The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
United States Government Accountability Office GAO Testimony Before the Subcommittee on Oversight and Investigations, Committee on Financial Services, House of Representatives For Release on Delivery Expected at 10:00 a.m. EDT Thursday, July 17, 2008 FAIR LENDING Race and Gender Data Are Limited for Nonmortgage Lending Statement of Orice M. Williams, Director Financial Markets and Community Investment GAO-08-1023T July 17, 2008 FAIR LENDING Accountability Integrity Reliability Highlights Race and Gender Data Are Limited for Nonmortgage Lending Highlights of GAO-08-1023T, a testimony before the Subcommittee on Oversight and Investigations, Committee on Financial Services, House of Representatives Why GAO Did This Study What GAO Found The Federal Reserve Board’s (FRB) Regulation B, which implements the Equal Credit Opportunity Act of 1974 (ECOA), generally prohibits lenders from collecting certain data from loan applicants, such as their race or gender, for nonmortgage loans (e.g., small business loans). FRB has stated that this provision of Regulation B minimizes the chances that lenders would use such data in an unlawful and discriminatory manner. However, others argue that the prohibition limits the capacity of researchers and regulators to identify possible discrimination in nonmortgage lending. GAO’s June 2008 report found that most research suggests that discrimination may play a role in certain types of nonmortgage lending, but data limitations complicate efforts by researchers and regulators to better understand this issue. For example, available studies indicate that African-American owned small businesses are denied loans more often or pay higher interest rates than white-owned businesses with similar risk characteristics. While the primary data source for these studies, a periodic FRB small business survey, provides important insights into possible discrimination, it also has limits compared to HMDA data. For example, the FRB survey data are collected from borrowers rather than lenders, which limit their usefulness as a means to assess lending practices. In addition, federal bank regulators that enforce ECOA said that HMDA data facilitates the identification of lenders that may be engaging in discriminatory mortgage lending. In the absence of such data for nonmortgage loans, regulators may rely on time-consuming and less reliable approaches to identify possible discrimination, such as assuming a loan applicant is Hispanic based on his or her last name. This testimony is based on the GAO report, Fair Lending: Race and Gender Data Are Limited for Nonmortgage Lending (GAO-08698, June 27, 2008). Specifically, GAO analyzes (1) studies on possible discrimination in nonmortgage lending and the data used in them, (2) FRB’s 2003 decision to retain the prohibition of voluntary data collection, and (3) the benefits and costs of a data collection and reporting requirement. For this work, GAO conducted a literature review; reviewed FRB documents; analyzed issues involving the Home Mortgage Disclosure Act (HMDA), which requires lenders to collect and publicly report data on personal characteristics for mortgage loan applicants; and interviewed FRB and others. While testimony from researchers and other information GAO collected did not fully agree with all aspects of FRB’s 2003 rationale for retaining the prohibition of voluntary data collection, there was general agreement that such voluntary data would have limited benefits. FRB did not adopt a proposal that would have allowed lenders to collect data, without any standards, because it said the proposal would have (1) created an opportunity for lenders to use the data for discriminatory purposes and (2) such data would not be useful since lenders may use different collection approaches. While some researchers and others agreed with FRB’s first rationale, others said that data collection alone would not necessarily create the risk for discrimination because, in some cases (e.g., small business lending), lenders may already be aware of applicants’ personal characteristics as such lending is often done on a face-to-face basis. Even so, a range of researchers, regulatory staff, and others agreed that voluntarily collected data would not likely materially benefit efforts to better understand possible discrimination because the data would be collected on an inconsistent basis or few lenders would participate out of concern for additional regulatory scrutiny of their nonmortgage lending practices and the potential for litigation. FRB did not take a position on this report’s analysis. In addition to restating its rationale for retaining the prohibition of voluntary data collection, FRB summarized GAO’s findings, including the potential benefits and costs of additional data for fair lending enforcement. To view the full product, including the scope and methodology, click on GAO-08-1023T. For more information, contact Orice M. Williams at (202) 512-8678 or williamso@gao.gov. Requiring lenders to collect and publicly report data on personal characteristics for nonmortgage