20 Dec. 2017 | Comments (0) Share Follow @Conferenceboard
Acceptance of LGBTQ people in all spheres of society – work, family, and community – has grown at a remarkable pace in the United States. A recent Pew Research Foundation study reported that 92% of all LGBTQ adults felt that society is more accepting of them than a decade ago, and 87% of adults report personally knowing someone who is gay or lesbian (up from 61% in 1993). Same-sex couples throughout the country can now get legally married after the 2015 Supreme Court decision in Obergefell v. Hodges. LGBTQ people are highly visible in the media, on television, in the movies, and in the C-suites of major companies like Apple, Google, and IBM. For LGBTQ people, it has certainly seemed as if, in the language of columnist Dan Savage’s 2010 campaign to combat the epidemic of LGBT youth suicide, “It Gets Better.”
Whether these massive changes have translated into improvements in workplace outcomes for the average gay man or lesbian, however, is not so obvious. There is, for example, no federal nondiscrimination protection on the basis of sexual orientation or gender identity. A natural question, then, is: have the shifts in approval of LGBTQ individuals corresponded to equivalent improvements in their paychecks?
Economists and management scholars have been crunching the numbers on this question for over 20 years, and until very recently, nearly all the studies have found an identical result: if you compare the earnings of two men with similar education profiles, years of experience, skills, and job responsibilities, gay men consistently earns less than straight men (between 5% and 10% less). The stability of this finding has been remarkable: it has been replicated across numerous datasets in several different countries (e.g., Canada, the United Kingdom, and the United States) and time periods. It seemingly did not budge.
Until now. In a recent paper, a PhD student and I analyzed data from a major federal survey in the United States that had not previously been used in this literature – presumably because it only recently began to ask about sexual orientation – and found that the gay male earnings penalty had disappeared. And not only had it disappeared, it had turned into a 10% premium, meaning that gay men in recent years earned substantially more than straight men with similar education, experience, and job profiles. We went back through the published literature to see if we were making new or strange measurement or specification choices. We were not. We double- and triple-checked the dataset for other patterns that would indicate some fundamental error or data problem. We found none. We subjected the gay male earnings premium to a host of extra tests to see if we could make the result go away. We could not.
Once we had accepted that the finding was not going anywhere – that it was “real” – we set about trying to understand and explain it.
The simplest explanation that came to mind first was the Dan Savage explanation: “It Gets Better.” One interpretation of the literature’s near-universal prior finding of a gay male earnings penalty was that it was a consequence of labor market discrimination against gay men. If that’s the case, then, naturally, improved attitudes toward LGBTQ people would reduce this penalty. Moreover, a few patterns in the literature support this possibility, including the fact that two recent well-controlled field experiments failed to find meaningful differences in employment outcomes for fake candidates whose profiles were manipulated to be either gay or straight (one fielded in 2013 where the candidate’s profile on a social network site was listed as either “interested in” men or women, and the other fielded in 2010 where the candidate’s resume listed a leadership position in an LGBT-related student group or a non-LGBT-related student group).
The null findings of these recent resume studies contrast sharply with an earlier 2005 controlled resume study that also used the LGBT-student group approach and found substantial differences in the likelihood of getting a callback for an interview in favor of the straight candidate, a difference about as large as the black/white callback difference in the well-known Bertrand and Mullainathan “Emily and Greg/Lakisha and Jamal” resume study. The patterns from these experiments certainly were consistent with the idea that better attitudes toward LGBTQ individuals could translate into better workplace outcomes for that group.
And yet there are also patterns that make the Dan Savage explanation difficult to square. One is that while “It Gets Better” seems reasonable for explaining the gradual disappearance of an earnings penalty, it does not seem well suited for explaining the emergence of an earnings premium (did it really get that much better?). Another is that although we find a very different result than prior work for relative earnings of gay men compared to straight men (a premium versus a penalty), our companion analysis for women found a nearly identical result to decades of published work. Previous studies have found that lesbians tend to earn more than straight women with similar education, experience, skills, and job characteristics, and our estimate using different data was right in line with those of prior work. Is it plausible that it gets better for gay men but not “even better” for lesbians?
In the end, we don’t have a great way to explain why the gay male earnings penalty disappeared and turned into a premium. But the finding does suggest several avenues for future study.
First, there are increasingly more large federal surveys with information on sexual orientation and workplace outcomes, as well as education, experience, and job characteristics. Scholars should see if the gay male earnings premium we have identified replicates in other recently fielded surveys.
Second, because it is clear that workplace dynamics associated with sexual orientation are different for sexual minority men than for sexual minority women (recall that there has been consistent evidence of a gay male earnings penalty and a lesbian earnings premium for most of the past two decades of study), more research is needed to understand the nature of workplace attitudes regarding sexual orientation and how these might differ between gay men and lesbians. It could be, for example, that historically strong associations between gay men and the HIV epidemic contributed heavily to negative attitudes toward gay men specifically and that reductions in these views benefited gay men relative to straight men but not lesbians relative to straight women.
Finally, it is possible that the changing nature of family lives is strongly linked to the changing nature of workplace chances for the LGBTQ community. Prior work has shown that sexual minority women enter into and formalize their same-sex relationships at a higher rate than sexual minority men. But fundamental changes in family opportunities and responsibilities brought about by recent nationwide same-sex marriage may be exerting very different influences in gay male households than in lesbian households, and this changing nature of household specialization – theorized by Nobel Prize-winning economist Gary Becker – could be producing some of the patterns we have documented. A gay male couple who gets married may have one partner select out of the workforce to focus on caregiving responsibilities; this might make the other partner more productive at work, resulting in relative improvements in gay men’s earnings relative to those of straight men. If the relatively lower earning partner systematically selects out of the labor market, this productivity effect would be compounded by a compositional change in the sample of relatively higher earning gay men we observe working. And if the effect of relationship recognition has less effect on women in same-sex couples – perhaps because they were more likely to be functioning as a household unit in the absence of formal recognition – then this could explain the large difference we see in relative male earnings compared to prior studies and the lack of difference we see in relative female earnings compared with prior work.
On the whole, our recent research study likely raises more questions than it answers. But in documenting that the gay male earnings penalty has not only disappeared but in fact has re-emerged as an earnings premium, our results challenge scholars to understand differential workplace experiences of sexual minority men versus sexual minority women and highlight the strong interconnections between the spheres of work and family for LGBTQ Americans.
This blog first appeared on Harvard Business Review on 12/04/17.