ons even when the nnovat ons are dea y adm n stered. Our purpose, however, s to exp a n near y un versa d fferences between nnovat on management and other management, so we have concentrated on exp a n ng the soc a structura requ rements of nnovat on management. We expect there to be s m ar tens ons n a manufactur ng f rms between bu d ng an account ng department to support cost reduct on p ann ng and bu d ng one to cert fy prof tab ty for the cap ta market, as ana yzed n Chapter 3. These tens ons, we argue, w tend to be a feature of the soc a structures of account ng departments n manufactur ng f rms, g ven that requ rements for us ng the same data for d fferent rout ne purposes are genera y ncompat b e. The var ab es to be exp a ned, then, are soc a structura features of nformat on and dec s on systems, because these can be detached from the part cu ar dec s ons be ng made and because t s eas er to genera ze across organ zat ons, and across subparts of organ zat ons, when the va ue of the var ab e one s exp a n ng s of the essence of the organ zat on, rather than an acc dent of the dec s on.

Units of Analysis and Variations to Be Explained within Organizations Each substant ve chapter of the book can be thought of as about a ma n type of soc a structure usefu for dea ng w th part cu ar k nds of uncerta nty: sk s n Chapter 2, manufactur ng subsystems n Chapter 3, autonomous d v s ons n Chapter 4, spec a zed structures for ntroduc ng nnovat ons n Chapter 5, nterorgan zat ona contracts n Chapter 6, personne and abor market systems n Chapter 7, categor es of workers sub ect to un form ncent ves and output measurement systems n Chap-

355 ter 8, and spec a zed scho ar y reputat ona systems n Chapter 9. That s, n each chapter a broad c ass of soc a structures s presented that responds to a d st nct ve amount and k nd of uncerta nty. These d fferent sorts of structures are part y autonomous, as they have to be to do s gn f cant y d fferent obs for the organ zat on. But they can a so operate s mu taneous y and together n determ n ng the deta s of a g ven soc a system. Thus, for examp e, a dr ng superv sor has a sk ed ob (as we ana yzed n Chapter 2) because he must respond to many cont ngenc es and must have a reperto re of responses ready for nstant use. He s n turn embedded n a system that gets most of ts nformat on from w th n the dr ng operat on n that part cu ar f e d, as d scussed n Chapter 3. To exp a n the comp ete soc a structure of a dr ng crew, we need to use the causes and structura adaptat ons to those causes that were out ned n the two chapters: why dr ers are sk ed n a way s m ar to mach n sts (Chapter 2) and why they are at the same t me very d fferent from mach n sts because the news about the uncerta nt es of the r ob does not come, as a mach n st s does, from other parts of the organ zat on than the dr ng crew (Chapter 3). The cost accountant who spec a zes n negot at ons w th n the same f rm about wh ch costs shou d be a ocated to the dr ers, wh ch to manufactur ng operat ons, a so needs to be sk ed, as mach n sts and dr ers are: to remember the precedents and the genera po c es on the matter, to understand the genera purpose beh nd the po c es so that d scret on can be gu ded by those purposes, to know whom to consu t, to make sure a the tra s of paper about a g ven expense end up n the p ace correspond ng to the cost a ocat on, w th the correct s gnatures, and so on. But the genera purposes beh nd the accountant s a ocat on are or ented to taxat on aw and to cap ta account ng pract ce and aw, to the measurement of execut ve performance n a fa r way for both dr ng and operat ons, to tra s of paper that ead from nvo ces from outs de the organ zat on through to the f na cost a ocat on, and n genera to nformat on systems that have to be un form both w th n the organ zat on and, to some extent, across organ zat ons (see Chapter 3). So the same bas c types of features of obs that make dr ers, mach n sts, and accountants sk ed are embedded n very d fferent types of manufactur ng nformat on systems, because the uncerta nt es the sk s respond to come from d fferent p aces. One k nd of sk ed person stens ma n y to other dr ers; another stens to the requ rements of the peop e us ng the mach ned part; the th rd stens to standards that spread throughout the organ zat on, throughout the soc ety, or, n the extreme, throughout wor d cap ta sm. Some themes recur throughout many types of nformat on systems.

356 For examp e, when some set of act v t es cannot eas y be decoup ed because the performance of one depends on the performance of another, because they cannot eas y be separate y measured, and because they have to be carr ed out under t me pressure, the nformat on for the act v t es needs to be processed by the same author ty system and the act v t es need to be sub ect to common ncent ves, because they have common measures. I have ca ed th s, n one part cu ar context, the "decoup ng pr nc p e"—that act v t es shou d be sub ect to separate contracts on y when they can be decoup ed, n the sense of ow nterdependence, separate measurement of performance, or oose schedu ng so that ate and mperfect performance of one does not ead to ate or neff c ent performance of the other (St nchcombe 1985e, 68–71). Th s decoup ng pr nc p e app es not on y to contract ng pract ces for construct on, but a so to wh ch act v t es shou d be adm n stered n a centra zed fash on by a d v s on of a mu t d v s ona company, wh ch shou d be adm n stered separate y by the separate d v s ons of that company (see Chapter 4); to why nnovat ons often have to be adm n stered as m n d v s ons (Chapter 5); to why systems of subcontracts often nc ude h erarch ca e ements (Chapter 6); to why the ma n dec s ons that depend on know edge of the reputat on of a g ven scho ar n h s or her research spec a ty have to be n t ated n the scho ar s own un vers ty department as superv sed by the cha r rather than e sewhere n the un vers ty (Chapter 9). The form of the book suggests that such bas c genera pr nc p es are not suff c ent for a substant ve ana ys s. They w not te us how ntegrated systems respons ve to d st nct ve types of news about uncerta nt es are formed. The centra zat on of d v s ons ana yzed n Chapter 4 s ana yzed there n the ght of the need to coord nate d fferent parts of the organ zat on to respond to a g ven market. The substant ve genera zat on concerns the number of markets and the r degree of uncerta nty; the number of uncerta n markets s used to pred ct the decentra zat on to d v s ons, because that s the way uncerta nty affects d v s ona zat on. In Chapter 5, n contrast, we are nterested n a certa n type of market—the market for an nnovat on—and how that market s connected to eng neer ng, market ng, and manufactur ng n a d st nct ve way n many d fferent k nds of organ zat ons. The argument of Chapter 5 concerns the common ways that d fferent organ zat ons are ke y to respond to the prob ems of the uncerta nty about how to deve op an nnovat on and to manufacture t cheap y so that t w f t ts deve op ng market. Pa nt at Du Pont, wh e not an nnovat on, requ red a separate d v s on that was nterna y centra zed because ts market was d fferent from that of exp os ves, wh ch were a so not an nnovat on. S nce the substance of the uncerta nty of an nnovat on eads to a pecu ar dependence of eng neer-

357 ng on market ng and manufactur ng on eng neer ng, t genera y requ res centra zat on w th n the nnovat ng subpart of the organ zat on and d fferent at on from the rest of the organ zat on. The po nt here, then, s that centra zat on of author ty n subparts of organ zat ons often forms a part of the response to uncerta nty when uncerta nt es are nterdependent. But t p ays qu te a d fferent ro e n determ n ng the resu t ng structure when the uncerta nty comes from, say, the d fference between package goods and tonnage goods than when t comes from adm n ster ng an nnovat on n a f rm most of whose products are not nnovat ons. So a though the decoup ng pr nc p e may be a fundamenta one n exp a n ng where we f nd centra zat on, exact y what ts mp cat ons are for any concrete sett ng var es w th the deta s of that sett ng. One has to be very carefu n mak ng the mapp ng from the components of the theory, such as the decoup ng pr nc p e, to the organ zat on of response to a part cu ar type of uncerta nty, such as d fferent at on of markets, nnovat on, or est mat on of scho ar y reputat on. S m ar y, sk pervades our ana ys s. In Chapter 2 the bas c descr pt on of what sk s t es t to uncerta nt es that have to be dea t w th at the eve of the workers and to the qua f cat ons requ red of workers to produce that response. It appears aga n n Chapter 5 n the exp anat on of why nnovat ng parts of an organ zat on have h gher sk eve s. In Chapter 7, a arge part of the ana ys s of why f rms part c pate the way they do n the abor market has to do w th the r prob em of f nd ng, cert fy ng, reward ng, and mot vat ng sk use and sk acqu s t on. But the personne system uncerta nty n Chapter 7 about who s sk ed s d fferent from the uncerta nty n Chapter 2 that requ res sk ; t has to be so ved w th a d fferent structure of nformat on process ng and dec s on mak ng, one that extends outs de the organ zat on and structures segments n the abor market (Pfeffer 1977). Sk eve s appear aga n as a bas s for the s m ar ty of abor contracts of art sans and other manufactur ng and construct on workers n Chapter 8, and consequent y as a bas s for c ass consc ousness. The uncerta nty that art sans and craft workers have about who s on the r s de, who on the other s de, and how that ana ys s s nf uenced s mu taneous y by the nformat on they get from the po t ca system and the nformat on they get from the ncent ve systems that are bu t nto the r craft abor contracts s ne ther the same uncerta nty as creates the need for the r sk nor the same as shapes the emp oyer s prob em of how to dec de whom to h re. Of course, the ncent ve systems that emp oyers des gn are nf uenced by both sk requ rements and h r ng uncerta nt es, so these can be regarded as u t mate determ nants of art sans c ass consc ousness. Thus, each chapter of the book can be thought of as spec fy ng an n-

358 terre ated set of var at ons n a comp ex of nterre ated dependent var ab es, th ngs to be exp a ned about why some parts of an organ zat on are d fferent from other parts of the same organ zat on, and a so why some organ zat ons are d fferent from other organ zat ons. We have p cked out the dependent var ab e comp exes accord ng to our argument about what sort of uncerta nty they are respons ve to.

Restructuring Research on Organizations The concept on out ned above about what organ zat ona ana ys s ought to be about has some mp cat ons for the pract ce of organ zat ona research. Let me summar ze them under three head ngs: (1) un ts of ana ys s, (2) types of var ab es, and (3) forms of theory.

Units of Analysis By un ts of ana ys s I mean those un ts on wh ch data are typ ca y co ected. That s, shou d one co ect data on nd v dua s, departments, transact ons, organ zat ons, systems of organ zat ons, or what? The centra cr ter on for determ n ng appropr ate un ts of ana ys s s that, w th respect to the theory at stake, they shou d be strong y causa y connected nterna y. The un ts of ana ys s have to be chosen so that they are the p ace where causes are connected to effects—funct ons to needs, dec s ons to s tuat ons, f ows to stocks, structures to env ronments. The centra argument of th s book s that un ts of ana ys s shou d be subparts of organ zat ons that dea w th d st nct ve sorts of uncerta nt es, that are respons b e for secur ng effect ve responses to d fferent sorts of news. Who e organ zat ons may somet mes be prox es for the r ght un ts of ana ys s when they are dom nated by a s ng e type of nformat on-process ng system, as when un vers t es are dom nated by departments n var ous branches of scho arsh p. But even n th s case one gets stronger and c eaner resu ts f one d st ngu shes facu t es from bu d ng and grounds departments. Ord nar y organ zat ons themse ves, because they have had to bu d d fferent substructures to dea w th d fferent sorts of uncerta nty, w themse ves prov de gu des to the r ght un ts. Un vers t es w tend to ocate bu d ng and grounds far from the facu ty n the adm n strat ve system, but w tend to ocate soc o ogy moderate y near to phys cs, because est mat ng the reputat on of a phys c st s the same genera sort of prob em as est mat ng the reputat on of a soc o og st. One has to go to a much h gher eve to f nd a common super or for bu d ng and grounds and soc o ogy than to f nd a common super or for soc o ogy and phys cs.