loan applicants could help address current data limitations that complicate efforts to better assess possible discrimination. However, such a requirement would impose additional costs on lenders that could be partially passed on to borrowers. These potential costs include those associated with information system integration, software development, data storage and verification, and employee training. Limiting a requirement to certain types of loans could help mitigate such costs but may also involve complexities that would need to be carefully considered. For example, to the extent that small business lending is more complicated than other types of lending, lenders may need to collect and report additional information on a range of underwriting standards in addition to data on personal characteristics so that informed judgments can be made about their lending practices. United States Government Accountability Office Mr. Chairman and Members of the Subcommittee: I am pleased to be here today to discuss the available research on the potential for discrimination in nonmortgage lending and the Federal Reserve Board’s (FRB) basis for largely retaining Regulation B’s prohibition against the voluntary collection of data on personal characteristics for nonmortgage loan applicants. As you know, the Equal Credit Opportunity Act (ECOA) of 1974 prohibits discrimination in lending based on an applicant’s personal characteristics, such as race, gender, color, religion, national origin, marital status, or age. 1 A provision of Regulation B, which implements ECOA, generally prohibits lenders from asking for, inquiring about, or documenting such information for individuals who apply for nonmortgage loans, such as small business, automobile, or credit card loans. In 1975, FRB established the general prohibition as a means of discouraging discrimination in lending, based on its belief that if lenders could not inquire about or note such information on applicants’ personal characteristics, they would be less likely to unlawfully consider it when making lending decisions. However, some members of Congress and consumer advocates argue that the prohibition on data collection has limited the ability of researchers, regulators, Congress, and the public to monitor nonmortgage lending practices and to identify possible discrimination. In response to such criticism, the FRB, in 1999, proposed and considered an amendment to Regulation B that would have removed the prohibition and permitted lenders to voluntarily collect data on personal characteristics, without any restrictions or standards, for nonmortgage loan applicants. However, in 2003, after reviewing more than 600 public comment letters on the proposed amendment and taking other steps, FRB ultimately decided to leave the basic elements of the prohibition intact. FRB did not adopt the amendment because the agency believed it would have (1) created an opportunity for lenders to use the data for discriminatory purposes; and (2) generated data that would not be useful or reliable because lenders would likely adopt inconsistent data collection approaches. However, some members of Congress and consumer advocates questioned FRB’s decision, particularly its conclusion that such data could be used for unlawful discrimination. To support their position, they argued that requiring lenders to collect and publicly report data on 1 Pub. L. No. 90-321, title VII, as added by Pub. L. No. 93-495, title V, § 503, 88 Stat. 1521 (Oct. 28, 1974) (codified, as amended, at 15 U.S.C. §§ 1691 et seq.). Page 1 GAO-08-1023T personal characteristics of mortgage loan applicants under the Home Mortgage Disclosure Act of 1975 (HMDA), as amended, has made lenders less likely to engage in discriminatory mortgage lending practices, and facilitated the ability of regulators to monitor and enforce compliance with fair lending laws. My comments today are based on findings from our June 2008 report entitled Fair Lending: Race and Gender Data Are Limited for Nonmortgage Lending.2 Specifically, I will discuss (1) available research on possible discrimination in nonmortgage lending and review the strengths and limitations of the data used in the studies, (2) FRB’s 2003 basis for largely retaining Regulation B’s prohibition against the voluntary collection of data on personal characteristic for nonmortgage loan applicants, and (3) the potential benefits and costs of a data collection and reporting requirement and options to mitigate such costs. To prepare our June 2008 report, we conducted a literature review to identify studies that used nationwide databases and statistical techniques to identify possible discrimination in nonmortgage lending and assessed the strengths and weaknesses of key data used to support the studies’ findings, particularly in comparison to HMDA data. Further, we reviewed relevant FRB documents pertaining to Regulation B and did a content analysis of a random sample of 90 from the more than 600 comment letters that FRB received in response to the proposed 1999 amendment to the regulation. We also conducted interviews with a range of researchers who have