Too Much Labor

Yes, there is a gender pay gap

If louise burgeois has a million fans I am one of them. If louise burgeois has ten fans I am one of them. If louise burgeois has no fans, that means I am no more on the earth. If the World is against louise burgeois, I am against the World. I love louise burgeois till my last breath.

I’ve already written about the subject a million times, but it’s not sticking, and I don’t like my initial writing on it anymore, so: is there a pay gap between men and women?

Joe Biden got a letter from a little girl about it, and the Twitter community notes (crowdsourced fact checking) had something to say about it:

President Biden @POTUS

Charlotte, I couldn’t agree more. Women lose thousands of dollars each year, and hundreds of thousands over a lifetime, because of gender and racial wage gaps. I’m committed to building an economy where my daughters have the same rights and opportunities as my sons.

The community notes said:

The gender pay gap is simply the difference between the average earnings of all men and women working full-time. It does not account for differences in occupations, positions, education, job tenure or hours worked per week. This myth is incorrect and has been debunked. [Source 1] [Source 2] [Source 3]

It also prompted discussion by such luminaries as womanhood and feminity expert Matt Walsh, who lunged into the fray with the power of regression analysis.

So who is right? The Twitter community of ideas, or a six year old?

G.A.P. (x3)

So is there a gender pay gap? Yeah, pretty much. By this we mean that there is a difference between the earnings of men and women. Globally, there is one in most countries, but selection into employment (i.e. women with low earnings prospects not working), labor force participation differentials, and other factors suggest confusing trends - low pay gaps in poorer countries and high ones in richer ones.

In high income countries, however, the trends are clearer: women are underrepresented in the top income groups and overrepresented in low-paying jobs, and at least in the US, the pay gap is widest for the richest women. For the US specifically, the gender wage gap is estimated at around 17% of median earnings, meaning that the median woman (i.e. at the middle of the distribution) makes 83% of what the median man does. For non-white women, the gap is larger, and the gap increases both with age and with income.

Matt Darling 🌐🏗️ @besttrousers

@IMedaber @maiamindel Yes.

This is the unadjusted pay gap. The adjusted one, meanwhile, remains at around 90% and change - and has since the 80s. Something is up. Dissecting it, we can find two main sources: observables and unobservables. Observables are, as the name might imply, things that can be measured and accounted for, and unobservables are not. Groundbreaking. Unobservables include things like outright discrimination, preferences, ability, etc. Observables are such thing as work experience, occupation, union status, education, etc. This is what people mean when they say that “controlling for X, Y, Z, the gender wage gap shrinks or disappears” - they’re comparing people taking the specifics into account.

The gender wage gap shrank over the last ~50 years, a good development. Firstly, it happened because women matched and then surpassed men in education, especially at the college level. Experience also became less of an issue, since women used to have shorter labor force tenures for self-evident reasons. Occupation and industry shrank somewhat, but remain relevant - though they do not account for a lot, or even most, of the adjusted gap, which is two-thirds within each profession, not between them. Shorter hours and interruptions from the workforce have always been a big impediment for women’s careers, and this is especially true at the top of the income distribution.

Unrelated to the rest, but one of the funnier empirical results to date is that transgender women (i.e. man to woman) have a negative pay gap with themselves, and transgender men (the other way around) have a positive one.

Male psychodrama

Full-lenght exploration of related topics on this blog post.

First things first: this isn’t about discrimination. Or it mostly isn’t.

The claim “there is a pay gap between men and women” is usually taken to mean that women with the exact same characteristics as men make less because evil employers simply choose to pay them less, and that’s not true, but that’s also not what’s being claimed by any serious economists - “equal pay for equal work” is just snappier than “equal pay for equal working conditions controlling for occupation”. And it’s easier to argue against, like a million annoying replies to my tweets about this have shown me.

