Women have a harder time than men securing a home mortgage and often pay higher interest rates. Yet, women repay their mortgages more reliably than men do, notes a recent study by the Urban Institute, a think tank in Washington. While it is unclear if a gender bias is responsible, at the very least the repayment predictors’ models that lenders use are flawed, says Jun Zhu, one of the study’s authors. She and Guy Cecala, CEO of Inside Mortgage Finance Publications, discussed the issue on the Knowledge at Wharton show on Wharton Business Radio on SiriusXM channel 111.
An edited transcript of the conversation follows:
Knowledge at Wharton: Why is it that women are seemingly better at paying off mortgages?
Jun Zhu: We don’t know right now. People will hopefully start looking at why [that is the case] now, because we know that there is definitely a difference.
Knowledge at Wharton: How does that occur?
Zhu: This is what we find. Even [considering] everything else like FICO [credit] scores, income, LTV (loan-to-value, a risk assessment ratio), etc., we do find out that a single woman performs much better than a single man. Right now, we don’t know why.
Knowledge at Wharton: Women are also dealing with higher interest rates and higher settlement payments than men do, among several other issues, correct?
Zhu: Right. We find that they pay higher interest rates. There are two reasons why they pay higher interest rates. One is because they have more subprime mortgages, which cost more. We don’t really address that in the paper. The other reason they pay more — the focus of this paper — is they have weaker characteristics. Their weaker characteristics predict that they will default more, but what we find in our research is that this is wrong. They do have weaker characteristics, but they do not default more. There is something about a woman that is not being picked up in the proxies we use to predict how well someone will do in paying their mortgage.
“Our research finds that despite their low incomes women are better at paying their mortgage.” –Jun Zhu
Knowledge at Wharton: I guess there’s a suggestion that relying more on FICO scores could be useful in determining the best route for people to qualify for mortgages.
Zhu: Right. FICO could be robust. Accurate measures of the risk can take into account the actual loan performance. The new version, FICO 8, which was introduced in July 2009, and [its competing product] VantageScore aim to achieve that. We have not looked at whether or not it might capture this difference better, but government agencies rely on the earlier version of FICO. The scores provided by the credit rating services were used in our paper. FICO can and should evolve.
Knowledge at Wharton: One other aspect to factor in is that when single women look for a mortgage, many of them are minorities who may live in lower-income areas, which is being counted against them, correct?
Zhu: That’s right. We did not find any discrimination at this point; we just find that there is definitely a flaw in the predictors. Our research showed that the indicators lenders have been using to determine whether or not someone would do a good job of paying their mortgage have been incomplete. It’s inaccurate. Definitely, there’s some impact on people living in low-income communities.
Knowledge at Wharton: Joining us is Guy Cecala, the CEO of Inside Mortgage Finance. I want your opinion on the data being brought by this study, because you’re in the industry and you see this on a daily basis.
Guy Cecala: The first impression is it’s not that surprising. The mortgage market is heavily reliant on credit scores, which don’t necessarily help single borrowers at all. Ideally, mortgage lenders like two borrowers just because they feel like they’re doubling their chances that somebody is going to pay a mortgage. Finally, there’s always been a disconnect between performance and the ability of a borrower to repay a loan, and what a mortgage lender decides, whether or not they give them the loan. That’s unfortunate. We’ve developed standards for handling millions of loans that were not very good on a case-by-case basis.
Knowledge at Wharton: What about involving FICO scores on a more regular basis to determine who exactly would be the best qualifiers for a mortgage?
Cecala: The mortgage market has always been, and continues to be overly dependent on credit scores. If your credit score is below 700 you’re considered suspect and you have to have compensating factors before you can get a mortgage at a good rate. That’s unfortunate.
Knowledge at Wharton: Could there potentially be a level of bias against women borrowers here?
Cecala: I think there can be. The mortgage market prides itself on being color blind, and essentially using a black box, but any sort of black box basically discriminates against single borrowers, lower-income borrowers and borrowers with lower credit scores. If those happen to be predominantly women, you have to assume that they are getting that kind of treatment from the mortgage market.
Knowledge at Wharton: The current market is a little bit more favorable than it was in the wake of the housing bubble and the recession. Are we seeing this type of approach even after the recession? Maybe part of it is that lenders have to scrutinize the data even more because of what happened in the past, correct?
Cecala: Yes. One of the legacies of the financial crisis is we’ve got significantly tougher mortgage underwriting standards. There is somewhat of a plain vanilla, one-size-fits-all mortgage underwriting standard, and that’s not very good at accommodating minority borrowers in general, or anybody with any sort of a non-typical, non-generic credit profile. Minority buyers in general are getting fewer mortgages than they did before. The good news is that they’re not getting subprime loans, because the subprime market has dried up completely, but they’re not getting mortgages at all in many cases.
Knowledge at Wharton: Jun, the data you brought looked at the pre-recession period, then the recession period, and post-recession. … Were there any major differences in terms of the success or lack of success of women servicing mortgages between those three periods of time?
Zhu: No. The results are consistent for those three different periods. For all of those three different periods we find that women were better than men.
Knowledge at Wharton: Is it surprising to you or not surprising that the data is fairly similar across the board?
“There is something about a woman that is not being picked up in the proxies used to predict how well someone will do in paying their mortgage.” –Jun Zhu
Zhu: It’s not surprising. The predictors definitely missed something that can make the behavior of women different from the behavior of men. It can be fixed, and it may already be fixed in the most recent FICO models, or other models. But since we used old-fashioned FICO scores, which were developed in 2003, maybe we did not pick up [the differences].
