Official PDF , 219 pages - World Bank - Documents & Reports [PDF]

Public Disclosure Authorized

The views expressed in these papers are those of the authors and do not necessarily reflect those of the Wlorld Bank.

SW P-1 96

Public Disclosure Authorized

Public Disclosure Authorized

Public Disclosure Authorized

INTERNATIONAL BANK FOR RECONSTRUCTION AND DEVELOP1INT Working Paper No.

196

Investment in Education; National Strategy Options for Developing Countries

John Simmons Arnold Harberger Dean Jamison Ernesto Schiefelbein Marcelo Selowsky Michael Todaro

Mark Blaug Samuel Bowles Martin Carnoy Remi Clignet Ronald Dore Edgar Edwards

The papers review the main feature of educational systems in developing countries and the investment issues. Options for national educational strategies are proposed. They were discussed at a workshop jointly sponsored by the Development Economics Department and the Education Department of the IBRD held in Washington in October, 1973. Reports of rapporteurs suggest the nature of the discussion.

Population and Human Resources Division Development Economics Department Development Policy Staff February 1975

Education Department Central Projects Staff

TABLE OF CONTENTS Page Contributors

iii

Summary Nutrition and Other Early Factors Marcelo Selowsky, Pre-School Age Investment in Human Capital Rapporteur's Comments

4 21

Migration and Fertility Michael P.Todaro, Education, Migration and Fertility

21

Rapporteur's Comments

31

Internal Efficiency Dean Jamiaon, Radio and Television for Education in Developing Countries

7

Ronald Dore, Deschool? Try Using Schools for Education First: The Educational Impasse in the Developing World

50

Ernesto Schiefelbein, The Impact of the Jencks Type Research on Developing Countries

70

Rapporteur's Comments

80

Employment Mark Blaug, Education and Employment

82

Martin Carnoy, Schooling and Unemployment

90

Remi Clignet, Education and Employment in Developing Nations: Fifteen Years Later

101

Rapporteur's Comments

109

Income Distribution and Social Mobility Samuel Bowles, Education, Class Conflict and Uneven Development

113

Page Intersectoral Allocation Edgar O.Edwards, Investment in Education in Developing Nations: Conflict Among Social, Economic and Political Signals

127

Rapporteur's Comments

144

Alternative National Strategies John Simmons, Education, Poverty and Development This paper has been issued as IBRD Working Paper 188 It is included here because in its original version entitled Investment in Education: National Strategy Options it formed the concluding paper in this series.

-

ii

-

147

LIST OF CONTRIBUTORS

Mark Blaug, Professor of Economics of Education and Head of Research Unit in the Economics of Education at the University of London Institute of Education. Samuel Bowles, Associate Professor, Department of Economics, Harvard University; and Research Associate, Center for International Affairs and Center for Educational Policy Research, Harvard University. Martin Carnoy, Associate Professor of Education and Economics, University.

Stanford

Remi Clignet, Professor, Department of Sociology, Northwestern University. Ronald Dore, Professor, Institute of Development Studies, University of Sussex. Edgar 0. Edwards, Foundation.

Economic Advisor, Asia and Pacific Program, The Ford

Arnold Harberger, Visiting Professor, Woodrow Wilson School, Princeton University; and Professor of Economics, University of Chicago. Dean Jamison, Research Economist and Chairman, Economics and Education Planning Group, Educational Testing Service. Ernesto Schiefelbein, Visiting Professor of Education, Graduate School of Education, Harvard University; and Professor of Education, Catholic University, Santiago, Chile. Marcelo Selowsky, Economist, Development Research Center, World Bank. John Simmons,

Economist, Development Economics Department, World Bank.

Michael Todaro, Associate Director, Social Sciences, The Rockefeller Foundation. Rapporteurs: World Bank Staff Rachel Brandenberg Jean Pierre Jallade E. Philip Jones

Ricardo Moran Marcelo Selowsky

-

l

-

SUNMKARY The expansion of investment in formal education has been phenomenal in the last two decades. And only in a few developing countries are there signs that the rate of expansion may be slowing down. In country after country, education has been the only sector to meet or exceed the planned investment targets. Demand has been fueled by critical manpower needs, parental pressures for more years of schooling for their children and thus better job opportunities, and the conviction that education is a human right for all citizens. Because of the relatively low capital costs, including foreign exchange, of supplying new school places, at least compared to creating new jobs, the supply rapidly expanded. The successful expansion of formal education, however, has created a new generation of social and economic problems. Education has become the largest item in many countries' current budgets: as a result, unless present trends are damped, an increasing number of nations face serious financial problems. Foreign exchange costs become serious when the expansion of secondary and higher education forces the maintenance of ever larger numbers of expatriate teachers and imported school equipment and supplies. Yet reducing the rate of expansion creates serious political problems. Middle-class parents are the most outspoken when reductions are threatened, since their children, expecting a place at the secondary or university level, might not be admitted. And these parents are usually the main source of political support of the governments. Teachers, usually the largest single group of wage earners in a poor country, are also a politically potent element when changes in the education system are proposed. Finally, recent research suggests that schooling has not been as effective in promoting cognitive achievement, as measured by school tests, as we had expected. Research results for Eirope and the United States in the 1960's have recently been supported by results for a limited number of developing countries.l/ Non-school factors like parental behavior, nutrition, and peer group experience are more important than expected in predicting achievement scores. Traditional school inputs like class size, length of teacher training and expenditure per pupil, are usually insignificant in determining differences in student scores. This research has important implications for future investment decisions. Policy makers in countries that are considering increased efforts at reducing poverty through employment, income redistribution and welfare measures, are asking what is the contribution of education to individual and social welfare? Should resources now used for formal education be shifted to the other programs where they might have a higher marginal social product? Can allocation within the sector be made more efficient and effective? The papers of this two day workshop help to clarify these and related educational policv questions.

1/ John Simmons, "The Effectiveness o4 Schooling in Promoting Learning: A Review of the Research," (Sovtbowming IB4RD wo#41ng papq?.

- 2 While individuals in each country will have to undertake a thorough evaluation of their educational objectives and resources and promote broad consultation in preparing their strategy, three major areas for action appear significant. A.

To damp the social demand: (1)

Ensure that artificial wage differentials for different occupations be rapidly narrowed. International trade in human capital also requires better controls to limit the export of highly-skilled manpower from poor countries. Artificial distortions in factor and product prices should be reduced to encourage use of labor intensive technologies;

(2)

Ensure that certificates of schooling are no longer required for most occupations. When employers become more skill specific in their needs, a flexible education structure will emerge to meet the better specified demand;

(3)

Require that middle- and upper-income students share a greater burden of the costs of their secondary and higher education. Explore the advantages of establishing an educational bank to transfer to the consumer the responsibility for both repayment and purchase of educational services. Time spent in public service would count as a form repayment, and should be encouraged. Localize school taxation and expenditure so that the costs and benefits are better understood locally. Low-income groups should have sufficient subsidies to permit them to overcome the private costs of schooling; and

(4)

In countries with an over supply of school leavers, introduce a system of quotas whereby the absolute number of students in the formal system at the secondary and higher levels is restricted to the rate of population increase or even less. It can be combined with existing exam systems. In some countries, rationing is already taking place on a basis of geographic location, which serves as a proxy for parental social status. It should be made explicit and fair. Finally, this should encourage the share of the national budgets and GNP going for schooling to fall for many countries and level off for others. Misallocation will be reduced.

B. Increasing the economic relevance of learning alternatives is more difficult than it may seem. Some countries like Cuba and China have made more sustained and comprehensive efforts than others. Rich countries have also made considerable efforts to improve learning among low-income groups. Some countries are considering nonformal education as a panacea for the illnesses supposedly caused by formal education, but we know virtually nothing about either how effective of how costly this is. To improve the opportunity for the development of alternative learning systems:

- 3 (1) Eliminate the monopoly power of governments in the field of organized learning. If creative and economic alternatives to formal schooling are to develop, then incentives should be established to encourage more productive learning technologies. Experience is increasing with voucher systems where the parents pay the teachers. Apprentice systems may be the most efficient for other skills; (2) Increase the effectiveness with which the basic cognitive skills are learned, and thus reduce the phenomenon of primary schools producing semiliterates; and (3)

Explore the opportunities of informal education, using adults and more advanced students to teach.

