Arnold Kling  

Mortgage Lending and Discrimination

Hypothetical Agendas... Joe Barton is Wrong...

Felix Salmon writes

If you're a high-income Latino with a mortgage, you're almost twice as likely to be facing foreclosure than a high-income non-Hispanic white person. And in general, the foreclosure crisis is hitting blacks and Latinos much harder than it is whites, according to a startling new report from the Center for Responsible Lending.

Pointer from Mark Thoma.

About twenty years ago, the Boston Fed did a study which showed that minority mortgage applicants were turned down at a much higher rate than white applicants. This was deemed evidence for racism. However, economists who criticizes the study pointed out that if banks were more lenient with white applicants, then this should show up as a higher default rate among white borrowers, which was not observed.

Again, the economic argument was that those who benefit from racial disparity in lending standards should end up with higher default rates. So, my first thought in reading the results that Salmon cites is that minorities must have experienced favorable discrimination in lending standards.

Instead, Salmon suggests that minorities were given loans with more onerous terms. However, the main driver of defaults is not loan terms. It is the decline in home prices.

Let me summarize possible explanations for the foreclosure disparity, with the weight I give to each.

1. Discrimination against minorities by giving them more onerous loans. (3 percent)
2. Discrimination in favor of minorities by giving them lenient underwriting in order to meet CRA and Freddie/Fannie "affordable housing goals." (15 percent)
3. Aggressive lending aside, minorities might have had a higher propensity to jump into the housing market at the wrong time and to overpay for houses. This would mostly be due to lack of experience with the housing market, but it could also be a contagion effect as they saw friends and relatives taking the plunge into real estate. (80 percent)
4. Other explanations, which I cannot come up with at the moment. (2 percent)

Comments and Sharing

COMMENTS (20 to date)
Edward Harrison writes:

You are no doubt aware that even higher-income Blacks and Hispanics are concentrated in different areas of most metropolitan areas than do non-Hispanic whites.

Some of the areas in which house prices were hardest hit during 2008 and 2009 were just those areas with high concentrations of Blacks and Hispanics.

It makes sense that, in fact, Blacks and Hispanics did jump into the property market at precisely the wrong time. However, I would phrase my 80% weighting differently.

"Aggressive lending aside, minorities might live in areas that have experienced higher house price volatility inducing them to jump into the housing market at the wrong time and to overpay for houses. This would mostly be due to a contagion effect as they saw friends and relatives taking the plunge into real estate. Unfortunately volatility works in both directions - up as well as down.(80 percent)"

Steve Sailer writes:

Default rates for 1992 vintage FHA-insured mortgages after 7 years:
White: 4.27%
Black: 10.81%
Hispanic: 13.18%

Default rates for 1994 vintage FHA-insured mortgages after 5 years:
White: 4.10%
Black: 9.14%
Hispanic: 9.47%

Default rates for 1996 vintage FHA-insured mortgages after 3 years:
White: 3.34%
Black: 6.93%
Hispanic: 6.99%

The source is

Analysis of FHA Single-Family Default and Loss Rates
Prepared for:
U.S. Department of Housing and Urban Development
Office of Policy Development and Research
Prepared by:
Robert F. Cotterman
Unicon Research Corporation
March 2004

Even when adjusted for FICO scores and other objective factors related to credit risk, minority default rates still look worse. Cotterman concludes:

Blacks, Hispanics, and those in judicial foreclosure states and underserved areas have higher conditional loss rates, other things the same.

So, the changes in policy and worldview that led to the gigantic increases in mortgage lending to minorities seen over the last decade (with total mortgage dollars written per year increasing 691% for Hispanics and 397% for blacks from 1999 to the peak of the Housing Bubble in 2006) unsurprisingly led to world-historical levels of mortgage defaults in 2007-2009. After all, blacks and Hispanics were still defaulting at very high levels when they weren't getting as much mortgage lending. The law of diminishing marginal returns suggests that throwing more mortgage money at them wasn't going to improve their credit worthiness.

