In a previous post, I explored a central insight of feminist economics: the array of unpaid and underpaid labour women do, both at home and at work. In this post I will look at another central insight: the discrimination women face, particularly in employment, and the related issue of the gender pay gap. One does not have to look far to find self-reports of discrimination by women, and the broad statistics show that women are underrepresented in top paid professions such as lawyers, doctors, and tech executives; and even moreso in positions of power such as politicians, CEOs, and judges. These facts are enough to make us sit up and take notice and for some they may be enough to show that there is a problem. Nevertheless, it is worth diving into the issue in more detail to show beyond any reasonable doubt that women’s position in the economy is harmed by discrimination.
There are several clever approaches that have been taken in the literature to detect discrimination at each stage of employment and education, most of which involve modifying reported gender to see how it alters perceptions and outcomes. For example, researchers sent out CVs which were exactly the same except that one had a male while the other had a female name, and the CVs with a male name had a 50 percentage point higher chance of being offered a better job at a more expensive restaurant. Similarly, students taking an online course (where you cannot see or hear the professor) were misinformed that their professor was a woman, and they evaluated them more negatively than those who took the same course but were (correctly) informed their professor was a man. Though it is impossible to blind people to gender at every stage of most job application processes, an orchestra which was able to do so boosted female employment substantially.
Not only are women less likely to get employed; they are rated less favourably and offered lower salaries should they be employed. A study of science professors found that prospective academics were rated as more competent, hireable, and less desirable as mentees. Subsequently, the women were offered $26,500 on average while men were offered $30,200, and the women were less likely to be hired in the first place. Another study replicated this finding while showing that both men and women were just as likely to engage in this type of discrimination. My favourite example is quite meta: when presented with research like this, men are more likely to criticise it than women. This one is not as carefully controlled as other studies (and both men and women could be biased here), but it’s good example of motivated reasoning in this area.
One more thing to emerge from the literature recently is the Implicit Association Test (IAT), otherwise known as the unconscious bias test, which you can take online. It is a less direct test of discrimination in practice than the above examples, but it is worth mentioning because it attempts to measure the subtle biases that cause the behaviour above by associating men and women with career versus home, or science versus the arts (respectively). One could fill an entire post discussing the IAT because it pertains to more than just gender, is quite new, and has been subject to some debate. A short summary is that generally speaking IATs correlate with explicit measures of bias but retain additional predictive power for behaviour above and beyond these explicit measures, especially for socially sensitive topics. In the case of gender people are found to be biased toward traditional gender roles, with one study finding that using half a million IATs across 34 countries the IAT predicted national sex-differences on grades for science and maths.
As this post pertains to economics, it is worth mentioning the issues that academia and especially academic economics has with discrimination against women. A recent essay demonstrated in meticulous detail how, despite aggregate gains in women’s representation in academia, they are still less likely to be cited (including self-citations), more likely to be in junior positions, more likely to drop out, rated lower by students, more likely to be expected to nurture and support others, and more likely to be victims of harassment. Economics is one of the worst subjects, with widespread reports of harassment and discrimination and underrepresentation which increases at every level. According to a 2018 AEA report, in the US females make up 36% of undergraduate economics students, 27% of PhDs, 14% of full professors, and of course only 1 (out of 49) woman has ever won the biggest prize in the discipline. This has been dubbed the ‘leaky pipeline’.
Gender Pay Gap
What is the combined impact of these subtle but substantial setbacks on women’s position in the economy? One indicator of this is the gender pay gap, which is how much women are paid when compared to men. The gender pay gap can have different definitions, but the most cited statistic is that women are paid 79% as much as men, which comes from comparing women and men in full-time employment. One can use alternatives which swing this percentage in either direction: if one only compares men and women in exactly the same jobs, it is closer to 90%. If one simply compares how much the average woman is paid to the average man regardless of whether they work or not, the number goes down to 61%. How one views this depends on whether one views the choices made by women – around childcare, occupation, seniority, education and more – as free or as a result of discrimination. The above would suggest that discrimination plays a major role, though it’s difficult to put a number on its importance. Regardless, the question of which of these things drive the gender pay gap is an interesting empirical exercise.
The most prominent contemporary gender pay gap scholar is arguably Claudia Goldin, who has used unique data to come to some intuitive but unexpected conclusions. She uses US Census data to show that women sorting into different occupations and education levels is much less important than it used to be – the equalisation of education in particular has narrowed the pay gap, which was $0.56 on the dollar as recently as 1980. Nowadays the gender pay gap is mostly driven by caring for children with an important mediator: the flexibility of work, and it seems to have stalled. Occupations such as law require constant availability, which are more difficult for a primary caregiver to offer, and as such women tend not to do as well in them. What’s more, sometimes this work is rewarded in a nonlinear fashion – somebody who can offer the law firm 70 hours a week will do more than twice as well as somebody who can offer 35 hours a week. Men are much more likely to be in the former category.
Goldin’s uses detailed data on MBA graduates – an example of a demanding and inflexible career path – to explore the lifetime earnings of men versus women in more depth. Her data show the pay gap is small (though not zero) among young women but widens around the time people usually have children. It then proceeds to grow even further, and Goldin conjectures that many women try to have both children and a career but gradually give up and drop out. As corroborating evidence, the gap is much smaller for women without children and is also smaller in occupations which allow flexibility, such as pharmacists. Goldin’s solution is surprisingly dovish: companies themselves need to make effort to make occupations less like law and more like pharmacy to reduce the gap. On the other hand, she does note that a greater role for men in childcare “wouldn’t hurt”.
The story is fairly similar across Western countries. In an IFS working paper the UK pay gap is shown to be about 10% before the arrival of children, widening to 33% by the time the first child is 12 years old. Furthermore, women who work part-time see less wage growth, though the data do not permit investigating Goldin’s theory about flexibility in as much detail as she does. In a (slightly older) review of 11 European countries from 1995-2001, researchers found a pay gap in every country even after adding a number of controls (though excluding occupation). The average is close to 20% but there was a lot of variation, with Ireland and Austria having a highest pay gaps and Denmark and the Netherlands having the lowest. The paper also considers the pay gap across the distribution and finds it wider at each end: both: a ‘glass ceiling’ and ‘sticky floor’ effect. The glass ceiling is consistent with Goldin’s data on MBAs, though she could fairly be accused of paying less attention to lower paid work.
Feminist economics is an entire school of thought which covers a wide range of topics and changes our understanding of existing topics. In both this post and the previous one I have empirically established the differing experiences of women and the effect this has on their role in the economy by focusing on unpaid labour, discrimination and pay. There are topics I’ve not covered including the feminist critique of economic rationality and the role of women in development (I accept that these posts have been Western-centric), and you can find out about these and more on both the Exploring Economics and Economics Education websites. Despite some progress in recent years, these topics are not covered enough in economics and there is a lot we’ve yet to understand.