assessed potential discrimination in nonmortgage lending, staff involved in fair lending law enforcement from bank regulators, representatives from banking organizations and consumer groups, and officials from organizations that represent minority and women-owned businesses. We conducted the audit work underlying the report from September 2007 to June 2008 in Washington, D.C., in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives. 2 GAO, Fair Lending: Race and Gender Data Are Limited for Nonmortgage Lending, GAO-08-698 (Washington, D.C.: June 27, 2008). Page 2 GAO-08-1023T In summary, we found that most studies suggest that discrimination may play a role in certain types of nonmortgage lending, but data limitations have complicated efforts by researchers and regulators to understand the extent to which possible discrimination occurs. For example, available research on minority business lending generally indicates that AfricanAmerican business owners are denied loans more often or pay significantly higher interest rates than white-owned businesses with similar risk characteristics. However, the data used in these studies are collected from small business borrowers rather than lenders and, therefore, cannot be used to conduct in-depth analyses of the practices of individual lenders or the lending industry generally. In contrast, studies on possible discrimination in mortgage lending often use HMDA data, which are collected directly from a large population of lenders and thus provide for more in-depth research among other benefits.3 Further, we found that data limitations may also impede the relative efficiency of the bank regulators’ fair lending examination process for the nonmortgage sector as compared with the mortgage sector. While testimony from researchers and other information we collected did not reflect full agreement with all aspects of FRB’s 2003 rationale for retaining Regulation B’s general prohibition on collecting data on personal characteristics, most experts agreed with the agency’s overall conclusion that voluntarily collected data would offer limited benefits as a means of better identifying possible discrimination in nonmortgage lending. FRB’s conclusion that voluntary data collection could create some risk of discrimination, while supported by some interviewees, was challenged by a range of researchers, regulatory staff, and others we contacted. For example, several researchers said that voluntary data collection would not necessarily increase the risk of discrimination because, in certain cases— such as small business lending, which is often done on a face-to-face basis—lenders could already observe an applicant’s race and gender. Even so, a range of researchers, regulatory staff, and representatives from both consumer and banking groups we contacted generally agreed with FRB that lenders would likely adopt different approaches to collecting and using data on personal characteristics, potentially limiting the reliability and usefulness of the information. They also said that relatively few, if any, 3 However, as described in this testimony, studies that use HMDA data to assess possible discrimination in mortgage lending have been controversial because the data do not include key underwriting variables such as a loan applicant’s credit score. Some studies have used HMDA data in conjunction with underwriting data available from other sources to better detect potential discriminatory mortgage lending practices. Page 3 GAO-08-1023T lenders would likely choose to collect such data out of concern that their nonmortgage lending practices would become subject to increased regulatory oversight and potential litigation. Finally, we found that requiring lenders to collect and publicly report data on personal characteristics for nonmortgage loan applicants, similar to HMDA requirements, could help address current data limitations but would also involve costs and complexities that would need to be considered. In concept, such a requirement could facilitate efforts by researchers, regulators, and others to better assess potential discrimination in nonmortgage lending. However, such a requirement would also impose additional costs on lenders for items such as system integration, software development, and training that could be partially passed onto borrowers. One option to potentially mitigate some of these costs would be limiting data collection and reporting to specific types of lending, such as small business lending, but this option may also involve additional complexities and costs that must be considered. For example, to the extent that small business lending is more complicated than other types of lending, lenders may need to collect and report additional information on a range of underwriting characteristics in addition to data on personal characteristics so that informed judgments can be made about their lending practices. Alternatively, lenders could be required to collect data on personal characteristics and make such data available to regulators to facilitate the fair lending examination process and potentially decrease costs, but, in the