Justin writes:

If you believe that the primary explanation for the foreclosure disparity is that people timed the market incorrectly, surely a large part of the explanation is that large fractions of the minority population, and particularly the minority population that is wealthy enough to buy a home, happens to also be located in areas where the housing price bubble was at its peak. Californians are much more likely than average to be facing foreclosure regardless of their race but Californian home owners are more likely than average to be Latino. It doesn't appear from the methodology section of the report that the authors have corrected their data for these demographic differences. They do appear to have stratified the data by state, though, so if the underlying data was published, it should be relatively easy to see what fraction of the disparity disappears when geography is considered.

I'd wager that you are also discounting the affect of unemployment. Plenty of borrowers will continue making payments when their house is underwater so long as their income continues to support those payments. When unemployment hits, it becomes harder to make the mortgage payment regardless of the terms of the loan and it becomes far more costly to avoid recognizing that the house is worth far less than is owed. Since the unemployment rate of minorities is generally higher than that of white Americans, their foreclosure rate ought to be higher as well.

Tom writes:

Why do you never see credit scores discussed in the studies. Banks find them pretty interesting at time of the loan, you would think there may be part of the answer there. (You don't see it in the old Boston Fed study either.)

Funny to see the sharp disparity in the quality of comments between the two sites linked to.

Sara writes:

What about the obvious problem of Hispanics being concentrated in California, Arizona, Nevada, and Florida? Was geography controlled for in this study?

agnostic writes:

"Why do you never see credit scores discussed in the studies."

Because ideology trumps good statistical investigation. If the latter would clarify a pattern that must not be seen, its parts must be mangled and corroded enough so that the image comes out blurry enough to be made safe for public consumption.

For all of the complaining about "lying with statistics," by far the major taboo that causes people to lie with numbers is the one against using statistics to point out how group differences have an understandable basis rather than reflect hatred, prejudice, etc.

Steve Sailer writes:

On average, even with the same incomes, whites have higher net worths than blacks and Latinos. At the median, the difference has typically been about an order of magnitude. So, whites on average have more ability to dip into other assets to avoid foreclosure.

Moreover, whites are more likely to have relatives who can bail them out with a zero interest loan when the alternative is foreclosure. For example, over the last 20 years, I've twice lent money to in-laws to keep them from defaulting on their mortgages.

Is it fair that an individual black or Hispanic is less likely to have in-laws with some surplus cash to help them hang on to their houses?

I don't know. But that is how it is. On the other hand, any lender that would give an advantage to a white mortgage applicant because whites tend to have these kind of advantages that make them less likely to default would be keelhauled by the government for discrimination.

Steve Sailer writes:

Sara asks:

"What about the obvious problem of Hispanics being concentrated in California, Arizona, Nevada, and Florida? Was geography controlled for in this study?"

If you look just at California, where a majority of all the money was lost in the USA, minorities defaulted a sizable majority of it.

In California in 2006, subprime only accounted for 14 percent of all borrowing by non-Hispanic whites, but 47 percent of borrowing by Latinos and 52 percent by blacks.

When you calculate the correlation for the 20 biggest metro areas in California for minority subprime share of total mortgage dollars handed out in 2006 to q1-2009 foreclosure rates, the goodness of fit is r = 0.89, which is extremely high.

See the eye-opening graphs at

Steve Sailer writes:

Speaking of California, in Lending in Low- and Moderate-Income Neighborhoods in California: The Performance of CRA Lending During the Subprime Meltdown by Dr. Elizabeth Laderman and Dr. Carolina Reid of the San Francisco Fed, a study of defaults among 239,000 mortgages in California given out during the Bubble, you will find:

"We also find that race has an independent effect on foreclosure even after controlling for borrower income and credit score. In particular, African American borrowers were 3.3 times as likely as white borrowers to be in foreclosure,
whereas Latino and Asian borrowers were 2.5 and 1.6 times respectively more likely to be in foreclosure as white borrowers."

Milton Recht writes:

In 2007, the Fed issued a study of credit scores, "Report to the Congress on Credit Scoring and Its Effects on the Availability and Affordability of Credit" using a national statistical sample of 301,536 individuals.
From the executive summary of the report: "The analysis also finds that some groups perform worse (experience higher rates of serious delinquency) on their credit accounts, on average, than would be predicted by the performance of individuals in the broader population with similar credit scores. For example, on average, blacks perform worse than other racial and ethnic groups with similar credit scores. Similarly, single individuals and those residing in predominantly black or low-income census tracts perform worse on their loans than do their complementary demographic groups with similar credit scores. In contrast, the loan performance of Asians, married individuals, foreign-born individuals (particularly, recent immigrants), and those residing in higher-income census tracts was better than the performance predicted by their credit scores. The results hold after controlling for the other personal demographics of these individuals and for an estimate of the individuals incomes and locations; other factors that could be important, such as differences in employment experience, were not available."

