ByAndrew P. Meyer
The Home Mortgage Disclosure Act (HMDA) was enacted in 1975 and required that banks disclose certain information to the public regarding the home mortgages that they originated or purchased during the course of each year.
The original intent of HMDA was to determine whether mortgage lenders were discriminating against certain neighborhoods, particularly those with high minority concentrations. The law became a more powerful diagnostic tool following a 1989 amendment that required banks to collect information on the race and gender of each applicant. These data showed that minority applicants were more likely to be rejected than white applicants with similar incomes and loan amount requests.
Starting with the 2004 reporting year, lenders also were required to report loan pricing information to determine whether minorities who received loans were paying higher rates than white applicants. The lenders only report the rate if an applicant's annual percentage rate (APR) is more than a certain number of percentage points over a benchmark rate. A loan secured by a first lien with an APR more than 3 percentage points above a comparable-maturity Treasury security would be entered as a high-cost loan. For junior liens, the high-cost threshold is 5 percent above the comparable-maturity Treasury security.
These thresholds were chosen so that most loans in the subprime market would be flagged as high-cost loans and most loans in the prime market would not. As the subprime mortgage market proliferated, many lenders used higher prices to compensate themselves for added risk rather than rejecting marginal applicants outright. Thus, it became clear that pricing information was becoming increasingly important to any fair-lending analysis.
The HMDA evidence suggests that minorities are indeed paying higher rates. This caused regulators, lenders and the public to grapple with many of the same issues that arose after the initial rejection rate disparities were discovered.
The original HMDA data showed that the denial rate for minority loan applicants was roughly three times that of white applicants. When these data were released to the public, many people concluded that this disparity represented a clear case of racial discrimination. Before this release, most of the evidence for discrimination had been anecdotal, but now there were some hard numbers shoring up this conclusion.
Naturally, many in the banking community took exception to this interpretation. They argued that if minorities were rejected at a higher rate on average, then they must have weaker average credit quality.
In the early 1990s, economists at the Federal Reserve Bank of Boston tested this hypothesis by painstakingly gathering additional underwriting variables from banks in the Boston area.1 These additional variables included, among others, the debt-to-income ratio, the loan-to-value ratio, the number of past loan delinquencies and the number of public records on applicants' credit reports. These variables were critical to the underwriting decision and added important information to the simple loan amount and income data required by HMDA.
When the additional credit quality variables were considered, much of the rejection rate disparity disappeared, but there was still a statistically significant role for race. That is, other things equal, minority applicants were still more likely than white applicants to be rejected for a given loan. Criticism of the Boston Fed study centered on data quality issues and the fact that underwriting standards vary widely across banks and across loan product types, leading to potentially misleading conclusions.
We are now traveling down a similar path with the release of the 2004 HMDA data. The new data show that not only are minorities being rejected at a higher rate, they are paying higher prices for the loans that they do get.
According to a study by economists at the Federal Reserve Board of Governors, the incidence of these high-rate loans for minorities is higher than the corresponding incidence for whites.2 In the raw data, the differences are usually more than 20 percentage points for various loan products. More than two thirds of the difference, however, can be attributed to differences in the groups' incomes, loan amounts, other borrower-related characteristics included in the HMDA data, and the choice of lender. The choice of lender plays a role because, historically, minorities are more likely than white borrowers to borrow from a mortgage company as opposed to a bank, and mortgage companies tend to charge higher rates and fees than banks. When the authors analyzed the data on an institution-specific level, roughly 2 percent of lenders exhibited a statistically significant difference in the incidence of higher-priced loans between minorities and whites after accounting for factors included in the HMDA data.
An example of these issues can be found in St. Louis, which, in large part, mirrors the situation nationwide. A total of 726 local and nationwide financial institutions made at least one loan in the St. Louis Metropolitan Statistical Area (MSA) in 2006.
Of all the mortgages in St. Louis, the high-rate proportion has risen from 17.7 percent in 2004 to 29.6 percent in 2006. This increase is attributable in part to a flattening yield curve and some technical issues regarding its effect on the rate spread calculation, but also in large part to the burgeoning subprime mortgage market. In a pricing analysis, it is also important to distinguish between types of lenders. Figure 1 shows that mortgage companies (either independents or subsidiaries in a bank holding company) tend to have higher proportions of high-rate loans than banks, thrifts and credit unions.
The fair-lending issue becomes apparent in Figure 2. In 2006, of all the mortgage loans made to African American borrowers, 61.6 percent were high-rate loans, compared to 31.5 percent for Hispanic borrowers and 24.0 percent for white borrowers.
As outlined above, however, it is important to dig deeper into these numbers before concluding that these differences are a result of racial discrimination. For example, differences in income and the fact that African-Americans tend to borrow from mortgage companies more often than other borrowers explain some of this gap. A true picture of the underwriting implications can only be obtained with more data from the lenders.
The Federal Reserve and other regulatory agencies follow up aggressively whenever these disparities show up in an individual bank. We request additional underwriting variables and carefully model whether these variables explain differences in prices (or rejection rates) regardless of the applicants' race or ethnicity. This approach is in the spirit of the Boston Fed study, but for individual lenders. We work closely with lenders to ensure that the data are accurate and that we are modeling as closely as possible the criteria that were used to make the lending decision for each specific loan product. In this type of analysis, race and ethnicity cease to be statistically significant factors in the underwriting and pricing decisions of most lenders; but in the cases where they do not, we have imposed a variety of sanctions, including referrals to the Department of Justice for criminal prosecution.
In short, because the HMDA data alone are insufficient to draw definitive conclusions, the Federal Reserve and other supervisory agencies work carefully to gather enough further information to be fair to all parties involved. The disparities in the raw data show a clear need for further study, but without knowing more information about the creditworthiness of the individual applicants, one cannot draw any definitive conclusions about discrimination.