FOR MANY YEARS, explicit discrimination in mortgage lending was part of a broader pattern of racial discrimination and segregation in residential housing markets. For example, until 1950 the Code of Ethics for Realtors prohibited real estate agents from being "instrumental in introducing into a neighborhood ... members of any race, nationality, or any individual whose presence will clearly be detrimental to property values in that neighborhood." As recently as 1970, Prentice-Hall published a textbook for real estate appraisers which declared that "the mixing of residents with diverse historical backgrounds within a neighborhood has immediate and depressing influence on value."
Discrimination against people of color by mortgage lenders has been a long-standing civil rights concern in the United States. Both Title VIII of the Civil Rights Act of 1968 and the Equal Credit Opportunity Act of 1976, commonly called the Fair Lending laws, prohibit discrimination against mortgage applicants on the basis of race or national origin. While these statutory prohibitions have been the law of the land for many years, there has been scant administrative or judicial action to challenge lending practices that suggest discriminatory behavior by mortgage lenders. A major reason for this inaction was the lack of publicly available data showing the disposition of mortgage loan applications by race of the applicant - the type of data that may be strongly suggestive of discrimination, and used to exert pressure on regulatory agencies to take action and to facilitate civil rights litigation.
However, in 1989 Representative Joseph Kennedy, D-Massachusetts, succeeded in tacking on to Savings & Loan bailout legislation an amendment that requires mortgage lenders to include information on the race and income of mortgage loan applicants in public reports on their mortgage lending activity. These mortgage activity reports are required by the Home Mortgage Disclosure Act (HMDA), a disclosure law enacted by Congress in 1975, to provide information on the extent to which individual banking institutions are "redlining" - referring to a pattern of discrimination in which financial institutions refuse to make mortage loans, regardless of credit record of applicant, on properties in specified areas - in inner city neighborhoods. Prior to the Kennedy Amendment, lenders were required by the HMDA to report only the location (by census tract) of their mortgage loans. Under the Kennedy Amendment, lenders now have to report not only the number and location of loans made, but also the number and location of applications, the race and income of the applicants and the disposition - meaning whether the loan was approved or denied - of the applications. In addition, the type of lenders subject to HMDA was expanded to include not only depository institutions and their subsidiaries, but also the larger mortgage companies.
The additional information required by the Kennedy Amendment has transformed HMDA into a more powerful tool for detecting discriminatory lending patterns. This new data, which first became available in late 1991 and is reported each year, has already provided a powerful impetus to strengthen Fair Lending enforcement and may ultimately result in serious efforts to root out discriminatory mortgage lending practices.
Fair lending?
While the Fair Lending laws have for the most part eliminated explicit forms of discrimination by mortgage lenders, the evidence suggests that people of color are still subject to subtle forms of discrimination in obtaining access to mortgage credit. Over the last 14 years, HMDA data have consistently shown that mortgage lending rates are dramatically lower in African-American and Latino neighborhoods than in white neighborhoods. Research examining the mortgage market in cities such as Boston and Washington, D.C. has found that the proportion of home sales financed by mortgages from banking institutions is considerably lower in minority neighborhoods than in white neighborhoods. This body of data on loan originations suggests that residents of minority neighborhoods do not enjoy the same access to mortgage credit as the residents of white neighborhoods.
During the 1980s, the federal banking regulators collected systematic data on mortgage loan applications and the disposition of such applications. Data that indicate the applicant's race can be used to determine whether loan application approval rates are lower for people of color than for whites. These data on approval (or denial) rates are generally not made available to the public in any great detail, and data for individual lending institutions are particularly inaccessible. However, in 1989 the Atlanta Journal-Constitution used the Freedom of Information Act to obtain from the Federal Home Loan Bank Board summary data on loan approval rates for a number of major metropolitan areas. The data show that on average, applications from African- American applicants were rejected by Savings & Loans at a rate twice as high as that for white applicants. The Atlanta Journal-Constitution story sparked increased interest in mortgage discrimination, raised new doubts about the adequacy of Fair Lending enforcement by the federal banking agencies and aided passage of the Kennedy disclosure amendment later in 1989.
Until recently, administrative and judicial actions to enforce the Fair Lending laws have been minimal to non-existent. Both the banking industry and banking regulators have argued that disparities in lending rates between minority and white neighborhoods and even disparities in application approval rates between minority applicants and white applicants do not provide probative evidence of discriminatory conduct by mortgage lenders. From this perspective, loan rate and approval rate data have only limited significance because they do not take into consideration possible differences in mortgage loan demand between minority and white neighborhoods or possible differences in the ability of people of color versus whites to meet credit underwriting criteria. Over the last 15 years, federal regulatory agencies have repeatedly employed this line of reasoning as grounds for refusing to initiate serious efforts to investigate individual mortgage lenders for Fair Lending violations. In particular, banking regulators have been unwilling to systematically review the loan files of individual banks to determine if race has been an underlying factor in their mortgage lending decisions. Bank regulators have also refused to implement testing programs under which testers would be employed to determine if minority and white applicants with comparable credit credentials receive different treatment when they apply for mortgage loans.