C. Compensate for the inequality effects of most of the existing formal school systems: (1) Establish quotas on the basis of social class, geographic location or school, to assure that the proportion of the low income students at secondary and higher levels is equal to or higher than their representation in the population; (2) Consider the use of automatic promotion from one grade to the next; (3) De-emphasize the role of examinations for promotion. Eliminate the social class bias to present examinations. Add the testing for wider range of behavioral skills; and

(4) Develop mechanisms to encourage students who dropped out to re-enter the formal system at an older age. These suggestions are a best guess as to possible guidelines. They are nothing more than a starting point for discussion among the producers and consumers of formal education to improve efficiency and equity. To increase the value of human capital with optimal use of resources will require significantly different approaches than are now being used in most poor countries. The evidence increasingly shows that the early factors which determine cognitive and noncognitive traits need increased understanding and investment. The traits required by occupation as well as leisure need to be more systematically considered than they have in the past. The question 'learning for what' is essential for both individuals and societies. What is called for is an integrated strategy for investment in human capital that considers the potential contribution of all human capital inputs. Such a strategy would assist in reduing past inefficiencies and ineauities and increasing the returns to scarce resources.

RE-SCHOOL AGE INVESTMT IN HUMAN CAPITAL

Marcelo Selowsky

"It is true that schools have 'inputs' and 'outputs', and that one of their nominal purposes is to take human 'raw material' (i.e. children) and convert it into something more 'valuable' (i.e. employable adults). Our research suggests, however, that the character of a school's output depends largely on a single output, namely the characteristics of the Everything else - the school budget, its policies, entering children. the characteristics of the teachers - is either secondary or completely irrelevant." Christopher Jencks, Inequality: A Reassessment of the Effect of FamilY and Schooling in America, p. 256 (Basic Books, 1972)

I.

INTRODUCTION

The above statement by Jencks is perhaps a suggestive way of introducing the subject of preschool investment in human capital into a It is not obvious that workshop concerning the economics of education. such statement could be less relevant for developing countries. The relevance of preschool investment in human capital - for the process of human capital formation in general and formal schooling in particular - rests heavily on the following set of evidence: (1)

The acceleration of the rate of enrollment in primary schools in less developed countries is and will be characterized by drawing an increasing number of children from lower income In other words the ratio of children from low income families. families in the elementary schooling system will increase over time.

(2)

There is a growing empirical evidence showing that preschool children from poorer segments of the population in developing countries have a lower performance in any type of mental test Large part than matching controls from higher income groups. of this evidence has been generated in medical studies attempting to isolate the effect of early malnutrition on mental development (Monckeberg, 1967; Kardonsky,et. al, 1971; Robles et. al,

1959).

-5 -

(3)

The recent literature in the field of education and psychology suggests that even heritability explains an important fraction of children's intelligence scores (Jensen, 1969) crucial in such explanation (Jencks, environment is still 1972) and particularly true for environment at early ages of life (Bloom, 1964).l/

(4)

There is a growing literature in the medical field attempting to show that early protein type malnutrition - a phenomenon that characterizes a large fraction of children in developing countries (see surveys by Woodruff, 1966; Revelle and Frisch, 1967; Frisch, 1966; Guzman, 1967) - adversely affects mental performance as well as children's psychomotor activity (for a survey see Berg, 1973; Craviotto and De Licarde, 1973).

(5)

Considerations (3) and (4) would suggest that the determinants of deficits in mental development in preschool age children as noted by (2) - are not completely exogenous to public policy.

If we accept (1) and (2) it will be true that the elementary schooling system in developing countries will be facing an increasing deterioration of its (entering) "raw input". On the other hand if the productivity of school inputs (in the production of abilities, however defined) is largely dependent on the quality of that "raw input" it will be true that the effect of schooling might be in the future highly sensitive to policies concerning preschool types of investment in human capital. As economists two interdependent questions would seem appropriate: first, to what extent are we overinvesting in schooling vis-a-vis preschool types of investment? Second, what are the types of investment in preschool age that can be manipulated by public policy and what is the "productivity" of such investments?

II.

FRAMEWORK

In order to focus on the economic questions of the earlier discussion, let us start with the following simplifying premises: (a) We want to analyze the contribution of preschool public actions to a vector of cognitive and noncognitive performance at adult age (or at the age the individual enters the labor force) as measured 1/ According to Bloom "by about age 4, 50% of the variation in intelligence at age 17 is accounted for. This would suggest the very rapid growth of intelligence in the earlier years and the possible great influence of the early environment on this development."? (Bloom, p. 68).

by available tests. (b) We are not going to worry at this stage - hopefully it will be discussed in another session of the seminar about the mechanism by which cognitive and non-cognitive characteristics of the individual affect economic productivity (see for example Gintis, 1971). (c) The "economic problem" become therefore how preschool investment in human capital can contribute to a "least cost solution" in the production of cognitive and noncognitive skills at adult age. A glance to the literature on education and psychology would suggest we can use some of the tools familiar to us: Assune the production of cognitive and non-cognitive performance at adult age (n) can be written as: (1)

I

=

F(A, S)

where A is the vector of abilities of the child when he enters the schooling system and S is a vector of school and (non-school) environmental inputs he is exposed to during schooling age. From now on let us interpret S as simply a vector of school inputs. However, it would seem that some of Jencks' finding also allow us to say something about the shape of (1): according to him not only A has an important independent contribution to R but it also determines the magnitude of the contribution of S.1/ This would mean we could write (1) as: (2)

=

AaS-a

From expression (2) it is clear that the contribution or marginal productivity of school inputs (MPS) becomes now a function of the level of ability of the entering child:

(3)

MPS =

as

=

(1-a)S

We can now interpret Jencks' statement:

1/

the more important is

"We have therefore abandoned our initial belief that equalizing educational opportunity would substantially reduce cognitive inequality among adults. This does not mean that we think cognitive inequality derives entirely from genetic inequality, or that test scores are ixirune to environmental influence. It simply means that variations in what children learn in school depend largely on variations in what they bring to school, not on variation in what schools offer them" (Jencks, op. cit., p. 53).

- the more - which implies a "largefl value of A in the production of A) in of value (the important becomes the quality of the "raw input" abilities. adult to determining the contributions of schooling On the other hand, points (3) and (4) of pages 1 and 2 suggest that A is itself a function:

(4)

A = Gf (0,E) where: G = genetic endowment 0 = index of inputs affecting organic and physical growth E = environmental inputs characterizing the "milieu" of the child

For policy purposes we can further break down some of the explanatory variables of the above expression: (5)

A=

Gf {N, H, Eh, EO}

where: index of the quality of the food intake (nutrition) during preschool age H = index of health services the child is exposed to during preschool age Eh= index of the quality of the home environment the child is exposed to Eo= index of the quality of "out-of-home"f environment during preschool age.

N

Our first question preschool investment in human It will be true if, given the situated in a point like Pl.

- are we overinvesting in schooling vis-a-vis capital - can now be summarized in figure 1. (vector) of relative costs of A and 4 we are Answering such a question will require:

(a) Information about a, or the contribution of A to n; to certain extent it can be obtained through "educational production function studies" (Bowles, 1970) where A is introduced explicitly into such a function. (b) Information about the coefficients of those explanatory variables of expression 5 that can be affected by public policy. (c) The resource cost of inducing changes in those variables affecting n, per unit change in n.

- 0

-

Section III presents some empirical evidence concerning (a) and (b) out of existing research.

Figure 1 Preschool index of abilityzA

_

_

_

+

S - Schoul inpuvs

III. A.

BRIEF REVIEW OF SOME OF THE EMPIRICAL EVIDENCE RELEVANT TO OUR HYPOTHESIS THE COEFFICIENT OF EARLY ABILITY (a)IN THE PRODUCTION OF TI

To my knowledge no research attemtpting to isolate the "value added of schooling " from the effect of the ability of the entering child has been undertaken in developing countries. Most of the research on this issue has been conducted in the United States by researchers from different disciplines. An estimate of a requires measurement of abilities in long follow-ups of individuals (at least 15 years) as well as data on all envircnmental variables in the interim period. Almost no study combines both requirements: some of them, particularly undertaken by psychologists (see Bloom, 1964), consist of long follow-up studies of IQ measurements recorded data on environmental over time; however, most of them have little variables the individual was exposed to between these measurements. We now review some of the research that, one way or another, is suggestive of the notion that early ability (at ages 4-6) is an important determinant of an adult's level of ability as measured by current IQ tests.

