174 M.C. Daly et al. Economics of Education Review 19 2000 169–178
education occurred. Thus, by 1985 a substantially larger percentage of men with little work experience also had
a deficient education for their current job. Finally, as Bu¨chel and Weißhuhn 1997a, 1997b found, a signifi-
cantly larger fraction of German women than German men are overeducated.
8. Economic value of surplus and deficit education
The previous two sections have compared the preva- lence of surplus and deficit education over time in the
United States and between the United States and Ger- many. In this section we examine the economic rewards
and penalties for having too much or too little education. To do this we report the results from estimated earnings
functions where an individual’s completed schooling is decomposed into the number of years required for his or
her current job and the number of years of surplus or deficit education. By entering required, surplus, and
deficit education separately we allow for each to have a different effect on earnings. Note that all information
about job characteristics other than the mismatch vari- able are considered to be endogenous in our analysis and
therefore do not enter the models.
In Table 3 we present the results from two specifi- cations of a conventional cross-sectional earnings
regression. In the first specification we include a variable representing the individual’s actual educational attain-
ment i.e. years of completed schooling. In the second specification we decompose completed schooling into
required, surplus and deficit education and enter each as a separate parameter in the model. The dependent vari-
able in both models is the natural logarithm of hourly earnings. The other independent variables are: years of
work experience since age 18, experience squared, and a dummy indicating residence in a city of 500,000 people
or more. Summary statistics for the model variables are reported in Table 5 in Appendix A.
Consistent with other work in this area we find com- pleted education and experience have a positive and sig-
nificant effect on earnings in all time periods and in both countries. As shown in the first and third columns of
Table 3, the rate of return on completed education was 5.9 and 6.5 per additional year of schooling for non-
black men in the United States in 1976 and 1985, respectively. Work experience earned a similar return
among these men, 5.3 in 1976 and 5.6 in 1985. In neither year was the rate of return on education signifi-
cantly different from the return on experience for non- black US men.
5
In contrast, among non-black women
5
This finding contrasts with Krueger and Pischke 1995. They find large differences between the coefficients on edu-
cation and experience among US workers. However, most of the difference between our and their results seems to be associa-
ted with differences in model specification. For example, we
with mean characteristics, the rate of return on education was significantly higher than the rate of return on experi-
ence in both 1976 and 1985.
6
For the average US woman, the rate of return on experience was about 36
of the rate of return on education in 1976, and about 54 of the return on education in 1985.
As is the case in the United States, German men and women enjoy positive and significant returns on edu-
cation and experience. The average German man earned 8.5 for each additional year of completed schooling
and 3.2 for each additional year of work experience. Thus, in contrast to US men, among German men, edu-
cation is relatively more important than experience.
7
Among German men with mean characteristics the rate of return on education was about 2.5 times the return
on work experience. The results are similar for German women. Similar to US non-black women, German
women get a larger return from additional years of schooling than from additional years of work experience.
Having established the general results on completed education, experience, and earnings we turn to the
decomposed version of the education variable. Like others we find that required, surplus, and deficit edu-
cation each have a significant effect on earnings. Surplus and required education increase hourly earnings, while
absolute years of deficit education decrease hourly earn- ings. Among both men and women in the US, the reward
for required education is larger than the reward for sur- plus education. Likewise, the penalty for deficit edu-
cation is smaller than the reward for either required or completed education. These differences are statistically
significant for men and women in both time periods.
The results for Germany reveal similar patterns. Sur- plus education earns a positive return for both men and
women, although the magnitudes are significantly smaller. The rate of return on surplus education is about
one half for men and two-thirds for women of the rate of return on required education. Among German men,
the penalty for deficit education is larger than the return for surplus education. This is not the case for German
women.
report results for non-black men and women separately, whereas Krueger and Pischke examine all men and women
black and non-black combined. When we re-estimate our models using the Krueger and Pischke specification we confirm
their results. However, given the goal of this paper and the dif- ferences in the results for men and women, we maintain our
specification throughout the analysis.
6
Tests for differences in the rates of return were significant at the 1 level.
7
Tests for differences in the rates of return were significant at the 1 level.
175 M.C. Daly et al. Economics of Education Review 19 2000 169–178
Table 3 Effects of education on ln hourly wage rate in the United States 1976, 1985 and Germany 1984 among working men and women
aged 18-64 years
a
United States Germany
1976 1985
1984 Men
Women Men
Women Men
Women Regression I
Completed education 0.059
0.092 0.065
0.095 0.085
0.079 0.004
0.005 0.004
0.005 0.003
0.004 Years of work experience
0.053 0.034
0.056 0.052
0.032 0.038
0.004 0.005
0.004 0.004
0.002 0.003
Work experience squared 20.0008
20.0006 20.0008
20.0009 20.0005
20.0008 0.0001
0.0001 0.0001
0.0001 0.0001
0.0001 Reside in city of .500,000
0.219 0.133
0.173 0.167
0.050 0.080
0.023 0.029
0.031 0.034
0.014 0.019
Adjusted R
2
0.325 0.273
0.257 0.265
0.369 0.32
Regression II Required education
0.061 0.090
0.078 0.109
0.090 0.090
0.004 0.005
0.004 0.005
0.003 0.005
Surplus education 0.045
0.061 0.054
0.086 0.049
0.066 0.005
0.007 0.006
0.007 0.008
0.008 Deficit education
20.034 20.036
20.016 20.025
20.078 20.038
0.009 0.016
0.008 0.011
0.014 0.022
Years of work experience 0.052
0.032 0.052
0.045 0.030
0.037 0.003
0.005 0.004
0.005 0.002
0.003 Work experience squared
20.0008 20.0006
20.0008 20.0008
20.0005 20.0007
0.0001 0.0001
0.0001 0.0001
0.0001 0.0001
Reside in city of .500,000 0.214
0.129 0.158
0.160 0.051
0.080 0.022
0.028 0.030
0.033 0.013
0.019 Adjusted R
2
0.339 0.265
0.294 0.306
0.379 0.333
a
Standard errors are in parentheses. Asterisks indicate significance at the 5 level. The US sample consists of non-black men and women who were either heads or wives partners in a PSID household in the interview year. The German sample consists of
West German men and women. Source: Panel Study of Income Dynamics 1976, 1985 and the German Socio-Economic Panel 1984.
9. Significance of differences over time and across countries