359 Qu te often, whatever s ca ed a "department" n an organ zat on w be the r ght un t of ana ys s; but f the uncerta nty n quest on has to do w th nvestments or other matters w th a onger t me hor zon, d v s ons or subs d ar es may be the r ght un ts. But s nce we have been argu ng that features of the soc a organ zat on of news process ng are to be the ma n var ab es, we do not want to have un ts that cannot have soc a organ zat on w th a h gh degree of cont nu ty through t me. We therefore do not want to have e ther nd v dua s or dec s ons as the centra un ts of ana ys s. S nce n the theory a soc a structure re ates dec s ons to news about uncerta nt es, the un ts chosen have to be those that t e news to dec s ons. Consequent y, we need to so choose the un ts that they have author ty (w th rev ew, to be sure) over a c ass of dec s ons that s served by a common news-co ect ng structure. S nce author ty s very genera y d v ded n much the same way spec a zed departments are d st ngu shed from each other (we rare y f nd departments that do not make—or at east recommend— mportant dec s ons), th s cr ter on usua y co nc des w th the one that te s us to choose as un ts of ana ys s spec a zed subun ts. But t draws attent on to the vo ume of the f ow of nformat on, short per ods between transm ss ons of nformat on, and the ke, as a cr ter on for what s a un t of ana ys s. Thus, f t s true that some drug compan es have a "v ce pres dent n charge of go ng to a ," who s respons b e for everyth ng affect ng the qua ty of drugs but gets very tt e nformat on about those th ngs and makes very few dec s ons about them (John Bra thewa te, persona commun cat on), we wou d not want the d fferent departments for wh ch he or she s "respons b e" to be a un t of ana ys s for our theory. H s ob s to go to a when ow qua ty s to be pun shed by the courts, not to un fy qua ty contro . Un ts of ana ys s are fundamenta aspects of the strategy of sc ent f c nqu ry and sc ent f c theor z ng. If peop e do not typ ca y ook at the un ts that connect a g ven k nd of uncerta nty to dec s ons, the r theory s un ke y to be about such connect ons. S nce our un ts do not have the natura sk ns that serve as boundar es around the un ts of ana ys s n much of psycho ogy, and do not necessar y have ega ex stence as separate organ zat ons as the un ts n eco og ca organ zat ona theory have, we have to te how to recogn ze those un ts n some deta . That makes the def n t on of un ts of ana ys s tse f nto a theoret ca quest on. Our suggest on that one norma y w not go too far wrong by us ng whatever s ca ed a "department" as a un t of ana ys s s mere y an emp r ca conven ence. But that emp r ca conven ence comes about because of an observat on that connects t to the core of our theory—that usua y departments connect news about a d st nct ve sort of uncerta nty to the author ty requ red to

360 make dec s ons, ust as a th ng w th a sk n s a un t n psycho ogy because t happens that th ngs w th sk ns are what connect mot ves to nd v dua act ons.

Types of Variables The argument of th s book mp es that we need to ook at two broad types of var ab es, one descr b ng the var at ons n the k nds of uncerta nty that affect the un ts of ana ys s, another descr b ng var at ons n the nformat on and dec s on structure of the parts of the organ zat on that form the un ts. For examp e, the serv ce prov ded by a un vers ty facu ty member s more va uab e and has a h gher pr ce f the un vers ty has a good reputat on, a fact that descr bes one type of uncerta nty, about how to est mate reputat on, wh e customer sat sfact on and the capac ty to get the ob done r ght are more mportant n a serv ce ke that of barbers or beaut c ans, a fact that descr bes a d fferent sort of uncerta nty, about adequacy of competence. It s var at ons n th s uncerta nty that gets un vers t es nto the prob em of est mat ng both the present and the future scho ar y reputat on n contract ng for the r facu ty, wh e no such prob ems ar se n contract ng for barbers or beaut c ans. The comp cated structure of peer eva uat on, and the system of pay ng facu ty members for research work that s mon tored not ns de the un vers ty but n the researcher s own f e d, ref ect (or so we have argued) the fact that t s uncerta nty of reputat on that determ nes the pr ce a un vers ty can charge for ts teach ng. We therefore expect peer rev ew when such rent ng of reputat ons s rat ona , because news of reputat ons cannot re ab y be co ected by other means (of course, t s not a that re ab e when co ected by these means e ther). S m ar y, hosp ta s that depend on referra s to the r phys c ans by pr mary-care phys c ans or by other hosp ta s shou d be expected to use peer rev ew and nvestment n future reputat ons of the r staff n much the way un vers t es do. Of course, such hosp ta s are often un vers ty teach ng hosp ta s and so f nd t easy to adopt var at ons of usua un vers ty personne procedures. But, to return to the po nt made above about un ts of ana ys s, we w not expect much peer rev ew n the h r ng of nurses n that hosp ta , or n the bu d ng and grounds department of the un vers ty, because ne ther serv ce becomes much more va uab e by ncreas ng the reputat on of the serv ce g vers. We have g ven many examp es of such var ab es: whether or not the structure of the program govern ng a worker s hand ng of uncerta nty s ana ogous to a batch program or to an nteract ve program, and the correspond ng sk eve of the worker (h gher for nteract ve, ower for

361 batch; see Chapter 2); whether the accounts have to prov de for compar sons among f rms by outs ders and honesty of report ng to those outs ders rather than mater a s to ana yze cost reduct on pro ects, and the correspond ng standard codes versus spec a zed deta and r g d ty versus f ex b ty of the account ng structures (Chapter 3); whether a f rm s n severa markets or a s ng e market ( n a sense descr bed n cons derab e deta n Chapter 4), and the correspond ng d v s ona zat on; whether the subpart of the organ zat on s try ng to ma nta n the monopo y pos t on der ved from hav ng ntroduced a product nnovat on, and the correspond ng network structure connect ng users to the market ng department and the c ose ntergrat on of eng neer ng w th market ng and manufactur ng (Chapter 5). These examp es suggest the generat v ty or fru tfu ness of the theoret ca approach. Many more examp es are scattered throughout the book.

Forms of Theory The def n t on of the var ab es and un ts of ana ys s nvo ve a s mu taneous or entat on of the theory to three ma n doma ns of fact: the sources of uncerta nty outs de the organ zat on; the organ zat ona ob ect ves that make the uncerta nty mportant to the organ zat on; and the vo ume, error, and b as of the f ow of nformat on about the uncerta nty. The theory connects the arger soc a structure and the ava ab e product ve techno ogy to the m crostructure of organ zat ons. But t does not connect every part of the arger structure to every m cropart of the organ zat ona structure. The cruc a nd cator that some uncerta nty outs de the organ zat on s shap ng a part of an organ zat on s that nformat on or news about that uncerta nty s f ow ng through that part and be ng reshaped by that part nto such a form that t can serve as the bas s of a dec s on. The c tat ons to a facu ty member s work n the scho ar y terature are processed nto an overa rank ng of the mpact that scho ar has on h s or her f e d, as requ red by the dean s off ce, that s, departments must demonstrate the d st nct on of a g ven facu ty member, as compared w th a ternat ve cand dates who m ght be h red nstead. Th s nd cates that the nformat on n the sc ent f c or scho ar y commun ty shapes the personne process of the un vers ty. S m ar y, the fact that the we p an for a g ven we s shaped n deta by the we reports for ne ghbor ng we s s what nd cates that dr ers care a ot about the nformat on from ne ghbor ng dr crews and dr ng eng neers, and makes t ess surpr s ng that they stamp Bu sh t! on nformat on from purchas ng on how they m ght save money buy ng spare parts.

362 The theory, then, s about a f ow of nteract ons between the env ronment and the subpart of the organ zat on. Further, the argument s that the ma n th ng that s go ng on n those nteract ons, the part that matters, s what w shape the ma n out nes of the subpart s structure. Somet mes the theory w be about var at ons n what matters n the nteract on. For examp e, the argument of Chapter 5 s that the th ngs that matter when one s try ng to preserve the monopo y advantage got from a product nnovat on are qu te d fferent from those that matter when one s se ng compet t ve products n a stra ghtforward way. Somet mes the theory w be about var at ons n the tempora aspects of the nteract on f ow, as when the nformat on for the genera off ce of a mu t d v s ona f rm s abstracted nto onger-per od measures of performance, because the tempora structure of nvestment and return s s ower than the tempora structure of market var at ons and response (see Chapter 4). Somet mes the theory w be about var at ons n the degree to wh ch one has unana yzed nformat on nd cat ng that everyth ng seems to be go ng a r ght, as when personne systems make great use of sen or ty cr ter a (Chapter 7). But the bas c presumpt on of a the subvar et es of the theory s that t s mass ve f ows of nteract on between an organ zat ona part and the uncerta nt es of the env ronment that shape organ zat ona structure. A the theoret ca structures, then, are not so much about cr t ca events as they are about f ows of nteract on, f ows of nformat on about uncerta nty, f ows of mpacts of uncerta nty on mportant cont nu ng ob ect ves, f ows of outcomes that show that th ngs are (or are not) be ng hand ed by the system. The overa theory of th s book, then, s about a the d fferent sorts of th ngs a f ow of nteract on between an organ zat on and ts env ronment br ngs n, and how that f ow of nteract on affects the structure of work f ow n d fferent subparts of the organ zat on.

363 Notes 6— Organizing Information Outside the Firm: Contracts As Hierarchical Documents 1 Mans e d and o hers apparen y do no pred c h s exp c y They argue o ow ng Arrow and o hers ha he pr va e re urn o he nnova ng rm w genera y be be ow he soc a re urn because s d cu o appropr a e a he re urns rom new n orma on (Mans e d e a 1977) H r ng as a subcon rac or o deve op new echno ogy resu s n he subcon rac or appropr a ng he va ue as s documen ed by Scherer (1964 381) or he re a ons be ween e ec ron cs manu ac urers and a r rame manu ac urers hus subcon rac ng w rare y be used or R&D Scherer (p 372) quo es da a wh ch show ha "on y 4% o he our b on do ars n company-sponsored research and deve opmen per ormed dur ng 1959 was con rac ed o ou s de organ za ons he res was per ormed n-house " o ows ha R&D shou d be n egra ed o ac a e appropr a on The assump on ha pr va e rms never con rac or R&D s mp c n he me hods o Mans e d e a (1977) See a so Teece 1981 where amoun s o n-house R&D n coa gass ca on and o sha e by o compan es are pred c ed on he bas s o a heory o appropr ab y o re urns 2 Scherer (1964 165 178) compares c v an n-house R&D w h weapons deve opmen con rac s 3 See A ch an and Demse z 1972 and Newhouse 1982 On he genera n erpene ra on o c en s and supp ers and he cond ons ha produce such n erpene ra on see Reve and Johansen 1982 4 W amson s argumen (1975) ha more han au hor y--spec ca y he capac y o aud per ormances and ncen ve sys ems based on s a uses and promo ons-- s requ red o make a h erarchy work s based on an ana ys s o S mon s paper (1957b) 5 See Scherer 1964 145-146 No e ha he heory ha pred c s he use o h erarchy ra her han marke s argues on he bas s o compe ve pressures on e c ency These compe ve argumen s do no app y o governmen s excep perhaps n he ong run n war ke per ods The exp ana on o why he Amer can governmen does no deve op s weapons n governmen arsena s does no de ver Med care and Med ca d hrough a na ona hea h serv ce or hrough coun y m ary and ve erans hosp a s and does no bu d s own pub c bu d ngs mus be ound n po ca h s ory Governmen s ha do a hese h ngs w h h erarch es show every ev dence o be ng v ab e n he modern wor d and n he med ca case n an mor a y s a s cs sugges hey are more e ec ve S nce we are here n eres ed ma n y n how h erarch es can be bu w h con rac s we w avo d ry ng o exp a n why he governmen uses con rac s when he econom c argumen s sugges h erarch es wou d be more e c en Perhaps governmen h erarch es are ess e c en or spec a reasons espec a y n he Un ed S a es 6 See Scherer 1964 108-109 a so K dder 1981 121-122 For a s m ar prob em n so ware des gn see A man and S onebreaker 1982 where where we can make work "

s es ma ed ha "perhaps on y a h rd o he o a e or s requ red o ge a arge sys em o he s age

7 A ch an and Demse z (1972) de ne "manage" as "renego a e w h " wh ch makes he same po n 8 See Scherer 1964 308 Scherer (p 159) g ves re urns on con rac s by ype o r sk shar ng or cos uncer a n y and documen s (p 226) he dependence o con rac or w ngness o share he r sks o cos uncer a n y on he amoun o uncer a n y presen 9 See a so Moe e a 1980 324 329 or he cos o hav ng o recru and ra n an unexpec ed y arge number o new workers on a pro ec 10 Three b ds s a rough es ma e based on m sce aneous sources on he cons ruc on ndus ry n severa coun r es 11 Compare he reason ng beh nd he prov s on o Dan sh aw ha con rac ors assume he r sks o cos uncer a n es unknown o he c en s " he con rac or w uncer a n es and can by v r ue o h s pro ess ona know edge es ma e he r sk n advance o some degree" my rans a on rom Kr ener 1976 127

n many cases be n a be er pos on han he c en or m ng he mpac o he cos

12 See K e n Craw ord and A ch an 1978 on oppor un es o "renego a e" he erms o cons ruc on con rac s under way see Kre ner 1976 119 The ab y o change he erms o a con rac a ready n ex s ence due o barga n ng advan ages deve oped dur ng execu on mp es ha "repea ed-p ay" shor - erm con rac s a so wou d no so ve he prob ems nvo ved Th s s why K e n and o hers pred c ver ca n egra on n such c rcums ances 13 See Scherer 1964 111 Wes ern E ec r c ns s ed ha componen s or m ss es be sh pped o hem or measuremen and es ng be ore be ng sen o m ss e s es or assemb y even hough he componen s were be ng made on con rac s be ween he supp ers and he governmen and some pr me con rac ors accep ed a zero-pro overr de o have componen supp ers be subcon rac ors ra her han have hem con rac w h he governmen so he supp ers wou d repor o and coord na e w h hem Bo h ac s nd ca e subcon rac s w h h erarch ca e emen s 14 Kre ner 1976 91-119 has a good accoun o he var e y o h ngs ha are hard or a c en o observe n cons ruc on con rac ng 15 "Second wh e s poss b e o conce ve he ee as be ng d rec y unc ona y dependen on he ac o he agen ra her han be ng dependen on he ou come--as n ee- or-serv ce agency re a ons ra her han a yp ca sa es agency re a on he heory oses much o s n eres because he agen by assump on w choose he ac ha s bes pa d orc ng a par cu ar ac " (Ross 1973) Phys c ans have ess roub e conce v ng o he n eres o such ee- or-serv ce arrangemen s han do econom s s 16 Scherer (1964 99) repor s an r-square o 862 Th s gure s no n agreemen w h ha g ven n Peck and Scherer (1962) repor ng on he same s udy where he repor ed r o 903 g ves an r-square o 815 E her one s ex reme y h gh and bo h show ha con rac or per ormance canno be d s ngu shed rom c en per ormance 17 Gran ck sugges s ha h s Sov e prac ce resu ed n grea er han propor ona ncreases n average wages w h ncreased produc v y per od c campa gns o rev se he ncen ve pay sca es reduced h s creep (1954 83-85 245) For a compar son o Hungar an and Amer can mach ne shop ncen ve sys ems see Burawoy 1979 On ncen ve sys ems as charac er s c o h erarchy see W amson 1975 145 and S nchcombe 1974 123-150 18 W amson has gone some way oward res a ng h s ma n hypo hes s n con nuous orm n W amson 1985

