Gender differences in the effect of education on the slope of experience-earnings profiles: National Longitudinal Survey of Youth, 1979-1988

American Journal of Economics and Sociology, The, Oct, 1996 by Kevin C. Duncan

I

Introduction

In a recent article O'Neill and Polachek (1993) attribute much of the narrowing of the gender wage gap during the 1980s to increases in women's work-related characteristics and to the increased monetary returns attributable to these characteristics. The conclusion that gender differences in skills underlie gender differences in earnings is shared by other researchers such as Smith and Ward (1989, 1984) and O'Neill (1985). Specifically, the results of the decomposition analysis used by O'Neill and Polachek indicate that the relative rise in women's returns to a year of work experience explains the largest share, between 35%-40%, of the approximately 1% per year narrowing of the gap since 1976. The relative increase in women's years-of-work experience accounts for 25% of the narrowing, while changes in the structure of the economy explains 20%. The authors also find that the relative increase in the level and returns to women's education contribute significantly to convergence.

O'Neill and Polachek have identified the return to female work experience as a significant determinant of the gender wage gap in the 1980s, however, it remains unclear what roles discrimination and choice play in determining this important factor over this period. For example, as the authors conclude "some of our results could be interpreted as circumstantial evidence that discrimination in the labor market declined in the 1980s. For instance, the steepening of women's age-experience profiles might in part reflect a greater willingness on the part of employers to train and promote women. However, changes in employer behavior may reflect a reassessment based on observed increases in women's work attachment rather than a decline in pure prejudice. Cause and effect are difficult to unravel."(1)

The purpose of the present study is to take a step towards unraveling this cause and effect by analyzing the extent to which women's skill investment decisions were limited in the 1980s. This analysis will involve a test of the complete skill investment model implied by human capital theory which suggests that education and on-the-job training are complementary. Jacob Mincer (1974) observes this relationship between schooling and on-the-job training when he notes that "individuals who invest more in human capital, invest more in both forms of it. Evidently, abilities to learn on the job are positively, though far from perfectly, related to abilities to learn in school."(2) Greater abilities to learn reduce the marginal cost of investments and allow individuals to produce more of both forms of human capital. Since jobs offering more training are characterized by greater earnings growth as work experience increases, human capital theory suggests that the rate at which earnings grow will be higher for those with more education (or, the slope of the experience-earnings profile will be an increasing function of the level of education). In the absence of wage and occupation discrimination and holding labor market attachment constant, the theory predicts that more educated men and women will hold occupations with steeper experience-earnings profiles. Testing this hypothesis is useful not only in determining how education affects earnings, but how premarket investments in human capital transfer into continued skill acquisition and earnings growth on-the-job.

A sample of young men and women who entered the labor market in the 1980s will be used to conduct this test to determine if women receive the same benefit from their education investment as men do. Results indicating that both more educated men and women have steeper experience-earnings profiles would not only support human capital theory, but would also suggest that labor market discrimination was not a significant influence on the skill investment decisions of women in the 1980s. However, results indicating gender differences (favoring men) among comparably educated and market attached men and women would suggest that, in the recent past, discriminatory barriers persist as a problem for educated women seeking to pursue further skill enhancement while working.

II

Data

The data for the study were derived from the National Longitudinal Survey of Labor Market Experience of Youth (NLSY) from 1979 to 1988. The survey contains human capital, earnings, occupational and other demographic information for over 12,000 black and white females and males. In 1979, members of the sample were between the ages of 14 and 22. Respondents serving in the armed forces were not included in this study.

The NLSY is well suited to address the issue raised here because of its focus on young men and women entering the labor market in the 1980s. This cohort has been noted in the literature as one expected to have continued narrowing of skill, market participation and earnings differentials (see Smith and Ward (1989), (1984) and O'Neill, (1985)).

III

Model

To compare the slope of the earnings profiles of men and women and to determine if the profiles are influenced by the level of education, the NLSY data were pooled between 1979 and 1988. The longitudinal sample was employed to estimate the following model.