It is well known that the raw wage data indicate that on average women earn less than 80% of what men earn in similar jobs. When economists do research to control for some factors that might explain the gap (such as differences in occupational choice or skills) some, but not all, of the gap goes away. We get to numbers less than 10% when we do that. In other words, we are not able to explain about 10% of the gender wage gap on the basis of measurable factors alone.
A basic explanation for this would then be that the remaining gap comes from things that are not measurable. And perhaps the least measurable thing would be discrimination. Whether the remaining 10% difference is due to discrimination for real can never be known. It could in fact either over- or under-state the degree of discrimination. It might understate it if for example women end up selecting occupations on the basis of pre-market discrimination, or if there is discrimination in the acquisition of skills throughout life. It might overstate it if there are significant impacts of female labor force membership that are not measured but which lead to lower wages. We might not be able to measure, for example, some aspects of productivity that are tied to time off from the job due to child-care.
But in typical discussions of these issues, one factor has gone terribly under-reported. I can see why, because this factor could not explain the earlier wage gaps or the narrowing of the wage gap over the last 50 years. But it might actually still have a say in why the gap persists or isn’t perhaps even negative. What is that? Could it be that the existence of anti-discrimination laws, though well intended, raise the costs of hiring women as compared to otherwise identical men? After all, the empirical work of Acemoglu demonstrated pretty convincingly that this indeed happens for the case of disabled Americans in response to the passage of the ADA in the early 90s. I am not familiar, offhand, with research that has looked into this aspect of the male-female differences in labor market outcomes, but if there is not work done here it would be a terrific topic for a research paper.
Why, by the way, might it be more costly to hire a similarly situated woman? If a man and a woman are each performing badly on the job, for whatever reason, it may be “easier” to fire a male employee because he is very unlikely to be able to claim he was being discriminated against when he was fired. There are lots of empirical ways to tease this out – perhaps there are papers out there that do it. When I get a breather I will look around. But this is certainly not something that came up in this interesting and informative interview.