absence of a public reporting requirement, this option would not enhance the ability of researchers, Congress, and others to better assess the potential for discrimination. FRB did not take a position on this report’s analysis. In addition to restating its rationale for retaining the prohibition of voluntary data collection, FRB summarized GAO’s findings, including the potential benefits and costs of additional data for fair lending enforcement. Background Regulation B imposes a general prohibition on collecting data on personal characteristics for nonmortgage loan applicants. But in 2003, FRB expanded its exceptions to this prohibition to include permitting lenders to collect data on race, gender, and other personal characteristics in connection with a self-test for the purpose of determining the effectiveness of the lender’s compliance with ECOA and Regulation B. A self-test is any program, practice, or study that is designed and used by creditors to determine the effectiveness of the creditor’s compliance with ECOA and Regulation B. The results of a self-test are privileged—that is, Page 4 GAO-08-1023T they cannot be obtained by any government agency in an examination or investigation in any lawsuit alleging a violation of ECOA. Although Regulation B prohibits creditors, except in limited circumstances such as conducting a self-test, from collecting data on personal characteristics for nonmortgage loan applicants, creditors are required to collect such data for mortgage loan applicants. Specifically, HMDA, as amended in 1989, requires certain financial institutions to collect and publicly report information on the racial characteristics, gender, and income level of mortgage loan applicants.4 In 2002, FRB, pursuant to its regulatory authority under HMDA, required financial institutions to report certain mortgage loan pricing data in response to concerns that minority and other targeted groups were being charged excessively high interest rates for mortgage loans. Authority for enforcing compliance with ECOA with respect to depository institutions, such as Federal Reserve System member banks, national banks, state-chartered banks, saving associations, and credit unions, lies with the five federal regulators—FRB, the Office of the Comptroller of the Currency (OCC), the Federal Deposit Insurance Corporation (FDIC), the Office of Thrift Supervision (OTS), and the National Credit Union Administration (NCUA).5 To carry out their responsibilities, the agencies may conduct periodic compliance examinations of depository institutions. These compliance exams generally assess depository institutions’ loan underwriting guidelines and credit decisions to detect possible discrimination in both mortgage and nonmortgage lending. FRB’s Survey of Small Business Finances (SSBF) is one of the principal sources of information available on the factors that affect the availability of credit for small businesses. FRB has conducted the SSBF about every 5 years from 1987 through 2003 from a nationwide sample of small businesses of varying sizes, locations, and ownership characteristics. In 2007, FRB decided to discontinue the SSBF due to its cost and other 4 Pub. L. No. 94-200, title III, 89 Stat. 1125 (Dec. 31, 1975) (codified, as amended, at 12 U.S.C. §§ 2801 et seq.). 5 Other agencies with enforcement authority under ECOA with respect to certain nondepository institutions include, among others, the Securities and Exchange Commission, the Small Business Administration, and the Farm Credit Administration. To the extent that ECOA does not assign to another federal agency responsibility for enforcing compliance with respect to a particular creditor, the Federal Trade Commission has enforcement authority for such creditors. Page 5 GAO-08-1023T considerations. However, according to FRB officials, FRB plans to include elements of the SSBF in another survey, the Survey of Consumer Finances (SCF), starting in 2010. Studies Suggest That Discrimination May Play a Role in Certain Types of Nonmortgage Lending, but Data Limitations Complicate Efforts to Better Understand the Issue The limited number of studies on nonmortgage lending that met our criteria for selection in our June report focused primarily on the small business sector, and suggested that certain minority-owned businesses may be denied loans more often or be offered higher interest rates than similar white-owned businesses. However, the key data source for most of these studies, FRB’s SSBF, has certain limitations compared with HMDA data, and this may limit the data’s usefulness as an analytical tool. The few studies we identified that addressed possible discrimination in automobile and credit card lending relied on SCF data, which has certain limitations similar to those of the SSBF data. Further, our report found that data limitations may also impede the relative efficiency of the bank regulators’ fair lending examination process for the nonmortgage sector as compared with the mortgage sector. Research Suggests That Possible Discrimination Exists in Small Business Lending, but the Data Used in Such Studies Have Limitations Primarily using