на главную поиск contacts

The Causal Effect of Education on Earnings

Опубликовано на портале: 31-12-2003
Amsterdam: North-Holland, 1999, cерия "Handbooks in Economics", vol. 3a
Education plays a central role in modern labor markets. Hundreds of studies in many different countries and time periods have confirmed that better-educated individuals earn higher wages, experience less unemployment, and work in more prestigious occupations than their less-educated counterparts.1 Despite the overwhelming evidence of a positive correlation between education and labor market status, social scientists have been cautious to draw strong inferences about the causal effect of schooling. In the absence of experimental evidence, it is very difficult to know whether the higher earnings observed for better-educated workers are caused by their higher education, or whether individuals with greater earning capacity have chosen to acquire more schooling.

Economists' interest in this issue was stimulated in the late 1950s by growth accounting exercises which found that rising education levels could explain much of post-war US productivity growth, leaving little room for technological change (see, e.g., Becker, 1964; Griliches, 1970). Skeptics noted that this conclusion was only valid if the observed cross-sectional earnings differences between education groups reflected true productivity differentials, rather than inherent ability differences that happened to be correlated with education (e.g., Denison, 1964). The emergence of large-scale microeconomic datasets in the 1960s led to an outpouring of research on education and earnings, much of it focussed on the issue of "ability bias" in the earnings differentials between more- and less-educated workers. In his landmark survey of the 1960s and 1970s literature, Griliches (1977) concluded that such biases were small - potentially even smaller than other biases that lead measured earnings differences to understate the causal effect of education. In his earlier review of the evidence, Becker (1964) had similarly concluded that ability biases were overstated by critics of the human capital paradigm.2 Despite the careful reasoning of these earlier surveys, however, many analysts continue to believe that the measured partial correlation between schooling and earnings significantly overstates the true causal effect of education, and that findings to the contrary are counter-intuitive.

1. Introduction and overview

2. The human capital earnings function
2.1. Functional form
2.2. Measurement of education
2.3. Which measure of earnings?
2.4. Summary

3. Causal modelling of the return to education
3.1. Theoretical issues
3.2. Observed schooling and earnings outcomes
3.3. Measurement error
3.4. Instrumental variables estimates of the return to schooling
3.5. Limitations of instrumental variables
3.6. Family background
3.7. Models for siblings and twins
3.8. Summary

4. A selective review of recent empirical studies
4.1. Instrumental variables based on institutional features of the school system
4.2. Estimators using family background as a control or instrument
4.3. Studies of education and earnings using twins
4.4. Direct evidence on the heterogeneity in returns to education

5. Conclusions

Appendix A
A.1. OLS estimation of a random coefficients model
A.2. Estimation of a random coefficients model
A.3. Measurement error in a bivariate regression model


Ключевые слова

См. также:
Richard B. Freeman
George Psacharopoulos, Harry Anthony Patrinos
World Bank Policy Research Working Papers. 2002.  No. 2881.
David Card, Alan B. Krueger
Journal of Political Economy. 1992.  Vol. 100. No. 1. P. 1-40. 
Edward Lazear
Journal of Political Economy. 1977.  Vol. 85. No. 3. P. 569-598.