- 9-

(a) Bloom's particular interpretation of the results of the m-ajor longitudinal studies undertaken in the United States in the last sixty years is worth mentioning. These results, suxmtarized in figure 2, show that the correlation coefficient between the IQ at any age (T-t) and the IQ score at maturity (T) increases for consecutive lower values of t. Moreover such a relationship appears to be quite similar across studies "done wiith different groups of children, in different parts of the country, with different examiners, and at different times." Figure 2

90

'

45

S C/

30-

15

+Age at n"tMrty --- 3zrl8~ey 18 | 1.:a =-en16 Mc k 18 ---A. ( rs) 16.4

I

0 t |

1

-

3

---

5

~~Ar.r3o r 7

Of) osn 16 4

9 Age

11

13

15

17

What is relevant to us is the particular interpretation Bloom gives to the behavior of the correlation coefficient.

His interpretation follows a line

in psychology originally developed by Anderson, (Anderson, 1939), called the Overlap Hypothesis: this hypothesis interprets the correlation coefficient among two (longitudinal) measurements of intelligence as the "percentage of elements which are common to the two sets of measures involved." This leads Bloom to conclude that "in terms of intelligence measured at age 17, at least 20% is developed by age 1, 50% by about age )j, 80% by about age 8 and 92% by age 13." (b) Jencks' "path model of intergenerational mobilityt" attempts to isolate the effect of an IQ measure at 11 years (IQ) from the educational attainment of the individual (E) on the Armed Forces Qualification Test (AFQT), (Jencks, 1973). The Beta (regression) coefficients between IQ and AFQT and between E and AFQT turned out to be .80 and .20 respectively; in other words, holding education constant,

-

10

-

one standard deviation change in the early IQ score would induce 0.80 of one standard deviation in the AFQT score. If Bloom is correct it is plausible however that the regression coefficient obtained by using an IQ score at age 11 would represent an upward bias estimate of the contribution of an earlier IQ score (at preschool age). (c) A research by John Conlisk using longitudinal data on students IQ scores found the following regression equations (Bowles, 1970): 1Q1 8

=

4.77 + .490 IQ1_ 5 + 1.514 years schooling (6.44) (.099)

IQl8 =

R

=45

(.358)

8.11 + .527 IQ6_g + 1.051 years schooling (5-74) (.093)

R

2 =49

(.367)

where the subscript of IQ denotes the age range at which the test was administered and where standard errors are in parenthesis. Figure 3 shows the trade-off between early IQ and years of schooling in the production of later IQ (IQ1 7 ) implicit in the above results.

Figure 3 Early IQ

IQ

isocuant"

17s

o

1QI15

years of schooling

-

11

-

According to those results 3 points of IQ at ages earlier than

5 are a substitute for one year of schooling; for IQ at later ages (between 6 .nd 8) 2 points of IQ are a substitute for one year of subsequent schooling. B.

EVIDENCE CW THE DErERMINANT OF PRESCHOOL ABILITY SCORES (A). NCN-GENErIC DETERMINANTS

1.

GCeneral Considerations

THE

For our purposes what is important is to identify variables (a) determining A and able to be manipulated by policy instruments usually Very possibly the "non-genetic variables" available to governments. are the ones which fall in this category of variables. By now it is well accepted that an important fraction of children's intelligence scores is explained by heritability. The recent discussion has centered on the order of magnitude of such explanation: in this respect we quote Jencks "Whereas Arthur Jensen and others have argued that 80 percent of the variance in IQ scores is explained by genetic factors, our analysis suggests that the correct figure is probably more like 45 percent." The question remains to what extent the result of such studies The reason is that would change if undertaken in developing countries. perhaps the variance of other explanatory variables colinear to genetic endowment is larger in developing countries than in the United States, If this is true where most of the above studies have been undertaken. the (net) explanatory power of heredibility on early IQ in developing countries would be lower than the estimates obtained for the United States. For our purposes this is an important conclusion: it means an increased importance of the explanatory variables able to be manipulated by policy, at least in the range of the value of the variableo relevant to developing countries. (b) At this stage it is perhaps worth asking ourselves a pragmatic question: is it worth to disentangle the net effects of the different non-genetic variables to which low income groups'children are exposed to and which - as a package - are contributing to low preschool

ability scores? If it is true that children from lower income groups - subject to a package of nutritional, health and environmental deprivation - score worse in mental tests, why not affect directly the whole package of these variables? 'wo kinds of considerations would reject such a view. First, some of the variables in that package are difficult to manipulate from an institutional point of view; and we do not know to what extent;

- 12 -

such variables are perhaps the "true" explanatory variables of children's performance; in such case manipulating the rest of the package would not have a major effect on such a performance. Second, it is realistic to assume that the resources available for such public programs will be highly constrained, particularly because they involve a sharp redistributive policy. In that case we are interested in identifying those variables who have the major impact per dollar spent in manipulating that variable; this requires necessarily an estimate of the net impact of such a variable. (c) We proceed now to review some of the literature that, one way or another, gives some information about the net effect of the nongenetic variables outlined in expression 5. 2. The Effect of Early Malnutrition The causal hypothesis a deficit in the intake of llhifh ing is basically a medical one_A. the central nervous system given process of protein synthesis.

by which early malnutrition, especially quality" proteins, affects mental functionNutrient deficiency produces a damage in that early brain growth is largely a

This has been confirmed in

experiment with

animals and by preliminary findings of reductions in the number of brain cells in dead children who suffered severe malnutrition (Winick, 1969, AmbrosiusP 1951; Brown, 1965). What becomes relevant is to what extent these organic changes affect learning and behavior as measured by available test scores. Craviotto and De Licardie present in their survey a review of experin,ents undertaken in a variety of countries which tend to support such a hypothesis; what we want to emphasize here,however, is the fact that some particular types of abilities that seem to be affected by malnutrition appear to be crucial for further learning; If this is true-early malnutrition would condition the effectiveness of school inputs at later ages. Table 1.presents a survey of some studies whose specific aim was to study the effect of malnutrition on different types of intersensory integration abilities crucial for basic learning.

I/

"High quality" proteins provide all of the so-called essential aminoacids. Proteins of animal origin provide all of such aminoacids.

-

13

-

Table 1

Authors

Deficit in Performance

Probable effect on consequent learning

Craviotto and De Licardie, 1969 (Mexico)

Auditory-visual integration

reading ability

Craviotto et. al, 1969 (Mexico)

Visual-kinesthetic integration

writing and drawing abilities

Chapakam et. al, 1968 (India)

Visual identification

reading abilities

Craviotto et. al, 1966 ) (a) Kinesthetic-visual Craviotto et. al, 1967 ) (b) Kinesthetic-haptic ) (Mexico, Guatemala) ) (c) Haptic-visual ) Guthrie et. al (Philippines) ) (d) Auditory-visual

Source:

Craviotto and De Licardie,

general learning abilities

1973 (see references).

- 14 -

Other causal mechanism by which early malnutrition affects learnThere is some evidence showing ing have been advanced in the literature. severe and less frequent in less that infectious diseases are likely to be extent infectious the to well nourished children (Scrimshaw, 1961); environment it affects his to diseases affect the child's responsiveness his cognitive development. Apathetic behavior has been one of the most clear effects of Craviotto and De Licardie have an malnutrition in almost all studies. "It shoild interesting hypothesis about the further effect of apathy: a considerto is infant be recognized that the mother's response to the .... reactivity of able degree a function of the child's own characteristic a as Apathetic behavior in its turn can reduce the value of the child stimulus and diminish the adult's responsiveness to him. Thus, apathy can provoke apathy and so contribute to a cumulative pattern of reduced adult-child interaction." 3.

The Effect of Early Environmnent

This is a difficult subject to summarize, with a huge literature and with contradictory interpretations of the same data by social the One phenomienon is striking: scientist of the same discipline. countries, developing in area research that has been done in this little vis-a-vis developed countries, particularly the United States. Let us start with a policy oriented approach: We can define basically two types of environment the child is exposed to at preschool The policy options age, home environment and out-of-home environn,ent. are to change the quality of both types of environment and/or change the nmix"', to change the total fraction of time the child is exposed to a particular environment.