Cecala: There are significant differences in the time bands that the study produced. As you will see, a female-only borrower had an average credit score of 684 in the 2004-to-2007 period, which is not considered a real strong one. You look at the more recent 2011-2014 period, and [the average credit score] jumped to 741, which is very much A-prime category.
If you also look at delinquencies in that period, for females or [other categories], they went from over 20% to just about 2%, which is a huge, huge drop. It shows you that the performance, particularly in the last three or four years, has been very strong, and there’s no reason whatsoever to treat women borrowers any worse than any of the other categories, including couples, males or females.
Knowledge at Wharton: How different is it for an individual borrower in terms of getting a mortgage compared to a married couple?
Cecala: It used to be pre-crisis that you averaged the credit scores of two borrowers. After the financial crisis, we went into paranoia mode a little more and [considered] the lower credit score of the two borrowers. That is pretty much the system being used now. So to some extent, women or single borrowers shouldn’t be at a disadvantage if they have a decent or strong credit score, and in some ways they can beat a couple who may have fluctuating credit scores.
Knowledge at Wharton: Taking into account the lower score is somewhat understandable, considering what we went through during the housing bubble, and you do want to bake in a little bit of protection in the process, correct?
Cecala: Yes, and we’re risk-averse coming out of the foreclosure crisis, and as you point out, for good reason. But perhaps you could have an averaging of credit scores, and not necessarily assume the worst or the lowest on the borrower. That’s one of the things the mortgage industry is working through.
Knowledge at Wharton: Will we see that return at some point down the road?
Cecala: We haven’t seen a significant change in loosening underwriting standards in nearly eight years. I’m shocked that it’s gone this long, [and] that we’ve stayed this way. Frankly, as long as mortgage volume — particularly home purchase volume — remains relatively strong, mortgage lenders are reluctant to loosen standards. They tend to loosen standards when volume starts going down, as you might imagine. It’s not necessarily the best time to do it, but that’s historically when we’ve seen it.
“There is somewhat of a plain vanilla, one-size-fits-all mortgage underwriting standard, and that’s not very good at accommodating minority borrowers in general.” –Guy Cecala
Knowledge at Wharton: Is it the expectation, Guy, that because women borrowers tend to be better, that this is pattern will continue despite roadblocks in front of women?
Cecala: Yes, I don’t think there’s any question of that. The data shows that this pattern has existed during very different mortgage origination and housing markets. There’s no reason to expect it wouldn’t continue going forward. However, that doesn’t mean the mortgage industry is going to wake up one day and say, “Gee, we’ve got to look at this point completely differently given these performance numbers.”
Zhu: We definitely find that if you just look at the characteristics, women have weaker characteristics — there is no doubt about that. You can find that single women borrowers have lower income and smaller mortgages, but their smaller mortgages eat up more of their income. Their FICO scores are marginally lower, and despite these weak characteristics, we should predict that they will default more. In our study, we show that they pay their mortgage more reliably — that’s the finding.
Knowledge at Wharton: In a couple of the tables that you have in this report, the percentage of minority borrowers has dipped quite a bit in the wake of the recession. But even in the runback from the recession, and from the housing bubble, those numbers haven’t markedly gone back up.
Zhu: Yes. We took a look at those minorities, between women and men. We find that non-white borrowers like African American, Hispanic and others are more likely to default than white borrowers. But within all of those groups, men are more likely to default than women among those minority borrowers.
Knowledge at Wharton: In your study, the income difference between men and women has increased. The pay gap between men and women continues to be a big issue in this country.
Zhu: Yes, that’s right. Women are better at paying their mortgage, but lower income means weaker credit characteristics, right? So it’s a prediction that a person with low income should not pay their mortgage as reliably. But our research finds that despite their low incomes women are better at paying their mortgage.
Knowledge at Wharton: Guy, in terms of people trying to get mortgages these days, what do you see as the trend now and how do you see it playing out over the next few years, assuming that we don’t have another major recession, and the economy maintains the incremental growth we’ve been seeing the last few years?
Cecala: In the last couple of years, we have seen a continuation of tight underwriting, which is not helping women-only or single-only borrowers. You can see from the study that in the most recent period of 2011-2014, women-only [borrowers] have high credit scores, relatively low loan-to-value ratios, and not significantly higher debt-to-income ratios even though their income is significantly lower. What that tells you is that the underwriting is very tight, and they’re getting mortgages, sure, but they’re not getting a lot of them.
Going forward I don’t think we’re going to see that change unless there’s some regulatory push, and there is likely to be a regulatory push. Federal regulators, including the Justice Department, are working on a number of cases against some large lenders talking about discrimination just like this in the case. They are saying, “Okay, you’ve tightened your underwriting, and the result is minorities across the board are getting fewer mortgages than they did before, and you’ve got to make a strong case on why that’s so.” I think we’re going to see some loosening of underwriting standards, at least some more aggressive push by lenders to accommodate minority borrowers and lower-income borrowers, which will help women too.
“We definitely need to give more robust and accurate measures of the risk, to try to ensure that we are not denying all of those mortgages to people, especially women.” –Jun Zhu
Knowledge at Wharton: It will continue to be a little tough for women to be able to get the mortgages, but once they have them, hopefully they will continue to do exceptionally well with them.
Zhu: The credit bureaus already did something on trying to incorporate some factors. They will capture the actual performance of the mortgage. We hope that this will definitely ensure that we are not denying those mortgages to women and some other minorities that can pay on their mortgage.
Knowledge at Wharton: I guess it will also change the idea of what kind of property or what kind of home individuals, especially women, could go for because the loan sizes women get are smaller than what men are getting.
Zhu: We have taken into account the loan size and the location. Even controlled for those characteristics, we find that women have better performance than men. So there is definitely something else that is not captured, despite all of those things that we are looking for to predict how well someone will do in paying their mortgage.