Bibliography Abbott, Andrew. 1988. The System of the Profess ons: An Essay on the D v s on of Expert Labor. Ch cago: Un vers ty of Ch cago Press. A ch an, Armen, and Haro d Demsetz. 1972. "Product on, Informat on Costs, and Econom c Organ zat on." Amer can Econom c Rev ew 62, no. 5 (December): 777-795. A dr ch, Howard E. 1979. Organ zat ons and Env ronments. Eng ewood C ffs, N.J.: Prent ce-Ha . A son, Graham T. 1971. Essence of Dec s on: Exp a n ng the Cuban M ss e Cr s s. Boston: L tt e, Brown. A man, Er c, and M chae Stonebreaker. 1982. "Observat ons on the Evo ut on of a Software System." Computer 15, no. 6 (June): 27-32. A varez, A. 1986a. "A Reporter at Large: Offshore--I." New Yorker (January 20): 34-70. ———. 1986b. "A Reporter at Large: Offshore--II." New Yorker (January 27): 39-83. Apter, Dav d Ernest. 1964. Ideo ogy and D scontent. New York: Free Press. Banf e d, Edward. 1981. Po t ca Inf uence. G encoe, I .: Free Press. Barber, Bernard, and Wa ter H rsch. [1962] 1978. The Soc o ogy of Sc ence. New York: Free Press. Westport, Conn.: Greenwood Press. Barnard, Chester I. 1946. "Funct ons and Patho ogy of Status Systems n Forma Organ zat ons." In Whyte 1946, 46-83. Becker, Gary S. [1957] 1971. The Econom cs of D scr m nat on. 2d ed. Ch cago: Un vers ty of Ch cago Press. ———. [1964] 1975. Human Cap ta . New York: Nat ona Bureau of Econom c Research. 2d ed. Ch cago: Un vers ty of Ch cago Press. Becker, Gary S., and George J. St g er. 1974. "Law Enforcement, Ma feasance, and Compensat on of Enforcers." Journa of Lega Stud es 3, no. 1 (January): 1-18. Becker, Howard S. 1960. "Notes on the Concept of Comm tment" Amer can Journa of Soc o ogy 66 (Ju y): 32-40.

364 Ben-Dav d, Joseph. 1960. "Sc ent f c Product v ty and Academ c Organ zat on n N neteenth-Century Med c ne." Amer can Soc o og ca Rev ew 25: 828-843. (Repr nted n Barber and H rsch [1962] 1978.) ———. 1968-1969. "The Un vers t es and the Growth of Sc ence n Germany and the Un ted States." M nerva 7, nos. 1-2: 1-35. Bend x, Re nhard. 1956. Work and Author ty n Industry: Ideo og es of Management n the Course of Industr a zat on. New York: John W ey; Berke ey and Los Ange es: Un vers ty of Ca forn a Press. Bendor, Jonathan B. 1985. Para e Systems: Redundancy n Government. Berke ey and Los Ange es: Un vers ty of Ca forn a Press. Ben ger, James. 1986. The Contro Revo ut on: Techno og ca and Econom c Or g ns of the Informat on Soc ety. Cambr dge, Mass.: Harvard Un vers ty Press. Bere ther, Har and D., and Jerry L. Sch nger. 1968. Un vers ty Space P ann ng: Trans at ng the Educat ona Program of a Un vers ty nto Phys ca Fac ty Requ rements. Champa gn-Urbana: Un vers ty of I no s Press. B au, Jud th R. 1984. Arch tects and F rms: A Soc o og ca Perspect ve on Arch tectura Pract ce. Cambr dge, Mass.: MIT Press. B au, Peter M. [1955] 1963. The Dynam cs of Bureaucracy: A Study of Interpersona Re at ons n Two Government Agenc es. Ch cago: Un vers ty of Ch cago Press. B au, Peter M., and Marsha W. Meyer. [1956] 1987. Bureaucracy n Modern Soc ety. 3d ed. New York: Random House. B auner, Robert. 1964. A enat on and Freedom: The Factory Worker and H s Industry. Ch cago: Un vers ty of Ch cago Press. Boswe , Terry. 1986. "A Sp t Labor Market Ana ys s of D scr m nat on Aga nst Ch nese Imm grants, 1850-1882." Amer can Soc o og ca Rev ew 51, no. 3: 352-371. Boswe , Terry, and D ane M tsch Bush. 1984. "Labor Force Compos t on and Un on Organ z ng n the Ar zona Copper Industry: A Comment on J ménez" Rev ew, Journa of the Fernand Braude Center 8, no. 1 (Summer): 133-151. Bourd eu, P erre. 1984. D st nct on: A Soc a Cr t que of the Judgement of Taste. Cambr dge, Mass.: Harvard Un vers ty Press. Braverman, Harry. 1974. Labor and Monopo y Cap ta : The Degradat on of Work n the Twent eth Century. New York: Month y Rev ew Press. Brooks, Freder c P. 1975. The Myth ca Man-Month: Essays on Software Eng neer ng. Read ng, Mass.: Add son-Wes ey. Browne, Joy. 1973. The Used Car Game: A Soc o ogy of the Barga n. Lex ngton, Mass.: D. C. Heath. Burawoy, M chae . 1979. Manufactur ng Consent: Changes n the Labor Process Under Monopo y Cap ta sm. Ch cago: Un vers ty of Ch cago Press.

365 ———. 1985. The Po t cs of Product on: Factory Reg mes Under Cap ta sm and Soc a sm. London: Verso-New Left Books. Cap ette, M che e. 1982. "Women n Book Pub sh ng: A Qua f ed Success Story." In Coser, Kadush n, and Powe 1982, 148-174. Cap ow, Theodore. 1964. The Soc o ogy of Work. New York: McGraw-H . Carr-Saunders, A. M., and P. A. W son. 1933. The Profess ons. Oxford: C arendon Press. Carro , G enn R. 1987. Pub sh and Per sh: The Organ zat ona Eco ogy of Newspaper Industr es. Monographs n Organ zat ona Behav or and Industr a Re at ons, vo . 8. Greenw ch, Conn.: JAI Press. Chand er, A fred D., Jr. 1962. Strategy and Structure: Chapters n the H story of the Amer can Industr a Enterpr se. Cambr dge, Mass.: MIT Press. ———. 1977. The V s b e Hand: The Manager a Revo ut on n Amer can Bus ness. Cambr dge, Mass.: Harvard Un vers ty Press. Char ton, Joy C. 1983. Secretar es and Bosses: The Soc a Organ zat on of Off ce Work. Ph.D. d ss., Northwestern Un vers ty. Coase, R. H. 1937. "The Nature of the F rm." Econom ca 4, no. 16: 386-405. ———. 1988. "R. H. Coase Lectures [on The Nature of the F rm ]." Journa of Law, Econom cs, and Organ zat on 4, no. 1: 3-47. Cohen, L nda. 1979. "Innovat on and Atom c Energy: Nuc ear Power Regu at on, 1966 to the Present." Law and Contemporary Prob ems 43, no. 1: 67-97. Cohen, M chae , James G. March, and Johan P. O sen. 1972. "A Garbage Can Mode of Organ zat ona Cho ce." Adm n strat ve Sc ence Quarter y 17, no. 1 (March): 1-25. Cohn, Samue . 1985. The Process of Occupat ona Sex-Typ ng: The Fem n zat on of C er ca Labor n Great Br ta n. Ph ade ph a: Temp e Un vers ty Press. Co e, Robert E., and Pau M. S ege . 1979. Work, Mob ty, and Part c pat on: A Comparat ve Study of Amer can and Japanese Industry. Berke ey and Los Ange es: Un vers ty of Ca forn a Press. Co eman, James S. 1973. The Mathemat cs of Co ect ve Act on. Ch cago: A d ne. Co eman, James S., E hu Katz, and Herbert Menze . 1957. "The D ffus on of a New Drug Among Phys c ans." Soc ometry 20: 253-270. Commons, John R. 1924. Lega Foundat ons of Cap ta sm. New York: Macm an. Converse, Ph p E. 1964. "The Nature of Be ef Systems n Mass Pub cs." In Apter 1964, 206-261. Cooper, Bryan, and T. F. Gaske . 1976. The Adventure of North Sea O . London: He nemann.

366 Coser, Lew s A., Char es Kadush n, and Wa ter W. Powe . 1982. Books: The Cu ture and Commerce of Pub sh ng. New York: Bas c Books. Cou am, Robert. 1977. I us ons of Cho ce: The F-111 and the Prob ems of Weapon Acqu s t on Reform. Pr nceton, N.J.: Pr nceton Un vers ty Press. Crec ne, John P. 1969. Governmenta Prob em So v ng. Ch cago: Rand McNa y. Cyert, R chard M., and James G. March. 1963. A Behav ora Theory of the F rm. Eng ewood C ffs, N.J.: Prent ce-Ha . Dah , Robert A., and Char es E. L ndb om. [1953] 1976. Po t cs, Econom cs, and We fare: P ann ng and Po t co-econom c Systems Reso ved nto Bas c Soc a Processes. New York: Harper and Row. Doer nger, Peter, and M chae J. P ore. 1971. Interna Labor Markets and Manpower Ana ys s. Lex ngton, Mass.: D.C. Heath. Drucker, Peter F. 1946. Concept of the Corporat on. New York: John Day. Ecc es, Robert G., and Harr son C. Wh te. 1988. "Pr ce and Author ty n Inter-Prof t Center Transact ons." In W nsh p and Rosen 1988, S17-S51. Edwards, R chard C., M chae Re ch, and Dav d M. Gordon, eds. 1975. Labor Market Segmentat on: Conference on Labor Market Segmentat on. Lex ngton, Mass.: D. C. Heath. E ster, Jon. 1983. Exp a n ng Techn ca Change: A Case Study n the Ph osophy of Sc ence. Cambr dge: Cambr dge Un vers ty Press. Enge s, Freder ck. [1882] 1935. Soc a sm, Utop an and Sc ent f c. Trans ated by Edward Ave ng. New York: Internat ona Pub shers. Esp ng-Andersen, Gøsta. 1985. Po t cs Aga nst Markets: The Soc a Democrat c Road to Power. Pr nceton, N.J.: Pr nceton Un vers ty Press. Etz on , Am ta . 1975. A Comparat ve Ana ys s of Comp ex Organ zat ons: On Power, Invo vement, and The r Corre ates. New York: Free Press. Farkas, George, and Pau a Eng and, eds. 1988. Industr es, F rms, and Jobs: Soc o og ca and Econom c Approaches. New York: P enum. Fau kner, Robert R. 1983. Mus c on Demand: Composers and Careers n the Ho ywood F m Industry. New Brunsw ck, N.J.: Transact on Books. Ferber, Robert, ed. 1966. The Measurement and Interpretat on of Job Vacanc es. New York: Co umb a Un vers ty Press. F n ay, W am. 1987. "Workers on the Waterfront: Un on, Labor, and Techno ogy n the West Coast Longshore Industry." Iowa C ty, Un vers ty of Iowa. Photocopy. F shman, Kathar ne D. 1981. The Computer Estab shment: The Ins de Story of Amer ca s Most Dynam c Industry. New York: McGraw-H . F tzg bbon, Heather. 1988. "The Bus ness of Sc ence: Emergent Careers n Industr a Chem ca Research." Ph.D. d ss., Northwestern Un vers ty. F ood, John A. 1982. Barr sters C erks: The Law s M dd emen. Manchester: Manchester Un vers ty Press.