BDC

F. _u*e 4

j,out~de ome |>ho=e \j Ee t6ide

HH e

H e

hours per year exposed to the environment

e

a t home Age

BDC

B1oom's Deeeloptz.ent Curve

- 15 -

Figure 4 shows, under a particular setting, a hypothetical distribution of the time spent in both environment; the older the child the larger the fraction of time he spends (or can spend) in an out-ofhome environment. On the same graph we plot an index of Bloom's development curve (BDC); the shaded area represents the limit of the variation that the quality of the environment can produce at different points of the development of the child.l/ 3(a)

Changing the Out-of-Home Ehvironment

The best examples in this field are the large scale programs of prsschool compensatory education (between age 3 and 5) undertaken in the United States, particularly the Headstart program. The Rand Corporation has undertaken a survey of about all the evaluation research of such programs (Averch et. al, 1972). Even though there is a wide disagreement about the interpretation of such evaluations it would seem the results have not been encouraging. Those preliminary findings concerning large scale compensatory the United Stattes have led researchers into two new areas of in programs to what exte-nt such findings are a result of the failure First, study: of such programs to adapt themselves to the characteristics of disadvantaged children. To a large extent wide scale programs like Headstart were originally based on the Montessori model of the nursery school, which originally was adopted by higher income families, and by the Kindergarten movement freeing young chidren from "mothers' strict disciplinary controls"l and whose aim was free play. But "Headstart is not synonymous with compensatory education; Investigations of compensatory compensatory education has not failed. to offer play school has little traditional education have now shown that inculcate to effort an made which programs but poor, the of the children cognitive skills, language skills and number skills, whether they be taught directly or incorporated into games, show fair success." (Mc.V. Hunt). The second line of research stresses the fact that current compensatory programs start too late, in the sense that - if Bloom is correct - the child by age 3 or h has already been conditioned by the environmental The main constraint deficits in the family (Weikart and Lawbie, 1970). there is a limit possibly very that is research of in pursuing this line the family from withdrawn physically be can child a to the minimum age program. institutionalized an to exposed in order to be 1/

One of the major conclusions of Bloom's study is that "variations in the environment have greatest quantitative effect on a characteristic at its most rapid period of change and least effect on the characteristic during the least rapid period of change"(Bloom, op. cit., p. vii).

-

16 -

Outside the United States we have found a piece of rejearch which must be mentioned (McKay M., McKay A., Sinisterra L., 1970). This research - undertaken in Cali, Colombia - attempts to identify the types of intkrvention (nutritional supplementation as well as behavioral stimulation) necessary to overcome specific mental deficit in preschool children (age 3) from the lowest economic level families of -hat city. The importance of this research stems from the fact that, (a) the children analyzed are not characterized by extreme malnutrition, the kind of cases usually studied in the T"pure malnutrition studies." They represent more a typical situation of low income families in (b) A particurban areas and therefore more relevant for our purposes. of particular effect separate the analyze to given being is ular emphasis of the child tasks mental specific on interventions types of stimulation that showed findings preliminary Their learning. further relevant for boost can 3 age at interventions nutritional and stimulation particular certain mental capabilities over and above the performance of well nourished children from similar income groups. 3(b)

Home Eavironment Before Age 3

Bloom's bypothesis and the fact that probably there are institutional constraints on the earliest age a child can be drawn from the family into an out-of-home compensatory program has led some researchers into new areas of study: the effect of different rearing practices and different iaother-child interactions before age 3. In this respect we quote Kagan (Kagan, 1970), "A final strategy, not exclusive of the first two (school and preschool years 2½-5) is to change the mother's relationship with her infant. The idea for this suggestion rests on the assumption that a child's experience with his adult caret aker during the firet 24imonths of life are major determinants of the quality of his motivation, expectancy of success, and cognitive abilities duiring the school years. In a moment we shall consider data in support of this position." In an experiment with 140 white infants (before 27 months) of dAffereat socioeconomic classes Kagan found significant differences In fixation time, vocalization and fear. In another experiment 60 tenmonth white girls from two different socioeconomic groups were studied.l/ l/ The socioeconomic group8 were characterized as follows: In the upper one, one or both parents had graduated from college and the father had a professional job. In the lower one, either one or both parents had dropped out of high school and the father was working at a semiskilled or unskilled job.

-

17 -

The families were visited and the child-mother interactions were recorded. In the higher socioeconomic group the mother "spent more time- in face-to-face posture, more time talking to her and issued more distinctive vocalization to the infant. They were more likely to entertain their children with objects, to encourage walking and to reward them for mastery." Behavior of the infants at the laboratory showed that those belonging to the upper group were better able to differentiate meaningful from wn-meaningful speech and its source. At the same time they showed a stronger will of resolving the discrepancy of acoustic differences between different voices. The Ypsilanti Carnegie Project has found an important effect on the mental growth of young infants of programs where specially trained teachers work with the mother and the infant at home. The main objective of the program has been for the teacher to affect the child via the mother (Weikart and Lambie, 1970). Other lines of research concerning the effect of very early environment refer to brain damage out of being reared in the dark. None of this research has been undertaken with humans yet; however experiment with different animals seems to show a very clear result (Mc. V.Hunt, 1969). IV.

WHRE DO WE GO FROM HERE7

The research implications for developing countries out of some of the earlier hypothesis are enormous and out of the scope of this particular vorkshop. However, some priorities ought to be pointed out. Given the scope of this workshop I have chosen the following criteria in such a selection:

(a)

Research that can be helpful in guiding policy instruments available in the short run.

(b)

Research requiring the cooperation of the existing logistics of the education system.

(c) Research whose policy .mplications are directly relevant to existing educational systems. 1.

Additional information concerning A in developing countries

Three questions concerning the levels of abilities of entering students (A) appear relevant for further exploration:

- 18 -

(i) How large are the differences in A across socioeconomic and ethnic groups on a country wide basis? Up to now this information has been obtained through isolated sarples and by a variety of scientists of different disciplines and therefore they are hard to compare. Is it possible to institutionalize a generalized common test of abilities to be administered, in groups and by the same schools, to entering children? What are the relevant questions to be asked in these tests? What can educators suggest in this respect?

2.

(ii)

What are the socioeconomic and ethnic groups that, in a particular country, will be incorporated into the elemnentary schooling system during the next decade? How do they perform in the test outlined in (i)?

(iii)

If it is true that the level of A will be changing given the information obtained through (i) and (ii), - what are the implications for changes in the quality and types of school inputs? Are certain types of school inputs better substitutes for A than other ones?

Affecting the determinants of A (i)

Policies to change the "out-of-home environment" through preschool compensatory programs on a wide scale basis would seem difficult to undertake in the short run; unless we think of those programs as simple extensions of the existing elementary schooling system in such a way they can simply draw children one or two years younger into Kindergarten programs. However the United States' experiences has shown that simple extensions of the t"kindergarten type" are not sufficient or properly designed to compensate for the environmental deprivation low income children have suffered before. A much more complex type of program appears to be required. A partial substitute for the above solution, at least in the short run, is a policy attempting to correct the fact that lower income children enter primary schools at a later age (one to three years later) than higher income students. This evidence is at least clear for Latin America. An interesting research topic in this respect will be to study what determines the parents decision concerning the age at which they send children for the first time to school.

(ii)

How can we use the logistics of the existing educational system to affect the determinants of A that take place at home before age 3?

