Editorial: The Pay Gap Revisited

Do women earn less than men? It’s more nuanced than you think.

Econometrics has literally changed my life. It is by far my favourite subject this year. What I enjoy the most about it is its applications to the real world: I love it because there is a beauty and logic in getting data to mean something.

Although, sadly, it tells you that you can probably never draw any conclusions about anything because of omitted variable bias. So the study you saw linking wine consumption to a decreased risk of heart disease is probably not robust, and neither is the one linking increased ice cream consumption to drowning (hint: what causes people to crave ice cream, as well as a swim?)

On Friday, I was working through a problem set which had the raw data for a research paper on the pay of male and female American MBA graduates twenty years since graduation. If you look at the averages, men in the study earn about 28 percent more than women—and in the UK, the gender pay gap among full-time workers is 9.1%. But is this a good way of looking at the issue? When controlling for MBA GPA, the fraction of finance classes taken, actual post-MBA experience, and work spells, along with weekly hours worked, job function and employer type, being female is associated with a 3.8 percent decrease in annual earnings, which is not statistically significant. In the UK, women make just 1% less than men who have the same rank and function at the same employer.

 When looking at men and women who have exactly the same characteristics, there is no significant difference in the wages they are paid. However, I wouldn’t say that this means that the difference in pay is not an issue. What this argument misses is that there is a societal reason why women are more likely to take time off (they have children), and for why they work fewer hours per week (more likely to take on the burden of housework and childcare), and for why they are less likely to take finance classes (women are not as encouraged to go into STEM). We need to think about why controlling for all those variables is necessary.

Although a woman with the same characteristics and the same work experience as a man is likely to earn a similar amount, women are overwhelmingly less likely to have the same work experience, by virtue of their being female and taking time off to have children—which is a gendered issue. In econometrical terms, controlling for the time taken off work when finding out whether women earn less than men is a bad control because it’s an outcome of the dependent variable—being female

So the main question is: what causes women to be in lower-paying jobs in lower-paying sectors? If girls are told that because of their gender, they aren’t very good at maths, this would push them to study courses that don’t involve so much maths, leading to lower-paying jobs (and this is not the case in Asia, where both men and women are pushed to study STEM). Then this is not necessarily the fault of the job market, but rather sexism in the system.

Of course, some women will prefer to take a few years off work to have children. But to what extent are these preferences what women actually want, and to what extent are they the easier option, because the alternative is going back into work when women are expected to shoulder most household responsibilities, and childcare is difficult and expensive to arrange?

All I know is that it’s more nuanced than saying “The gender pay gap is not statistically significant”.