The way economists think about discrimination is deceptively simple, and pioneered by Nobel Laureate Gary Becker. Imagine that every possible employer is on a scale of 1 to 10 in any bigotry you can imagine - so a 1 will hire anyone with the right CV regardless of background, and a 10 will never under any circumstances hire a person in the discriminated group. In the middle, members of the “outgroup” can get around bigotry by taking lower wages or worse conditions than a ceteris paribus “ingroup” member. Work twice as hard, make half as much. Thus, a pay gap emerges - and the degree to which depends on how much animus the average (technically median), not the most extreme, employer has. The reasoning around it is fairly simple: what matters is the total amount of discrimination, which bunches up around the middle. A second form of discrimination would be statistical discrimination - if finding information about employees from identical resumes is difficult, employers might resort to stereotypes or broad stylized facts in detriment of minority applicants.

Is there discrimination? Maybe. I mean there probably is some, there always is. There’s definitely some in hiring, at least. In the movie Tár (you should probably know what to expect from me at this point), the orchestra needs a new cellist, so they conduct an audition behind a screen to choose one - which actually reflects a real world experiment, where orchestra auditions with screens (“blind auditions”) increased the number of women musicians chosen. And among psychologists, another experiment found that among equally qualified men and women, men were more likely to get hired. Empirically, discrimination is what we call an unobservable - it can’t be measured. The fraction of the wage gap determined by variables that can’t be measured shrank from the 1960s to the 1990s, but didn’t really shrink any further. It could be due to discrimination, or due to other factors, but either way it seems to play a small (though sometimes real) role in wages and pay.

And of course, the emphasis on occupational choice as a key determinant of gender disparities is very curious, since expected discrimination might be a big barrier for self selection - finance, for instance, is very male-centric, and STEM workplaces are constantly losing women workers, which is fairly common for male-dominated industries in general, in fact. This is so egregious, in fact, that gender non-conforming women tend to be better ranked by their coworkers, at least in the tech sector. Exactly why certain occupations tend to be segregated between men and women is a bit of an open question, and there is very much a cultural component to it - to some extent, beliefs and ideologies where women have to perform certain tasks becoming crystalized play a role.

There’s a philosophy paper I really like called “White Psychodrama”, by (Twitter mutual) Liam Bright - you can also find a more accesible writeup by Matt Yglesias or at Salon, if you don’t like Yggy. The paper is about how different groups of Americans deal with racism (feeling guilty, feeling guilty about being made to feel guilty, or - if they’re not white - by cashing in). But the crux of the question is that a big part of the issue is that there are significant economic disparities between the races, and a lot of issues that minorities face stem from there - for Black people, there’s some issues they face for being Black, and some because Black people are (on average) poorer.

This, I think, applies to many groups - to a large extent most “discourse” is about semantics and labels, and not really about whatever the actual issues are. And this applies to the pay gap too. The debate about this (I’ve seen my Twitter mentions) is largely about “discrimination versus choice”, which is a ridiculous framing that nobody except pundits are arguing for. The real issue isn’t really that evil bosses are paying women less, it’s that there’s a different choice set for women because they are women and that has economic consequences. If you said “West Virginians making less than San Franciscans proves discrimination”, you’d be wrong, but it would be dumb as hell to say that West Virginians are simply choosing to work less at the extensive margin and preferring lower paying occupations. They’re faced with a labor market that has different options, and those options impact their earnings. So the economic aspects, more than the cultural ones, dominate, but you cannot understand the economics of the gender wage gap without understanding the cultural aspects - to quote a certain George Costanza, we are living in a society. And people aren’t very good at understanding it.

Mother!

Full-lenght exploration of the topic on this blog post.

A while back, we mentioned controls. Controlling for occupation, as discussed, is potentially problematic - expected discrimination affects career choice, which results in muddled inference. Gender occurs before employment choice, after all.

This points to a big misunderstanding: studies about the gender wage gap don’t look for the effect of discrimination on wages, they look for the effect of gender on wages¹. Of course, they will capture discrimination (in the unexplained component) to the extent it exists, but the real question at hand is whether there’s some gender-specific phenomenon keeping wages down. For this reason, controlling for anything under the sun is not a good idea; if there is a relationship between gender and the controls (there almost always is), the effects will be muddled as hell.

