THEWORLD BANK
Discussion Paper
EDUCATION AND TRAINING SERIES
Report No. EDT94
Time Trends of the Returns to Education
Cross-National Evidence
George Psacharopoulos
June 1987
Education and Training Department Operations Policy Staff
The views presented here are those of the author(s), and they should not be interpreted as reflecting those of the World Bank.

Discussion Paper
Education and Training Series
Report No. EDT94
TIME TRENDS OF THE RETUfIS TO EUWCATION:
CROSS-NTONAL mEDCE
George Psacharopoulos
Research Division
Education and Training Department
June 1987
The World Bank does not accept responsibility for the views expressed
herein, which are those of the author and should not be attributed to the
World Bank or its affiliated organizations. The findings, interpretations,
and conclusions are the results of research or analysis supported by the
Bank; they do not necessarily represent official policy of the Bank.
Copyright 0 1987 The International Bank for Reconstruction and Development/
The World Bank

I
I

Abstract
The paper presents evidence on the over-time behavior of the rate
of return to investment in education in a Large number of countries. The
emerging pattern is one of declining returns through time, a fact that is
interpreted in the context of alternative theories on the relationship
between education and earnings, such as human capital, screening, labor
market segmentation and the maintenance of the status quo from generation
to generation. The evidence lends support. to a human capital view of the
world. Also, the slow rate of decline of the returns to education over
time dispells fears that education might have overexpanded, especially in
developing countries.

TIME TREMDS OF THE RETURNS TO EDUCAON:
CROSS-ATOIAL EVIDENCE
by
George Psacharopoulos*
1. Introduction
Since the early days of the humarL capital literature, special
attention has been paid on the over-time behavior of the rate of return to
investment in education. 1/ The issue is Lmportant for several reasons.
First, if expenditure on educatinm is treated as investment, then
the yield of such investment should behave as that of more conventional
forms of capital. For example, and other things being equal, the rate of
return to investment in schooling should decline following educational
expansion.
Second, the behavior of the returns to schooling over time
constitutes a partial (although by no means sufficient) test of a number of
labor market theories related to educatiot. For example, a pattern of
declining private returns over time would be inconsistent with labor market
segmentation or radical (social class) explanations of earnings
differentials. (Gordon 1972, Bowles 1972). The reason is that such
theories assume a status quo or immobilitsy of labor with different
educational qualifications between income classes. Similarly, declining
social returns to investment in education would be inconsistent with the
screening hypothesis because the more educated would be unable to
persistently maintain their productivity-unrelated earnings advantage over
the less educated. (Arrow, 1973).
* The views expressed here are those of' the author and should not be
attributed to the World Bank.
1/ For a theoretical discussion and some! empirical evidence from the
United States and Israel, see Becker (1975), pp. 6-7, 75-77 and
205-2 13.

-2-
Third, for educational policy making, it is important to
know when the returns to investment in education have fallen enough to
not justify further expansion, or when the returns have increased
above the alternative discount rate to recommend expansion. (Blaug,
1967). 1/
Fourth, in the theory and practice of educational planning,
it is crucial to know the elasticity of substitution between different
types of educated labor. (Dougherty, 1971). A pattern of near
constant returns over time would be consistent with a near infinite
elasticity of substitution between more and less educated manpower --
an empirical fact that would invalidate attempts to forecast the
precise future needs in skilled manpower for economic development.
Lastly, a pattern of declining returns to investment in
education, would be consistent with the equalizing effect of
educational expansion on income distribution, i.e, other things being
equal, an increase in the number of more educated relative to the less
educated would narrow the reward structure and hence lead to a lower
index of income inequality. (Marin and Psacharopoulos, 1976).
1/ This proposition is based on the assumption that economic
considerations is the only criterion for planning. Of course
other values, e.g., universal literacy or political expediency may
override the narrow economic criterion.

