The gender wage gap and sample selection in Japan

•I examine the convergence in the gender wage gap in Japan from 1992 to 2002.•The changes in female labor participation behavior are taken into account.•I consider a three-choice framework: no work, non-regular work, and regular work.•The wage structure for female regular workers has become more ability based.•The behavioral changes explain 63% of the observed convergence for regular workers.

This paper examines how much the observed convergence in the gender wage gap in Japan from 1992 to 2002 is affected by changes in female labor participation behavior. The existing literature focuses on full-time workers, but the replacement of low-paid regular full-time workers by non-regular workers and the introduction of the Equal Employment Opportunity Law may have changed the selection of females in full-time work force. I consider a three-choice framework for a woman based on the Roy model: a woman chooses either no work, non-regular work, or regular work. This framework shows that large (potential) wage inequality within a gender can draw high-earning people into the workforce and push low-earning people out. I apply Lee’s (1983) method to the microdata from the Employment Social Surveys 1992 and 2002. The results show that female selection in regular workers became stricter in 2002 and women with low-earning ability tended to be pushed out from regular work force. The wage structure for female regular workers has become more ability based. This change in the female selection explains 63.27% of the observed convergence in the mean log wage gap between female regular workers and male workers.