Also see this fair newspaper summary of the report:

The study was done before the housing crisis and looked at all types of credit.

Lenders by law must lend equally to all borrowers with similar incomes and credit scores. If lenders include demographics, such as race or ethnicity, in their credit decisions, or adjust credit scores by race, etc., they would be found to have unlawfully discriminated.

The Fed study shows that while credit scores work well within ethnic and racial groups to predict default, different ethnic and racial groups with similar credit scores have different default rates.

Both the Fed and banks know that some groups with similar credit scores have higher default rates than other groups with the same credit scores. The banks however cannot impose more stringent credit terms on those with the higher default rate because it would be illegal discrimination.

Prior to all the anti-discriminatory lending laws, bank lending practices were based more on actual default rates than credit scores, which is why the Boston Fed study of 20 years ago did not find higher default rates among whites due to lax lending standards. After the passage of all the anti-discriminatory lending laws, banks were forced to treat all borrowers with identical credit scores equally, even if their expected default rates were different, which is why we now see higher default rates among some minority groups.

Despite protests by the the media, politicians and and back of the envelope calculations to the contrary by several liberal economists, much of the blame for the severity of the current foreclosures and housing mess can be traced back to well-intentioned laws that interfered with mortgage lending and housing decisions.

James A. Donald writes:

I was right in the middle of the housing boom and bust, and I saw an endless stream of cat eating no hablo English unemployed wetbacks fresh from crossing the Rio Grand buying houses worth near a million dollars no money down. I conjecture their income statements were completely fraudulent and wholly fabricated by the loan officer without consulting them, since the loan applications were in English, and generally described jobs where English would seem to be required.

Hence the higher default rate.

Steve Sailer writes:

The Center for Responsible Lending study almost certainly underestimates how much more likely blacks and Hispanics were to default on mortgages, as the report itself admits:

Like all estimates, ours has limitations. Most importantly, our method only captures differences in
foreclosure rates between racial and ethnic groups that are due to differences in distributions of
loans along the four dimensions that we use when calculating foreclosure rates (i.e. loan year, state,
occupancy and loan segment). That is, our estimated foreclosure rates will reflect differences in
where different demographic groups live, when they received loans and what types of loans they
received, but they will not capture any differences in foreclosure rates that are due to other factors.
Within the list of potential factors not included in our analysis, several tend to be associated
with both higher foreclosure rates and borrowers of color. As a result, our results are likely to
underestimate racial and ethnic disparities for at least three reasons.

First, not only were borrowers of color more likely to receive subprime loans than white borrowers,
but within the subprime market, borrowers of color were more likely to receive the most expensive
loans and were more likely to receive subprime terms associated with increased default risk, such as
prepayment penalties. Previous research has shown that African-American and Latino borrowers
were about 30% more likely to receive the highest-cost subprime loans relative to white subprime
borrowers with similar risk profiles and that subprime loans in communities of color were more
likely to carry prepayment penalties than subprime loans in majority communities.16 Our methodology,
however, applies the same foreclosure rate to all borrowers within a given state-cohort-occupancy-
loan type stratification. While our estimates will reflect differences in the foreclosure rates
between loan types, it will not reflect differences within a loan type that may be correlated with
race or ethnicity.17
Second, while the foreclosure crisis began with massive defaults on subprime mortgages, loan defaults
are increasingly being associated with unemployment. Our methodology does not incorporate the
higher unemployment rates of African Americans and Latinos, relative to non-Hispanic whites, and,
therefore, is likely to understate the disparate foreclosure rates among these groups.18
Third, on average, families of color have fewer resources for fending off foreclosure. As noted
earlier, families of color generally have lower levels of wealth than white families. In addition, the
composition of wealth between racial and ethnic groups is quite different. Non-Hispanic white
families tend to have a wider distribution of asset types than other families, with a higher proportion
of non-Hispanic white families holding every category of financial and non-financial asset. So, in
addition to the lower levels of wealth, lower levels of asset diversification makes families of color
more vulnerable to downturns in the housing market.19
All of these factors suggest that the disparities in foreclosure rates between racial and ethnic groups
are likely to be greater than indicated by our estimates.