data obtained from FRB’s SSBF, all eight studies we identified on minority business lending generally found that lenders denied loans to minority-owned businesses (seven of the eight specifically refer to African-American-owned businesses) or required them to pay higher interest rates for loans significantly more often than white-owned small businesses. This finding generally remained consistent after considering a variety of risk factors, such as borrower creditworthiness, industry sector, and other firm characteristics (e.g., business location, assets, and profits). In addition, studies have found that Hispanic-owned businesses were denied credit or charged higher interest rates more often when compared with white-owned businesses with similar risk characteristics. On the other hand, some studies we reviewed did not identify evidence that women-owned businesses face credit denials or higher rates significantly more often than male, white-owned businesses. While studies using SSBF data have provided important insights into possible discrimination in small business lending, researchers and FRB officials also pointed out a number of limitations: Page 6 GAO-08-1023T • SSBF data are collected from individual small business borrowers rather than lenders, which limit their analytical value.6 For example, SSBF data do not allow researchers to assess the overall small business lending underwriting standards or lenders’ performance by type of institution, by size, or by geographic or metropolitan region. • SSBF survey data are self-reported and are not verified by FRB. For example, FRB relies upon survey respondents to accurately report their race, gender, and other characteristics, as well as requested information on their business and their financing. Since the survey may be conducted long after the survey respondent applied for credit, the timing of the SSBF increases the risk that respondents may not accurately recall and report information from the time when the credit decision was made. • FRB conducts the SSBF about every 5 years rather than annually and, therefore, the survey results may not be timely. To illustrate, most of the studies that we reviewed were based on data that are about 10 years old from surveys conducted in 1993 and 1998. Researchers and FRB officials that we spoke with said it may also take FRB a significant period of time to review and process the SSBF data prior to releasing it to the public. In contrast, HMDA data offer certain advantages over SSBF data as a research tool to assess possible discrimination in mortgage lending. In particular, HMDA data are collected directly from a large and identified population of mortgage lenders on a consistent and annual basis. Researchers have used HMDA data to conduct analyses of possible discrimination by type of lending institution, size of the institution, and geographic or metropolitan area. FRB also requires that lenders help verify the HMDA data they report, such as applicant data on personal characteristics and the interest rates charged on certain types of mortgages. Despite these advantages, we noted that analyses of HMDA data as a basis for conducting research on possible discrimination in mortgage lending have been criticized for not including key loan underwriting variables, 6 It should be noted that data collected from borrowers can have distinct advantages. For example, survey respondents would know better than lenders whether they had been discouraged from applying for credit and could more accurately describe their race or gender. Page 7 GAO-08-1023T such as the borrowers’ credit scores or mortgages’ loan-to-value ratios.7 Some argue that such underwriting variables may account for many apparent discrepancies between minority and white mortgage borrowers. To compensate for the lack of underwriting variables in the HMDA data, several researchers have collected such data from proprietary sources and matched it with HMDA data.8 The Few Studies That Have Identified Possible Discrimination in Automobile and Credit Card Lending Use Data That Have Strengths but Also Limitations According to a study on auto lending, racial discrimination could play a role in differences between the treatment of minority and white borrowers.9 The study relied on data from FRB’s SCF, which asks a nationwide sample of about 4,500 U.S. consumers to provide detailed information on the finances of their families and on their relationships with financial institutions. Because SCF data is also collected from borrowers rather than lenders, like SSBF data, it cannot be used as a basis for assessing individual lenders’ lending practices or lending practices industrywide (i.e., by type of institution, size of institution, or geographic or metropolitan area). The two studies we identified that also relied on SCF data had mixed results with respect to possible discrimination in credit card lending. One study found that minorities were likely to pay higher interest rates on credit card debt than white credit cardholders even after considering the payment history and financial wealth of each group.10 Another study did 7 Steven R. Holloway and Elvin K. Wyly, “The Color of Money Expanded: Geographically Contingent Mortgage Lending in Atlanta,” Journal of Housing Research 12, no.1. (2001): 55-90; and Robert Avery, Kenneth P. Brevoort, and Glenn B. Canner, “Opportunities and Issues in Using HMDA Data,” Journal of Real Estate Research 29 (2007): 351-379. 