HMDA Data Offers Clues on Discrimination Against Gays

JUN 4, 2014 4:46pm ET
Comments (13)

Among Home Mortgage Disclosure Act wonks (there are more of us than you think), all eyes were on a recent kerfuffle between Fannie Mae and the Mortgage Bankers Association over loan denials to African-Americans. But what caught the eye of this HMDA geek was another set of figures much less frequently looked at than denials to minorities: denials to same sex-couples.

These were down in all categories in 2012, for male and female same-sex borrowers, but still higher than denials for mixed-sex couples having a male as the primary applicant.

Higher-priced lending to same-sex couples also dropped in all categories but one in 2012 compared to 2011.

Looking at analyses done by the Federal Reserve and by ComplianceTech, which has a HMDA analysis program called LendingPatterns, denial rates for same-sex couples applying for mortgages in 2012 were lower than denials for one-person applicants of either gender. (Having two incomes versus one might explain this.)

Compared to mixed-gender applicants, though, same-sex couples did not do as well, although two-male couples were denied at about the same rate as mixed-gender couples where a woman was the primary applicant.

Name of graphic

The Fed report shows denials increased for most categories between 2011 and 2012, while the ComplianceTech report, which includes all conforming and jumbo loans, showed all categories with a smaller percentage of denials.

Definitively extrapolating trends for gay and lesbian loan denials from HMDA data is impossible, since lenders do not survey for sexual orientation for their annual HMDA reports. Same-sex couples are not necessarily gay or lesbian; two male relatives could buy a home, for instance, or two women business partners looking for an investment could get a loan. LGBT (lesbian, gay, bisexual and transgender) applicants certainly can obtain home loans and be counted in the single-borrower categories as well.

But the figures are intriguing.

Denials of same-sex couple applicants were down in 2012 versus the year before in conventional (Fannie Mae, Freddie Mac and jumbo) home purchase and refinancing, and in nonconventional (FHA-, VA- and USDA-guaranteed) purchase and refinancing as well, according to the Fed. In some cases, especially on refis, they were down significantly.

For conventional refis, the denial rate dropped from 24.5% to 21.7% for male same-sex couples and from 24% to 20% for female couples.

On the nonconventional side, the drops were even greater. Male couples were denied in 31% of refi apps in 2011, but just 22% in 2012. Female couples saw denial rates drop from 33.3% to 21.7%.

For conventional purchase mortgages, denials were down for both male and female couples, though not as much as on the nonconventional side. Two-male couples saw 17.2% denials in 2012, 150 basis points better than in 2011. Two-female couples dropped to 16.2%, from 18.5% in 2011.

There is, no doubt, prejudice against LGBT applicants. But there’s not a lot of data to show how much.

"We've never studied it. It has never come up for any of our clients," says Michael Taliefero, managing director of ComplianceTech, a fair lending software vendor and consultant in Arlington Va., that specializes in HMDA analysis and supplied me with several custom tables of HMDA data for this column.

Lisa Rice, vice president of the National Fair Housing Alliance, notes that sexual orientation is not a protected class under the Fair Housing Act or the Equal Credit Opportunity Act. (Some states and municipalities have banned LGBT discrimination on their own.)

"We’re trying to get the law expanded," she says, to ban discrimination on the basis of sexual orientation or against veterans. "It’s been a very long campaign for us to get it done."

LGBT discrimination cases can be brought on the basis of gender, she says. The Department of Housing and Urban Development has stipulated by rule there can be no discrimination on the basis of sexual orientation in HUD-funded facilities or for government-insured mortgages.

HUD has charged and settled its first case under its Equal Access rule, alleging LGBT mortgage discrimination at Bank of America. The bank agreed to pay $7,500.

The agency last year released its first study on same-sex housing discrimination, in the rental market rather than the mortgage side. It found that "same-sex couples experience unequal treatment more often than heterosexual couples when responding to Internet ads for rental units, and findings show that gay male couples experience more discrimination than lesbian couples." Discrimination was shown in every metropolitan area it studied.

Rice wonders if the drop in denials to same-sex applicants has any correlation to the increasing number of states that have approved marriage equality.

"People may feel more comfortable applying for a mortgage now" in those states, she says, generating a larger number of qualified candidates.

Same-sex couples make up only a small percentage of total borrowers, and that can skew percentage changes. According to ComplianceTech, there were about 92,000 mortgages made to male couples in 2012, and 82,000 to female couples, out of a total of 9.8 million loans. That's a little less than 2% of all mortgages.

There were about 320,000 applications made by same-sex couples that year, according to ComplianceTech. More than 62% of male and female couples' apps were funded. That is better than the male-alone or female-alone categories, and it matches the mixed-couple "female primary" category. Both same-sex categories are considerably below the mixed-couple "male primary" category, which saw more than 69% of loans originated.

Denials are not the only way mortgages are not funded. There is also what ComplianceTech calls the fallout category, meaning applications withdrawn or incomplete. Male couples had a 17.5% fallout rate in 2012, while female couples had 17%. Again, these were better than fallout percentages for single borrower applications, and about the same as female primary, but considerably higher than the male primary category.

For incidences of higher-priced lending (defined as 150 basis points above the average), the only category where same-sex couples did worse in 2012 was in female-couple nonconventional home purchase lending. That category saw an eight basis point increase to 3.33%. All other categories fell for women, and all categories fell for men.