Lance writes:

I could be reading the study incorrectly, but the Center for Responsible Lending only controlled for income variation among the entire population of the different ethnic groups, not the individuals who obtained the loans.

So, if Non-Hispanic, White income rose 2.1%, they controlled for that as opposed to the variation in mortgage-holder specific income. That's hardly a suitable control variable to reduce the impact of income shocks on the model.

Nevertheless, the record foreclosures are the result of two strong winds: Large income shocks and a record depreciation in home prices (especially in areas such as Phoenix and other urban areas). It's not a matter of securitization making modification more difficult or "unaffordable mortgages" at the time of origination (a high monthly mortgage payment relative to gross income).

Lance writes:

Here's the link to the Boston Fed study on what factors predict foreclosure:

Steve Sailer writes:

I don't think the right model is to assume that the government "forced" Kerry Killinger of Washington Mutual to announce a $375 billion pledge in 2001 to lend to minority and lower income communities or Angelo Mozilo of Countrywide to announce a $1 trillion (with a T) CRA-style pledge in 2005.

Instead, the government, assisted by the press, worked hard for 40 years to change the culture of mortgage lending, to pester and punish and drive from the business skeptics about the , leaving behind only the True Believers in the Diversity Dogma like Killinger and Mozilo. Washington Mutual, for example, passed 29 Community Reinvest Act reviews from 1990 onward in making 29 purchases of other banks. It even won a CRA pledge bidding war with two more conservative banks in its disastrous foray into Southern California. Government pressure worked more as a selection effect, selecting for executives who thought they'd get rich lending to illegal aliens and others favored by the government.

James A. Donald writes:

Steve Sailer writes:

Government pressure worked more as a selection effect, selecting for executives who thought they'd get rich lending to illegal aliens and others favored by the government.

Quite so. Bank executives really believed that they were evil sinners guilty of racial discrimination against blacks and Mexicans, and therefore were prepared wash their sins away with rivers of money. If you don't believe, your career will not go far. And so, by conversion and selection, everyone believed.

Steve Sailer writes:

Dear Milton Recht:

Thanks, that Federal Reserve study is very important. I hadn't previously been aware of it, as it has largely vanished down the Memory Hole.

Boonton writes:

I think social capital is more at play here. Whites have been well established in the US's upper and middle classes for quite a while. This provides, on average, two benefits if you're white:

1. You're more likely to have sophisticated family and friends members. This is basically the "Rich dad, poor dad" idea. Even if you're poor you're more likely at some point in your life to have a 'rich dad' figure have a conversation with you about smart financial decisions. When you got a mortgage broker, then, trying to sell you, say, a 50 year sub-prime mortgage with a 2-year frozen rate you probably will have some intelligent parties to turn to for advice. If you're the first person in your social network to have a shot at home ownership, though, the only person you can turn to is the 'professional' mortgage broker.

2. Likewise if you do get in trouble you're more likely to have family or friends who will help you out. If you're the first in your network to have a mortgage, though, you're on your own. You have no margin for error or mistakes. I've seen this first hand among the people I know. Many whites I know seem to act like a circus performer who has a high net underneath him. Latinos and blacks I know, though, seem like they are walking a tight line and want to give the appearance of 'doing everything right'.

There's a reason for this high flying behavior by whites. A few months ago I was visiting a family friend. She was mad at her daughter. Why? Because her daughter had called her ex-husband asking for help to avoid foreclosure as she had gotten behind. The friend was aghast that the daughter would demonstrate such weakness to the ex-husband and told her 'we will help you but you must come to us first and not go to him'. This was clearly not the first time her daughter had run into trouble paying her mortgage. Her mother had not only bailed her out but also, for a time, took over her checking account putting her on an allowance to try to stop her overspending. I suspect on average fewer Blacks and Latinos have family who are equipped for such intensive, long term hand holding.

logical writes:

[Comment removed for supplying false email address. Email the to request restoring this comment. A valid email address is required to post comments on EconLog.--Econlib Ed.]

Steve Sailer writes:

For more on this topic, see:

Comments for this entry have been closed
Return to top