8 Alicia H. Munnell, Geoffrey M.B. Tootell, Lynn E. Browne, and James McEneaney, “Mortgage Lending in Boston: Interpreting HMDA Data,” American Economic Review, 86, no. 1 (1996); Debbie Bocian, Keith S. Ernst, and Wei Li, “Race, Ethnicity and Subprime Home Mortgage Pricing,” Journal of Economics and Business. 60, nos. 1 and no. 2 (2008); and Kenneth P. Brevoort and Glenn B. Canner, “Opportunities and Issues in Using HMDA Data,” Journal of Real Estate Research, 29 (2007): 351-379. 9 Darryl Getter, “Consumer Credit Risk and Pricing,” The Journal of Consumer Affairs 40, no.1 (2006): 41-63. Other research has looked at possible discrimination in the prices charged for new automobiles, as opposed to studies that analyze interest rate pricing for automobile loans. See: Ian Ayres and Peter Siegelman, “Race and Gender Discrimination in Bargaining for a New Car,” The American Economic Review, 85, no. 3 (1995): 304-321; and Ian Ayres, “Fair Driving: Gender and Race Discrimination in Retail Car Negotiations,” Harvard Law Review, 104, no. 4 (1991): 817-872. 10 Getter, “Consumer Credit Risk and Pricing.” Page 8 GAO-08-1023T not find that minority credit cardholders paid higher interest rates as compared with white credit cardholders after controlling for creditworthiness factors.11 These studies showed the strength of the SCF as a data source (e.g., the ability to consider data on personal characteristics and loan underwriting factors), as well as its limitations (e.g., the data are collected from borrowers rather than lenders). Data Limitations May Also Impede the Efficiency of the Fair Lending Examination Process for Nonmortgage Lending Representatives from the four federal bank regulatory agencies we contacted (FRB, OCC, FDIC, and OTS) said that the availability of HMDA data has facilitated the fair lending law examination process. In particular, agency staff said that the analysis of HMDA data provided insights into lenders that might be at high risk of engaging in potentially discriminatory practices in mortgage lending. While agency staff said that HMDA data were only a first start in the investigative process (because they must evaluate a range of underwriting criteria and practices that may help explain disparities in a lender’s mortgage lending patterns), HMDA data allowed them to prioritize their examination resources. We found that in the absence of similar race, gender, and other data on personal characteristics for nonmortgage loan applicants, examiners may rely on time-consuming and possibly unreliable techniques to assess lenders’ compliance with fair lending laws. Under the Interagency Fair Lending Examination Procedures, examiners can use established “surrogates” to make educated guesses as to the personal characteristics, such as race or gender, of nonmortgage loan applicants to help determine whether the lenders they regulate are complying with established laws and regulations in extending credit to minority and other individuals targeted for loan applicants. For example, examination guidance allows examiners, after consulting with their agency’s supervisory staff, to assume that an applicant is Hispanic based on the last name, female based on the first name, or likely to be an African-American based on the census tract of the address. While these techniques may help identify the racial or gender characteristics of loan applicants, they have potential for error (e.g., certain first names are gender neutral, and not all residents of a particular census tract may actually be African-American). 11 Amberly Hazembuller, Britton Lombardi, and Jeanne Hogarth, “Unlocking the Risk-based Pricing Puzzle: Five Keys to Cutting Credit Card Costs,” Consumer Interests Annual, 53 (2007): 73-81. Page 9 GAO-08-1023T As a result of the limitations of the data on personal characteristics for nonmortgage loan applicants, as well as regulatory guidance directing examiners to consider using surrogates, federal oversight of lenders’ fair lending law compliance in this area may be less efficient than it is for mortgage lending. According to a comment letter submitted by a Federal Reserve Bank to FRB as it considered amending Regulation B in 1999, its examiners were unable to conduct thorough fair lending examinations or review consumer complaints alleging discrimination for nonmortgage products due to the lack of available data. Moreover, our reviews of agency fair lending examination guidance and discussions with some agency staff (OCC, FDIC, and OTS) suggest that, due in part to HMDA data availability, agencies focus most of their resources on possible discrimination in mortgage lending rather than nonmortgage lending. We plan to further explore the issue of fair lending enforcement in future work, including the impact of potential data limitations on regulatory