Pay gaps [updated]

Lots of folks seem very upset about a government press release about a paper that doesn't seem yet to be publicly available.  If the press release comes out without the paper, my prior is that they want the headline but don't want anybody looking too closely 'till the media's lost interest.  Like when ASH was touting the numbers from its study on the costs of smoking but refused for two months to give me a copy.

The press release says that the Ministry of Women's Affairs has found that, comparing students with the same Bachelor degree (Commerce, for example), men earn more than women right after graduation and that the gap rises over time.

Using ocular least squares, peering into the main classes for the different Commerce subdisciplines, I can't help but wonder whether they've missed something rather important.  Namely, Economics and Finance are male-dominated while Marketing and Management are not.  Peering down the photos for our core calculus based intermediate micro course - the one you need for honours in econ or for a finance degree - about 40/150 students are female.  Peering down the same list for the second year Management "Consumer Behaviour" course, it looks about 50-50.  Suppose that we have good reason to think that an undergraduate degree in management just isn't worth as much in the marketplace as an economics or finance degree.  Some would go so far as to argue that an undergrad degree in management is next to useless: that it's best done as an degree after a main degree in the field in which you hope to manage (engineering followed by management, accounting followed by management, etc), or as an MBA post-experience.  But if we've got gender differences across majors, and if we've got reasonable pay differences across majors, then it would be shocking NOT to find substantial pay differences across genders if we aggregated at the degree level.

Of course, there are plenty of other reasons for differences in salaries across individuals: the total compensation bundle for any job includes both pecuniary and non-pecuniary factors, and if there are gender differences in the evaluation of those factors, then we'd expect to see differences in the more readily observed element of the pay packet.  Things like work hours flexibility are typically more valued by women.  Moreover, if you anticipate time out of the workforce down the track, you'll choose a different career path from day one than you'd pick if you didn't expect to be taking time out.  Those are the kinds of things that are hard to control for in a regression.  Luckily, most of the wage gap goes away once you condition on the kinds of things that are observable: part time and full time status, industry, and so on.  Hit the "wage gap" tab below....

Update: Kiwi Particle Physicist (in comments below) notes the study is now online. They aggregate degrees by broad degree type then compare mean income for males and females: just a straight comparison of means, no regressions with conditioning variables.

After one year of employment the income gap between male and female graduates averages around 6 percent. This average masks quite significant variations within different fields of study. Women leaving with information and technology or agricultural and environmental qualifications, for example, earned more on average after one year than men. On the other hand, the income gap for males and females with a health qualification averaged more than 20 percent within a year.8

8.The large gap in incomes in health is likely to be partly explained by the high level of occupational segregation in the sector: lower paid nursing jobs are dominated by women, while the highest paid jobs (e.g. surgeons) are dominated by men.

Footnote 8 indeed! Similarly for education: the proportion of males choosing a diploma in the lower-paid early childhood segment as compared to high school is much smaller. But my worries about commerce being confounded were misplaced: economics and econometrics are instead bundled with "Political Science and Policy Studies; Human Welfare Studies and Services; Librarianship, Information Management and Curatorial Studies; Philosophy and Religious Studies; Economics and Econometrics; and Sport and Recreation." as part of "Society and Culture: Additional". Now, I've always thought economics best fit with social science rather than commerce, but also bundled with sport and recreation? Hmmm....