Immutability, Inimitability and Autonomy: The 3 little paradoxes of the Toyota (Re)Production System CMS4 Stream 27: Technology and Power Akbar M. Saeed The Richard Ivey School of Business University of Western Ontario 1151 Richmond St. North London, Ontario Canada N6A 3K7 [email protected]
Phone: (519) 636-3156 Fax: (519) 661-3485 Abhijit Gopal The Richard Ivey School of Business University of Western Ontario [email protected]
Phone: (519) 661-4112
Abstract The Toyota Production System (TPS) has been the focus of much attention in both the academic and practitioner worlds, as it has long been recognized to be the primary source of Toyota’s stellar performance as a manufacturer. However, traditional research approaches have fallen short of rendering the system known, as evidenced by the ambiguous results of would-be imitators. In this paper, we use Actor Network Theory (ANT) and Foucault’s notion of disciplinary power to address three apparent paradoxes of TPS. We then propose that the secret of Toyota’s success may not lie in the system itself, but instead in the system of reproduction that faithfully creates and recreates an entire informational ensemble of human and non-human actants. We conclude that it is the fidelity of this reproduction that has eluded researchers and practitioners alike. Such fidelity is critical in ensuring the continued success of TPS and consequently the exceptional performance of the Toyota Motor Company.
In the global marketplace, the incredible success of the Toyota Motor Company is perhaps unmatched. By most conventional benchmarks, Toyota has repeatedly outperformed its competitors and may in fact be the ‘world’s best manufacturer’ (Liker, 2004). For instance, Toyota’s annual profit at the end of its 2002 fiscal year was larger than the combined earnings of GM, Chrysler and Ford. Accordingly, the markets have handsomely rewarded Toyota. Today, its market capitalization is also higher than the combined market capitalization of the ‘big three’ (Liker, 2004; Spear, 2004a). How does a company so young, in comparison to its bigger and older brothers, come to be dominant in a business that favors longevity and experience? Toyota has overtaken Ford to be the second largest producer of cars in the world and even more surprisingly it has recently surpassed Daimler-Chrysler to become the third-largest North American car producer (Liker, 2004; Spear, 2004a). The Toyota Production System (TPS) has been the focus of the bulk of attention from both academics and practitioners, as it has long been recognized to be a key source of Toyota’s stellar performance as a manufacturer (Adler, 1993; Liker, 2004; Spear, 2004a; Spear & Bowen, 1999; Womack & Jones, 2003; Womack, Jones, & Roos, 1990). However, despite all the research and documentation, despite the apparent transparency of the Toyota Motor Corporation and despite the many attempts at imitation, there have been few companies that have been able to match Toyota’s success (Spear, 2004a). In a landmark article on TPS, the authors claimed to be ‘decoding the DNA of the Toyota Production System’ (Spear et al., 1999). The terms ‘decoding’ and ‘DNA’ not only caught the attention of readers, but also conjured up images of a biological lab experiment. When institutions are the focus of our research, the implication of this level of scientific precision can be problematic and misleading. Even though Toyota has been extraordinarily open about its practices, merely copying them seems not to be enough (Austin & Devin, 2003; Spear, 2004b). Empirical evidence suggests that observers seem to confuse the tools and practices they see on their plant visits with the system itself (Spear & Bowen, 1999). Researchers have realized that the system is more than just the manufacturing method; it is a combination of philosophy, processes, people and problem solving behavior (Liker, 2004). If we consider that TPS could be ‘a new paradigm in manufacturing or service delivery - a new way of seeing, understanding and interpreting’ (Liker, 2004: 24), then we would suggest that conventional research approaches may be ill-equipped to unravel the power of such a system. More specifically, we contend that the inherent complexities of TPS are underappreciated within traditional research perspectives that assume an ontological separation between the social and the technical. In this paper, we simultaneously consider both the social and the technical aspects of TPS. We begin by using the lens of Actor-Network Theory (ANT) to reveal the many heterogeneous elements that constitute the system (Latour, 1999a). From this perspective, we are able to recast our conception of TPS and then use this understanding to address the paradoxes of immutability and inimitability. We then employ Foucault’s notion of disciplinary power (Foucault, 1977) to address the paradox of autonomy. We show how the worker is actually located in a very disciplined way by the heterogeneous elements of the system. Through this conceptualization, we propose that the secret of Toyota’s success may not be in the Toyota Production System itself but in the Toyota system of re-production that faithfully creates and recreates an entire informational ensemble of human and non-human entities. We will conclude that it is the fidelity of this reproduction that has eluded researchers and practitioners alike. Such fidelity is critical in ensuring the continued success of TPS and consequently the
exceptional performance of the Toyota Motor Company. In the next section, we depict TPS through the lens of ANT.