-3-
II. The Evidence
When the field of the economics of education started, over
time evidence on the returns to educationi existed only for the United
States. Although such evidence continues to be scarce today, there
have been sufficient estimates for a number of countries to dictate a
stock-taking exercise.
Tables A-1 and A-2 present the existing evidence on the
returns to education estimated within a given country for more than
one point in time. The tables correspoEd to the two basic
methodologies of estimating the returns to education, i.e., the
elaborate method, and the earnings function method (see
Psacharopoulos, 1981). The first gives returns by level of education,
whereas the second gives the returns to the average year of schooling
irrespective of the level of education.
The evidence appearing in the Appendix Tables is
indiscriminate, in the sense that no efEort has been made to render
the rates comparable between years. This is not an easy task because
the rates in each year may have been based on different samples or
methodologies. Also, the over-time changes in the social returns to
education are affected by changes in the unit (resource) cost of
education between the two years.
Bearing in mind this limitation, the overall trend in the
returns to education is an over-time decline. Out of the 71 pairs of
end year estimates in the two Appendix Tables, the returns to
education have declined in 45 cases between the earlier and later
year.

-4-
In order to have a more accurate picture of the over-time
behavior of the returns to education, Table 1 reports a set of more or
less, if not strictly, comparable estimates. All rates reported in
Table 1 are based on the same methodology, which is the Mincerian
semi-logarithmic earnings function with three arguments: Years of
schooling, years of experience, and the square of the latter. (Mincer,
1974). Also, all rates are based on a sample of similar nature in the
two years, which in most cases is the country's population census. In
all but one case (Colombia) the rates in the two years have been
estimated by the same author, which further assures consistency of
sampling and methodology in the two years.

5-
Table 1
Comparable Over-Time Returns to Investment in Education,
Selected Countries
Annual Change
Rate of Return Years of Rate of Return Years of
Country Year (percent) Sehooling (percentage points) Schooling
Developing Countries
Brazil 1970 16.5 3.9
1980 14.5 5.7 -0.20 +.18
Chile 1974 8.3 5.9
1980 9.6 8.1 +0.22 +0.19
Colombia 1965 17.3 2.8
1974 12.8 5.0 -0.05 +0.18
Venezuela 1975 13.7 4.6
1984 11.2 6.6 -0.28 +0.22
Cyprus 1975 12.5 8.4
1984 11.0 9.5 -0.17 +0.12
Industrial Countries
Canada 1/ 1972 16.0 10.0
1980 11.6 11.8 -.04 +0.18
France 1962 16.3 9.4
1977 15.4 11.4 -0.06 +0.13
United Kingdom 1972 9.7 10.5
1975 8.0 10.5 -0.57 0.00
United States 2/ 1950 7.1 9.7
1960 10.5 10.6 +0.34 +0.09
1970 8.4 12.1 -0.21 +0.15
1980 6.4 12.5 -0.20 +0.04
Source: Rates of return from Table A-2 unless iitated otherwise. Mean years of schooling
for Brazil, Venezuela, Cyprus, Canada and the UK from the original source of the
respective rate of return estimate. Chile (1969, 1981) and Colombia (1964, 1978)
from Psacharopoulos and Arriagada (1986), Table 2. United States 1950, refers to
the median year of schooling of the population from United States Bureau of the
Census (1975), H617, 1960 onwards from United States Bureau of the Census (1986),
Table 216. France, from INSEE, "Formation Qualification Professionelle,"
respective years.
Note: 1/ Rates refer to university education computed on the basis of mean earnings of
40-44 year old workers in Dooley (1986), Table 1.
2/ Rates refer to college education computed on the basis of rental prices in
Waldorf, Kunze, Rosenblum and Tannen, (1987) Table 8.
3/ Estimated on the basis of end years corresponding to the schooling means rather
than the rate of return years on the left.