In the short rum ad-hoc educational programs for young women concerning child rearing practices would provide one type of solution; it is one kind of education for which we economists never computed rates of return! From our earlier sections it is clear what ought to be emphasized in such type of educational programs: (a) Infant nutrition, with a particular emphasis on breast feeding

The decline in breast feeding practices is one of the main determinants of early protein malnutrition.l/ On the other hand the resource cost of substituting breast milk appears to be quite large: Preliminary estimates by Berg (Berg, 1973) suggest that if 20% of the mothers in the urban areas of developing countries do not breast feed the loss in breast milk is $365 million. If half of the other 80% do not continue to breast feed after the first six months the total loss reaches $780 million. This is an important area to be researched: what determines the length of breast feeding and how could it be lengthened through educational prograns? (b)

Child rearing practices concerning early stimulation Educational programs concerning this issue involve some preliminary research which, to my knowledge, has not taken place on a wide large scale in developing countries. How different are child rearing practices across families .in developing countries? What are the factors determining these differences? Are they related to income groups or to particular ethnic groups of the population?

and .

cars~ Extel21 of Breast Feeding in .7elected Cotnrtries

atId Years, 19?40-71 Country

Year Perce,if of babies breast-fed 0 50

?/tile: at 13 1?on0ths

1960 I 068 -

,Ak.ricv: at 6 monis

1960 1966

Philippines.: at 12 ,tio,llits

1958

1968 Signapore: ot 3 mr,11/1rho

1951

1971 United States. ort leasil?g 11ospita! 1946 1966 Amirnp! RA~_ro.

n-n

ri

'St.

_|

-

20 -

REFERENCES 1. Ambrosius, K., "El Comportamiento de Algunos Organos en Ninios Con Desnutricion de Tercer Grado", Bol. Medico Hospital Infantil, Mexico, 1951. 2. Anderson, J., "tThe Limitations of Infant and Preschool Tests in the Measurement of Intelligence" in Journal of Psychology, 8, 351-379, 1939. 3.

Averch, H. et. al., How Effective is Schooling? A Critical Review and Synthesis of Research Findings, the Rand Corporation, 1972.

4.

Berg, A., The Nutrition Factor, Its Role in National Developmnent, MIT Press, 1973.

5.

Bloor., B., Stability and Change in Human Characteristics, Wiley, 196h.

6.

Bowles, S., "Toward an Educational Production Function" in (Hansen, ed.) Education, Income and Wealth, N.B.E.R., 1970.

7. Brown, R., "Decreased Brain Weight in Malnutrition and its Implications" in East Africa Medical Journal, 42, 1965. 8.

Craviotto, S. and De ticardie, E., "The Effect of Malnutrition on the Individual" in (Berg, et. al., ed.) Nutrition, National Development and Planning, MIT Press, 1973.

9.

Frisch, R., "World Food Supplies", in World, War and Hunger, Hearings before the Committee on Agriculture, House of Representatives, Serial W, Part I, Washington, 1966, pp. 41-51.

10.

Gintis, H., "Education, Technology and the Characteristics of Worker Productivity", A.E.R., 61, 1971.

11.

Guzman, M., "Impaired Physical Growth and Maturation in Malnourished Populations", in Scrimshaw and Gordon (eds.), Malnutrition, Learning and Behavior, MIT, 1967.

12.

Jencks, C., Inequality: A Reassessment of the Effect of Schooling in America, Basic Books, 1972.

13.

Jensen, A., "E&vironment, Heredity and Intelligence", Harvard Educational Review, June 1969.

14.

Kagan, J., "On Class Differences and Early Development" in Denenberg (ed.), Education of the Infant and the Young Child, Academic Press, 1970.

- 21 -

15.

Kardonsky, V., et. al., "Cognitive and Emnotional Problems of Chilean Students (7 to 10 Years) in the Northern Section of the City of Santiago", Department of Psychology, University of Chile, 1971, (mimeo).

16.

McKay, H., MacKay, A., and Sinisterra, L., "Behavioral Interventions Studies with Malnourished Children", Western Hemisphere Conference on Assessment of Tests of Behavior from Studies of Nutrition, Puerto Rico, October 1970.

17.

Mc. V. Hunt, J.., "Has Compulsatory Education Failed? Has it Been Attempted?", Harvard Educational Review, Spring 1969.

18.

Monckeberg, F., Donoso, F., Valiente, S., Arteaga, A., Maccioni, A. and Merchak, N., "Analisis de las Condiciones de Vida y Estado Nutritivo de la Pbblacion Infantil de la Provincia de Curico", Revista Chilena de Pediatria, 38, 522, 1967.

19.

Robles, B. et. al., "Influencia de Ciertos Factores Ecolo'gicos sobre la Conducta del Nino en el Medio Rural Mexicano", IX Reunion, Asociacion de Investigacion Pediatricas Cuernavaca, Mexico, 1959.

20.

Revelle, R. and Frisch, R., "Distribution of Food Supplies by Level of Income", in The World Food Problem, The White House, 1967.

21.

Scrimshaw, N., "Nutrition and Infection", in Brock, J., Recent Advances in Human Nutrition, London, 1961.

22.

Weikart, D. and Lambie, D., "Early Ehrichment in Infants", in Denenberg (ed.), Education of the Infant and the Young Child, Academic Press, 1970.

23.

Winick, M. and Rosso, P., "Head Circumference and Cellular Growth of the Brain in Normal and Marasmic Children", Journal of Pediatrics, May 1969.

24.

Woodruff, C. W., "An Analysis of the I.C.N.N.D. Data on Physical Growth of the Pre-School Child", in Preschool Malnutrition, National Academy of Sciences, National Research Council, Washington, D.C., 1966.

- 22 -

RAPPOCRTER'$S COMMENTS

The main focus of Mr. Selowsky's paper centered on the importance of preschool age investment in human capital in determining the effectiveness of further investment in formal schooling. If there is an acceleration of the rate of enrollment in primary schools in LDC's, very probably it will be characterized by drawing an increasing number of children from lower-income families; if this is trye, the quality of the "raw input" (or abilities of the entering child) faced by the primary school system will decline over time. This lowering of A (abilities) is attributable to the observation that "there is a growing empirical evidence showing that preschool children from poor segments of the populatior. in LDC's have a lower performance in any type of mental test than matching controls from higher income groups.", If this low (initial) level of A can be attributed to preschool nutritional and environmental deficits - able to be manipulated by public policy - the hypothesis arises to what extent the effectiveness of future investment in schooling might be highly sensitive to today's preschool age investment in human capital. The discussion of the paper then led to the following remarks: A. The Role of Schools and Nutrition What role can schools play and also relate their activities 1. to overall broader policies? (a) Schools can be used as food production elements and as instruments to reach other groups; e.g., mothers. (b) Schools may be viewed as the vehicle through which compensatory programs may be introduced. However, this approach may place too high a burden on schools and it is not clear exactly what the trade-offs are between what the family can provide vis-v-vis the school. In short, are we asking the schools to do too much? How can pre-school support in nutrition actually be carried 2. out? As indicated by many medical researchers, measures to improve the nutritional status of individuals must be undertaken before 24 months of age. In operational terms, if the schools are to play an effective 3. role in improving the nutritional status of school-age children, then school programs utilizing teachers and possibly women from the local areas might contribute to improving the nutritional level of the target group. The participation of the aforementioned groups might provide for the development of linkages between school children and the community.

- 23 -

What influence have school food pro.,rams had in improving food

4.

habits? habits?

Can schools really change the attitude of persons toward food

The discussion then moved to the effect of nutrition on ability. It was pointed out that raising the cognitive ability of persons may be a waste since in fact, cognitive ability explains little individual success measured by lifetime earnings. Several participants felt it was ludicrous to rationalize the value of nutrition. It was z obvious that healthier mothers have healthier offspring. Thus, th4_part of the discussion became very disjointed and the participants did not address themselves to the topic. This was further demonstrated by the discussions centering around the reasons why lower class women refuse to participate in welfare programs whereas their middle class "sisters" do. The Bank, however, is hardly interested in "give-away" programs! B.

Policy Implications In conclusion,

some of the participants suggested that:

(a) Schools should adopt both a production and distribution function. Thus it follows that teachers must do other jobs to accommodate other roles since the education resources should not be designed only for educational purposes; (b) Nutrition components should be introduced within teacher training institutes; and (c)

Research should be undertaken to clarify the relationship between feeding habits - parental behavior and achievement. This would also include an investigation into the inefficient use of

income and the effect of

improved information systems and resource allocation on mothers.

