Ages ago I supervised a superb Honours thesis, which turned into a Masters, looking at the lesbian wage premium. It showed up regularly in the US data: homosexual women earned more than heterosexual women - the opposite of the pattern that obtains for men.
I was curious whether the difference could in part be due to employers' expectations about the costs of accommodating maternity leave. And it looked like that mattered. Hayden Skilling did superb work on it, helped in part by Ron Oaxaca's visit to Canterbury while Hayden was writing.
I'd wanted to use New Zealand data but that seemed to be impossible. We could, with a few clicks, get US data from the ACS without any bother. If we wanted to use NZ data, it would have been impossible. Stats New Zealand just makes it too hard to access NZ microdata. So we wind up with NZ researchers using American data and helping advance global understanding of what's going on in the US.
Hayden's thesis was out in 2014.
It's 2026. Stats NZ just put out a couple of releases looking at earnings among LGBT+ populations.
I wanted to check whether the lesbian wage premium held up in NZ data.
So I went to have a look.
And it's just a big mess.
First, we get a press release highlighting substantial differences in age-adjusted average annual personal disposable income between the (lumped together) LGBT+ population and the non-LGBT+ population, with the transgender and non-binary population having the lowest age-adjusted annual disposable income.
That sounds like discrimination right?
But then you check the second page. The one listed as a "related page" that folks might not click on. That one notes that the LGBT+ population reports disability rates of 25.6%, compared to 15.8% for others. 37.9% of the transgender and non-binary population were identified as disabled.
Disability will mean lower earnings. Stats NZ's big headline press release figures adjusted for age but didn't adjust for disability. That seems like an important omission. People might chalk differences up to identity that are at least partially differences by ability. A cross-tab so you could compare earnings by disability status across the categories could help, but this is one of those 'if we didn't pre-supply the cross-tab, it is unknowable' things.
I downloaded the excel sheet, hoping that I might be able to check for differences in earnings between heterosexual and homosexual men and between heterosexual and homosexual women.
But that appears to be impossible.
In their gender splits by row in the cross-tabs, male and female include the transgendered identifying with each category. Heterosexual males (including the cisgendered and transgendered) earn more than males (including the cisgendered and transgendered) who identify with a sexual minority. That latter category lumps together people identifying as gay, those identifying as asexual, and many others. There could be substantial differences within that bundling.
So I can't have a clean read on earnings differences between non-trans straight men and non-trans homosexual men.
And among females (including cisgendered and transgendered), mean personal disposable income for heterosexuals is suppressed. Stats NZ does this when reported numbers are too low. But they do report earnings among females reporting as sexual minorities. Which is difficult to understand. The sample size tends to be smaller for minority groups. But even if it were not suppressed, it wouldn't be helpful. Because I wouldn't be able to get difference between cisgendered heterosexual women and cisgendered homosexual women.
I don't know if US data has gotten any worse over the period, but NZ data sure hasn't gotten any better.
I've emailed SNZ asking whether any of this is knowable.
There could be good reasons for lumping groups together as they have; they have a lot of potential categories, and splitting out each one would just mean everything would wind up being suppressed. But splitting out the main obvious categories would seem pretty possible.
No comments:
Post a Comment