Mobility: Race & Gender
[ Part of a sequence of posts on income mobility ]
Causality
What does it mean for gender or race to cause something?
One a one level, you might define gender and race as genetic phenomena. In this view, the only possible confounders are the genes of your ancestors, and not even that for gender.
On another level, maybe you define gender and race in socially constructed terms. If women are socialized to not pursue STEM jobs, is this their gender causing a pay gap? Is it their gender causing a difference in preferences? Or are the difference in preferences what their gender actually is in the first place?
Such philosophical questions are beyond the scope of this sequence of posts, so I will instead focus on just one phenomenon that I find very interesting.
Race x Gender
It is well know that large income differences are seen between the sexes Gender pay gap in the United States. In Wikipedia and between races Racial pay gap in the United States, with most scholarship on the later focusing on the difference between blacks and whites.
However, what is less well known is that, after controlling for parental income, the racial income gap completely disappears between white and black women, but not between white and black men Race and economic opportunity in the United States: An intergenerational perspective:
What drives this discrepency? Well, on one level, we can decompose the overall gap by considering the gap in wages and the gap in hours worked Race and economic opportunity in the United States: An intergenerational perspective. Black women earn about 1% less per hour than white women and work about 1% more. Meanwhile, black men earn about 7% less per hour and work 9% less.
This kind of analysis kind of passes the buck: why do black men earn less per hour and work fewer hours? At least part of this gap is probably caused by differences in hiring. For instance,
The hypothesis that felony convictions drive the black-white income gap via its effect on hiring is very tidy: there is an intuitive causal relationship, felony convictions statistically explain the vast majority of the black-white hiring gap (conditioned on submitting resumes), and the hypothesis explains the dramatic difference between males and females.
The main problem is that (as far as I can tell) there is no good empirical evidence relating the observed effects on hiring with effects on earnings. Without this link, the most we can say with confidence is that this kind of statistical hiring discrimination causes some non-zero amount. Claiming it drives the bulk of the gap is consistent with the evidence, but not implied by it.
Another important piece of the puzzle is that, controlling for parental income, the gap for college attendance is smaller between black and white women (~3pp) than between black and white men (~7pp) Race and economic opportunity in the United States: An intergenerational perspective and the difference in these gaps is even larger in college graduation rates Table 104.20. Could this education gap be driven by felonies? I don't know.
A more controversial for the racial earning gap was infamously put forward by Herrnstein and Murray: that the bulk of the gap in black-white income is driven by the gap between their average IQs Herrnstein. Even granting the hypothesis, it can't be the whole story, since it predicts a large and equal gap for both sexes, rather than just for men. At a minimum, proponents need to explain why black women earn as much as white women after controlling for parental income. Still there's no denying that the racial IQ gap throws a wrinkle in the discussion of the racial income gap.
For more analysis, see the Wikipedia page Racial pay gap in the United States.