Mrs. Rachael Brandenberg

- 24 -

EDUCATION, MIGRATION AND FERTILITY

Michael P. Todaro

Basic Issues and Summary of Arguments Education appears to be an important factor influencing both ruralurban migration and levels of fertility. The relationship between education and migration, however, is apparently more powerful than that between education and fertility. Moreover, existing empirical evidence reveals that the mechanism through which education influences labor mobility is more direct (i.e., through its impact on higher income expectations) than in the case of fertility. Numerous studies of migration in the less developed countries have documented the positive relationship between the educational attainment of an individual and his propensity to migrate from rural to urban areas. Basically, individuals with higher levels of education face wider urban-rural real income differentials and higher probabilities of obtaining modern sector jobs than those with lower levels of education. The probability variable in particular accounts for the continual influx of the more educated rural migrants in the face of rising levels of urban unemployment among the less educated. With regard to the education and fertility relationship, the evidence is less clear. While most studies reveal an inverse relationship between the education of women and their fertility, particularly at lower levels. of education, the mechanism through which education influences family-size subject to considerable speculation. decisions is still Assuming that lower levels of urban unemployment (especially among the .dueaated) and lower levels of fertility are important policy objectives for both LDC governments and international assistance agencies, the basic issue is whether or not the continued linear expansion of the formal educational system ( and the resource allocation decisions implicit therein) will ameliorate or exacerbate the twin problems of rising urban unemployment and rapid population growth. It is our contention that, given limited government and donor agency resources, the further rapid quantitative expansion of school places beyond perhaps basic education is both undesirable and unwise for the following two reasons: 1.

Any rapid expansion of the formal primary system creates inexorable pressures on the demand side for the expansion of secondary and tertiary school places. This is because of the widening divergence between private and social returns to education, the tendency for scarce jobs to be rationed by progressively higher educational certification, and the political dimensions of educational investment decisions (see Edwards and Todaro, 1973a and 1973b). The net result

-

25 -

is the widespread phenomenon of over-expansion of school places in terms of real resource needs in many LDCs and the concommitant dilemma of rising levels of urban unemployment among a cadre of increasingly more educated and more politically-vocal migrants; 2. If, as many have argued, the education of women does affect their fertility behavior primarily through the mechanism of raising the opportunity cost of women's time in childrearing activities, then it follows that, unless sufficient employment opportunities for women (as well as men) can be created, the reliance on educational expansion as a policy instrument for lowering fertility will be weak, if not totally ineffective. Education aid Migration The accelerated influx of rural migrants into urban centers is a phenomenon common to almost every less developed nation. The extraordinary growth of urban populations during the past decade can be attributed primarily to the large and, in many cases, growing pace of rural-urban migration. Many factors influence the decision to migrate but, by and large, almost all informed observers, regardless of their disciplinary training, agree that the economic Sample surveys in Africa, Asia, and Latin America factors predominate. invariably reveal that the overwhelming reason why people migrate is because of the existence of a wider range of higher paying job opportunities in urban areas. (Byerlee, 1973; Todaro, 197-!; Herrick, 1969; IERD, 1971). Furthermore, individual country surveys consistently show that people living in rural areas who have been to school are far more likely to migrate than people who have had little or no formal education (see, for example, Lin and Chen, 1973; Remple, 1970; Sabor, 1972; and Herrick, 1969). In fact, there seems to be a consistent, positive correlation between an individual's level of education and his propensity to migrate. In a series of articles, I have attempted to provide a theoretical framework for understanding the phenomenon of high, and in some cases, accelerating rates of rural-urban migration in the face of rising levels of urban unemployment (Todaro, 1968, 1969, 1971, 1973; and Harris and Todaro, 1970). By focusing on "expected"' income differentials, encompassing both the absolute urban-rural differential and the probability of obtaining an urban job, this framework facilitates, I believe, a better and more accurate analysis of the dynamics of rural-urban migration. This is especially so when analyzing differential rates of migration by educational levels. Individuals with higher levels of educational certification tend to have a higher propensity to migrate than those with less education because: (1) the income disparity between what they can earn in urban areas as compared to their rural opportunity cost is larger, and (2) the probability of their being successful in obtaining a lucrative modern-sector job is higher than for those with less education.

- 26 -

The reason why the more educated are in an advantageous position and have a greater probability of finding higher paying urban jobs than the less educated is due not so much to their superior training as it is to the widespread "filtering down" or"occupational displacement" phenomenon which pervades the labor markets of LDCs. (See Edwards and Todaro, 1973a). Faced with an excess of job applicants over job openings, governments and private employers tend to select individuals on the basis of their level of educational certification. As the urban labor supply continues to outpace demand, two forces are set in motion. On the demand side, jobs which were formerly filled by, say, primary school graduates now require a secondary school certificate. On the supply side, job aspirants now need to get some extra years of further education in order to "qualify" for jobs which only a few years earlier were being filled by those with less education. The net result of these two forces is the burgeoning of private higher levels of formal education in order to meet more ever for demand (and the consequent explosive demand for primary requirements job stringent education as a necessary first stage), the emergence of rising levels of unemployment and underemployment among the more educated, the growing proportion of the more highly educated among the migrant populations, and the gradual drying-up of employment opportunities for those with limited education. Finally, given the strong political pressures for governments to try to satisfy the aggregate private demand for education (even though the social return on such investment is likely to be considerably lower than net private benefits), there is an inexorable tendency to expand formal educational facilities well beyond the point where such investment can be said to be economically justified. In order to break this vicious cycle of accelerated rural-urban migration leading to rising unemployment and underemployment, causing job rationing by educational certification which, in turn, causes burgeoning private demands for education which place severe pressures on governments to linearly expand school facilities, etc., inmmediate steps must be taken by LDC governments to change the prevailing system of economic incentives and political constraints (see Edwards and Todaro, 1973b). Restructuring educational systems to minimize the inherent (and, in many cases, inherited) urban bias and to orient the curriculum more towards the real development needs of the nation (e.g., towards rural development), will have limited success in either curtailing rural-urban migration or changing students' attitudes in the absence of necessary fundamental changes in incentives outside the educational system. Education and Fertility Educational attainment is thought to influence many forms of the case of women's education, fertility behavior. including,in behavior, is the well-known direct relation between a women's there On the one hand, and her propensity to enter the labor force. On attainment educational evidence of an inverse relationship between growing is there other, the

-

27 -

a woman's education and her fertility (see, for example, Miro and Mertens, 1968; Ben-Porath, 1973; and Rich, 1973). The apparent empirical linkages between educational attainment, labor force participation, and fertility have been widely interpreted by economists to be the consequence of a causal mechanism in which education increases the opportunity cost of a woman's time through the positive effect which it has on her earning opportunities in the labor force. Since it is assumed that child rearing represents a very time-intensive activity for the mother compared with other goods, it follows that the price of children relative to that of other goods moves together with the price of a woman's time. By linking fertility decision3 within the framework of recent reformulations of the micro-theory of household behavior. (see, for example, Becker, 1965; and Lancaster, 1966), the conclusion is drawn that the rise in the price of a woman's time (or alternatively, the opportunity cost of children) causes her to reallocate her time from nonmarket activities and leisure (i.e., being a mother) to market labor force activities. Her desired and actual completed fertility will, therefore, decline. In short, education's effect on the shadow price of a woman's time is thought to account, within the so-called "New home economics," for the widely observed inverse empirical association between a woman's educational attainment and her fertility behavior. The most comprehensive attempt to test this education - cost of time - fertility hypothesis is Yorami Ben-Porath's recent paper on education and fertility in Israel (Ben-Porath, 1973). Using both longitudinal and cross-sectional data, Ben-Porath finds that the evidence does support the expected negative association between fertility and the education of women (whereas the association between fertility and the education of men is However, when the data is disaggregated by level of education, ambiguous). the inverse relation between fertility and education turns out to be steep only at the very low levels of education (one to four years) and tends to flatten and, in some cases, even become positive at the top levels. BenPorath concludes that "the relation among education of women, the wage rate, and labor supply indicates that at the low levels of education, where the decline in fertility is large, the differentials in labor supply are modest, while at the top educational categories, where differential fertility is modest, the differences in labor supply are large. . . Thus, my impression is that the simple cost-of-time hypothesis, while consistent with some of the evidence, leaves some important aspects of the fertility-education relation unexplained" (Ben-Porath, pp. S204-5). Clearly, there are many ways in which education might influence fertility behavior through other economic and taste effects. Unfortunately, an appropriate methodology for untangling these lines of causation has not yet adequately been formulated or tested. Until this is accomplished, the promotion of expanded women's education simply as an effective anti-natal policy instrument should be approached with considerable caution.