agencies’ oversight and enforcement of the fair lending laws for mortgage and nonmortgage lending. Voluntary Lender Collection of Data on Personal Characteristics Would Likely Offer Limited Benefits in Better Understanding Possible Discrimination in Nonmortgage Lending While some individuals we contacted generally agreed with FRB’s 2003 conclusion that permitting lenders to voluntarily collect data on personal characteristics for nonmortgage loan applicants could create some risk of discrimination, many other individuals we contacted expressed skepticism about this argument. Even so, a range of researchers, regulatory staff, and representatives from both consumer and banking groups we contacted generally concurred with FRB that voluntarily collected data might not be useful or reliable and that very few banks would choose to collect it. Consequently, the benefits of permitting lenders to voluntarily collect data on personal characteristics as a means for researchers, regulators, and others to better understand possible discrimination in nonmortgage lending would likely be limited. Page 10 GAO-08-1023T Researchers and Others Had Mixed Views on FRB’s Conclusion That Voluntary Data Collection Could Create Some Risk for Discrimination in Nonmortgage Lending Some researchers, staff from a bank regulatory agency, and representatives from banking and business trade groups we contacted generally agreed with FRB that permitting voluntary data collection on personal characteristics could create a risk that the information would be used for discriminatory purposes. These officials told us that the best way to protect borrowers against discrimination is to minimize the availability of information to lenders about their personal characteristics. However, many other researchers, staff from some regulatory agencies, and officials from consumer groups expressed skepticism on this conclusion. First, a staff member from a regulatory agency, several researchers, and representatives from consumer groups said that, in certain cases, lenders were already aware of the race and gender or other information on personal characteristics of nonmortgage loan applicants. Therefore, simply collecting data on personal characteristics on applicants in such cases would not necessarily create a risk of discrimination. Other researchers and officials from banking institutions disagreed. They noted that, in some cases, lending decisions may be made by officials who do not interact directly with loan applicants. Second, lenders’ voluntary collection and use of data on personal characteristics for nonmortgage loan applicants, outside of the ECOA selftest privilege, would also be subject to varying degrees of regulatory scrutiny, which could serve to deter lenders from using such data for discriminatory purposes. Similarly, all lenders that chose to collect and use such data for discriminatory purposes would face the risk of public disclosure of such practices through litigation. Further, according to a variety of researchers and officials we contacted, as well as FRB documents we reviewed, there is no evidence that lenders have used HMDA data for discriminatory purposes. These officials generally attributed the transparency of the HMDA program, through regulatory reviews and public reporting requirements, as serving to help deter lenders from using the data to discriminate in mortgage lending.12 Finally, FRB could potentially have mitigated some of its concerns that voluntarily collected data could be used for discriminatory purposes by 12 We recognize that there are differences in the level of transparency between HMDA’s data collection and reporting requirements and the voluntary data collection proposal that FRB considered in 1999 for nonmortgage loan applicants. In particular, FRB did not propose that lenders who chose to collect such data report it to the public whereas lenders are required to report HMDA data. Page 11 GAO-08-1023T including, as part of its 1999 proposal, minimum procedures for the collection and use of such data. FRB established such procedures for federally regulated lenders that choose to conduct a self-test. These procedures include developing written policies describing the methodology for data collection and keeping data on personal characteristics separate from loan underwriting data that are used to make credit decisions. Imposing such minimum procedures and requirements for a voluntary program could serve to enhance regulators’ oversight of lenders’ data collection, processes, practices, and uses of the data, and further deter possibly discriminatory practices. Many Researchers and Others Agreed That Voluntarily Collected Data May Not Be Reliable or Useful in Helping to Better Identify Possible Discrimination in Nonmortgage Lending Even so, many researchers, regulatory staff, and representatives from consumer groups and banking trade groups agreed with FRB’s conclusion that the reliability of voluntarily collected data may be limited in identifying possible discrimination in nonmortgage lending. In particular, they agreed with FRB that, due to potentially inconsistent