The sustained efforts by federal regulators and the banking industry to downplay the significance of race-based disparities in the HMDA and other lending data has also worked to discourage attempts to enforce the Fair Lending laws through private litigation. As a consequence, unlike other major civil rights areas such as discrimination in regard to housing sales and rentals, employment, education and voting rights, where there has been extensive litigation, there have been virtually no major judicial decisions defining the nature of discriminatory practices in regard to discrimination in mortgage credit markets.
A central tenet of U.S. civil rights law is that practices which do not intentionally or directly discriminate against people of color may nonetheless be unlawful under certain circumstances if they have discriminatory effects. Because of the dearth of judicial decisions on the issue of mortgage credit discrimination, no progress has been made in carving out an "effects test" doctrine for this important branch of civil rights law.
Byting the banks
In late 1991 the Federal Reserve Board (FRB) published the first extensive analysis of the expanded HMDA data mandated by the 1989 Kennedy Amendment. The FRB analyzed the entire HMDA database for 1990, the first year of data with information on applications, disposition, borrower income and borrower race. The analysis found that white applicants for conventional loans were rejected only 14.4 percent of the time, compared to 33.9 percent for African American applicants and 21.4 percent for Latino applicants. With respect to government-insured mortgage loans, the respective denial rates were 12.1 percent for whites, 18.4 percent for Latinos and 26.3 percent for African Americans.
Public dissemination of the HMDA data for 1990 triggered a spate of newspaper stories analyzing race-related disparities in loan approval and denial rates within individual metropolitan areas and for individual mortgage lending institutions. During 1992, the vast scope of the HMDA database, the marked disparities it revealed, the extensive media coverage of the issue and, undoubtedly, the increased attention paid to the oppressive conditions of inner-city neighborhoods after the Los Angeles riots in April, combined to create a political climate in which the long-standing presumption that such disparities are not indicative of discrimination and do not warrant investigation has begun to erode. Yet it is within the last month that two important events occurred to dramatically destroy the presumption of nondiscrimination.
In October, the Federal Reserve Bank of Boston (FRB Boston) released a comprehensive study of mortgage lending discrimination in the Boston metropolitan area. HMDA data for Boston for 1990 had shown that on average African-American and Latino applicants were rejected at a rate 2.7 times greater than white applicants. The FRB Boston designed a statistical study to determine the extent to which this disparity remained if all the factors employed by lenders in reaching credit decisions were taken into consideration. In other words, the FRB Boston sought to test the validity of the presumption of nondiscrimination.
To do this, the FRB Boston examined loan application files for a total of 4,500 mortgage loan applications at 131 Boston area financial institutions. From each file the FRB Boston collected data on 38 variables which lenders have indicated as relevant to reaching credit decisions - for example, various aspects of the applicant's economic situation and credit history and the appraisal of the property to be acquired.
The FRB Boston's analysis found that inclusion of all the credit decision variables reduced the disparity in the white and minority denial rates from the average ratio of 2.7 to 1 shown by the HMDA data to a ratio of 1.6 to 1. This dramatic result indicates that when the relevant credit decision variables are taken into consideration, minority applicants in the Boston area will still encounter a denial rate that is 60 percent higher than white applicants. According to a statement released by Richard Syron, president of the FRB Boston, "The racial disparity found in the HMDA data is substantially reduced when additional economic factors are considered, but it remains significant and it must be faced directly. Unfortunately, race plays a role, perhaps an unconscious and unintended role, but a role nonetheless, in mortgage lending decisions."
Another key finding of the FRB Boston study goes a long way toward explaining how discrimination enters the loan-decision process. The study found that the majority of loan applicants, both white and minority, had some flaw in their credit credentials and that in many cases these flaws were overlooked. This shows that even where lenders adopt detailed underwriting standards, they exercise tremendous discretion in applying these standards - a situation ripe for the introduction of subtle prejudice. According to the study, "Given the same imperfections in a mortgage application, whites seem to enjoy a general presumption of creditworthiness that blacks and Hispanics do not. Lenders seem more willing to overlook flaws for white applicants than for minority applicants."
The fundamental message of the FRB Boston study's findings undercut the presumption that major disparities in mortgage loan approval rates between whites and people of color are not the result of discriminatory behavior. Since the 1990 HMDA data reveals disparities for other U.S. metropolitan statistical areas (MSAs) that are roughly the same as those shown for Boston, the clear implication of the FRB Boston study is that a major federal initiative to strengthen Fair Lending enforcement is required. Of equal importance, the HMDA data reveals that many individual mortgage lenders in many different cities exhibit disparities in approval and denial rates that are far greater than the average rates for their metropolitan area. As the Office of the Comptroller of the Currency, the federal regulator of national banks, has observed, "This study is definitive. It changes the landscape."