ComplianceTech did a sort for the total universe (including jumbo loans) of male versus female denials (applications with at least one male versus one or more females and no male) and the difference is significant: Females were denied at a rate of 22% for 2012 while males were denied at 17.5%.

Does that indicate discrimination? And what about that figure of less than 2% of mortgages going to same-sex couples—is that simply reflective of their percentage in the population?

My colleague Victoria Finkle has done an excellent job of sorting out the debate over access to credit by minorities as revealed (or not revealed) by HMDA data. That is an important discussion. Gender questions, whether they involve same-sex couples or not, could use a thorough airing out as well, and I welcome any reflections on this in the comments box below.

Mark Fogarty, Editor at Large at National Mortgage News, writes analysis and commentary based on his 30 years covering the mortgage industry.

Comments (13)
Wow, This discrimination thing has me shaking my head. Mack in the early 90s, the Feds made a big issue about minorities getting denieed at a larger percentage then the rest of the population. Ti me, as a banker, I agreedthat everybody had something in common. Their money is green so the only thing keeping one from getting a loan was the 3 Cs. (Credit, Capacity, and Capital). But the almighty Fed told the bankers that we ere still discrimantiong, so we had to lower our standards with no doc loans to overcome the Capacity issue. Subprime loans to overcome the Credit issue. And "0" down loans to overcome the Capital issue. Well, lo and behold, many people who obtained these loans didn't repay, and we had the foreclosure issue and the Feds told us that minorities were getting foreclosed upon at a higher rate then the rest of the poulation. Imagine that???? So, when I see a stupid a$$ article lime this, I think..."Here we go again"
Posted by Bill A | Thursday, June 05 2014 at 9:03AM ET
Not all joint same gender couples are homosexual couples, more often than not Male A is on the loan to help Male B qualify for the loan but Male B has too many problems with their own credit to qualify.
Posted by | Thursday, June 05 2014 at 11:06AM ET
This article is really terrible. People take statistics and then make a case for discrimination. Ridiculous! I've been writing mortgage loans for 20 years and have never seen discrimination. From a practical standpoint, the LO is the only contact who would know they applicant may be gay. I've never seen an LO who was so prejudice they didn't want a commission from lending someone elses money! Could it be the higher turn down rates on gay couples could be attributed, generally speaking, to lifestyle choices which contribute to a lower than average credit score? I'm not prejudice but if I had to guess, that would be it. My experience it that on average gay couples have had lower scores than others. That doesn't mean that every gay couple has low scores but that on average scores tend to be lower, contributing to higher turn down rates. I've never seen discrimination in writing mortgage loans in 20 years experience.
Posted by Eddie S | Thursday, June 05 2014 at 3:15PM ET
One cannot soundly analyze disparities in lending outcomes without understanding the ways that measures of differences between outcome rates tend to be affected by the frequency of an outcome. I explained that reductions in denial rates tend to increase relative difference in denial rates while reducing relative differences in approval rates as long ago as April 1992 [1] and as recently as May 2014 [2], with close to twenty times in between (as collected in reference 3). Still, few attempting to interpret data on lending disparities understand these issues.

A corollary to the described pattern is a pattern whereby, when denial rates generally decline for two groups, the group with the lower baseline rate will tend to experience a larger proportionate decrease in its denial rate than the other group, while the other group will tend to experience a larger proportionate increase in its approval rate. One observes this pattern in comparisons of the changes experienced by Male-Primary and Male-Male applicants in the Table (based on my estimates of the rates graphically represented there). Male-Primary applicant experienced the larger percentage decline in denial rates (10.4 percent for Male Primary versus 7.9 percent for Male-Male), while Male-Male applicants experienced the larger percentage increase in approval rates 2.2 percent for Male-Male versus 2.0 percent for Male Primary). To the extent that we can actually compare changes in denial/approval prospects experienced by the two types of applicants, the changes would appear to be exactly the same (.06 standard deviations according the methodology employed in Tables 1 and 2 of reference 4).

Not all the changes reflected in the table conform to the described pattern. But one can only soundly interpret the data with an understanding of the patterns by which standard measures tend to be affected by the frequency of an outcome and an understanding a way to measure differences in the circumstances of two groups (or difference in the circumstances of one group at two points in time) reflected by a pair of outcome rates that is unaffected by the general frequency of an outcome. See also reference 5 and 6.
One, of course, must not lose sight of the likelihood that any actual (properly measured) changes in the comparative situation of two groups from one year to the next is random variation.
References (links are prohibited but each item may be located by a name search)

1. "Bias Data Can Make the Good Look Bad," American Banker (Apr. 27, 1992):

2. "The Perverse Enforcement of Fair Lending Laws," Mortgage Banking (May 2014)

3. Lending Disparities page of jpscanlan.com

4. Educational Disparities page of jpscanlan.com

5. "The Mismeasure of Discrimination," Faculty Workshop, University of Kansas School of Law, Sept. 20, 2013

6. "The Mismeasure of Group Differences in the Law and the Social and Medical Sciences," Applied Statistics Workshop at the Institute for Quantitative Social Science at Harvard University, Oct. 17, 2012
Posted by James S | Thursday, June 05 2014 at 3:41PM ET
The conclusions/implications made in this article indicate a fundamental misunderstanding of HMDA and data analysis by the author.
Posted by | Friday, June 06 2014 at 11:11AM ET
Add Your Comments:
Not Registered?
You must be registered to post a comment. Click here to register.
Already registered? Log in here
Please note you must now log in with your email address and password.
MultimediaSee All »