367 Freeman, John, G enn R. Carro , and M chae T. Hannan. 1983. "The L ab ty of Newness: Age Dependence n Organ zat ona Death Rates." Amer can Soc o og ca Rev ew 48 (October): 692710. Freeman, R chard. 1976. "Ind v dua Mob ty and Un on Vo ce n the Labor Market." Amer can Econom c Rev ew. Supp ement (Papers and Proceed ngs) 66, no. 2 (May): 361-368. Freeman, R chard, and J. Medoff. 1984. What Do Un ons Do? New York: Bas c Books. Ga e, Duncan. 1983. Soc a Inequa ty and C ass Rad ca sm n France and Br ta n. Cambr dge: Cambr dge Un vers ty Press. Geertz, C fford. 1960. The Re g on of Java. New York: Free Press. ———. 1963a. Agr cu tura Invo ut on: The Process of Eco og ca Change n Indones a. Berke ey and Los Ange es: Un vers ty of Ca forn a Press. ———. 1963b. Pedd ers and Pr nces: Soc a Change and Econom c Modern zat on n Two Indones an Towns. Ch cago: Un vers ty of Ch cago Press. ———. 1973a. "Deep P ay: Notes on the Ba nese Cockf ght." In Geertz 1973b, 412-453. (Or g na y n Deda us 101 [1972]: 1-37.) ———. 1973b. The Interpretat on of Cu tures: Se ected Essays. New York: Bas c Books. G enn, Eve yn Nakano, and Ros yn L. Fe dbert. 1977. "Degraded and Desk ed: The Pro etar an zat on of C er ca Work." Soc a Prob ems 25 (October): 52-64. Go dner, Fred H. 1965. "Demot on n Industr a Management." Amer can Soc o og ca Rev ew 30, no. 5 (October): 714-724. Gou dner, A v n W. 1954. W dcat Str ke: A Study n Worker-Management Re at onsh ps. New York: Harper Torchbooks. Gran ck, Dav d. 1954. Management of the Industr a F rm n the USSR: A Study n Sov et Econom c P ann ng. New York: Co umb a Un vers ty Press. ———. 1967. Sov et Meta Fabr cat ng and Econom c Deve opment. Mad son: Un vers ty of W scons n Press. Granovetter, Mark. 1973. "The Strength of Weak T es." Amer can Journa of Soc o ogy 78, no. 6: 1360-1380. ———. 1974. Gett ng a Job: A Study of Contacts and Careers. Cambr dge: Harvard Un vers ty Press. ———. 1988. "The Soc o og ca and Econom c Approaches to Labor Market Ana ys s: A Soc a Structura V ew." In Farkas and Eng and 1988, 187-216. Green, A bert Edward, and A.J. Bourne. 1972. Re ab ty Techno ogy. New York: W ey Intersc ence. Gr ches, Zv . 1957. "Hybr d Corn: An Exp orat on n the Econom cs of Techno og ca Change." Econometr ca 25, no. 4: 501-522. Ha d, Anders. 1952. Stat st ca Theory w th Eng neer ng App cat ons. New York: W ey. Ha e, Dav d. 1984. Amer ca s Work ng Man: Work, Home, and Po t cs

368 Among B ue-Co ar Property Owners. Ch cago: Un vers ty of Ch cago Press. Harast , M k os. 1978. A Worker n a Worker s State. New York: Un verse Books. He mer, Caro A. 1984. "Organ zat ona and Ind v dua Contro of Career Deve opment n Eng neer ng Pro ect Work." Acta Soc o og ca 27, no. 4. (Repr nted n St nchcombe and He mer 1985, 257295.) ———. 1985a. React ve R sk and Rat ona Act on: Manag ng Mora Hazard n Insurance Contracts. Berke ey and Los Ange es: Un vers ty of Ca forn a Press. ———. 1985b. "A ocat ng Informat on Costs n a Negot ated Informat on Order: Interorgan zat ona Constra nts on Dec s on Mak ng n Norweg an O Insurance." Adm n strat ve Sc ence Quarter y 30, no. 3 (September): 395-417. ———. 1985c. "Subst tutes for Exper ence-Based Informat on: The Case of Offshore O Insurance n the North Sea." In St nchcombe and He mer 1985, 172-224. ———. 1986a. "Produc ng Respons b e Behav or n Order to Produce O : Br ng ng Ob gat ons, R ghts, Incent ves, and Resources Together n the Norweg an State O Company." Bergen, Nor.: Inst tute of Industr a Econom cs. Photocopy. ———. 1986b. Rev ew of The Process of Occupat ona Sex Typ ng by Samue Cohn. Amer can Journa of Soc o ogy 92, no. 3: 763-766. Herskeda , St g, and Frode Kr st ansen. 1987. "O epr ser og Mu g Produks onsstans for Fe t på Norsk Sokke " (O pr ces and poss b e product on ha t for f e ds on the Norweg an cont nenta she f). Notat no. 100. Bergen, Nor.: Inst tute for Industr a Econom cs. H rschman, A bert O. 1963. Journeys Toward Progress: Stud es of Econom c Po cy-Mak ng n Lat n Amer ca. New York: Twent eth Century Fund. Hobsbawm, Er c. 1984. Workers: Wor ds of Labor. New York: Pantheon Books. Hobsbawm, Er c, and Joan W. Scott. 1980. "Po t ca Shoemakers." Past and Present: A Journa of H stor ca Stud es 89 (November). (Repr nted n Hobsbawm 1984, 103-130.) Hochsch d, Ar e. 1983. The Managed Heart: Commerc a zat on of Human Fee ng. Berke ey and Los Ange es: Un vers ty of Ca forn a Press. Homans, George Caspar. 1950. The Human Group. New York: Harcourt Brace Jovanov ch. ———. 1961. Soc a Behav or: Its E ementary Forms. New York: Harcourt Brace Jovanov ch. Homans, George Caspar, and Dav d M. Schne der. 1955. Marr age, Author ty, and F na Causes. New York: Free Press. Huber, R chard. 1985. "Sears and the Mu t -d v s ona Structure." Evanston, I .: Northwestern Un vers ty. Photocopy.

369 Hughes, Everett C. 1958. Men and The r Work. New York: Free Press. Inke es, A ex, and Dav d H. Sm th. 1974. Becom ng Modern: Ind v dua Change n S x Deve op ng Countr es. Cambr dge, Mass.: Harvard Un vers ty Press. Jackson, Robert Max. 1984. The Format on of Craft Labor Markets. Or ando, F a.: Academ c Press. Jaques, E ot. [1956] 1972. The Measurement of Respons b ty: A Study of Work, Payment, and Ind v dua Capac ty. New York: W ey. Jones, Ca v n, Pau B. Sheats ey, and Arthur L. St nchcombe. 1979. Dakota Farmers and Ranchers Eva uate Crop and L vestock Surveys. Ch cago: Nat ona Op n on Research Center. Kahn, Lawrence M., and Peter D. Sherer. 1986. "Race D scr m nat on of Profess ona Basketba ." Urbana: Un vers ty of I no s Industr a Re at ons. Photocopy. Kanter, Rosabeth Moss. 1977. Men and Women of the Corporat on. New York: Bas c Books. Katzne son, Ira, and Ar st de R. Zo berg, eds. 1986. Work ng-C ass Format on: N neteenth-Century Patterns n Western Europe and the Un ted States. Pr nceton, N.J.: Pr nceton Un vers ty Press. Ke ey, Jonathan. 1972. "A Resource Theory of Soc a Mob ty: Modern zat on and Mob ty." Paper presented at the annua conference of the Amer can Soc o og ca Assoc at on, New Or eans. K dder, Tracy. 1981. The Sou of a New Mach ne. New York: Avon; Boston: L tt e, Brown. K e n, Ben am n, Robert G. Crawford, and Armen A. A ch an. 1978. "Vert ca Integrat on, Appropr ab e Rents, and the Compet t ve Contract ng Process." Journa of Law and Econom cs 21, no. 2: 297-326. Knorr, Kar n, R. Krohn, and R chard Wh t ey, eds. 1980. The Soc a Process of Sc ent f c Invest gat on. Soc o ogy of the Sc ences , vo . 14. H ngham, Mass.: Re de . Kocka, Jürgen. 1986. "Prob ems of Work ng-C ass Format on n Germany: The Ear y Years, 1800-1875." In Katzne son and Zo berg 1986, 279-351. Kre ner, Kr st an. 1976. The S te Organ zat on: A Study of Soc a Re at onsh ps on Construct on S tes. D ss., Techno og ca Un vers ty of Denmark. Kusterer, Ken C. 1978. Know-how on the Job: The Important Work ng Know edge of "Unsk ed" Workers. Bou der, Co o.: Westv ew Press. LaBonte, Joanne. 1987. "Cu tura Constra nts on Advert s ng Content, or Th s Bud s for Whom? : A Case Study of Amer can Beer Ads from Co on a T mes to the Present." Evanston, I .: Northwestern Un vers ty. Photocopy. Latour, Bruno. 1980. "Is It Poss b e to Reconstruct the Research Process?: The Soc o ogy of a Bra n Pept de." In Knorr, Krohn, and Wh t ey 1980, 53-73.

370 Lawrence, Pau R., and Jay W. Lorsch. 1967. Organ zat on and Env ronment: Manag ng D fferent at on and Integrat on. Boston: Graduate Schoo of Bus ness Adm n strat on, Harvard Un vers ty. Le dner, Rob n. 1988. "Home Work: A Study n the Interact on of Work and Fam y Organ zat on." In S mpson and S mpson 1988, 69-94. L cht, Wa ter. 1983. Work ng for the Ra road: The Organ zat on of Work n the 19th Century. Pr nceton, N.J.: Pr nceton Un vers ty Press. L eberson, Stan ey, and G enn V. Fugu tt. 1967. "Negro-Wh te Occupat ona D fferences n the Absence of D scr m nat on." Amer can Journa of Soc o ogy 73, no. 2: 188-200. L entz, Bennet P., and E. Burton Swanson. 1980. Software Ma ntenance Management. Read ng, Mass.: Add son-Wes ey. L ndb om, Char es E. 1959. "The Sc ence of Mudd ng Through. " Pub c Adm n strat on Rev ew 19 (Spr ng): 79-88. ———. 1977. Po t cs and Markets: The Wor d s Econom c Systems. New York: Bas c Books. L ndzey, Gardner, ed. 1954. Handbook of Soc a Psycho ogy. Read ng, Mass.: Add son-Wes ey. L pset, Seymour Mart n, Pau Lazarsfe d, et a . 1954. "The Psycho ogy of Vot ng: An Ana ys s of Po t ca Behav or." In L ndzey 1954, 2: 1124-1175. Lockwood, Dav d. 1958. The B ackcoated Worker: A Study n C ass Consc ousness. London: A en and Unw n. Macau ay, Stewart. 1963. "Non-contractua Re at ons n Bus ness: A Pre m nary Study." Amer can Soc o og ca Rev ew 28 (February): 55-66. ———. 1966. Law and the Ba ance of Power: The Automob e Manufacturers and The r Dea ers. New York: Russe Sage Foundat on. Mann, M chae . 1973. Workers on the Move: The Soc o ogy of Re ocat on. Cambr dge: Cambr dge Un vers ty Press. Mannhe m, Kar . 1929. Ideo ogy and Utop a: An Introduct on to the Soc o ogy of Know edge. New York: Harcourt, Brace. Mansf e d, Edw n, John Rapoport, Anthony Romeo, Samue Wagner, and George Beards ey. 1977. "Soc a and Pr vate Rates of Return from Industr a Innovat ons." Quarter y Journa of Econom cs 91, no. 2: 221-240. March, James C., and James G. March. 1978. "Performance Samp ng n Soc a Matches." Adm n strat ve Sc ence Quarter y 23, no. 3: 434-453. March, James G. 1978. "Bounded Rat ona ty, Amb gu ty, and the Eng neer ng of Cho ce." Be Journa of Econom cs , 9, 2 (Autumn): 587-608. ———. 1981. "Dec s ons n Organ zat ons and Theor es of Cho ce." In Van de Ven and Joyce 1981, 205-248. March, James G., and Johan P. O sen. 1976. Amb gu ty and Cho ce n Organ zat ons. Bergen, Nor.: Un vers tetsfor aget.