-6-
Out of the nine pairs of end-year estimates appearirng in
Table 1, there are seven declines. (The less than 1 percentage point
increase in the Chilean case may well be due to sample variability or
particular labor market conditions in one of the two end years). In
most cases the rate of decline amounts to less than one half
percentage point per year. The last column of Table 1 documents the
deepening of human capital stock in the countries in question by means
of the change in the educational attainment of the labor force. 2/
The evidence on the over-time behavior of the returns to
education is consistent with what Tinbergen (1975) described as the
"race between technology and education," namely, educational expansion
shifts the supply curve of educated persons to the right, whereas
technological advances also shift the demand curve for skilled persons
in the same direction. Therefore, the eventual observed intersection
between supply and demand over time is an empirical matter.
Figure 1 presents the outcome of such race for one
industrial and one developing country. Thus, in the United States
during the decade of the fifties, demand shifts dominated supply
shifts and the returns to education increased between 1950 and 1960.
Thereafter, however, the expansion of education has dominated demand
with the result of a falling rate of return pattern. The Venezuelan
case is also one of falling returns between 1975 and 1984.
2/ It is reminded that the translation of an increase in enrollments
(flow measure) to a higher level of educational attainment of the
labor force (stock measure) is a very slow process. (For a
documentation of this point see Psacharopoulos and Arriagada,
1986).

-7-
r RATE oF 5;
(% ~) aE%tru/-
8/, Rrz-r N S
loI I
D
DIS
D
9 1 1 3
MEAN YEARS oF SCHOOLiNN
RATE OF
(/o) |RETUjN
14~~~~~~17
IL ~~~~~~~D
VENEZUELA 4
o -D
4 5 7
MENN YEARS oF SCkiOOL)NC
FiGURE I THE RkCE 5ET.WEEN EDUCATiON AND TECHNo%.oGY
IN T7O covN�tES

-8-
III. Conclusion
The evidence presented above lends support to the economic
interpretation of the over-time behavior returns to education, i.e.,
the rates decline over time following educational expansion.
Is the magnitude of the decline such as to raise doubts
about the economic profitability of further educational expansion?
The answer to this question must be negative on two main grounds:
First, the returns to education, especially in developing countries,
continue to be above a reasonable measure of the opportunity cost of
capital, say 10 percent. The decline of the returns to education
recorded above has taken place during a period of rapid global
educational expansion.
Also, the evidence seems to be more consistent with the human
capital theory of the relationship between education and earnings,
rather than the alternative theories of screening, segmentation, or
the maintenance of the status quo from generation to generation.

-9-
Arrow, K., "Higher Education as a Filter," Journal of Public Economics,
July 1973.
Becker, G., Human Capital, 2nd ed. (NBER), 1975.
Blaug, M., "Approaches to Educational Planning," Economic Journal, No. 306,
June 1967: 262-87.
Bowles, S., "Schooling and Inequality from Generation to Generation,"
Journal of Political Economy, May-June (Supplement) 1972.
Demetriades, E. and Psacharopoulos, G., "Educational Expansion and the
Returns to Education: Evidence from Cyprus," International Labour
Review, 1987 (Forthcoming).
Dooley, M.D., "The Overeducated Canadian? Changes in the Relationship
Among Earnings, Education and Age for Canadian Men: 1971-81," Canadian
Journal of Economics, 19, (1), 1986: 142-159.
Dougherty, C.R.S., Optimal Allocation of Investment in Education, in
Chenery, H.B. (Ed.) Studies in Development Planning, Harvard
University Press, 1971.
Gordon, D., Theories of Poverty and Unemployment, Lexington, 1972.
Hansen, W. L., "Total and Private Returms to Investment in Schooling,"
Journal of Political Economy, ApriL 1963.
Hines, F., Tweeten, L. and Redfern, M., "Social and Private Rates of
Return to Investment in Schooling by Race-sex Groups and Regions,"
Journal of Human Resources, vol. 3, 1970.
Mincer, J., Schooling, Experience and Earnings, Columbia University Press,
1974.
Mingat, A. and Jarousse J.P., "Earnings Functions, Generation Effects
and Filtering: A Test on French Data," Centre National de la Recherche
Scientifique, IREDU, Faculte des Science Mirande, B.P. 138, Dijon
France (Mimeo), 1986.
Marin, A. and Psacharopoulos G., "Schooling and Income Distribution," The
Review of Economics and Statistics, vol. LVIII, No. 3, August 1976.
Psacharopoulos, G., "Returns to Education: An Updated International
Comparison," Comparative Education, vol 17, No. 3, 1981:321-41.
Psacharopoulos, G., "Returns to Education: A Further International Update
and Implications," Journal of Human Resources, Fall 1985.

