Decomposing two values of a binary variable: application to the unemployment differential in Hungary

This paper presents a study of the individual determinants of unemployment in Hungary. It seeks to understand the reasons why Hungarian women were, in 1992, less likely to be unemployed than their male counterparts. It estimates first, the probability of being in the stock of unemployment and secondly, one particular inflow into unemployment. Adapting standard discrimination analysis to the case of binary variables, the paper decomposes the differential in average unemployment rates by gender. The paper shows that differences in individual endowments account for very little of the gender gap in the stock of unemployment. On the other hand, slightly less than half the unemployment gap could come from differences in one digit occupations and sectors of employment between male and female employees.