Statement by Richard F. Syron, President, Federal Reserve Bank of Boston, before the Committee on Banking, Housing, and Urban Affairs, U.S. Senate, February 24, 1993 - Statements to Congress - Transcript

Federal Reserve Bulletin, April, 1993

Thank you for this opportunity to discuss the Federal Reserve Bank of Boston's recent study of mortgage lending patterns and the report's implications for combatting discrimination in mortgage lending.

As the committee knows, the Home Mortgage Disclosure Act (HMDA) data for 1990 showed substantially higher denial rates for black and Hispanic applicants than for white applicants. This was true in all the major metropolitan statistical areas, and it was certainly true in Boston. Approximately 30 percent of black and Hispanic mortgage applicants were denied loans in the Boston metropolitan statistical area in 1990, compared with only 11 percent of white applicants. The 1991 data for Boston, which became available in fall 1992, show a narrower but still sizable gap, with 24 percent of black and Hispanic applicants denied loans, compared with 11 percent of white applicants.

When the 1990 HMDA were released, the implications of the racial disparities in denial rates were unclear. Although the HMDA date included information on applicant income, no information was collected on applicants' credit histories, loan-to-value ratios, debt-to-income or so called obligation ratios, and other factors that lenders commonly consider when they make mortgage loan decisions. Some felt that this missing information could explain the high denial rates experienced by minorities. Others argued that even if all relevant information were included, substantial bias in mortgage lending still existed. This disagreement has made it difficult to formulate solutions to improve credit flows to poor and minority neighborhoods.

The Federal Reserve Bank of Boston, with the support of the Federal Reserve Board and other supervisory agencies and the cooperation of mortgage lenders in the Boston area, undertook a major study of mortgage lending in an effort to clarify this issue. Racial disparities in mortgage lending patterns have been a concern in Boston for some years, and in 1989 the Boston Fed had undertaken a study of mortgage lending within the city of Boston. Although that study had found that housing and mortgage markets were functioning in a way that hurt black neighborhoods, the data available at that time could not distinguish the role played by lenders form the actions of buyers, sellers, realtors, and other market participants, After the 1990 HMDA data on applications were released, the Boston Fed was able to improve upon its earlier research and focus on the activities of the mortgage lending industry. I would like to submit for the record a copy of the Boston Fed's study, "Mortgage Lending in Boston: Interpreting HMDA Data."

The 131 financial institutions that had been the most active mortgage lenders in the Boston metropolitan area were asked to provide additional information on thirty-eight financial, credit history, and employment variables for all 1,143 of their black and Hispanic mortgage applicants and for a random sample of 3,300 white applicants. To protect the confidentiality of borrowers, we assured the lenders that all information collected would remain with the Federal Reserve and other bank regulatory agencies. The response from lenders was excellent, although missing information, errors in recording the original data, and withdrawn applications resulted in a final sample of 722 black and Hispanic applicants and 2,340 white applicants.

The additional variables collected were chosen after numerous conversations with underwriters, examiners, and others familiar with the mortgage lending process. We attempted to include all the variables that lenders view as relevant to their mortgage decisions. The information collected from the financial institutions was then combined with information on neighborhood characteristics from the 1990 Census and used to develop a model of mortgage lending decisions in the Boston area. Using this model, it was possible to test whether race was a significant factor in the lending decision once financial, credit history, employment, and neighborhood characteristics were taken into account.

The analysis revealed that the additional information about each applicant substantially reduced the disparity in denial rates but did not eliminate the gap. Black and Hispanic mortgage applicants in Boston, on average, had larger debt burdens, higher loan-to-value ratios, and weaker credit histories, and in other respects did not fare as well according to the evaluation criteria used by mortgage lenders. But after having taken all these factors into account, black and Hispanic mortgage applicants were still more likely to be turned down than white applicants. Minority applicants with the same financial, credit history, employment, and neighborhood characteristics as the white applicants in Boston would have experienced a denial rate of 17 percent rather than the actual white denial of 11 percent.

The information gathered in this study provides some insight into how this outcome occurs. Many observers have difficulty accepting that discrimination exists because they do not believe that rational lenders would turn down a perfectly good application simply because the applicant was black or Hispanic. The problem is that few applications fit a narrow definition of perfect. Most applicants, white as well as minority, exceed some guideline for obligation or loan-to-value ratios or credit history; or some possess a characteristic that requires additional documentation, such as self-employment or the fact that they are purchasing a two- to four-family home. As a consequence, the mortgage decision is not a purely mechanical process. Loan originators must exercise judgment, and they have considerable discretion in the way they evaluate these deviations from perfection and in the degree to which they take compensating factors into consideration.