371 March, James G., and Gu e Sevon. 1982. "Goss p, Informat on, and Dec s on Mak ng." In Sprou and Larkey 1982, vo . 1. March, James G., and Zur Shap ra. 1982. "Behav ora Dec s on Theory and Organ zat ona Dec s on Theory." In Ungson and Braunste n 1982, 92-115. March, James G., and Herbert A. S mon. 1958. Organ zat ons. New York: W ey. Markoff, John. 1975. "Governmenta Bureaucrat zat on: Genera Processes and an Anoma ous Case." Comparat ve Stud es n Soc ety and H story 17, no. 4: 479-503. Marschak, Thomas, Thomas K. B ennan, Jr., and Robert Summers. 1967. Strateg es for R & D: Stud es n the M croeconom cs of Deve opment. Ber n: Spr nger Ver ag. McKenz e, Robert Tre ford, and A an S ver. 1968. Ange s n Marb e: Work ng-C ass Conservat ves n Urban Eng and. Ch cago: Un vers ty of Ch cago Press. Merton, Robert K. 1973a. The Soc o ogy of Sc ence: Theoret ca and Emp r ca Invest gat ons. Ch cago: Un vers ty of Ch cago Press. ———. [1942] 1973b. "The Normat ve Structure of Sc ence." In Merton 1973a, 267-278. (Or g na y pub shed as "Sc ence and Techno ogy n a Democrat c Order." Journa of Lega and Po t ca Soc o ogy .) Meyer, John W., and Br an Rowan. 1977. "Inst tut ona zed Organ zat ons: Forma Structure as Myth and Ceremony." Amer can Journa of Soc o ogy 83, no. 2: 340-363. M ntzberg, Henry. 1973. The Nature of Manager a Work. New York: Harper and Row. Moe, Johannes, et a . 1980. Kostnads ana ysen: Norsk kont nta sokke , de I og II (Cost ana ys s: Norweg an cont nenta she f, parts I and II). Os o: Roya M n stry of O and Energy, 29 Apr . Moore, Barr ngton. 1966. Soc a Or g ns of D ctatorsh p and Democracy: Lord and Peasant n the Mak ng of the Modern Wor d. Boston: Beacon Press. Ne son, R chard R., and S dney G. W nter. 1982. An Evo ut onary Theory of Econom c Change. Cambr dge, Mass.: Harvard Un vers ty Press, Be knap Press. Newhouse, John. 1982. The Sporty Game. New York: Knopf. Nob e, Dav d F. Forces of Product on: A Soc a H story of Industr a Automat on. New York: Knopf. Northwestern Un vers ty Off ce of the Provost. 1986. Northwestern Un vers ty Stat st ca and Facu ty Summar es, 1985-86. Evanston, I .: Northwestern Un vers ty. Okun, Arthur. 1981. Pr ces and Quant t es: A Macroeconom c Ana ys s. Wash ngton, D.C.: Brook ngs Inst tut on. Padgett, John. 1980. "Bounded Rat ona ty n Budgetary Research." Amer can Po t ca Sc ence Rev ew 74, no. 2: 354-372.

372 ———. 1981. "H erarchy and Eco og ca Contro n Federa Budgetary Dec s on Mak ng." Amer can Journa of Soc o ogy 87, no. 1: 75-129. Page, John S., and J m G. Nat on. 1976. Est mator s P p ng Man-Hour Manua . 3d ed. N.P.: Gu f and Western Pub sh ng. Parsons, Ta cott. 1939. "The Profess ons and Soc a Structure." In Parsons 1966, 34-49. ———. 1966. Essays n Soc o og ca Theory. 2d ed. New York: Free Press. Peck, Merton J., and Freder c M. Scherer. 1962. The Weapons Acqu s t on Process: An Econom c Ana ys s. Boston: D v s on of Research, Graduate Schoo of Bus ness Adm n strat on, Harvard Un vers ty. Perrow, Char es. 1981. "Markets, H erarch es, and Hegemony." In Van de Ven and Joyce 1981, 371-386; w th re o nders, 387-406. ———. 1984. Norma Acc dents: L v ng w th H gh-R sk Techno og es. New York: Bas c Books. Pfeffer, Jeffrey. 1977. "Towards an Exam nat on of Strat f cat on n Organ zat ons." Adm n strat ve Sc ence Quarter y 22, no. 4: 553-567. P ore, M chae J. 1975. "Notes for a Theory of Labor Market Strat f cat on." In Edwards, Re ch, and Gordon 1975, 125-150. Powe , Wa ter W. 1978. "Pub sher s Dec s on-mak ng: What Cr ter a Do They Use n Dec d ng Wh ch Books to Pub sh?" Soc a Research 45, no. 2: 227-252. Reve, Torger, and Eg Johansen. 1982. "Organ zat ona Buy ng n the Offshore O Industry." Industr a Market ng Management 11, no. 4: 275-282. Rogers, Everett M., and Jud th K. Larson. 1984. S con Va ey Fever: The Growth of a H gh Techno ogy Cu ture. New York: Bas c Books. Rosenbaum, James E. 1981. "Careers n a Corporate H erarchy." In Tre man and Rob nson 1981, 95-124. Ross, Stephen A. 1973. "The Econom c Theory of Agency: The Pr nc pa s Prob em." Amer can Econom c Rev ew 63 (May): 134-139. Ross , Ino, ed. 1982. Structura Soc o ogy. New York: Co umb a Un vers ty Press. Sabe , Char es F. 1982. Work and Po t cs: The D v s on of Labor n Industry. Cambr dge: Cambr dge Un vers ty Press. Sche ng, Thomas C. 1960. The Strategy of Conf ct. Cambr dge, Mass.: Harvard Un vers ty Press. Scherer, Freder c M. 1964. The Weapons Acqu s t on Process: Econom c Incent ves. Boston: D v s on of Research, Graduate Schoo of Bus ness, Harvard Un vers ty. ———. 1980. Industr a Market Structure and Econom c Performance. 2d ed. Boston: Houghton M ff n. Schroeder, Sandra J., and W am F n ay. 1986. "Interna Labor Markets, Profess ona Dom nat on, and Gender: A Compar son of Laboratory Emp oyees n Hosp ta s and Chem ca -O F rms." Soc a Sc ence Quarter y 67, no. 4: 827-840.

373 Schumpeter, Joseph A. 1942. Cap ta sm, Soc a sm, and Democracy. New York: Harper Torchbooks. Schwartz, Barry. 1975. Queue ng and Wa t ng: Stud es n the Soc a Organ zat on of Access and De ay. Ch cago: Un vers ty of Ch cago Press. ———. 1978a. "Queues, Pr or t es, and Soc a Process." Soc a Psycho ogy 41, no. 1: 3-12. ———. 1978b. "The Soc a Eco ogy of T me Barr ers." Soc a Forces 56, no. 4: 1203-1220. Se zn ck, Ph p. 1949. TVA and the Grass Roots: A Study n the Soc o ogy of Forma Organ zat ons. Berke ey and Los Ange es: Un vers ty of Ca forn a Press. ———. 1957. Leadersh p n Adm n strat on: A Soc o og ca Interpretat on. Evanston, I .: Row Peterson; New York: Harper and Row. Repr nt, 1984. Berke ey: Un vers ty of Ca forn a Press. Se zn ck, Ph p, Ph ppe Nonet, and Howard M. Vo mer. 1969. Law and Soc ety and Industr a Just ce. New York: Russe Sage Foundat on. Sewe , W am, Jr. 1980. Work and Revo ut on n France: The Language of Labor from the O d Reg me to 1848. Cambr dge: Cambr dge Un vers ty Press. ———. 1986. "Art sans, Factory Workers, and the Format on of the French Work ng C ass, 1789-1848." In Katzne son and Zo berg 1986, 45-70. Shack-Marquez, Jan ce, and Ivar Berg. 1982. "Ins de Informat on and Emp oyer-Emp oyee Match ng Processes." Fe s D scuss on Paper, no. 159. Ph ade ph a: Un vers ty of Pennsy van a, Schoo of Pub c and Urban Po cy. Shap ro, Susan. 1984. Wayward Cap ta sts: Target of the Secur t es and Exchange Comm ss on. New Haven, Conn.: Ya e Un vers ty Press. ———. 1987. "Soc a Contro of Impersona Trust." Amer can Journa of Soc o ogy 93, no. 3: 663-658. S mon, Herbert A. 1957a. Mode s of Man, Soc a and Rat ona : Mathemat ca Essays on Rat ona Human Behav or n a Soc a Sett ng. New York: W ey. ———. 1957b. "A Forma Theory of the Emp oyment Re at on." In S mon 1957a, 183-195. ———. [1947] 1976. Adm n strat ve Behav or: A Study of Dec s on-mak ng Processes n Adm n strat ve Organ zat on. 3d ed. New York: Free Press. S mpson, Ida, and R chard S mpson. 1988. Research on the Soc o ogy of Work. Vo . 4: H gh Tech Work. Greenw ch, Conn.: JAI Press. Skocpo , Theda, and Ann Or off. 1983. "Why Not Equa Protect on?: Exp a n ng the Po t cs of Pub c Soc a We fare n Br ta n and the Un ted States, 1880s-1920s." Paper present at the annua conference of the Amer can Soc o og ca Assoc at on. S cher Van Bath, B. H. 1964. The Agrar an H story of Western Europe, A.D. 500-1850. Trans ated by O ve Ord sh. New York: St. Mart n s Press.

374 S chter, Sumner H., James J. Hea y, and E. Robert L vernash. 1960. The Impact of Co ect ve Barga n ng on Management. Wash ngton, D.C.: Brook ngs Inst tut on. Sme ser, Ne J. 1959. Soc a Change n the Industr a Revo ut on: An App cat on of Theory to the Br t sh Cotton Industry. Ch cago: Un vers ty of Ch cago Press. So tan, Karo . 1987. A Causa Theory of Just ce. Berke ey and Los Ange es: Un vers ty of Ca forn a Press. Somers, G., and M. Tsuda. 1966. "Job Vacanc es and Structura Changes n Japanese Labor Markets." In Ferber 1966. Sp erman, Seymour. 1986. "Organ zat ona Ru es and the Features of Work Careers." In Research n Soc a Strat f cat on and Mob ty , 5: 41-102. Greenw ch, Conn.: JAI Press. Sprou , Lee S., and Patr ck D. Larkey, eds. 1982. Advances n Informat on Process ng n Organ zat ons , Greenw ch, Conn.: JAI Press. Stephens, John. 1986. The Trans t on from Cap ta sm to Soc a sm. Urbana: Un vers ty of I no s Press. St nchcombe, Arthur L. 1959. "Bureaucrat c and Craft Adm n strat on of Product on." Adm n strat ve Sc ence Quarter y 4, no. 2: 168-187. (Repr nted n St nchcombe 1986a, 177-195.) ———. 1964. Rebe on n a H gh Schoo . Ch cago: Quadrang e Books; New York: New York T mes Books. ———. 1968. Construct ng Soc a Theor es. Ch cago: Un vers ty of Ch cago Press. ———. 1973 [1967]. "Forma Organ zat ons." In Soc o ogy: An Introduct on , ed ted by Ne J. Sme tzer, 23-65. 2d ed. New York: W ey. ———. 1974. Creat ng Eff c ent Industr a Adm n strat ons. Or ando, F a.: Academ c Press. ———. 1979a. "Ana ys s of Errors." In Jones, Sheats ey, and St nchcombe 1979, 164-198. ———. 1979b. "Soc a Mob ty n Industr a Labor Markets." Acta Soc o og ca 22, no. 3: 217-245. (Repr nted n St nchcombe 1986a, 86-121.) ———. 1982. "The Deep Structure of Mora Categor es, E ghteenth-Century French Strat f cat on, and the Revo ut on." In Ross 1982, 62-95. (Repr nted n St nchcombe 1986a, 145-173.) ———. 1983. Econom c Soc o ogy. Or ando, F a.: Academ c Press. ———. 1984. "Th rd Party Buy ng: The Trend and the Consequences." Soc a Forces 62, no. 4: 861-884. ———. 1985a. "Author ty and the Management of Eng neer ng on Large Pro ects." In St nchcombe and He mer 1985, 225-256. ———. 1985b. "Contracts as H erarch ca Documents." In St nchcombe and He mer 1985, 121-169. ———. 1985c. "De ays n Government Approva s n Norweg an Offshore Deve opment." In St nchcombe and He mer 1985, 296-321.

