Measuring and testing for gender discrimination in professions: the case of English family doctors

In 2004 the income of female GPs was 70%, and their wages (income per hour) were 91%, of those of male GPs. We compare estimates of gender discrimination from Oaxaca decompositions using models of wages (income/hours), OLS and 2SLS models of income, and propensity score matching. We propose a new direct test for within workplace gender discrimination based on a comparison of the differences in income of female and male GPs in practices in which all GPs are of the same gender with the differences in male and female income in mixed gender practices. We find that the coefficients on log hours in the log income models are positive but significantly less than 1, so that log wage models are misspecified. Discrimination, as measured by the unexplained difference in mean log income varied between 21% to 28%. However, our direct tests could not reject the null hypothesis of no within workplace gender discrimination.