Actor-Network Theory ‘Boeing 747s do not fly, airlines fly'1 Actor-Network Theory forces us to shift our natural frame of reference, in order to allow us to perceive those networks of relations that constitute our world. With this lens, we are able to appreciate that it is not the Boeing 747 that flies us but the airline itself, with its entire ensemble of willing actants2. Purposeful action becomes the property of institutions and not of objects or humans (Latour, 1999). ANT allows us to bring technical objects and people into a process of reciprocal definition in which objects are defined by subjects and subjects defined by those objects with which they have relations (Akrich, 1992). The social and the technical are therefore established simultaneously, mutually constituting each other (Bijker and Law, 1992). It is through such movement that the network emerges; it is not given a priori (Latour, 1992). Using this relational perspective, we can focus on the emergence, construction and maintenance of such networks. We can also become more attuned to how networks compete with other networks and how they are made more durable over time (Latour, 1996). With this brief introduction to ANT, we can now depict the network of relations that constitute what we objectively label as TPS. The Toyota Production System The conception of TPS as a network of relations allows us to bring both human and non-human actants into focus. Most importantly, we can consider the non-human actants as full-fledged participants in the collective, an important condition in maintaining a symmetrical approach (Latour, 1999b), whereby human and non-human are given equivalent importance. This conceptualization seems to be aligned with Toyota’s own perspective, as evidenced by some of their institutional practices. For instance, Toyota chooses to keep much of the critical machine design functions in-house, unlike many of its North-American competitors (Toyota, 1998). According to ANT, the design engineer has a critical role in inscribing non-human actants with a defined range of competencies (Akrich, 1992). Such competencies, in turn, define the nature of the exchanges in which an actant is able to engage. Therefore, control over the inscription process is crucial to the continued sustainability of TPS as the actants, human and non-human, define the framework of action in which they act (Akrich & Latour, 1992). The machines, having been recognized as co-producers in the system, are also treated as such. They are not abused but instead treated with respect (Hayes, 1981). Those machines with long service records are used with even greater care than are new machines (Ohno, 1988). The TPS principle of muri recognizes natural limits of both human and machine (Toyota, 1997), contradicting conventional thinking that values the new and often discards or discounts the old (Hayes, 1981). 1 2
We use the term actant to denote both human and non-human elements of the network
Within the TPS system, both sets of actants enjoy a highly linked co-existence. Production machines are equipped with a device that automatically stops operations when an abnormality is detected (Toyota, 1998). Toyota calls this autonomation, one of two core TPS principles (Ohno, 1988). The machines are therefore invested with both intelligence and agency. As ANT would suggest, this approach has resulted in a change in the nature of the relationship among human and machine. Machines assume the responsibility of monitoring their own production and thereby alerting the worker if there is a problem. This allows the worker to concentrate on other activities that can more directly contribute to major productivity gains (Ohno, 1988). Consequently, one worker is able to oversee the production of several machines, as each machine is trusted to perform as inscribed.