375 ———. 1985d. "The Funct ona Theory of Soc a Insurance." Po t cs and Soc ety 14, no. 4: 411-430. ———. 1985e. "Pro ect Adm n strat on n the North Sea." In St nchcombe and He mer 1985, 25-120. ———. 1986a. Strat f cat on and Organ zat on: Se ected Papers. Cambr dge: Cambr dge Un vers ty Press. ———. 1986b. "Norms of Exchange." In St nchcombe 1986a, 231-267. ———. 1986c. "On Soc a Factors n Adm n strat ve Innovat on." In St nchcombe 1986a, 221-230. ———. 1986d. "Transform ng Informat on Systems n Organ zat ons: The Norweg an State O Company Beg ns Operat ons." Bergen, Nor.: Inst tute of Industr a Econom cs. St nchcombe, Arthur L., and T. Robert Harr s. 1969. "Interdependence and Inequa ty: A Spec f cat on of the Dav s-Moore Theory." Soc ometry 32, no. 1: 13-23. (Repr nted n St nchcombe 1986a, 58-69.) St nchcombe, Arthur L., and Caro A. He mer. 1985. Organ zat ona Theory and Pro ect Management: Adm n ster ng Uncerta nty n Norweg an Offshore O . Bergen, Nor.: Norweg an Un vers ty Press. ———. 1988. "Interorgan zat ona Re at ons and Careers n Computer Software F rms." In S mpson and S mpson 1988, 179-204. St nchcombe, Arthur L., Ca v n Jones, and Pau Sheats ey. 1981. "Non-Response B as for Att tude Quest ons." Pub c Op n on Quarter y 45: 359-375. Stockf sch, Jacob A. 1973. P owshares nto Swords: Manag ng the Amer can Defense Estab shment. New York: Mason and L pscomb. Stone, Kather ne. 1975. "The Or g ns of Job Structures n the Stee Industry." In Edwards, Re ch, and Gordon 1975, 27-84. Sturmtha , Ado ph. 1953. Un ty and D vers ty n European Labor. New York: Free Press. Teece, Dav d J. 1976. "Vert ca Integrat on and D vest ture n the U.S. O Industry. Stanford: Stanford Inst tute for Energy Research. ———. 1981. "Asset Reva uat ons and the Appropr ab ty of Returns from Invent ve Act v ty." Stanford Bus ness Research Paper no. 425. Stanford. Thompson, E. P. 1963. The Mak ng of the Eng sh Work ng C ass. New York: V ntage Books. Thompson, James D. 1967. Organ zat ons n Act on: Soc a Sc ence Bases of Adm n strat on. New York: McGraw-H . T y, Char es. 1986. The Content ous French: Four Centur es of Popu ar Strugg e. Cambr dge, Mass.: Harvard Un vers ty Press, Be knap Press. To stoy, Leo. [1886] 1982. "The Three Herm ts." In The Ra d and Other Stor es , trans ated by Lou se Maude and Ay mer Maude. Oxford: Oxford Un vers ty Press. Toong, Hoo-M n, and Amar Gupta. 1982. "Persona Computers." Sc ent f c Amer can 247, no. 6: 86-107.

376 Tos , Henry, Ramon A dag, and Rona d Storey. 1973. "On the Measurement of the Env ronment: An Assessment of the Lawrence and Lorsch Env ronmenta Uncerta nty Subsca e." Adm n strat ve Sc ence Quarter y 18, no. 1: 27-36. Tre man, Dona d J., and Robert V. Rob nson, eds. 1981. Research n Soc a Strat f cat on and Mob ty. Vo . 1. Greenw ch, Conn.: JAI Press. Tuchman, Gaye. 1978. Mak ng News: A Study n the Construct on of Rea ty. New York: Free Press; London: Co er Macm an. Ungson, Gerardo, and Dan e Braunste n, eds. 1982. New D rect ons n Dec s on Mak ng: An Interd sc p nary Approach to the Study of Organ zat ons. Boston: Kent. Van de Ven, Andrew, and W am Joyce, eds. 1981. Perspect ves on Organ zat on Des gn and Performance. New York: W ey. Veb en, Thorste n. [1919] 1954. The Eng neers and the Pr ce System. New York: V k ng Press. Vernon, Raymond. 1960. Metropo s 1985: An Interpretat on of the F nd ngs of the New York Metropo tan Reg on Study. Cambr dge, Mass.: Harvard Un vers ty Press. ———. 1966. "Internat ona Investment and Internat ona Trade n the Product Cyc e." Quarter y Journa of Econom cs 80, no. 2: 190-207. Wa der, Andrew G. 1986. Commun st Neo-trad t ona sm: Work and Author ty n Ch nese Industry. Berke ey and Los Ange es: Un vers ty of Ca forn a Press. Wa ker, Char es R., and Robert H. Guest. [1952] 1979. The Man on the Assemb y L ne. Cambr dge, Mass.: Harvard. Repr nt New York: Arno. Wa ace, Anthony F. C. 1972. Rockda e: The Growth of an Amer can V age n the Ear y Industr a Revo ut on. New York: Norton. Wa erste n, Immanue . 1974. The Modern Wor d-System: Cap ta st Agr cu ture and the Or g ns of the European Wor d-Economy n the S xteenth Century. New York: Academ c Press. ———. 1980. The Modern Wor d-System II: Mercant sm and the Conso dat on of the European Wor d-Economy, 1600-1750. New York: Academ c Press. Wa sh, John. 1988. "Techno og ca Change and the D v s on of Labor n Reta Trade: The Supermarket Industry from WWII to Present." Ph.D. d ss., Northwestern Un vers ty. Watson, James D. [1968] 1981. The Doub e He x: A Persona Account of the D scovery of the Structure of DNA. Norton Cr t ca Ed t on, ed ted by Gunther S. Stent. New York: Norton. We ntraub, E. Roy. 1979. M crofoundat ons: The Compat b ty of M croeconom cs and Macroeconom cs. Cambr dge: Cambr dge Un vers ty Press. We tman, Sasha. 1968. Bureaucracy, Democracy, and the French Revo ut on. Ph.D. d ss., Wash ngton Un vers ty.

377 Wh te, Harr son. 1970. Cha ns of Opportun ty: Systems Mode s of Mob ty n Organ zat ons. Cambr dge, Mass.: Harvard Un vers ty Press. ———. 1981. "Where Do Markets Come From?" Amer can Journa of Soc o ogy 87, no. 3: 517-547. Wh t ey, R chard. 1984. The Inte ectua and Soc a Organ zat on of the Sc ences. Oxford: C arendon Press. Whyte, W am Foote. 1948. Human Re at ons n the Restaurant Industry. New York: McGraw-H . ———. 1961. Men at Work. Homewood, I .: Dorsey Press. ———, ed. 1946. Industry and Soc ety. New York: McGraw-H . W amson, O ver. 1975. Markets and H erarch es. New York: Free Press; London: Co er Macm an. ———. 1979. "Transact on Cost Econom cs: The Governance of Contractua Re at ons." Journa of Law and Econom cs 22, no. 2: 233-261. ———. 1985. The Econom c Inst tut ons of Cap ta sm: F rms, Markets, Re at ona Contract ng. New York: Free Press; London: Co er Macm an. W nsh p, Chr stopher, and Sherw n Rosen, eds. 1988. Organ zat ons and Inst tut ons: Soc o og ca and Econom c Approaches to the Ana ys s of Soc a Structure. Amer can Journa of Soc o ogy , Supp ement, vo . 94. Ch cago: Un vers ty of Ch cago Press. Wo fe, Tom. 1979. The R ght Stuff. New York: Bantam. [As c ted by Perrow 1984, 359.] Zetka, James. 1988. "The Hegemon c Foundat ons of U.S. Automob e Industry Deve opment: Market Structures, Workp ace Author ty Structure, and Labor M tancy." Ph.D. d ss., Northwestern Un vers ty. Zuckerman, Harr et. 1977. Sc ent f c E te: Nobe Laureates n the Un ted States. New York: Free Press.

379

Author Index A Abbott, Andrew, 46 , 363 A ch an, Armen, 194 , 204 , 210 , 214 , 220 , 237 -238, 363 , 369 A dag, Ramon, 12 , 376 A dr ch, Howard E., 346 , 363 A son, Graham T., 352 , 363 A man, Er c, 89 , 233 , 237 , 363 A varez, A., 67 , 78 , 81 , 363 Apter, Dav d Ernest, 363 , 365 Arrow, Kenneth, 194 , 236

B Banf e d, Edward, 179 , 363 Barber, Bernard, 363 Barnard, Chester I., 90 , 200 , 224 , 363 Becker, Gary S., 203 , 227 , 264 -265, 363 Becker, Howard S., 204 , 363 Ben-Dav d, Joseph, 187 , 333 , 364 Bend x, Re nhard, 44 , 57 -59, 285 -288, 364 Bendor, Jonathan B., 208 , 364 Ben ger, James, 61 , 130 , 364 Bere ther, Har and D., 321 , 325 , 330 , 364 Berg, Ivar, 264 , 373 B au, Jud th R., 266 , 364 B au, Peter M., 12 , 246 , 364 B auner, Robert, 9 , 67 , 299 -300, 364 Boswe , Terry, 262 , 364 Bourd eu, P erre, 304 -306, 364 Bourne, A. J., 77 , 220 , 367 Braverman, Harry, 307 , 364 Brooks, Freder c P., 212 -213, 216 , 364 Browne, Joy, 307 , 364 Burawoy, M chae , 205 , 226 , 239 , 246 , 285 -287, 295 , 307 , 364 -365 Burke, Kenneth, 27 Bush, D ane M tsch, 262 , 364

C Cap ette, M che e, 254 , 365 Cap ow, Theodore, 203 , 365 Carro , G enn R., 256 , 346 , 348 , 365 , 367 Carr-Saunders, A. M., 262 , 365 Chand er, A fred D., Jr., x , 8 , 13 , 19 , 22 -23, 30 , 56 -57, 61 , 64 , 99 , 100 -151, 155 , 166 , 170 -172, 186 , 190 -191, 346 , 365 Char ton, Joy C., 245 , 365 Coase, R. H., 194 , 198 , 234 , 365 Cohen, L nda, 212 , 365 Cohen, M chae , 113 , 344 , 365 Cohn, Samue , 261 , 365 Co e, Robert E., 179 , 205 , 253 , 258 -259, 265 , 365

380 Co eman, James S., 184 , 365 Commons, John R., 153 , 166 , 181 , 365 Converse, Ph p E., 319 , 365 Cooper, Bryan, 76 , 365 Coser, Lew s A., 365 -366 Cou am, Robert, 69 , 212 -214, 366 Crawford, Robert G., 204 , 219 , 238 , 369 Crec ne, John P., 4 , 366 Cyert, R chard M., 30 , 366

D Dah , Robert A., 194 , 198 , 366 Demsetz, Haro d, 194 , 214 , 220 , 237 , 363 D gest of Educat ona Stat st cs , 254 -255 Doer nger, Peter, 270 , 366 Drucker, Peter F., 100 , 366

E Ecc es, Robert G., 343 , 346 , 366 Edwards, R chard C., 244 , 366 E ster, Jon, 108 , 366 Enge s, Freder ck, 173 -174, 366 Eng and, Pau a, 366 Esp ng-Andersen, Gøsta, 288 , 366 Etz on , Am ta , 250 , 366

F Farkas, George, 366 Fau kner, Robert R., 49 , 366 Fe dbert, Ros yn L., 261 , 367 Fe dman, Arno d S., x , 65 Ferber, Robert, 366 F n ay, W am, 182 , 186 , 261 , 264 , 346 , 366 , 372 F shman, Kathar ne D., 68 , 159 -160, 164 -165, 168 , 189 , 310 , 366 F tzg bbon, Heather, 264 , 366 F ood, John A., 46 , 366 Freeman, John, 256 , 367 Freeman, R chard, 61 , 266 -267, 297 , 367 Fugu tt, G enn V., 254 , 370

G Ga e, Duncan, 288 , 367 Gaske , T. F., 76 , 365 Geertz, C fford, 241 , 367 G bert, W. S., 290 G enn, Eve yn Nakano, 261 , 367 Go dner, Fred H., 169 , 367 Gordon, Dav d M., 244 , 366 Gou dner, A v n W., 93 , 180 -181, 367 Gran ck, Dav d, 69 -70, 180 , 226 -227, 239 , 246 , 286 , 367 Granovetter, Mark, 268 , 367 Green, A bert Edward, 77 , 220 , 367 Gr ches, Zv , 82 , 166 , 181 -182, 367 Guest, Robert H., 64 , 376 Gupta, Amar, 211

H Ha d, Anders, 323 , 367 Ha e, Dav d, 245 , 367 -368 Hannan, M chae T., 256 , 367 Harast , M k os, 286 , 368 Harr s, T. Robert, 8 , 375 Haynes, W am, 140 -141 Hea y, James J., 374 He mer, Caro A., x , x , 16 , 20 , 76 , 81 , 85 -87, 89 , 168 , 187 , 193 , 198 , 211 , 225 , 229 , 248 , 252 , 261 , 264 , 345 -346, 368 , 375 Herskeda , St g, 5 , 368 Heum, Per, x H rsch, Wa ter, 363 H rschman, A bert O., 175 , 368 Hobsbawm, Er c, 277 , 288 , 368 Hochsch d, Ar e, 305 , 368 Homans, George Caspar, 108 , 368 Huber, R chard, 133 , 368 Hughes, Everett C., 303 , 369

I Inke es, A ex, 285 , 369 Inst tute of Industr a Econom cs, x

381

J Jackson, Robert Max, 262 , 267 , 369 Jacques, E ot, 12 , 204 , 369 Johansen, Eg , 194 , 237 , 372 Jones, Ca v n, 319 , 369 , 375

