In the US, there are persistent income gaps between Whites and Blacks . They are generally explained by historical and current discrimination. Similar gaps exist between other ethnic groups and in other countries and of course between men and women.
Readers of this blog know, that there are equally persistent gaps in some cognitive abilities or behavioral patterns that might easily explain big parts of these gaps.
But there are other reasons to be skeptical of tales of labor market discrimination. Discrimination on the labor market creates arbitrage opportunities. If 99% percent of business owners are not willing to pay Blacks as well as Whites for the same work done, the 1% willing to do so have the opportunity to scoop up great workers for little money. It’s just free money, and if labor costs are a significant part of business expense this kind of discrimination quickly becomes unaffordable.
Now, labor market discrimination of course is in principle possible. In the above example, having a racist (or sexist) majority of 99% is surely enough to leverage social sanctions against the 1% and to consistently pay some groups less for the same work. But as soon as social sanctions against non-racists are no longer plausible, labor market discrimination becomes equally implausible.
But wait a second, isn’t there the well publicized phenomenon that fields into which women went when they entered the workforce saw a drop in wages? 
Well, when the iron curtain fell and Poles went into the business of harvesting asparagus in the West, the wages in this beautiful and traditional endeavor also declined precipitously. This is called wage dumping and it is the opposite of labor market discrimination. Poles were just willing to do the job for less.
And of course the same is true for women. Women are willing to accept lower wages for jobs that are compatible with family formation. I.e. low stress, part time, no travel, close by kinda jobs. They got into these jobs by wage dumping.
But how do we explain different call-back rates for equal resumes? There is a large literature on the fact that resumes that only differ in the identifiable ethnicity can have quite different call-back rates when applying for the same job . This is a very rich and mathematically interesting question, that touches on key concepts like Simpson’s paradox, the normal distribution and regression to the mean, which is one reason why it is never accurately portrayed in the media. We are going to close that gap in the next post.
 Female wage dumping
 Different call-back rates by ethnicity.