The Paradoxes A paradox refers to a statement that appears self-contradictory, though possibly well-founded or essentially valid (Quine, 1976). Although the term has been used in many different ways, it invariably involves the notion of polar opposite conditions that can simultaneously exist, or can at least be potentiated, within the same thing (Mick & Fournier, 1998). In this paper, it is this idea of opposites that we wish to engage. We begin by briefly describing the opposing and often perplexing views that constitute each of the paradoxes. We contend that such paradoxes may often emerge as a result of traditional sociological perspectives that fail to fully consider the contribution of the nonhuman (Latour, 1992). We will now proceed to show how an ANT conceptualization of TPS, as a network of heterogeneous elements, might help us unravel each paradox. The Paradoxes of Immutability and Inimitability In the investigation of TPS, researchers have assumed a certain degree of immutability in the system even though there is much evidence that such systems are highly mutable and socially constructed (Liker, 2004). Coming from a mainly functionalist paradigm, researchers search for regularities, relationships and structures. These structures are often taken as objective phenomena, constituted quite independently of the subjects with whom they interact. The apparent structure is seen as shaping the activities of organization members in fairly deterministic ways. As a result of all this aggregated scientific knowledge, the researchers have determined, codified, controlled and represented what they believe to be TPS. Often such modes of investigation result in the belief that the essence of TPS can be captured in particular rules and that the mastering of this rigid specification is the very thing that will make TPS-like flexibility and creativity possible (Spear & Bowen, 1999). Unfortunately, the process of ‘objectification’ usually results in a false sense of control. The produced manuscripts can become prescriptive devices that serve to instruct companies how to achieve TPS-like production. These ‘immutable mobiles’ serve as combinable textual representations of phenomenon that are otherwise quite mutable, immobile and distinct (Latour, 1987a). Companies that have used such texts to imitate TPS have achieved ambivalent results, thereby leading to a rather paradoxical conclusion. Either the system is actually inimitable and therefore any reproduction will necessarily be inadequate. Or, the other possibility is that the representation itself is somewhat unhelpful, lacking and deficient. Functionalist scientists would choose the latter possibility as uj stification for more research, thereby maintaining the belief that a good representation can eventually be
found. However, in practice, the system seems to be quite mutable and therefore rather resistant to being captured in such a canonical fashion. Using the ANT lens, we can now begin to see the core issue. As a network of heterogeneous elements, TPS is constantly in flux. If we understand ‘being’ to be strictly composed of fluid, discontinuous and ever-changing processes , then entities are nothing but arrested movement (Bergson, 1998). Any attempt to capture TPS will be a snapshot at a particular point in time and therefore may not adequately represent the system itself. Indeed, researchers have become sensitive to this reality, realizing that Toyota’s success is not in the tools and tactics (what is visible) but instead they have focused on the application of the underlying principles (Spear, 2004b). We will take this notion further by saying that it is the faithful reproduction of the system, by both human and non-human elements, that leads to its success. As a techno-economic network (TEN), TPS is represented by a coordinated set of heterogeneous actants that interact to develop, produce and distribute methods for generating goods and services (Callon, 1992). TPS becomes strongly convergent when it is faithfully reproduced by those enrolled actants that ‘put it into play’. It is this network characteristic of TPS that leads researchers to falsely assume the immutability of the system. However, the network can be seen as a shifting system of alliances and exchanges among actants formed out of the aggregation of many underlying actor-networks (Underwood, 1998). Such highly complex systems often contain a vast number of heterogeneous elements that are not easily captured using conventional methods (Gopal & Prasad, 2000). The Paradox of Autonomy Even though TPS has a high degree of specification and structure, the system seems to not result in the ‘command and control’ environment that would be expected (Spear & Bowen, 1999). To further understand this paradox, we invoke the work of Foucault on power (Foucault, 1977). Foucault does not see power as the possession of some who exert it over others, thereby dominating and constraining them. Instead, he sees power as relational and productive. In particular, he sees modern disciplinary power as employing the panoptic gaze to produce the self-controlling subject. This form