K Kadush n, Char es, 366 Kahn, Lawrence M., 242 -243, 369 Kanter, Rosabeth Moss, 178 , 369 Katz, E hu, 184 , 365 Katzne son, Ira, 369 Ke ey, Jonathan, 252 , 369 Keynes, John Maynard, 1 K dder, Tracy, 68 , 166 , 169 , 187 -189, 214 , 218 , 233 , 237 , 369 K e n, Ben am n, 216 -217, 219 , 221 , 238 , 369 Knorr, Kar n, 369 Kocka, Jürgen, 288 , 369 Kre ner, Kr st an, 216 -217, 219 , 238 , 221 , 369 Kr st ansen, Frode, 5 , 368 Krohn, R., 369 Kusterer, Ken C., 43 , 256 -257, 369

L Labonte, Joanne, 306 , 369 Larkey, Patr ck D., 374 Larson, Jud th K., 165 , 372 Latour, Bruno, 319 , 369 Lawrence, Pau R., 12 , 370 Lazarsfe d, Pau F., 302 , 370 Le dner, Rob n, 70 , 370 Lév -Strauss, C aude, 288 L cht, Wa ter, 302 , 370 L eberson, Stan ey, 254 , 370 L entz, Bennet P., 82 , 370 L ndb om, Char es E., 194 , 198 , 366 , 370 L ndzey, Gardner, 370 L pset, Seymour Mart n, 302 , 370 L vernash, E. Robert, 374 Lockwood, Dav d, 27 , 262 , 274 -311, 370 Lorsch, Jay W., 12 , 370

M Macau ay, Stewart, 194 , 208 , 210 , 226 , 229 , 234 , 370 Mann, M chae , 164 , 370 Mannhe m, Kar , 175 -176, 370 Mansf e d, Edw n, 194 , 236 -237, 270 March, James C., 240 , 370 March, James G., 1 , 30 , 73 , 113 , 240 , 344 -345, 352 , 365 , 370 -371 Markoff, John, 278 , 371 Marschak, Thomas, 4 , 194 , 196 , 208 -209, 215 -216, 371 McKenz e, Robert Tre ford, 305 , 371 Medoff, J., 61 , 266 -267, 297 , 367 Menze , Herbert, 184 , 365 Merton, Robert K., x , 187 , 371 Meyer, John W., 86 , 371 Meyer, Marsha W., 12 , 364 M ntzberg, Henry, 5 , 170 , 371 Moe, Johannes, 209 -210, 217 , 220 , 238 , 371 Moore, Barr ngton, 285 -288, 371

N Nat on, J m G., 217 , 372 Ne son, R chard R., 20 , 30 Newhouse, John, 67 , 166 , 176 , 194 , 205 , 211 , 217 -218, 220 , 237 , 350 , 371 Nob e, Dav d F., 52 -53, 71 , 371 Nonet, Ph ppe, 229 -230, 295 -296, 373 Northwestern Un vers ty Off ce of the Provost, 327 , 371

O Okun, Arthur, 371 O sen, Johan P., 113 , 343 -344, 352 , 365 , 370 Or off, Ann, 281 , 373

382

P Padgett, John, 108 , 114 -117, 371 -372 Page, John S., 217 , 372 Parsons, Ta cott, 208 , 372 Payne, Char es, x Peck, Merton J., 221 , 238 , 372 Perrow, Char es, 53 , 372 Pfeffer, Jeffrey, 357 , 372 P ore, M chae J., 270 , 366 Powe , Wa ter W., 67 , 145 , 366 , 372

R Re ch, M chae , 244 , 366 Reve, Torger, 194 , 237 , 372 R cardo, Dav d, 283 Rob nson, Robert V., 376 Rogers, Everett M., 165 , 372 Rosen, Sherw n, 377 Rosenbaum, James E., 260 , 372 Ross, Stephen A., 221 , 227 , 238 , 372 Ross , Ino, 372 , 374 Rossow, Pau a, x Rowan, Br an, 86 , 371

S Sabe , Char es F., 9 , 21 , 30 , 32 -72, 81 , 120 -122, 169 , 280 , 301 , 304 , 372 Sche ng, Thomas C., 372 Scherer, Fred M., 137 , 194 , 205 , 209 , 215 -217, 219 -222, 225 , 227 -228, 233 , 237 -238, 372 Sch nger, Jerry L., 321 , 325 , 330 , 364 Schne der, Dav d M., 108 , 368 Schroeder, Sandra J., 261 , 264 , 346 , 372 Schumpeter, Joseph A., 20 , 23 -24, 30 , 151 , 152 -193, 296 , 373 Schwartz, Barry, 38 , 373 Scott, Joan W., 288 , 368 Seeman, Me v n, 319 Se zn ck, Ph p, 229 -230, 295 -296, 373 Sevon, Gu e, 371 Sewe , W am, Jr., 54 -55, 277 , 373 Shack-Marquez, Jan ce, 264 , 373 Shap ra, Zur, 371 Shap ro, Susan, 16 , 248 , 373 Sheats ey, Pau B., 319 , 369 , 375 Sherer, Peter D., 242 -243, 369 Sher dan, James, x S ege , Pau M., 179 , 205 , 253 , 258 -259, 365 S ver, A an, 305 , 371 S mon, Herbert A., 1 , 73 , 200 , 214 , 237 , 344 -345, 371 , 373 S mpson, Ida, 373 , 375 S mpson, R chard, 373 , 375 Skocpo , Theda, 281 , 373 S cher Van Bath, B. H., 184 , 373 S chter, Summer H., 374 Sme ser, Ne J., 41 , 175 , 374 Sm th, Adam, 331 Sm th, Dav d H., 285 , 369 So tan, Karo , 296 , 374 Somers, G., 374 Sp erman, Seymour, 251 , 374 Sprou , Lee S., 374 Stat st ca Abstract of the Un ted States , 41 , 130 , 134 , 254 -255, 263 , 266 , 303 -304 Stephens, John, 288 , 374 St g er, George, 203 , 227 , 363 St nchcombe, Arthur L., 8 , 22 , 24 , 67 , 74 -75, 81 , 84 -87, 90 , 96 , 113 , 116 , 138 , 154 , 166 -168, 172 , 178 -179, 187 , 193 , 206 , 212 , 218 , 226 , 228 , 239 , 258 , 272 , 281 , 299 , 319 , 350 , 356 , 369 , 374 -375 Stockf sch, Jacob A., 196 , 208 , 213 , 233 , 375 Stone, Kather ne, 55 , 61 , 299 , 375 Stonebreaker, M chae , 89 , 233 , 237 , 363 Story, Rona d, 112 , 376

383 Sturmtha , Ado ph, 288 , 375 Su van, Arthur, 290 Swanson, E. Burton, 82 , 370

T Teece, Dav d J., 194 , 198 , 233 , 237 , 375 Thompson, E. P., 27 , 40 , 54 -55, 61 , 65 , 274 -311, 375 Thompson, James D., 64 -65, 345 , 349 , 375 T y, Char es, 288 , 375 To stoy, Leo, 291 , 375 Toong, Hoo-M n, 211 , 375 -376 Tos , Henry, 12 , 376 Tre man, Dona d J., 376 Tsuda, M., 374 Tuchman, Gaye, 351 -352, 376

U Ungson, Gerardo, 376

V Van de Ven, Andrew, 376 Veb en, Thorste n, 178 , 343 , 376 Vernon, Raymond, 67 -68, 145 , 183 , 376 Vo mer, Howard M., 229 -230, 295 -296, 373

W Wa der, Andrew G., 285 -287, 376 Wa ker, Char es R., 64 , 376 Wa ace, Anthony F. C., 41 , 376 Wa erste n, Immanue , 184 , 376 Wa sh, John, x , 300 , 324 , 376 Watson, James D., 4 , 6 , 13 -14, 318 , 376 Weber, Max, 205 We ntraub, E. Roy, 1 , 376 We tman, Sasha, 278 , 376 Wh te, Harr son, 299 , 343 , 346 , 366 , 377 Wh t ey, R chard, 335 , 369 , 377 Whyte, W am Foote, 179 , 181 , 377 W amson, O ver, 2 , 163 , 194 , 198 , 226 , 229 -230, 233 , 237 , 239 , 346 , 377 W son, P. A., 262 , 365 W nsh p, Chr stopher, 377 W nter, S dney G., 20 , 30 , 232 , 371 Wo fe, Tom, 53 , 377

Z Za d, Mayer, x Zetka, James, x , 126 , 377 Zo berg, Ar st de R., 369 Zuckerman, Harr et, x , 317 , 333 , 377

385

Subject Index A Account ng, 12 , 15 , 348 ; aud t ng, 94 , 202 , 228 ; cap ta , 77 -78, 86 -87; cost, 62 , 109 -110, 218 -219; cost compar sons, 84 -85, 94 ; cost reduct on, 85 -86, 94 ; f nanc a , 86 -87, 94 Actuar a va ues, 321 Adm n stered pr c ng, 197 , 226 -228, 230 -231 Adverse se ect on, 16 Agency, 15 -16, 201 -203, 322 -324 A r force, 52 -53, 212 -213 A rp anes, 66 -67 Appare ndustry, 68 , 146 App e Computer, 174 , 211 App ances, 67 Arb trat on. See Conf ct reso ut on Arch tectura draw ngs, 328 Arch tecture, 329 -330 Ar zona, 128 Art sans, 33 , 40 -44; n manufactur ng, 53 -57. See a so Craftsmen; Sk ed workers Assemb y nes, 39 -40. See a so Ford sm Ath et cs, 240 -243 Aud tab e performance measurement, 245 Aud t ng. See Account ng Author ty: arb trary, 295 ; author zat ons and, 89 -93; conf ct over, 61 -62; n contracts, 197 , 222 -226; n nvestment n nnovat ons, 176 -179; n abor contract, 200 -201 Automob es, 67 , 295 , 301

B Banks, 313 -315 Barker, James M., 133 Batch computer rout ne, 34 -38, 62 Beer product on, 64 Be Te ephone, 164 , 182 Benef ts, d v s on of, 185 -189 Bett ng, on new know edge, 315 -320 B g f rms, 258 -265 B atera monopo y, 204 B s of quant t es. See Adm n stered pr c ng B rm ngham, 40 Bott eneck factors, 6 Bragg, Lawrence, 4 Budget ng, 89

386 Bu d ngs and grounds nformat on, 328 -329 Bureaucrat zat on, 277 -278, 295 , 298 -301 Bus ness cyc es, 152 -153, 217 Buyers, 128 -131

C Ca forn a, 133 Cap ta equ pment, comm ss on ng of, 92 -93, 225 Cap ta sm, corporat st, 287 -288 Cap ta market, 201 -203. See a so Genera off ce Careers, 185 -189, 206 , 258 -265, 298 -299. See a so Status systems Carneg e, Andrew, 61 Carpenters, 41 , 49 -50 Caveat emptor, 203 Caesar an sect ons, 221 Center for Urban Affa rs and Po cy Research, x Cert f cat on, 17 -18, 27 , 248 -252 Cha enger, 347 Chem ca s, 11 , 69 , 140 -141, 295 Church and K ng mobs, 283 C tat ons, 339 C ass consc ousness, 26 -27, 274 -311; co ect ve abor contracts and, 298 -301; const tut ona sm and, 295 -299; n corporat st cap ta sm, 287 -288; cu ture and, 288 -295; def n t on of, 279 -285; and ndustr a zat on, 276 -278; and ega zat on of suffrage, 278 -279; se ng status symbo s and, 304 -306; n serv ce sector, 301 -308; n Sov et soc et es, 285 -287; and urban zat on, 279 C ass f cat on soc et es, 211 C ergy, 303 C osed corporat ons, 289 Coa m n ng, 69 , 293 Cohorts, n abor markets, 252 -258 Co ect ve barga n ng, 282 , 287 Co eg a ty, 202 Comm ss on ng cap ta equ pment, 92 -93, 225 Comm ss ons, 307 Comm tment, 203 Commod ty ne rat ona ty, 128 -131 Computer ndustry, 68 Conf ct reso ut on, 197 , 199 , 207 , 229 -230 Const tut on, 206 , 295 Construct on ndustry, 8 -9, 41 -43, 49 -50, 65 , 67 , 211 -212; costs, 216 ; s te adm n strat on, 236 Cont nuous exchange, 205 Contracts, 91 , 194 -239; academ c teach ng abor, 337 -338; change orders n, 222 ; f xed pr ce, 222 , 227 ; h erarch a e ements n, 221 -232; dea -type, 195 ; nterdepartmenta , 329 ; abor, 199 -201, 298 -301, 307 , 337 -338; marr age, 203 ; s mu at on of h erarch es, 231 -232. See a so Subcontract ng Corporat st cap ta sm, 287 -288 Cost pr c ng, 197 , 209 , 230 -231 Cost reduct on, 179 -181. See a so Learn ng curve Costs, uncerta nty of, 196 , 207 , 215 -219 Cottage ndustry, 41 Courts, 206 Craftsmen, 41 -49, 120 -121, 250 -251, 261 -265 Cr ck, Franc s, 4 , 318 Cu ture, 288 -295

