A Note on the Analysis of Male-Female Income Differences.
Argues that wage discrimination studies using hedonic wage equations must be viewed with skepticism because of the sensitivity of this technique to omitted-variable bias. Greenberg and McCall's R-1343 indicates that women teachers in Michigan receive $539 less on average than comparable men. However, only degree level, and not the exact number of college credits accumulated, was known. For San Diego teachers, this same regression specification yields a $188 average differential. When the six specific classes to which teachers are assigned on the basis of college credits are used in the regression analysis, however, all but $7 of the differential disappears. Apparent sex discrimination in Michigan may thus be a statistical artifact based on omission of the detailed educational information. Without better information on personal characteristics, or more sophisticated statistical techniques, economists stand little chance of isolating the true effects of racial and sex discrimination on market earnings using hedonic wage equations. 8 pp. Ref.