of sophisticated surveillance has effectively taken the place of more traditional forms of bureaucratic control (Barratt, 2002). TPS includes certain elements that promote the panoptic gaze. For instance, kaizen is a core principle of TPS that requires the employees to continuously make improvements in their work environment. It is a highly monitored activity that allows the improvement related efforts of an employee to be measured quantitatively. Management, which no longer needs to monitor the employee, instead closely monitors such numbers (Spear, 2004b). This form of control, control from a distance, allows the employee to experience a high level of perceived autonomy in the structuring of their work. We contend that the many practices of TPS enable the methodical disciplining of the time and space in which the employee conducts his/her work, thereby locating the employee in very controlled ways. For instance, the tempo of the factory is determined by the speed with which the automobile moves through it. This pace of production is matched with the rate of customer demand through the monitoring of takt time, a principle of TPS that means meter or rhythm in German. Management is able to vary the rate of production, as needed, rather than maintaining any particular given rhythm. Through heijunka, the workload is evened out in a very controlled way, thereby
eliminating the traditional unevenness in the production schedule (Liker, 2004). Also, spatial control is achieved through the specification of highly structured production paths. Every process of production depends on and is depended on by another process. ‘When operations are linked together, there is more teamwork, rapid feedback on earlier quality problems, control over the process, and direct pressure for people to solve problems and think and grow’ (Liker, 2004: 101). Problems become visible immediately, due to the continuous flow and the lack of excess inventory. Then, the ‘scientific method’ becomes the institutionalized way to deal with such problems (Spear et al., 1999). Perhaps, it is this possibility of making the hidden visible that has intensified the managerial gaze at Toyota and led to the creation of the self-controlled employee. Consequently, space and time are disciplined even as autonomy appears to be the order of the day. The Toyota System of Re-production In the information systems field, black-boxing is understood to mean the analysis of inputs and outputs with little to no regard for internal complexity. ANT defines blackboxing as the way that scientific and technical work is made invisible by its own success (Latour, 1987a, 1999b). A network becomes durable when actants feel no need to spend time opening and looking inside the black boxes they encounter, but just accepting them as given (Tatnall & Gilding, 1999). Thereafter, the joint production of actants and artifacts becomes entirely opaque (Latour, 1999b). However, when the network becomes unreliable or unstable, these black boxes are opened and their contents reconsidered (Callon, 1986). Such instability could be induced by the addition of new actants, the desertion of existing actants, or changes in the alliances among actants. Closing the black box of disputes allows people to take the work of others as a resource and move on, rather than continually reproducing and questioning it (Latour, 1987b). There is certain efficiency in this process. At Toyota, the basic structural unit of the production team is 5-7 employees with a leader. There are about four teams in a group, with a group leader (Adler, 1993). These cells are the equivalent of black boxes within the production operation. The teams are charged with the responsibility of delivering a particular output based on a particular input. They have the flexibility to deliver in the way they please. Management treats these cells as black boxes by not interfering with the innovation that occurs within them (Ohno, 1988). Eventually, black boxes combine with other black boxes to form even bigger black boxes and the network begins to stabilize. As a result, black boxes become dependent on other black boxes both for their proper functioning and for the proper functioning of the network (Callon, 1986). The communication amongst such black boxes is accomplished through intermediaries. An intermediary is anything that passes between actants and inevitably defines the relationship between them (Bijker & Law, 1992; Callon, 1992). They become the language of the network and the way that intentions are translated. In TPS, the kanban plays such a role. This system , that took ten years to establish (Ohno, 1988), serves as a means of conveying production order information, as it is tied through synchronization not only to each production process within Toyota but also to the production processes of cooperating firms. It always accompanies goods and is therefore the critical communications tool for the just-in-time approach. It links each process organically to the preceding and following ones as well as being linked to real demand. As an actant in TPS, the kanban is enrolled as a key element.
