This paper applies single and simultaneous equation fixed-effects (FE)
and random-effects (RE) panel data estimation techniques to obtain ma
le and female earnings function parameters. Using the Panel Study of I
ncome Dynamics (PSID), the paper finds that earnings appreciation with
experience and depreciation with labor market intermittency are compa
rable for men and women. Further, skill atrophy rates increase not dec
rease once one controls for heterogeneity and endogeneity. Finally the
unexplained male-female wage differential declines from 40 percent to
20 percent when one adjusts for heterogeneity. Adjusting for endogene
ity depends very much on the choice of instruments. However, when adju
sting for endogeneity the gender earnings gap falls and approaches zer
o percent. These results hold for two separate subsamples so that the
estimates appear robust independent of sample selectivity.