Essential Information's Banking Research Project analyzed the 1990 HMDA data for 21 metropolitan areas to determine how the mortgage loan denial and approval rates in these areas compare with those in Boston. In assessing racial disparities in the disposition of mortgage loan applications, it is important to examine the loan-approval rate as well as the loan-denial rate. In many instances, non-approval results from withdrawal of the application rather than denial by the lender. However, such withdrawals may be the consequence of excessive delay by lenders in processing an application - a subtle way for the lender to signal the borrower that the application is not likely to be approved.
A second landmark event in the evolution of Fair Lending enforcement occurred in September when the U.S. Justice Department filed suit against one of the largest mortgage lending institutions in Atlanta, Decatur Federal Savings and Loan Association, for having engaged in a "pattern or practice" of discrimination against African-American mortgage applicants. This judicial action represents the first time that the federal government has brought a pattern or practice suit against a mortgage lender for violations of the Fair Housing Act and the Equal Credit Opportunity Act.
Before filing the suit, the Justice Department conducted a comprehensive statistical review of Decatur's loan files, very similar to that conducted by the FRB Boston. HMDA data for Decatur showed that African-American applicants were rejected at a rate almost three times greater than that of white applicants. The Justice Department's analysis of over 4,000 Decatur loan files revealed that even after controlling for all underwriting variables, race was a significant factor. The Justice Department's suit also charged that Decatur had pursued marketing policies that sought to limit the volume of mortgage loan applications from African Americans and that such marketing policies violated the Fair Lending laws. According to the Justice Department, all of the branch offices that Decatur had opened in the Atlanta area were located in white-majority areas. Decatur also defined its lending area to exclude most of the African-American neighborhoods in the Atlanta area. Decatur aggressively solicited mortgage loan referrals from real estate brokers who are part of the white community, but rarely solicited such business from African-American realtors who serve African-American neighborhoods. Further, Decatur has not advertised its mortgage loans in media oriented to the African-American community and does not publicize the availability of FHA- or VA- insured mortgages, which are commonly sought by African- American applicants.
Decatur's marketing strategy clearly depressed the volume of loan applications that it received from African Americans. According to the Justice Department, of the 24,300 mortgage loan applications received by Decatur during the 1985-1990 period only 6 percent were from African-American applicants.
Decatur entered into a consent decree with the Justice Department under which it agreed to hire an outside auditor to monitor its processing of mortgage loan applications for discriminatory conduct, adopt new marketing programs to affirmatively reach out to the African-American community, open a branch office or a regional loan office in a predominantly African-American section of Atlanta and provide $1 million in damages to 48 African-American individuals whose mortgage loan applications had been rejected between 1988 and 1992.
Civil rights
The FRB Boston study and the Justice Department suit against Decatur have greatly elevated the importance of the new HMDA data. Mortgage lenders whose HMDA data reveals a pattern of substantial disparity in loan denial or approval rates between white and minority applicants or a depressed volume of applications from people of color relative to their position in the local housing market must now be considered suspect under the Fair Lending laws. In this situation, the banking regulators will be under increasing pressure to thoroughly investigate their lending behavior to determine whether unlawful discrimination does in fact exist, and private litigation by civil rights organizations is likely to develop.
The new HMDA data should serve as a powerful spur to Fair Lending enforcement on two counts. First, the fact that such data is made public should exert strong pressure on the federal banking regulators and the U.S. Department of Housing and Urban Development to begin serious enforcement activity. Second, the public availability of the data enables civil rights and local community organizations to identify lenders with suspect lending patterns, strengthens their hand in negotiating remedial agreements and facilitates litigation where negotiation fails.
Census tracts classified as "Minority: 50 percent-75 percent" are those in which all people of color (not African Americans alone) comprise 50 percent to 75 percent of the census tract population. A comparable definition applies to those in the "Minority: 25 percent-50 percent" category. Census tracts are classified according to 1980 Census data. The charts do not provide loan flow data for census tract categories which had less than 1,000 owner-occupied housing units.
In some metropolitan areas, the relatively high mortgage loan flows shown for the various lower income census tract categories reflect a considerable level of gentrification. For example, the 1990 HMDA data show that of the 279 home purchase loans made in Atlanta's "lower income" black census tracts 62 percent were made to upper and middle income borrowers and 40 percent were made to whites. In Los Angeles, the 1990 HMDA data indicate that 55 percent of the total of number of 2280 home purchase loans made in the "lower income" census tracts were made to upper income persons.
Chart 5 is based on the same data as the cover map, except the home mortgage loan approval rate is calculated for census tract categories rather than individual census tracts. For example, the approval rate for all home mortgage loans made in African- American middle income census tracts was only 62 percent, compared to 80 percent for white middle income census tracts. Approval rates were not calculated for census tract categories for which the total number of applications was less than 50.