No abstract

Using data from the 1991 National Longitudinal Survey of Youth, the authors estimate earnings equations for each of seven occupational categories and the aggregate sample. When fringe benefits are excluded from the compensation measure, a gender coefficient is statistically significant (that is, women are found to have received significantly lower compensation than men) within six of the seven occupational categories, the exception being the most female-dominated category. When an index of compensation that includes fringe benefits is used, however, a gender coefficient is significant in only one category, which contains relatively heterogeneousjobs. Gender-specific regressions are used to estimate what part of the earnings gap between men and women is due to differences in traits. The results indicate that occupational assignment is the primary determinant of the pay gap, a result that is consistent with a "crowding" explanation of that gap. M ost studies of the gender pay gap have used the wage rate to measure compensation. Non-pecuniary dimensions of work and fringe benefits are usually ignored or are included as explanatory control variables. It is possible that the gender pay gap is actually smaller than is indicated by traditional analysis if the jobs chosen by women have higher non-pecuniary rewards than those chosen by men. That is, women may choosejobs, asJoni Hersch (1991) has