From New Results on the Disparities between Same-Sex and Different-Sex Couples in the Home Mortgage Market by Nir Eilam and Yeonjoon Lee. From the Abstract:
Despite improving public sentiment towards same-sex couples, research suggests that they still face discrimination in various markets. We empirically estimate disparities between same-sex and different-sex couples in the home mortgage market, an under-studied, yet important, market. Improving previous research, we use confidential administrative data on the universe of home mortgage applications in the US from 2018 until 2021. We identify same-sex and different-sex couples according to the gender of the mortgage applicant and co-applicant. Then, controlling for a rich set of lender, borrower, and loan characteristics, some of which are important in mortgage decisions but were not available in previous research like credit scores, we find that same-sex couples are 8.8 percent more likely to be denied a home mortgage than similar opposite-sex couples and conditional on being approved, are quoted an interest rate that is 0.8 percent higher. We explore heterogeneity by regions, by acceptance of same-sex marriage, and pre- and post-COVID. Interestingly, we also find that same-sex couples default significantly more (53.9%) than similar different-sex couples, which suggests an unobserved characteristic that causes same-sex couples to default more, and could explain a part of observed disparities in mortgage approval, undermining results in previous research.
The vicimtologists went searching for victims and found them. Same-sex couples are discriminated against in comparison to opposite-sex couples, paying higher mortgage rates and being denied mortgages at nearly a 10% greater rate.
Woof! After all the anti-discrimination legislation and after all the apparent social progress, old evil habits die hard.
But that is not, of course, the story at all. Once again the data tells a different story than the researchers apparently wanted to tell.
They claim the different outcomes despite the comparison groups being "similar" and despite "controlling for a rich set" of characteristics.
But the two groups, gay and straight couples, are not similar at all.
We also find that same-sex couples default significantly more (53.9%) than similar different-sex couples.
Since whether mortgages are extended and at what rate they are made are almost entirely a function of the probability of the lender being repaid, a 54% difference in default rates is no minor issue. Further, if all else were similar, I would expect same-sex couples, having a 55% greater default rate, to have a much higher denial rate than 8.8% and to pay much more than a 0.8% interest rate premium.
Stating the blindingly obvious at the very end that the difference in default rates:
Suggests an unobserved characteristic that causes same-sex couples to default more, and could explain a part of observed disparities in mortgage approval, undermining results in previous research.
In fact, the researchers have not found that same-sex couples "still face discrimination in various markets." What they have found is that couples with poor loan repayment prospects "still face discrimination in various markets."
A finding that comports with economic theory, the law and with experience.
It is disappointing to see researchers from the US Federal Reserve and the University of North Carolina demonstrate such disregard for the truth. They are flat out misleading about what they found.
And it is doubly disappointing because their finding suggests further interesting lines of research.
It could actually be that same-sex couples are being discriminated against despite the 54% higher rate of default. I would have thought that the researchers might have been able to impute the actual denial rate and mortgage interest rate for varying levels of default. If same-sex couples are being denied at a greater rate and paying higher mortgage interest rates than the formula would indicate for a 54% higher default rate, then clearly they are being discriminated against (though issues of uncertainty about risk assessment and effect size might obviate that conclusion.)
Since the calculation should be relatively straight-forward, I can't help but suspect that they did do it and what they probably found was that in fact same-sex couples are more likely to receive a mortgage and at lower interest rates than otherwise comparable opposite-sex couples with a similarly 54% greater probability of default.
Exactly the opposite of what they wanted to find.
The second research arena that their finding opens up: why do same- and opposite-sex couples have such different default rates when supposedly they are in all other ways similar according to the researchers" What is/are the "unobserved characteristic that causes same-sex couples to default more"?
Especially since the researchers explicitly say that their research included credit scores and that they took into account credit scores. In other words, if we take the researchers at face vale, two populations with an identical credit score had a 54% difference in default rates.
Since credit scores are pretty explicitly intended to capture inability to repay loans, that is a pretty surprising finding.
The only obvious answer I can come up with is that the existing mortgage origination process is so rigorous that the actual default rate is so small that a 54% differential is inconsequential compared to other factors. E.g. if, under the existing mortgage origination process, population X has a mortgage default rate of 0.001% and population Y has a mortgage default rate of 0.00154%, the differential in default rates is indeed 54% but in absolute terms it is relatively inconsequential.
Maybe that is true. But when I think about the history of the mortgage origination process in the US, I don't have the impression that it has been nearly that good at screening down default rates to those low levels. As best I can tell, mortgages run 5-10% in delinquencies (behind payment) and 1-2% in defaults averaged over time. If the baseline default rate were 1% then 1.54% would be a very material difference.
The only obvious "unobserved characteristic" that I can think of which might not have been taken into account is relationship dissolution rates. I assume that relationship dissolution would be highly correlated with a sharp increase in financial default on the mortgage.
The challenge, of course, is how does one measure relationship dissolution rates, especially since legal sanction of same-sex marriages is relatively recent anywhere in the world?
I am somewhat startled by the data in Wikipedia on the matter. They cover the Netherlands, Scandinavia, the US and the UK. Same-sex marriages in these nations tend not to go back more than a couple of decades at most.
Broadly, what the data suggests is that same-sex marriages dissolve at the same or a lower rate than opposite sex marriages but that lesbian marriages dissolve at twice the rate of gay marriages. Hmmm. I'm not certain that I had any firm convictions as to what the data might show but I did not anticipate those numbers.
The biggest residual question is how reliable is the data given that formal marriage is still a relatively recent phenomenon. That introduces two separate issues 1) what will the longer term steady state numbers look like once gay marriages are firmly and widely established/accepted? and 2) Is there currently some sort of distorting demographic effect given the recency of gay marriage and do studies normalize for that?
In other words, as an extreme example, divorce rates among 40-year marriages tend to be very low. You'll have a reasonable number of 40-year marriages in the opposite-sex population. You'll have zero 40-year marriages among the same-sex population. How do those demographic differences, purely a function of legislative history, affect the results?
I don't know.
OK, (yeah, rabbit hole), do we have data for relationships as opposed to "marriage" given that same-sex marriage is relatively recent.
Are Same-Sex or Heterosexual Relationships More Stable? by Theresa E. DiDonato asks all the right questions but does not provide the answers. More accurately, she gives comparative answers but without the absolute numbers which does us no good.
The more I search the more muddled it gets. Same-sex gays differing from same-sex lesbian, differing from opposite-sex couples. Living together or not. Bi-sexual complications in the data. Differences in age gaps between the different groups, etc.
So I don't know whether same-sex relationship dissolution is behind the greater 54% differential in default rates.
Some other "unobserved characteristic"? Health status perhaps? I am grasping at straws.
Interesting questions.
But fundamentally what the researchers in this paper report but do not readily admit, is that there is no clear evidence of discrimination against same sex mortgage borrowers on a like-for-like basis. Rather, there is a 54% differential in default rates which is apparently not due to credit scores. There is some factor associated with same-sex borrowers which is causing a 54% higher default rate but which the researchers and the lenders cannot yet identify.
That is pretty interesting and I would have thought that would be lede rather than the Woke victim-seeking misleading write-up which was done instead.
Posts
No comments:
Post a Comment