WASHINGTON A battle over social media about how well the current housing finance system treats African-Americans and other minorities has spurred a critical discussion about racial disparities in the mortgage market.
The protracted fight erupted on Twitter last week and lasted for several days, centering on a statistic quoted by David Stevens, president and chief executive of the Mortgage Bankers Association, at an industry conference. Joshua Rosner, a prominent analyst at Graham Fisher & Co., challenged the validity of the data point, which suggested more than half of African-American applicants are denied a conventional mortgage.
The dustup prompted a larger debate about credit access while the two government-sponsored enterprises remain in conservatorship. Below we break down the details of the dispute, the limitations on industry data and the broader questions raised for the mortgage market.
So, what happened?
Speaking at an MBA conference in New York on May 20, Stevens claimed that housing data show "a 56% denial rate on GSE purchase loan applications for African-Americans." The statistic was part of a larger argument he made about the problems with the current housing market and why reform was necessary. The trade group has been vocal in supporting efforts to unwind Fannie Mae and Freddie Mac, including a recent Senate bill designed to bring more private capital into the market.
The data came from the Home Mortgage Disclosure Act database, the group said. The law requires financial institutions to collect detailed information from prospective borrowers who apply for a mortgage loan.
Rosner sent his first shot soon afterward, asking for details about the origins of the statistic, which launched a multiday back-and-forth between the two industry experts.
joshua rosner (@JoshRosner) May 20, 2014
Others in the industry chimed in along the way as the fight gained increasing attention. Critics had difficulty reproducing the figure using the publicly available data and raised questions with how the statistic was calculated.
(For a look at some of the debate's highlights, see here.)
Even Fannie Mae jumped into the fray, citing its own concerns.
"We believe there are serious flaws with how this HMDA data was analyzed and characterized," said Andrew Wilson, a spokesman for the enterprise. "We are committed to working with all stakeholders to ensure access to safe, affordable mortgage credit in every community."
A spokesman for Freddie Mac declined to comment.
Stevens said in a statement on Thursday that the trade group "stands by its analysis of HMDA data," though the paragraph containing the statistic was removed from the text of the speech that now appears on the MBA's website.
What exactly is up for debate?
The statistic has raised several questions about the methodology the MBA used for calculating the denial rate.
Some pointed to earlier reports by the Federal Reserve and Zillow, which found significantly lower denial rates using the same 2012 dataset. The Fed reported a 32% denial rate for blacks seeking conventional loans, while Zillow found 25.4% of black applicants were denied.
Critically, the MBA included both manufactured housing data and pre-approval data in its calculations, which some opponents argue is inappropriate.
"MBA includes manufactured housing (MH) data in its calculation because MBA members lend on this collateral, and both GSEs have MH loan programs and tout these programs as means to provide affordable homeownership to lower and moderate income homebuyers," Stevens said in the May 22 statement.
He added that the group looks at pre-approvals as well because "being denied at this stage is an important signal on availability of credit. A denial is a denial, regardless of where in the loan process it occurs."
But others note that Fannie Mae and Freddie Mac can only guarantee manufactured housing loans when a borrower also purchases the land underneath it, and it's not clear that HMDA data can be dissected narrowly enough to split off those loans from so-called chattel loans.
"Given limitations that prevent the GSEs from non-real estate lending, inclusion of chattel loans is questionable," said Rosner in a statement released Friday, pointing to a Consumer Financial Protection Bureau factsheet suggesting that more detail about the types of loans for manufactured housing offered would be beneficial.
Observers also noted that pre-approval denials shouldn't necessarily be considered, because the data can be difficult to interpret.
"The inclusion of pre-approval rates is something I've never seen, and I'm concerned it could artificially inflate" denial rates, said Debbie Bocian, a principal researcher at the Center for Responsible Lending.
Buyers may shop around for pre-approvals and not all lenders offer it, she added. If a buyer is pre-approved but doesn't take the loan, institutions also have the option of reporting the data or not, which could also complicate the calculations.
Why should we care?
Stepping back from the specific data dispute, concerns about racial disparities in the mortgage market aren't going away.
"There's always going to be data disputes and debates. People can twist themselves up disputing the data, but there are disparities and they are statistically significant," said Joshua Silver, vice president of research and policy at the National Community Reinvestment Coalition.
In fact, recent legislation to overhaul the mortgage finance market stalled in part because of ongoing questions by some on the left over how the new system would serve low-income and minority borrowers. The bill by Sens. Tim Johnson, D-S.D., and Mike Crapo, R-Idaho, to unwind the government-sponsored enterprises passed out of the Senate Banking Committee earlier this month but is not expected to get a vote on the chamber floor this year.
"I'm heartened by the inclusion of mortgage access in this debate, because I think it's an important one to have and I'm troubled by the current disparities in the system," said Stan Humphries, chief economist at Zillow. "Mortgage access is not where we want it to be in the current system. We can debate if Johnson-Crapo is bringing us closer or further away, but it's not where we want it to be."
In an interview, Stevens says the statistic was meant to help make a larger point.
"You can argue until you're blue in the face on the methodology. This was one line in a speech designed to point to the fact that this belief that the GSEs provide some extraordinary help to underserved borrowers is overstated," he said. "Progressives feel they are playing an extraordinary role on access to homeownership, particularly for some minority demographics, and that just doesn't bear out when you look at actual purchase transactions."
He pointed to separate data which says that 71% of mortgage loans to African-Americans were Ginne Mae products.
The fight also exposes some of the big limitations of HMDA data for examining lending trends. The CFPB, which now has authority over the law, announced earlier this year that it was weighing whether to significantly expand reporting requirements, which could help improve industry oversight.
But industry experts were also quick to note that the HMDA data don't provide enough detail to control for differences in income, credit score, debt levels, geographic location and other factors besides race that could explain at least some of the discrepancies that show up in denial rates across difference races.
"The broader issue raises questions about social bias and the sad reality of wealth disparities among borrowers of different races," Rosner said in an interview. "The whole industry, from the mortgage bankers to the enterprises, needs to do everything it can to ensure that, even with those economic and social disparities, borrowers who are equally qualified are treated equally. But the problem is we don't have any hard data demonstrating whether they are or are not at this point."
A new analysis from the Urban Institute due out later this week could help shed additional light on at least some of these lingering questions.
Researchers used 2012 HMDA data and information from analytics firm Corelogic to look more closely at mortgage application denial rates, separating out potential borrowers with poor credit from the wider population of applicants.
"If you don't look at the credit profiles of the applicants, then you're missing an important piece," said Wei Li, a senior research associate at the Urban Institute's housing finance policy center, and lead author on the study.
They found, for example, that the total population of black mortgage applicants had a denial rate of 40%, compared with a 75% denial rate for African-Americans in the low-credit-profile group. By comparison, non-Hispanic whites were found to have an overall denial rate of 14%, while low-credit white borrowers were denied 50% of the time. The study also finds that denial rates for the low-credit profile groups across all races have jumped dramatically since 2006.
Li added that even this new data excludes an important dimension of the debate over access to credit, pointing to the unknown number of potential borrowers who don't pursue a mortgage at all because of concerns about being denied. He likens it to efforts to discussions about the number of discouraged workers who are out of the labor force because they have given up looking for a job, which some argue leads to an underestimation in the official unemployment rate.
"You ignore those who are deterred by the tightness of the credit market," he said.