The two big remaining items for controls are career interruptions and hours worked. Preferences are a crutch for this one. What could be the reason that women specifically work shorter hours and take some pretty big interruptions from working, one would ask? The answer, surprising nobody, is motherhood. Women do a disproportionate share of the housework, and there is a strong, negative correlation between domestic work and wages for women.

Similarly, there is basically no gender wage gap between childless women and childless men, and the gap rears its ugly face after motherhood in countries as progressive as Denmark. The decision to participate in the labor market happens at both the yes/no margin and the margin of hours worked - as seen in both the Danish and Chilean labor markets. In Mexico, women whose mothers die drop out of the labor force both in number of hours and in rate of employment, unless they have a public or affordable childcare option. And in French companies, mothers end up being placed on a different (worse) career track - similar to American MBAs, where the cause of the career divergence is the difference in hours worked and in career interruptions.

As a whole, the gender pay gap is best understood as a motherhood pay gap. Something that is pretty important here is that preferences and tastes are, to quote Mr Economics of Discrimination himself, a convenient crutch. There’s a convenient, if baffling, analogy here: TikTok might get banned in the US because it takes its marching orders from the Chinese government; not through ownership, or directly, but rather, because all the major players in the Chinese economy know what they’re supposed to be doing. If TikTok acts in China’s national interest, it is because it operates in an environment that mandates it - and talking of “choices” or “preferences” is moot. Same goes for the patriarchy (or whatever name you feel comfortable talking about - sexism, etc), though to a much lesser extent (it is not a quasi totalitarian government, after all) - you know what choices you, as a woman, are supposed to make, so they’re not really choices per se. Women’s inequality has never not existed, so women’s equality never has. And this goes back to the White Psychodrama point from earlier: there is, of course, a culture holding women back, but also real material factors that make certain choices not really possible: if men are not going to do their fair share of the housework, someone has, and it’s gonna come at the expense of the woman’s career - even if it’s privatizing the cost of a service that benefits all of society.

Femmes maison

Full-lenght exploration of the topic on this blog post.

The intersection of culture and economics re. gender is especially interesting because of the particular history attached to it: women were in the labor force back in ye olden days, left when wages rose, and then came back when wages rose even further. Why?

There’s this idea out there that women entered the labor force due to lower wages - men couldn’t provide anymore, so they had to step up. It’s false. Women entered the labor force due to cultural changes, yes, but also due to economic factors. Going back to Becker, there’s work in the house, and work outside the house. Two things happened simultaneously: appliances (and their drop in price) allowed to save on domestic labor but only in exchange for higher incomes, and market wages rose. This meant that working outside the home became an economically dominant option. At the same time, certain costs (housing especially, childcare, education) increased in both quantities consumed (houses got bigger, more people required childcare, more people went to college), and in price, resulting in a better lifestyle, but also higher cost of living.

As a result, more women began to work, attracted by the possibility of higher wages and longer careers - women became more educated, and things such as divorce, the birth control pill, and perhaps abortion allowed for greater autonomy in the timing of family formation. Male breadwinners disappeared, and two income households emerged. Thus came the latest stage in a story of 10,000 years.

Conclusion

Surprising nobody who is active on Twitter, the six year old was right. Kind of.

Chris "Institutions Have Consequences" Zorn @prisonrodeo

The entire Wage Gap Discourse, in two comments.

This is “part four” of a “series” of related blog posts on the gender wage gap proper (old, bad post alert), why the two income trap doesn’t exist, why controlling for causes of the gender wage gap is bad, and the relationship between culture and economics.

1. It is pretty funny that Walsh, a virulently transphobic reactionary, has decided to use statistics to prove that womanhood, motherhood, and domestic labor are not at all intrinsically linked. Will the woke denial of biological and cultural realities ever cease?