In 2003, a landmark settlement was reached in a mortgage-lending action brought by multistate attorneys general and financial regulators against Household Finance Corporation (HFC) and Beneficial Finance Corporation (BFC). HFC/BFC, which were charged with engaging in various predatory-lending practices, agreed to make available up to $484 million to resolve consumer complaints.1
That predatory lending exists is not disputed. What is at issue is the degree to which it is widespread. Activists who champion the rights of consumers claim that predatory lending is pervasive in our lending marketplace. However, industry officials maintain that it is not so prevalent. Notwithstanding the previously mentioned settlement, the accurate documentation of the extent of predatory lending remains a bone of contention. What followes is a summary of a recent study on mortgage-lending practices that focuses on this issue and an opposing view of the study's methods and conclusions. The discussion underscores the difficulty in reaching a consensus on the magnitude of predatory lending in the U.S.
In December 2003, the National Community Reinvestment Coalition (NCRC) released a study of subprime lending in 10 large metropolitan areas.2 On the basis of its analysis, the NCRC contends that "the credit system is broken and discrimination is widespread in America." The NCRC offers several recommendations covering proposed legislative initiatives and actions that should be undertaken by lenders, community groups, and consumers. However, the study did not go unchallenged. Economists at the Mortgage Bankers Association (MBA) took issue not only with the general findings but also with the nature of the analysis used in the study and the consequences of the proposed recommendations.
The NCRC begins the report on its study by making a distinction between a subprime loan and a predatory loan. The former is a loan that has a higher-than-market rate of interest to compensate for the risk of lending to a borrower with less-than-perfect credit. The latter is defined as a subset of subprime loans and, as stated in the report, includes one or more of the following attributes: a lender charges more in interest and fees than is required to cover the added risk of lending to borrowers with credit imperfections, does not take into account the borrower's ability to repay the loan, or violates fair-lending laws by targeting women, minorities and communities of color; or the loan contract contains abusive terms and conditions that trap borrowers and lead to increased indebtedness. Then, the NCRC summarizes its primary finding: "Using the best available industry data on creditworthiness, NCRC uncovered a substantial amount of predatory lending involving rampant pricing discrimination and the targeting of minority and elderly communities."
The MBA takes exception to this characterization of the analysis, since it believes the NCRC's results merely describe the presence and attributes of subprime lending without proving the existence of its predatory lending criteria. This fundamental difference is the basis of the disagreement between the two organizations' views of the validity of the study.
The NCRC conducted separate analyses for home-purchase and refinance lending in 10 metropolitan statistical areas (MSAs) selected from different regions of the U.S. It used information for home lending from the 2001 Home Mortgage Disclosure Act (HMDA) database, 1999 credit-scoring data from one of the three large credit bureaus, and demographic information from 1990 census tracts. To identify a loan as subprime, the NCRC relied on a list of subprime lenders compiled by the U.S. Department of Housing and Urban Development (HUD). For a lender to be included on the list, more than 50 percent of the loans in its portfolio had to be subprime.
A key set of variables used in the NCRC's study were credit scores. These scores measure the likelihood of future delinquencies and foreclosures. In the NCRC's study, scores ranged from 0 to 1000, and lower scores indicated a lower risk of borrower default.3 For use in the analysis, the NCRC constructed five equal categories of risk or quintiles: very low, low, moderate, high, and very high. The NCRC adjusted the database for each census tract to include the number and percent of neighborhood residents in each of the five risk categories.
The NCRC used multivariate regression analysis to explain the percentage of subprime loans in a census tract while controlling for various demographic, economic, and risk factors. It found that after it took into account risk, housing-market conditions, and other factors, the level of subprime lending for refinance increased in nine of the 10 MSAs as the proportion of African-Americans in a neighborhood increased but in only six of the 10 MSAs for home-purchase lending. Similarly, the portion of subprime refinance lending increased in seven of the 10 MSAs when the number of residents over 65 increased in a neighborhood but in only three of the 10 MSAs for home-purchase lending.
The MBA raises several concerns about the data and analytical approach used by the NCRC. First, it questions the use of data (HMDA, credit scores, and census) collected more than 12 years apart by various sources. The MBA doesn't believe the implicit assumption that nothing substantive changed in credit conditions, markets, neighborhoods, or individual financial conditions during this period. Since the NCRC cannot identify any specific loan as a subprime loan, the MBA questions the NCRC's use of HUD's list of subprime lenders that self-reported that over 50 percent of their loans are subprime. According to the MBA, this implies that up to 50 percent of the loans of some of the lenders classified as subprime are really prime loans. Likewise, some of the lenders not classified as subprime (i.e., less than half of their business consists of subprime loans) also make subprime loans but are excluded from the analysis. Thus, the data may well misrepresent the true marketplace.
The MBA also has reservations about the NCRC's decision to separate credit scores into quintiles. Such a division suggests that one of the quintiles contains credit scores between 600 and 800. The MBA indicates that based on its information, this range contains not only the majority of prime loans but also a large share of subprime loans. Consequently, the MBA doubts that the credit-score variable adequately distinguishes between the prime and subprime markets.
Finally, the MBA suggests that when the NCRC interprets the results of its analysis, it confuses correlation with causality. Correlation indicates a relationship between two variables. Causality implies a cause-and-effect relationship between two variables such that a change in one variable causes a change in the other variable. Thus, the MBA maintains that the NCRC is overstating its findings if it claims to have demonstrated causality when it is presenting only a correlation.
NCRC's Recommendations On the basis of its analysis, the NCRC makes several recommendations, among them: Congress should enact comprehensive anti-predatory lending legislation; bank regulatory agencies should expand coverage of the Community Reinvestment Act (CRA) to include independent mortgage companies and all non-depository affiliates of banks; the Federal Reserve Board of Governors should intensify its anti-discrimination and fair-lending oversight; and federal agencies should be more diligent in enforcing the fair-lending laws in the case of nonbank lenders.
The MBA contends that the implementation of the NCRC's recommendations would result in deterioration in the efficiency of credit markets, which, in turn, would lead to a rise in the cost of credit for everyone, including those the policy proposals are designed to help. Thus, a whole group of borrowers might be locked out of the market because they can't afford the cost of credit necessary to compensate for the risk of lending to them.
No doubt the NCRC would have some responses to the criticisms raised by the MBA. However, the exchange, as presented, is informative. One important lesson learned from the preceding discussion is that although predatory lending might be a subset of subprime lending, the two should not automatically be considered the same. Specific actions by lenders must be identified as abusive before their behavior is considered predatory. Nonetheless, it is fair to say that the issue about the extent of predatory lending in the U.S. has yet to be resolved.