The Statistical Fairy Tale They Call The Gender Pay Gap

The worst time of year for anyone with an intense allergy for abused statistics is election times: of all the many things economists and math teachers have failed to convey to the public, the difference between percent and percentage points must be the worst and most prevalent one – the amount of headache I suffer from listening to journalists reporting election outcomes or polling results is incredible! Below in occurance, but definitely equalled in annoyance, are the hopelessly repeated claims about a gender pay gap. They generally occur around March 8 (or October/November, after which women apparently work for free), when good virtue signalling commentators across the political spectrum try to illustrate how women are oppressed, by reaching for the alleged Gender Pay Gap. So, pre-empting all those sensationalist claims you are about to witness tomorrow, here is a critical refutation. Similar to debates over 'Austerity', you have to see the simple truth: there is no gender pay gap.

  

First, to be clear, there are many ways in which we could discuss gender differences in the work force: labour force participation, working conditions, sectors, professional careers, educational attainment etc. The one factor we seem to almost always talk about, which has somehow become synonymous with "Gender Discrimination" or "Gender Inequality", is the often-cited earnings for men vs women. Media figures given in the U.S. normally say 77 cents on the dollar, with Australia and the U.K. reporting numbers around 82%. Such numbers are created by adding up all incomes earned by women in a pile and comparing it to the size of men's equivalent pile. Quite useless.

Of course, everybody understands that earnings are given by factors other than gender (productivity, working hours, overtime, profession or sector, flexibility and a whole range of other variables) making the randomness assumption behind these arguments – expecting an equality between earnings of the two groups – quite remarkable. Why would two differing groups receive similar earnings in the first place?

As many economists and commentators have explained over and over again, the fact that we observe such a number between unadjusted and aggregated groups of people is unsurprising and doesn't indicate discrimination or that anything sinister is at play – after all, there are "Pay Gaps" between highly-educated Asian men and low-skilled Hispanic women, basketball players and office workers, rural and urban populations, perfectly acceptable differences carrying no basis for moral outrage. Instead economists have tried to show the reasons behind those differences in order to address allegations about discrimination and unfair pay for women. Once economists start to adjust the Pay Gap to account for these factors, it almost completely disappears: when we compare apples-to-apples, women and men in same positions (within as well as across professions) working same hours, we hardly ever find any Gap left to argue over.

In the British context, even the Office of National Statistics accepts this and reports an adjusted figure before the sensationalist one picked up by the media (18-20%), where the gender gap in "median earnings for full-time employees" was 9.4% last year. Indeed, The Guardian recently even reported how unadjusted figures for the British workers aged 22-29 showed that women earn more than men in aggregate. The Fairy Tale of 82p/£, or 77c/$ that we'll read about tomorrow should be understood in this context.

So what is the reason for the "unexplained" residual some studies find (9.4%; or the infamous 8% Motherhood Wage Penalty)? The critics may have a point here: that could be due to discrimination. But it would be a mistake to assume it, according to Harvard economist Claudia Goldin, probably the most well-known scholar on income inequality and gender differences. She suggests that the very small residual we sometimes find after having controlled for age, working hours, position etc is explained by "temporal flexibility", referring to the desire or need for mid-career women to trade high-performing, competitively-paid jobs for jobs that ensure flexibility, presumably with child-rearing in mind.

I explore what happens when employees cannot “hand off” clients, patients and customers in a costless fashion. The framework fits into a model of compensating differentials and provides the foundations for the costs of providing a worker amenity such as flexible work hours. [...]. The framework suggests that “non-linearity” in labor value arises when it is costly to employers to allow workers to be off the job temporarily, when it is difficult to hand off clients to colleagues, and when interdependent teams must coordinate schedules – as in many finance and legal occupations. Note that nonlinearity here means that a lawyer working 30 hours a week is worth less than half what a lawyer working 60 hours is worth. (pp. 29-30). 

Another, more controversial reason, is a kinda of risk-premia in the kinds of jobs men and women do. Here is Mark Perry of the AEI:

men represented 92.3% of workplace fatalities in 2014 (and the male share of job-related deaths has been consistently that high in every previous year) because men far outnumber women in the most dangerous, but higher-paying occupations like logging, mining and roofing that have the greatest probability of job-related injury or death. In contrast, women, more than men, show a demonstrated preference for lower risk occupations with greater workplace safety and comfort, and they are frequently willing to accept lower wages for the greater safety and reduced probability of work-related injury or death.

Economists have searched high and low for it, but are still unable to find evidence for discriminatory pay practices (including a U.S. Department of Labor report from 2009). Instead, the literature consensus are Goldin-style explanations of flexibility or Motherhood penalties.

Critics (for example Joshua Holland here) have generally objected either of two things: Women's choices are shaped by biological constraints (giving birth) and social stereotype (socially acceptable fields of work/conditions/working hours); and the relatively more hardcore argument that regardless of the reason, there's still an aggregate difference that is damaging. These "Fallback positions" are unconvincing for three reasons:

1. Present-day women are, as Christina Hoff Sommers argued in Time Magazine last year, "among the best informed and most self-determining human beings in the world" (and in history, we should add); taking time off to give birth is very much a deliberate consideration rather than outcome of some arcane oppressive force. 
2. We can't determine where unfairness occurred simply from where the statistics were gathered; if we can't find evidence that employers or working practices intentionally harm/discriminate against women, it's not clear why the differences matter (maybe origin elsewhere, but then go find the source rather than condemning employers' "unfair practices"). As stated above, most people don't object to income differentials between basketball players and bus drivers – in essence, minor gender pay differences are of a similar nature, so why should we care?
3. If you have a problem with the kind of choices women make, attacking employers or publicly complaining about pay seems both futile and childish. Instead, perhaps, you should tell women that you disapprove of their lifestyle choices? Of course, that would lose you the moral highground and the virtue signalling purposes of ranting about pay on March 8. 

In summary, there is very little reason to believe that there is anything sinister going on with the minor (or nonexistent) pay differentials we can observe between aggregated groups of men and women. The "Gender Pay Gap" most virtue signalling news outlets will rant about tomorrow has very little research supporting it, making it one of those annoying statistical fairy tales people keep believing.