Now that we have considered the network that is TPS, we turn our focus towards understanding the relative strength of that network. ANT has often been referred to as the sociology of translations (Callon, 1986; Law, 1992; Tatnall et al., 1999). The operation of translation is triangular as it involves a translator, something that is translated and a medium in which that translation is inscribed (Callon, 1992). When translation and its circulation of intermediaries leads to agreement, the network is said to be highly convergent. Convergence can be thought of in terms of alignment and coordination (Callon, 1992). In an aligned network, actants speak in the same way about themselves, about one another and about the intermediaries that link them (Callon, 1992). They engage in successful translations and thereby generate a shared space. At Toyota, the ‘scientific method’ is inculcated in employees (Spear & Bowen, 1999). It aligns. For instance, employees are trained to ask the ‘why’ question five times to get to the root of a problem (Ohno, 1988). Unsuccessful translation means that players are no longer able to communicate and they often reconfigure themselves in separate spaces with no common measure. Sometimes there is conflict and the translation is rejected as betrayal. Translations thus flow through, and are held in place, by intermediaries. A strongly aligned network is one in which translations are successful and relatively similar. The network becomes normalized. Normalization makes links predictable, limits fluctuations, aligns actants and intermediaries, cuts down the number of translations and the amount of information in circulation (Callon, 1992). At Toyota, every connection must be standardized, direct and unambiguously specified (Spear & Bowen, 1999). Actants understand the kanban and consequently all other actants with which they interact. The reduction in ambiguity allows little room for unsuccessful translations among actants due to the highly specified nature of the interactions. In fact, TPS uses codifying regulations to limit the number of translations that can be easily stabilized. These codifying regulations lead to a higher degree of co-ordination, as employees quickly come to ‘know-what’ they are supposed to be doing and also they come to ‘know -how’ they are supposed to do it. In this way, the rules of coordination become constraining norms that create and control any attempt at deviance. The universe of possible translations is relatively restricted, and therefore network behavior also becomes relatively predictable. The boundaries of a network are determined by the effect an element has on the network’s level of convergence. An element can be treated as lying outside the network if it weakens the convergence of the network when moved into it. Suppliers are linked through kanban to the manufacturing plant and therefore lend stability to the TEN by falling within its boundaries. Perhaps not so Incidentally, the Japanese believe that the objective of all partnerships is a mutually beneficial long-term relationship, what many Japanese companies refer to as co-destiny (Hayes, 1981). The next step in network evolution is punctualization. As the network becomes strongly convergent, it is punctualized into a black box. Its behavior becomes relatively known and predictable. It may then link itself to either one or more external actornetworks with which it can exchange intermediaries. Under such circumstances it is punctualized within such networks as it is converted into a single point or node in another network. From this perspective, TPS can be said to depend on the successful punctualization of networks. In the factory, the distinct work areas are punctualized into black boxes within the production line which in turn become punctualized within the larger factory network. The entire factory then becomes punctualized into a black box within the greater production system. The solidity of such a system results from a structure where each point is at the intersection of two networks , “one that it simplifies and another that simplifies it” (Callon, 1987: 97). Thus, the power of this network lies in the predictable and consistent behavior of its component black boxes.
Conclusion Economists study technical networks, sociologists study social networks and management academics study both. However, few consider the network as a composite of humans and non-humans. From an ANT perspective, we investigated three apparent paradoxes of TPS and thereby tried to show that the Toyota system of re-production is in fact more essential to Toyota’s success than is TPS itself. Through the fidelity of such a reproduction, TPS is able to sustain a continued success. The higher the network convergence, the more its actants work together and the less their status as actants is in doubt. Each actant in a convergent network is able to identify and mobilize the skills within that network without having to get involved in costly adaptation, translation or decoding. As a result, the network operates