387

D Data Genera , 187 , 188 Debureaucrat zat on. See Bureaucrat zat on Decentra zat on, 114 -117; def n t on of, 100 -101, 104 -107. See a so D v s ona zat on Decoup ng pr nc p e. See Teamwork Deep-sea d v ng, 81 Defense Department. See Weapons Deference, 305 -306 Democracy, 282 Democrat c party, 302 Dependence, 163 -164 Des gn, pro ect, 87 -89, 221 D s nterestedness, 202 . See a so Zone of nd fference D spute reso ut on. See Conf ct reso ut on D v s ona zat on, 12 , 100 -151; as dependent var ab e, 119 ; as nnovat on, 170 -172; and ntroduc ng nnovat ons, 189 -191. See a so Genera off ce D v s on of benef ts, 185 -189 D v s on of abor, 21 -22, 52 -53 D v s ons. See D v s ona zat on; Genera off ce du Pont, Irénée, 110 , 113 -114 du Pont, P erre, 61 , 109 -110, 113 , 170 Du Pont (company), 20 , 22 -23, 61 , 101 , 109 -119, 150 , 190 , 347 Durant, W am C., 120 -122, 150 Dye product on, 111

E Ecc es astes, 207 , 326 Eff c ency, 57 -58 Effort, 297 E nste n, A bert, 316 -317 E ect on, 202 E ectr c ans, 50 E ectr c ty product on, 66 Eng neer ng adequacy, 180 Eng neers, 46 -48, 52 -53, 63 , 221 Eng and, 187 , 336 En ghtenment, 291 -292 Entrepreneurs, 265 -266 Ep stemo ogy, 315 -320 Exped t ng, 225 Exper ence, w th a product, 213 Expert se, 16 -17 Exp o tat on, 292 -295

F Fam y recru tment, 265 -266 Ferm , Enr co, 317 Feuda sm, 278 F duc ary re at ons, 199 , 201 -203, 312 -315 F restone, 141 F ex b ty, 40 , 83 , 94 , 195 F111, 212 -213 Food ndustry, 66 , 139 , 141 , 143 Ford, Henry, II, 170 Ford sm, 39 -40, 62 -71 France, 187 , 206 Franch se system, 200 , 210 Freeborn Eng shman, 290 -291 Funct ona exp anat on, 2 , 17 -20 Fung b ty, 322 -325 Furn ture ndustry, 67

G Gender, 254 , 259 , 263 , 290 , 292 , 296 , 301 -302, 303 -304 Genera Motors, 11 , 20 , 62 , 64 -65, 101 -102, 120 -126, 144 , 150 , 204 , 301 Genera off ce, 12 , 124 -126 Germany, 52 -53, 187 , 336 Government product on, 237 ; careers n, 258 -265

388

H Haske , Harry, 111 , 190 -191 He copter p ot ng, 81 H erarchy, 194 -239; contract s mu at on of, 231 -232; def n t on of, 199 -207; short, effect on c ass consc ousness of, 306 -307 Homestead str ke, 302 Hong Kong effect, 160 -162 Hosp ta s, 39 , 221 , 295 , 299 , 360 Housepa nters, 42 -43, 49 -50 Hous ng and Urban Deve opment (HUD), 114 -117

I IBM, 10 , 91 , 159 -170, 181 , 183 , 188 , 211 , 310 Ice cream, 143 Idea -type method, 198 , 234 -235 I no s, 128 Imm serat on, 293 Incent ve systems, 197 , 199 , 203 -205, 226 -228 Independent var ab e, n Chand er, 136 -151 Industr a eng neer ng. See Tay or, Freder ck W. Industr a products, 143 -146 Inher tance aw, 202 Innovat on, 6 , 12 , 152 -193, 348 ; doctr ne of, 166 -167; nvestment n, 176 -179; mu t d v s ona structure as, 170 -172; n schoo s, 174 -175, 188 -189; soc a pred ctors of success of, 172 -173; n software, 81 -84, 94 ; zero resources, 168 -170 Inspect on. See Comm ss on ng cap ta equ pment Inspectors, 246 Insurance, 211 , 322 -323 Interact ve computer rout ne, 34 -38 Invent ons. See Innovat on Investment, n nnovat on, 176 -179

J Jacob n sm, 283 Japan, 52 -53, 287 Jur sd ct ons, of occupat ons, 49 -50

K Ka ser-Fraser, 20 Know edge, 318 -320

L Labor contracts, 199 -201; nd v dua z ng, 298 -301 Labor party (Br ta n), 302 Lat n Amer ca, 314 -315 Lawyers, 46 Learn ng curve, 154 -162, 179 -181, 217 -218 Lega persona ty, 199 , 203 -205 L ne management, 64 London Correspond ng Soc ety, 289 Lyd a P nkham, 20

M Mach ne too s, 66 Ma order, at Sears, 127 -134 Ma ne, 131 Ma ntenance, 7 , 63 -64, 75 , 87 -88; prevent ve, 76 -78; n software, 205 , 213 Manchester, 293 Manufactur ng: nformat on systems, 73 -99; and market uncerta nty, 136 -148 Marcos, Ferd nand and Ime da, 183 Market: for cap ta , 201 -203; and h erarchy, 194 -239; networks, 162 -165; uncerta nty, 136 -148 Markov cha n, 128 Marr age contract, 203 Mass product on: and ford sm, 71 ; deo ogy of, 57 -61 Med c ne, 261 . See a so Hosp ta s Method sm, 283 , 289 -290 Methodo og ca nd v dua sm, 19 ; n Chand er s work,

389 109 -119; def n t on of, 101 , 107 -109; un ts of ana ys s and, 359 -360 Monopo y, 152 -193, 204 Montgomery Ward, 20 Mora hazard, 16 Mov e ndustry, 68 Mu t d v s ona organ zat on. See D v s ona zat on

N NASA, 347 Networks, 154 , 162 -165 New Or eans, 131 New York, 183 News, 3 -5, 7 N ches, 82 No se reduct on, 9 -11 Norweg an State O Company (Stato ), 22 , 73 -79 Numer ca contro , 52 -53

O Observab ty. See Performance, measurement of O ndustry, 11 , 22 , 73 -99, 139 , 143 ; and North Sea p atforms, 67 , 224 ; of Norway, 209 -210 (see a so Stato ); uncerta nty n, 4 -5, 14 OPEC, 5 Operat ng nformat on systems, types of, 78 -95 Oppress on, 292 -295 Oregon, 133 Owen te soc a sm, 283

P Pa ne, Tom, 54 , 56 , 291 Pa nters, 42 -43, 49 -50 Parad gms, 335 Par s, 183 Peer cert f cat on, 250 -251 Peer rev ew, 320 Peop e dr v ng, 74 -75 Performance: measurement of, 196 , 199 , 203 -205, 219 -222; pred ct on, 247 ; of work, 240 -248 Petro eum. See O ndustry P card, Freder ck W., 111 P ecework, 307 P ttsburgh P ate G ass, 8 , 145 P ace, Franc s, 56 P ant ng the L berty Tree, 291 -292 P umbers, 50 Po c ng, 91 , 94 Preferences. See Uncerta nty, of spec f cat ons Product nformat on, segregat on of, 139 -143 Product v ty: measurement of, 243 -244; un on workers and, 266 -269 Profess ona s, 43 -49, 221 , 250 -251, 261 -265 Programs for dec s ons, 345 . See a so Rout nes; Sk s Pro ect des gn, 87 -89, 221 Pro ect on of demand, 122 -123 Pub c ty, 202 Pub sh ng, 145 -146

R Raskob, John J., 113 Rat ona ty, 1 -4 Reform of 1832, 278 Re ab ty, 77 Renovat on nformat on, 328 -329 Repa r parts, automob e, 123 -124 Research, 312 -340 Research and deve opment, 12 Respons b ty, 112 -114 Reta shopp ng centers, 127 -134 Revo ut onary po t ca act on, 283 R cardo, Dav d, 282 R sk management, 314 -315 Rout nes, 39 -40. See a so Sk s Rug manufactur ng, 67

390

S Safety, 89 Sa es, 12 , 251 , 304 -307 Scan on P an, 182 , 186 Schedu es, 228 Scho ar y reputat ons, 331 -338; rent of, 332 -333; teach ng product v ty of, 331 -332 Sc ence, 13 -14 Sc ent f c nstruments ndustry, 69 Sears, 9 , 20 , 102 -103, 126 -135, 140 , 150 Segmentat on of abor markets, 240 -273, 300 Sem sk ed workers, 62 -63 Sen or ty, 203 , 204 , 249 , 252 -258 Serv ce market, at Genera Motors, 123 -124 Serv ce sector, 134 -135, 301 -308 Sexua harassment, 245 Shah of Iran, 183 Sheff e d, 40 , 55 Sk ed workers, 52 -53; as ma ntenance workers, 63 -64. See a so Art sans; Craftsmen Sk s, 11 , 21 -22, 25 -26, 32 -72; nformat on on, 240 -273; p ant spec f c, 81 ; rout nes and, 33 -47; and sk ed work, 12 S oan, A fred P., 102 , 122 -126, 150 , 170 -172 Sma f rms, 265 , 306 -308 Soc a sm, Owen te, 283 Soc a structure, 344 -345, 350 -354 Software, 7 , 69 , 187 , 205 ; spec f cat ons of, 212 -213; systems, 81 -84 Sov et Un on, 69 -70, 226 -227, 285 -287 Space adm n strat on, 326 -330 Spec f cat ons, uncerta nty of, 196 , 206 -223 Standard operat ng procedures, 198 -199, 205 -206, 228 -229 Stanford Graduate Schoo of Bus ness, x Stato , 22 , 73 -79 Status symbo s, 304 -306 Status systems, 199 , 203 -205 Stee ndustry, 61 , 136 -139, 145 , 147 -148, 224 , 302 Stock market, 10 Store nventory rat ona ty, 128 -131 Subcontract ng, 218 , 300 -301 Subs d ar es, 121 , 203 Suffrage, 278 -279 Supermarkets, 324 Supp er contracts, 218 , 300 -301

T Tay or, Freder ck W., 37 , 44 , 52 , 57 -62 Tay or sm. See Tay or, Freder ck W. Teach ng, 12 , 17 -18; oads n un vers t es, 333 -338 Teamwork, 196 , 207 , 219 -222, 271 Techn ca uncerta nty, 73 -99, 216 -217 Techno og ca utop an sm, 172 -176, 178 Tempora abstract on, eve s of, 11 -13 Tenure: precar ous, 202 ; secur ty of, 204 . See a so Un vers t es Text e ndustry, 67 TFX, 212 -213 Theor es of organ zat on, 341 -347, 361 -362 T ps, 307 Tokyo, 183 Transact on-spec f c nvestments, 163 Trust, 13 -16, 164 -165, 294 Trusts, 201 -203

391

U Uncerta nty: of costs, 196 , 207 , 215 -219; def n t on of, 4 -7; degree of, 13 ; f duc ary re at ons and, 201 -203; and ford sm, 64 -70; and h erarchy, 222 -223; of abor contracts, 240 -243; market, 136 -148; of sc ent f c research, 315 -320; of spec f cat ons, 196 , 207 -215; techn ca , 73 -99, 216 -217; tota , 146 -149; zone of nd fference and, 200 -201 Un ons, 266 -269, 301 -308 Un ts of ana ys s, 8 -9, 358 -360 Un vac, 160 , 165 Un versa sm, 278 , 289 Un vers t es, 27 -29, 312 -340, 342 -343, 349 , 364 ; nternat ona compars ons of, 187 ; abor contract n, 298 -301; as organ zed anarch es, 352 -353; peer rev ew n, 271 , 320 ; reg strar s off ce n, 36 -37; space nformat on systems n, 327 -330; space prob ems of, 322 -323; and tenure, 16 , 202 , 204 Urban zat on, 279 USSR. See Sov et Un on Utop an sm, techno og ca , 172 -176, 178

V Var ab es, types of, 360 -361

W Wage rates, 281 -282, 292 -295 Wash ngton, 133 Weapons, 68 , 212 -214, 222 , 224 , 227 , 236 -237 We reports, 78 -81 We s, comp et on of, 80 Westm nster, 278 Who esa ng: genera prob em of, 132 -134; reg ona nformat on n, 126 -128 Workers r ghts, 229 -230, 295 -296

Z Zone of nd fference, 200 -201 Preferred C tat on: St nchcombe, Arthur L. Informat on and Organ zat ons. Berke ey: Un vers ty of Ca forn a Press, c1990 1990. http://ark.cd b.org/ark:/13030/ft338nb1zq/