Flaws in the HMDA Data

Over the past several years there have been numerous articles and press reports suggesting that there is racial/ethnic discrimination in residential mortgage lending. Most of these publications have been based on the data gathered under the provisions of the Home Mortgage Disclosure Act of 1975 (HMDA). The purpose of the act was to determine whether or not lenders were discriminating against minorities in making home purchase mortgage loans, home improvement loans and refinancing loans. Legislators hoped that disclosure of any racial/ethnic bias would curb such bias. Just as one example, the Association of Community Organizations for Reform (ACORN) used the HMDA data exclusively in its recent report, Separate and Unequal.

The problem of missing data

However, there is a significant flaw in the HMDA data when used to address racial lending patterns, and the deficiency has been growing over recent years. In his excellent analysis of the HMDA data, Paul E. Huck, Senior Economist with the Federal Reserve Bank of Chicago, begins by summarizing the shortcomings of the HMDA data:

"The Home Mortgage Disclosure Act of 1975 (HMDA) was designed to further the goal of fair access to mortgage credit and required lender information including the location, loan amount, income and the race/ethnicity and sex of the applicant(s) for each application taken by the lender. However, the race/ethnicity of applications taken by mail or phone is not provided in a significant proportion of applicants. (Emphasis added). Given the widespread use of HMDA data by lenders, community groups, researchers, and regulators and the importance of mortgage lending as a policy issue, the strengths and shortcomings of the data must be clearly understood."

This table has significant implications for the charge of racial/ethnic discrimination in mortgage lending. The association of income with racial/ethnic characteristics (i.e., minorities comprise a higher percentage of low income households than high income households) is already well documented. The table demonstrates that, regardless of the type of mortgage, loans to low-income consumers have a higher level of non-reporting than loans to consumers with higher incomes. This is especially true for refinance loans. Given this incidence of under-reporting, it is quite likely that HMDA data consistently under-report mortgage loans obtained by minority/ethnic consumers.

Increasing problem of missing race/ethnicity

Furthermore, the prevalence of missing data has been increasing over the years. The accompanying table shows that the incidence of missing data on loan applications rose dramatically between 1993 and 1997. For example the proportion of missing data on home improvement loans rose to 29.7 percent from 14.9 percent in just four years.

The reasons for the rise in missing data are increasingly prevalent in 2001. Huck attributes much of the change to the marketing of mortgage loans in a manner that seeks to make it easy for consumers to apply for loans. Thus, potential applicants are encouraged to apply by mail, telephone or, in these days, even by e-mail. While these marketing methods make it more convenient for consumers to apply for and obtain loans, they do not generate HMDA data on the race/ethnicity of the applicants.

It is worth noting that the issue of missing data applies not just to the mortgage industry, but also to individual lenders. Different lenders will have different levels of missing data, depending in large part on the methods that they use to market their loans.

The bottom line

While it is useful to identify the serious flaws in the HMDA data and the increasing prevalence of those flaws, we cannot close the analysis without reminding our readers that, even if acceptance rates were complete and accurate for all types of mortgage loans, acceptance rates cannot be used as a basis for charges of racial/ethnic discrimination. Whether or not an application is accepted depends upon the credit quality of the applicant as revealed by credit reports, bankruptcy records and other factors that are captured in a credit score. Those critical variables are not captured in any way in the HMDA data. This is not to deny that some lenders may discriminate. However, as a Nobel-prize winning economist, Gary Becker, has observed, lenders with a "taste for discrimination" will ultimately not survive in a competitive market. The market for home purchase mortgages, refinancing loans and home improvement loans appears generally to be competitive. Therefore, a lender who chooses to discriminate is choosing to fail in the long run.