more efficiently. Conversely, weakly convergent networks may render the actants status in question and make it more difficult to mobilize other parts of the network. By describing TPS in these terms, we hope to take account of the mutability and inimitability of the system. TPS is far more complex than can be captured in a set of rules, procedures and practices. There are multiple dimensions to the system. There are multiple types of actants enrolled in a dynamic and fluid network that cannot be captured by simply studying the system at a given point in time. Perhaps, this is why many companies have found it exceedingly difficult to implement such a system. Decoding the DNA of Toyota does not necessarily mean that you can easily replicate it (Spear, 2004b). Strongly convergent networks only develop after long periods of investment, intense effort and coordination (Callon, 1992). And Toyota seems to understand this very well: ‘The key to the Toyota Way and what makes Toyota stand out is not any of the individual elements…But what is important is having all the elements together as a system. It must be practiced every day in a very consistent manner – not in spurts’3
(Fujio Cho, President of the Toyota Motor Company as quoted in Liker, 2004)
Adler, P. S. 1993. Time-And-Motion Regained. Harvard Business Review : pp 97-108. Akrich, M. 1992. The De-Scription of Technical Objects, Shaping Technology/Building Society: Studies in Sociotechnical change: 205-224. Akrich, M., & Latour, B. 1992. A Summary of a Convenient Vocabulary for the Semiotics of Human and Nonhuman Assemblies, Shaping Technology/Building Society: Studies in Sociotechnical change: 259-264: The MIT Press. Austin, R., & Devin, L. 2003. Artful Making: What Managers Need to Know About How Artists Work: Financial Times Prentice Hall. Barratt, E. 2002. Foucault, foucauldianism and human resource management. Personnel Review , 31(1/2): 189. Bergson, H. 1998. Creative Evolution (A. Mitchell, Trans.). New York: Dover Publications. Bijker, W. E., & Law, J. (Eds.). 1992. Shaping Technology/Building Society: Studies in Sociotechnological Change. Callon, M. 1986. Some elements of a sociology of translation: domestication of the scallops and the fishermen of St. Brieuc Bay. In J. Law (Ed.), Power, Action and Belief: 196-233. London: Routledge and Kegan Paul. Callon, M. 1992. Techno -economic networks and irreversability. In J. Law (Ed.), A Sociology of Monsters: Essays on Power, Technology and Domination: 132161. Foucault, M. 1977. Discipline and Punish: The Birth of the Prison. New York: Vintage Books. Gopal, A., & Prasad, P. 2000. Understanding GDSS in symbolic context: Shifting the focus from technology to interaction. MIS Quarterly, 24(3): 509-547. Hayes, R. 1981. Why Japanese factories work. Harvard Business Review, 59(4): 56-66. Latour, B. 1987a. Science in Action: How to Follow Engineers and Scientists Through Society: Open University Press. Latour, B. 1987b. Science in Action: How to follow scientists and engineers through society. Stratford, England: Open University Press. Latour, B. 1992. Where are the Missing Masses? The Sociology of a Few Mundane Artifacts, Shaping Technology/Building Society: Studies in Sociotechnical change: 225-258: MIT Press.
Latour, B. 1999a. On Recalling ANT. In J. Hassard (Ed.), Actor Network Theory and After: 15-25. Oxford: Blackwell Publishers / The Sociological Review. Latour, B. 1999b. Pandora's Hope: Essays on the Reality of Science Studies. USA: Harvard. Law, J. 1992. Notes on the Theory of the Actor-Network: Ordering, Strategy and Heterogeneity. Systems Practice, 5(4): 379-393. Liker, J. 2004. The Toyota Way. New York: McGraw-Hill. Mick, D. G., & Fournier, S. 1998. Paradoxes of technology: Consumer cognizance, emotions, and coping strategies. Journal of Consumer Research, 25(2): 123. Ohno, T. 1988. Toyota Production System: Beyond Large Scale Production: Productivity Press. Quine, W. 1976. The Ways of Paradox and other Essays: Harvard University Press. Spear, S. 2004a. The Toyota Production System: Rules for Activity, Connection, and Pathway Design and Improvement. Spear, S., & Bowen, H. K. 1999. Decoding the DNA of the Toyota Production System. Harvard Business Review, 77(5): 96. Spear, S. J. 2004b. Learning to Lead at Toyota. Harvard Business Review , 82(5): 78. Tatnall, A., & Gilding, A. 1999. Actor-Network Theory and Information Systems Research. Paper presented at the 10th. Australasian Conference on Information Systems. Toyota. 1997. Team Member Handbook: Toyota Motor Corporation. Toyota. 1998. The Toyota Production System: Leaner Manufacturing for a Greener Planet. Underwood, J. 1998. Not Another Methodology: What ANT tells us about systems development. Paper presented at the 6th. International Conference on Information Systems Methodologies, Salford, UK. Womack, J., & Jones, D. T. 2003. Lean Thinking : Banish Waste and Create Wealth in Your Corporation: Free Press. Womack, J. P., Jones, D. T., & Roos, D. 1990. The Machine that changed the World: The story of Lean Production: Harper Perennial.