- 10 -
Psacharopoulos, G. and Arriagada, A.M., "The Educational Composition of the
Labour Force: An International Comparison," International Labour
Review, vol. 125, No. 5, pp. 561-574, September-October 1986.
Psacharopoulos, G., "Economic Development and Investment: The Role of the
Work Force," Journal of Educational Planning and Administration,
(India), May 1987a.
Psacharopoulos, G. and Arriagada, A.M., "Earnings and Education Among the
Self-employed in Colombia," EDT Series No. 70, 1987.
Psacharopoulos, G., "Earnings and Education in Brazil: Evidence from the
1980 Census," EDT Series No. 90, 1987.
Psacharopoulos, G. and Steier, F., "Education in the Venezuelan Labor
Market, 1974-1985," EDT Series No. 93, 1987.
Shapiro, D.M. and Stelcner, M., "Earnings Regressions and Decompositions of
Earnings Differentials," Earnings Disparities in Quebe After Bill
101, Concordia University, Working Paper No. 14, 1985.
Tinbergen J., Income Distribution: Analysis and Policies, North-Holland,
1975.
Waldorf, W.H., Kunze, K., Rosenblum, L.S. and Tannen, M.B., "New Measures
of the Contribution of Education and Experience to U.S. Productivity
Growth," Bureau of Labor Statistics, U.S. Department of Labor, 1987
(Mimeo).
United States, Bureau of the Census, Historical Statistics of the United
States, Colonial Times to 1970, US Printing Office, 1975.
United States, Bureau of the Census, Statistical Abstract of the United
States, US Printing Office, 1986.

Table A-1
The Returns to Investment in Education by Year (pertent)
Social Private
Country Year Prim. Sec. Higher Prim. Sec. Higher
Africa
Kenya 1971 19.2
1980 13.0
Malawi 1978 15.1
1982 15.2
Upper Volta 1970 25.9 60.6
1975 27.7 30.1 22.0
1982 20.1 14.9 21.3
Asia
India 1952 44.0 1.0.0 57.0 30.0
1960 17.0 11.0 9.0 23.0 12.0 11.0
1965 13.4 15.5 10.3 17.3 18.8 16.2
1978 29.3 15.7 1.0.8 33.4 19.8 13.2
Iran 1972 34.0 11.5 15.0
1976 15.2 17.6 L3.6
Pakistan 1975 20.0 11.0 27.0
1979 14.6 6.7 9.4
Philippines 1971 8.5 9.5
1977 8.5 16.0
South Korea 1967 9.0 5.0
1969 11.0 9.5
1971 14.6 9.3
1973 12.2 8.8
1980 8.1 11.7
Taiwan 1970 26.5 15.0 17.6 18.4
1972 12.3 17.7 12.7 15.8
Thailand 1970 30.5 13.0 11.0
1972 63.2 30.9 18.4
Continued

Table A-1 (Continued)
Social Private
Country Year Prim. Sec. Hither Prim. Sec. Higher
Latin America
Brazil 1970 23.5 13.1 24.7 13.9
1980
Colombia 1973 20.7
1976 24.9
Peru 1972 46.9 19.8 16.3
1974 34.3 9.0 15.0
1980 41.4 3.3 16.1
venezuela 1957 82.0 17.0 23.0 18.0 27.0
1975 16.4 14.6 11.2 24.5 20.2 21.5
1984 17.5 10.5 8.7 24.3 12.4 15.0
OECD/Europe
Australia 1969 14.0 13.9
1976 8.1 21.1
Cyprus-1 1975 15.0 11.2 14.8
1979 8.6 8.1 14.1
1984 15.5 4.5 14.1
Cyprus-2 1975 10.5 9.7 11.6 8.6
1979 6.8 7.6 7.0 5.6
Greece 1962 6.3 13.7 7.2 14.0
1977 5.5 4.5 6.0 5.5
France 1962 14.3 11.5 9.3
1969 16.2 12.0 9.6
1976 13.5 10.8 9.3
Germany 1964 4.6
1978 10.5
Continued

