A broad range of studies -- spanning methodologies, years, populations, and nations -- have found that gay men earn less than comparable heterosexual men and lesbians earn either more than, or similar to, comparable heterosexual women.   Although these gender differences in the earnings effects of sexual orientation have not been explained, three hypotheses have been suggested: (1) unmeasured gender-linked characteristics that affect earnings are distributed differently by gender and sexual orientation; (2) strength of attachment to market work, relative to household work, vary by gender and sexual orientation; and (3) anti-gay bias is greater for men.  

We use gender composition of occupations and sharing of market work within households to proxy, respectively, the variation in the unmeasured characteristics and differences in sharing of household duties by gender and sexual orientation.   Variation in measured sexual orientation discrimination by gender over time provides some evidence on the role of gender differences in this discrimination. 

Using data from the 5% Public Use Microdata Samples of the 1990 and 2000 U.S. Census and the American Community Survey for 2001-2007, we find that the unmeasured characteristics explain about half of the gender differential in the gay pay gap, although differences in the household division of labor have no effect on the gay pay gap among women.  We also find that the gay pay gap declines for men, disappearing by 2007 but does not change for women, when gender composition of occupation and market participation of mate are controlled.  This time pattern is consistent with, at least historically, greater gay bias against men.