-

28 -

There are other reasons why one should be wary of the education First, if the expansion of women's employment opporfertility hypothesis. tunities is in fact the vehicle through which education affects fertility, it may be important to identify the relative contribution of investment in job creating activities as compared to the simple expansion of educational facilities (Edwards and Todaro, 1973). Furthermore, as Tobin has pointed out, it is one thing to find that educated women have fewer children than their less-educated contemporaries; it is another thing to expect that this difference will predict how much a general increase over time in women's education in a country will diminish overall fertility. Clearly, the second effect will be weaker than the first (Tobin, 1973, p. S278). Second, the education - labor market - fertility hypothesis implies that investments in women's education may be less effective in reducing their fertility in rural areas where few women find jobs regardless of their educational attainment. If reductions in fertility are to be viewed as an important social externality of educating women, it is essential that we know more about how this change in behavior is influenced by education and how it interacts with other household and community forces. Finally, since educated women comprise a growing proportion of the urban migrant stream, the "filtering down" and "occupational displacement" phenomena described in the preceding section will apply equally to them, perhaps, even more so, in determining their probability of successfully securing wage employment in highly competitive urban labor markets. Those with limited education (i.e., the ones which Ben-Porath found to exhibit the sharpest fertility declines) will have little or no chance of competing successfully for scarce urban jobs. We may note, in conclusion, that the social value and importance of a purposeful expansion of education opportunities for women may be significant for a variety of economic and non-economic reasons. However, the rationale for such a differential expansion from the narrow perspective of potential fertility reductions is, in our opinion, simply not compelling, at least until better and more convincing evidence can be brought forward. This is clearly an important topic about which further research is urgently needed.

-

29 -

REFERENCES

Ben-Porath, Yoram, 1973, "Economic Analysis of Fertility in Israel: Point and Counterpoint." Journal of Political Economy, 81, 2, Part II, S202-S233. Becker, Gary, 1965, 75:4930517.

"The Theory of the Allocation of Time."

Economic Journal,

Byerlee, Derek, 1972, "Research on Migration in Africa: Past, Present, and East Lansing: Department Future." African Rural Employment Paper No. 2. of Agricultural Economics, Michigan State University. Edwards, E.O., and Todaro, M.P., 1973a, "Educational Demand and Supply in the Context of Growing Unemployment in Less Developed Countries." World Development, 1, (3-4): 107-117. , 1973b, "Education and Society in Developing Nations: Conceptual Relationships and Emerging Opportunities." Paper prepared for Conference on Education and Development Reconsidered, Bellagio, Italy, November 1973. 197.3c, "Education and Employment in Developing Nations: Some Main Themes and Suggested Strategies." Paper prepared for Conference on Education and Development Reconsidered, Bellagio, Italy, November 1973. _,

Harris, John R., and Todaro, Michael P., 1970, "Migration, Unemployment and Development: A Two Sector Analysis." The American Economic Review, 60, 1: 126-142. Herrick, B.H., 1970, "Urbanization and Urban Migration in Latin America: Economist's View," The Latin American Urban Annual. IBRD, June 1971, "Migration to Urban Areas." No. 107.

An

Economic Staff Working Paper

Lancaster, Kelvin J., 1966, "A New Approach to Consumer Theory." Political Ebonomy, 74: 132-157. Lin, T.L., and Chen, H.H., 1973, "Rural Labor Mobility in Taiwan."

Journal of

(mimeo)

Miro, Carmen, and Mertins, Walter, 1968, "Influences Affecting Fertility in Urban and Rural Latin America." Milbank Memorial Fund Quarterly, 46, 3. Rempel, Henry, 1970, Labor Migration into Urban Centers and Urban Unemployment in Kenya. Unpublished Ph.D. dissertation, University of Wisconsin.

- 30

-

Rich, Willian, 1973, Smaller Families Through Social and Economic Progress. Overseas Development Council Monograph No. 7, Washington, D.C. Sabot, Richard H., 1972, "Education, Income Distribution and Urban Migration in Tanzania." Economic Research Bureau, Dar es Salaam, Tanzania (mimeo). Tobin, James, 1973, Comment on paper by T. Paul Schultz, Journal of Political Economy, 81, 2, Part II, S275-S278. Todaro, Michael P., 1968, "An Analysis of Industrialization, Employment and Unemployment in LDCs." Yale Economic Essays, 8(2): 329-402. , 1969, "A Model of Labor Migration and Urban Unemployment

in Less Developed Countries."

The American Economic Review, 59(l): 138-148.

, 1971, "Education and Rural--Urban Migration:

Theoretical

Constructs and Empirical Evidence from Kenya." Paper presented at a conference on Urban Unemployment in Africa, Institute for Development Studies, University of Sussex. , 1973, "Rural-Urban Migration, Unemployment, and Job Probabilities: Recent Theoretical and Empirical Research." Paper prepared for International Economic Association Conference on "The Economic Aspects of Population Growth," Valescure, France.

-

31 -

RAPPORTEUI S COMMENTS

The major part of the session was devoted to a discussion of the relationship between education and migration and the derived policy The remainder focussed on the relationship between education implications. and fertility. Todaro's model of the relationship assumes that individual decisions to continue education, to migrate - are detenrined by two functions: 1) the differentials between the prevailing average wage rates which obtain for each educational level and for each sector - rural, urban subject to the inherent characteristics of the individual (sex, age, etc.), and 2) the objective probabilities for an individual with particular inherent and acquired characteristics of obtaining employment at the associated wage rate in each of the two sectors. The individual is assumed to maximize some intertemporal function of expected income, which is equal to the prevailing average wiage rate multiplied by the objective probability of obtaining such employment for each set of characteristics and for each sector. As applied to the normal condition of a developing economy, the model explains the high demand for education and the high level of rural-urban migration as follows: 1) the base wage for less educated labor in the rural sector is very low, especially as compared to the prevailing wage for more educated labor in the urban sector,* 2) the base wage for less educated labor in the urban sector may be somewhat higher, but the probability of obtaining such employment is very small, 3)

therefore, the ea:nings foregone by remaining in school are

minimal, 4) therefore, students will remain in school until the individual financial constraint becomes binding azid/or the expected income from entering the labor force subjected to the individual intertemporal preference exceeds that expected from remaining in school, 5) in most cases, the probability of obtaining employment in the will be lower than. in the rural sector, but the wage differential sector urban / The average marginal product. of labor in agriculture is often asserted to be zero.

-

- 32 -

will be sufficiently large that the expected income in the urban sector will exceed that in the rural sector, 6)

therefore, there will be substantial rural-urban migration.

These relationships result in the perverse dynamic feedback system which Dore refers to as the credentialization effect, whereby higher-level intrinsic degrees serve to ration the insufficient number of jobs with little operates relation to the objective requirements of the job. The process as follows: 1)

if

the wage differentials remain sufficiently large,

2) then the number of graduates from each level of education and the number of such graduates who migrate from rural to urban areas will increase as the population grows and educational opportunity is expanded, 3) therefore, if employment does not keep pace with the supply of graduates at the higher levels,

4) then higher level graduates will take lower level jobs 5) therefore, the prevailing average wage and expecially the probability of obtaining employment at this wage will diminish for lower level graduates, 6) the negative effect on the expected income of lower level graduates will be more significant than the negative effect on the expected income of higher level graduates,

7) therefore, students will desire to remain in school longer, to urban sector employment will become more restricted access and the graduates,* level to higher Thus, the facts that urban-rural wage differentials are actually widening and that employment at higher levels is not keeping pace with the supply of graduates is especially worrisome.** *

**

So that we are faced with the dilemma that increased educational opportunity will only accelerate the divergence of the educational pyramid from, any realistic job requirements pyramid (hence the return in terms of economic growth to such investment is likely to be small) while failure to increase educational opportunity will lead to decreased social mobility, given the increasing amount of education required for employment. A complicating factor is the disparity between wages in the LDC's and the developed countries for more skilled personnel. In order to retain such manpower, the real wage offered to more skilled personnel must be moderately competitive. Even then, the emigration of high-level manpower is often substantial. This factor suggests that, with respect to certain categories, it will be difficult to reduce or retard the advance of the differential between high-level and less skilled manpower. The implication is that society must exact an obligation in',return for the education it gives to such personnel, if the social investment is not to be lost.

