All serious researchers agree that the true gender pay gap in the United States is nowhere close to the sensational figure of 20 percent now current in the popular mind and media. Yet most serious researchers do agree that, after controlling for (most of) what seem to be the relevant economic factors, some small gap does remain. It’s approximately five percentage points (give or take a couple). So what explains this small gap?
Prejudice, irrationality, or systematic error are possible explanations. But in competitive markets – which most labor markets are, over time – even a such a small gap seems unlikely to persist for reasons such as these. Reasons such as these, after all, are profit opportunities.
So what else might explain this small gap separating the earnings of women from those of men? In response to this recent Cafe Hayek post, two readers have suggested (in e-mails) plausible candidates.
The first is from Steve Hardy:
The one factor that doesn’t seem to be accounted for when you adjust women’s pay is the litigation risk. This may account for the 4 or 5% difference. The rational decision for an employer would always be to hire a young white man over a woman, minority or old white guy with all things being equal. The young white guy is the only one who can’t sue you for discrimination if fired.
The second is from UMSL economist Dave Rose:
Much of the remaining 5% can be explained by the greater likelihood that women marry-up the occupation distribution than men, which increases their transaction costs in the labor market and therefore reduces their ability to effectuate best matches for their human capital. I came up with this argument years ago based on the observation that the highest paid male in our department was paid significantly less than the lowest paid husband of any female department member. To see how this theory squared with the facts I did an empirical paper with Anne Winkler to get a sense of the strength of the effect. See “Career Hierarchy in Dual-Earner Families,” Research in Labor Economics, Vol. 19, 2000, 147-172.
Each of these hypotheses is quite plausible. Both effects might operate in tandem. Other forces also, of course, might be in play.