Table A-1 (Conltinued)
Social Private
Country Year Prim. Sec. Higher Prim. Sec. Higher
OECD/Europe
Great Britain 1971 11.0 7.0 14.0 27.0
1972 3.6 8.2 11.7 9.6
1973 8.0 8.0 6.0 16.0
1975 7.0 7.0 9.0 22.0
1977 8.0 6.0 9.0 17.0
1978 9.0 7.0 11.0 23.0
Japan 1967 10.5
1973 4.6 6.4 5.9 8.1
1976 8.6 6.9 10.4 8.8
1980 5.7 8.3
United States 1939 18.2 10.7 16.0 14.5
1949 14.2 10.6 14.5 10.1
1959 10.1 11.3 19.5 13.6
1969 10.7 10.9 18.8 15.4
1970 11.3 8.8
1971 12.5 8.0
1972 11.3 7.8
1973 12.0 5.5
1974 14.8 4.8
1975 12.8 5.3
1976 11.0 5.3
Source: Unless stated otherwise, PsLcharopoulos (1985), Appendix
Table A-1. India 1952 and L960, Psacharopoulos (1987a),
Appendix. Venezuela 1975 andl 1984, Psacharopoulos and Steier
(1987). Cyprus 1984, DemetrLades and Psacharopoulos (1987).
United States, private rates, 1939, Becker (1975), p. 206;
1949, Hansen (1963), Table 5; 1959, Hines (1970).

Table A-2
The Coefficient on Years of Schooling:
Mincerian Rate of Return, by Year
Country Year Coefficient
(percent)
Asia
India 1954 9.4
1970 6.7
1978 7.4
1980 9.7
Hong Kong 1976 6.3
1981 6.1
Malaysia 1970 14.0
1978 22.8
South Korea 1976 6.5
1980 11.1
Latin America
Brazil 1970 16.5
1980 14.5
Chile 1974 8.3
1975 9.8
1976 10.0
1977 11.3
1978 8.6
1979 8.8
1980 9.6
Colombia 1965 17.3
1971 16.7
1973 17.6
1974 12.8
1975 15.4
1977 14.4
1978 14.4
1984 11.0
Guatemala 1975 10.8
1977 11.3
Venezuela 1975 13.7
1984 11.2
Continued

Table A-2 (Continued)
The Coefficient on Years of Schooling:
Mincerian Rate of Return, by Year
Country Year Coefficient
(percent)
OECD/Europe
Canada 1970 7.8
1980 8.6
Cyprus 1975 12.5
1979 9.8
1981 5.2
1984 11.0
France 1962 16.3
1970 19.2
1977 15.4
Germany 1974 12.1
1977 12.6
Greece 1960 9.2
1964 8.6
1977 5.8
United Kingdom 1972 9.7
1975 8.0
United States 1959 10.7
1969 7.8
1973 8.2
Source: Unless otherwise stated, Psacharopoulos (1985), Appendix
Table B.1. Brazil, Psacharopoulos (1987b). India,
Psacharopoulos (1987a), Appendix. Cyprus 1984,
Demetriades and Psacharopoulos (1987). Colombia, 1974
and 1984, Psacharopoulos, Arriagada and Velez (1987),
Table 4. Venezuela, Psacharopoulos and Steier (1987).
France, MiLngat and Jarousse (1986), Table 2. Canada
(Quebec), Shapiro and Stelener (1985), Tables Rl and AII
R-1.