- 33 -

Moreover, as Blaug noted, even a general increase in urban employment might not suffice to counter the perversity of this mechanism, since this would increase the probability of obtaining employment and induce rural-urban migration which would restore the old unemployment rates. As such, the rural sector may be compared to a "bottomless sink." The most significant implication of this is that the wage differentials are the crucial variables which must be controlled.* For example, since causality runs from the wage structure as an independent variable to education as a dependent variable, rural orientation of education is not likely to alter migration patterns. Rural development and the creation of opportunities to earn an income comparable to those in the urban sector must take first place. As to the validity of the model and its implications, it was noted that, even controlling for the increase in average educational advance, the rural-urban migrant stream has become more concentrated over time in the higher education categories. This increase in the correlation between education and migration was seen as the expected result of widening urban-rural differentials and increasing credentialization (so that whether or not an individual possesses a certain degree can be seen as a proxy for the probability variable). Moreover, even when one controls for the fact that there is some migration from rural to urban areas in order to obtain an education (thus raising the education-migration correlation and imolying that causality may run from education to migration), it is found that the partial coefficient of correlation is significant and reaffirms that education leads to migration in the manner narrated above. It was suggested that the relationship between education and migration might also operate by a change in the individual's intertemporal preference function because of his education. Certainly this would be difficult to determine empirically. In any case, the dominant considerations would be the differential and probability variables. To illustrate the way in which these two variables interacted to determine migration according to education level, the Tanzanian experience was note(d. On the one hand, rural-urban wage differentials had widened, which in itself would lead to increased migration among all education categories. On the other hand, the probability of obtaining employment at each leve:L decreased, so that higher-level graduates took jobs which would otherwise have been filled by lower-level graduates. Therefore, given the resulting very small probability of employment for lower-level graduates, migration ofisuch lower-level graduates stagnated. The phenomenon of temporary migration was explored within the construct of the model. Two kinds of temporary migration were distinguished. * Along with population growth.

-

34 -

One kind of temporary migration is seasonal migration, where the demand for labor varies cyclically between areas and individuals migrate so as to maximize income. The other kind of temporary migration starts off as an intention to migrate for good. Once the irmmigrant has been in the new location for some time without finding a job, however, his subjectively perceived probability of obtaining employment decreases and he may return to the rural sector. This accounts for the observed reverse (gross) migration from urban to rural areas. In this regard, finance problems and the length of unemployment may be seen as the significant variables. Of the two kinds of migration - seasonal and (intended) permanentFor example, seasonal migrants seasonal migration is of decreasing importance. in Upper Volta were of all ages, whereas (intended) permanent migrants were primarily the young. Given rates of population growth and shifts in the age structure of the population, this is becoming a more serious problem. Harberger examined the relation of the Todaro model and the concept of filtering down. He posed a two-group model (high and low education, income) where supply exceeded demand for both groups. If education transferred someone from the low group to the high group, did this: (a) only redistribute income by changing probability variables; (b) sift between groups according to differential abilities, implying increased productivity (but given an institutional wage); (c) also change wages outside the, for example, 0-20 per cent unemployment fixed-wage range, so that 80 per cent of wage change would represent additional product. Todaro responded that the losses in the low group would offset the gains in the high group, because one has to go through the low group to get to the high group and because of the dynamics of increasing certification. Harberger responded that, in the second and third cases, better qualified persons at each level implied higher product, although wages were unchanged in Case 2. Carnoy noted that the correlation of nutrition, education, socioeconomic status, etc., implied that, over time, it takes increasingly longer periods to produce the same gain in product. Blaug described Hargerber's first case as the "screening hypothesis," while the second and third cases implied a positive marginal product of education. He said it was crucial to know the nature of this

- 35 -

additional productivity. He saw Todaro as relating education to cognitive requirements, but felt that instead non-cognitive requirements were more important. This implies that the marginal productivity of non-cognitive outputs of education increased aS the education level advanced. Edwards observed that the relation of wages to credentials versus marginal productivity related to the cost of education and the job-creation/education tradeoff. The importance of the family with respect to migration was noted. In India, kinship and friendships links were a major determinant of the localities to which individuals would migrate. Previous immigrants could help new immigrants when they arrivedi. Moreover, there is an incentive for families to educate their children, so that educated members of the family can support the others by sending money home. The West African experience was noted in this regard. As for policy conclu

e

10.51L

Xote: Yig.e-r edtcaticn begE-im .>urces: Tatles I and 2.

2.29 L.2? 8.15

6.29

2.L9

13.09

8.82

13 .1

13.82

w

1960.

6.15 13.32 ]J.±.30

2.11 10.01 li.37

6.66

1L.85 18.11

2.75 10.01 9.18

UGANDA 5.07 4.71 10.99

2.9IL O.:

3.:'I

- 201

-

Annex Table h

Public expenditure on education as % of GNP 1960 and 1965-1968 (unweighted averages) 'No. of countries covered

1960

1965

1966

1967

1968

World

(85)

3.02

3.75

3.83

4.08

4.24

Developed countries

(31)

3.52

4.45

4.55

4.73

4.80

Developing countries

(54)

2.73

3.35

3.42

3.72

3.91

Sourcoe

Goldstone, p23

- 202 -

Annex Table 5

-

Uxc PUblic US dollars

-

±t-xC-A nn education, 1960 and 1965-1968 (millions in current prices) (revised)

1960

1965

1966

1967

1068

VIOiLD TOTAL 1)

54 350

96 360

106 980

119 090

131 640

AFRICA NORTHEIRN AMFERICA LATIIN AMERICA ASIA 1)

1 22 1 3

EUROPE AND USSR

24 380

Majoi regions

110 670 880 710

DEVELOPED COUNTRIES DEVELOPING COUNTRIES

660 740 230 380

1 44 3 8

920 050 670 370

2 49 4 9

2 56 4 10

170 940 000 470

370 510 430 660

43 280

47 660

52 110

56 220

1 070 (1 020)

1 310 (1 100)

1 400 (1 210)

1 450 (1 310)

49 450

88 190

97 820

108 670

120 290

4 900

8 170

9 160

10 420

11 350

600 (700)

OCEANIA (ARAB STATES)

1 39 3 7

1) Not including China (mainland), Democratic People's Republic of Korea, and Democratic Republic of Viet-Nam. national currbncies are converted Ii to the US dollar 2) When effects in the other ourrercies are minimized. inflation the

Source;

Goldstone,

p25

Annex TRble 6 AVERAGE WAGES BY SECTOR FOR SELECTED COUNTF.IES MlDEXED TO LOWEST SECTOR WAGE IN COUNTRY (-100) SECTORS BRANC1ES IN YANUFACTURE

AGRICULTURE

MAWJFACT-iE

Ghana Kenya Tanzania Latin America

100 100 100

231 244 208

127 216

455 542

257 237

131 100

-

-

-

-

Colombia Guatemala Peru

-100*

155

132 113 132 115

126 234 202

137 116

10

FOOP, PRO.

TOBACCC CHENIICALS

FU?I.'TURc

T?ANSPOJ

r IST-

EQUIPMENT M1,iI'.Iz

R'JCTI1' ?ORTAr: TA

'7. A,.

Africa

-

138 -

Venezuela

1-0

25 220 -

225 279

1IL 285

261 661

226

21.9

171

262

223

121

150

253

100

16L

-

-

-

113

-

100

150

261

252

238

160

100

225

26-

-

-

425

316

460 1Ll

100

126 100

119

173

-

162

150 191

-

-

-

-

-

157

-

Asia Indj4 Korea (ROK) Philippines

100*

)02

-

100* -

137 173

11.2 132

10L

159 167

100

2OE

164

164

-

-

-

Europe Austria

*

Estimated by multiplying daily rate by 200 working days..

FRENCH ACHIEVEIENT: PERCENTAGE CORRECT --4SPOi4SES BYI GRkDE

Axnex Table 7:

Total Correct Responses .' 2 _ __s_ ___ _ 1 __ 1) 18 1I ;;fiU.r, j R.'ry

14

CVi'lP'!i~iwl;.l'

4'