data collection standards, it would be difficult to use voluntarily collected data to compare fair lending performance across different lenders. Additionally, there may be data inconsistency problems for any given lender that chooses to collect data on personal characteristics for nonmortgage loan applicants. For example, a lender could “cherry pick,” or collect racial, gender, and other data on personal characteristics on applicants only for certain loan products that they felt would reflect favorably on their fair lending practices and not collect data for other products. Just as FRB could potentially have mitigated some of its concerns about the possibility that lenders would use voluntarily collected data for discriminatory purposes by adopting minimum procedures, as mentioned previously, it could also potentially have considered adopting data collection standards. Such standards could have served to better ensure the consistency of the data and enabled regulators and others to use the data to assess individual lender performance and compare lending practices across different financial institutions. However, according to a senior FRB official, a researcher, and a bank industry trade association official, the imposition of such standards would have undermined the voluntary nature of the data collection proposal. For example, FRB could be required to conduct examinations to help ensure that federally regulated lenders were collecting the data in a manner consistent with any such standards. Moreover, the establishment of such data collection standards might also have further diminished lender interest in a voluntary program, which researchers, FRB officials, and others said was already limited due to the potential for increased regulatory and public scrutiny of Page 12 GAO-08-1023T their lending practices. According to bank regulators and banking trade groups, very few, if any, lenders choose to conduct self-tests out of concern that the results of such tests would be subject to regulatory review even though they are privileged. Finally, while some officials we contacted and documents we reviewed said that any data that was collected and potentially reported by lenders would provide important insights into nonmortgage lending practices that are not currently available, other researchers and researchers suggested that such data would be prone to substantial selection bias. That is, the data would likely be skewed by the possibility that only lenders with good fair lending compliance records would choose to collect such data. Consequently, although voluntarily collected data on personal characteristics could provide some benefits, it would not likely materially assist the capacity of researchers, regulators, and others to better understand possible discrimination in nonmortgage lending. A Data Collection and Reporting Requirement Could Further Efforts to Better Understand Possible Discrimination in Nonmortgage Lending but Would also Involve Complexities and Costs That Would Require Consideration In concept, a requirement that lenders collect and publicly report data on the personal characteristics of nonmortgage loan applicants, similar to HMDA requirements, could help address some of the existing data limitations that complicate efforts by researchers, federal bank regulators, and others to identify possible discrimination. However, mandatory data collection and reporting would impose some additional costs on the lending industry, although opinions differed on how burdensome these costs might be. While options exist to potentially mitigate some of these costs, such as limiting data collection and reporting to specific types of lending, these options also involve additional complexities and costs that must be considered. Page 13 GAO-08-1023T Researchers and Regulators Could Benefit from Mandatory Data Collection and Reporting, but Lender Costs Would Increase Required data collection and reporting for nonmortgage loan applicants, similar to HMDA’s requirements, could help address some of the existing limitations of available data and facilitate the efficiency of the fair lending examination process for nonmortgage lending. Such data would be more timely than SSBF data, and the implementation of data collection standards could help ensure its reliability. For example, researchers and financial regulators would be able to analyze the practices of specific lenders and compare practices across lenders, assessing lending practices by type, size, and location of the institutions, similar to analyses done currently with HMDA data. While such analyses would represent only the first step in determining whether or not particular lenders were engaging in discriminatory practices, they could potentially help regulators prioritize their examinations and better utilize existing staff and other resources. While it is not possible to quantify the potential costs associated with a reporting requirement, in part because the requirements could vary, banking organizations and banks that we contacted identified a variety of additional costs that lenders might face. These officials also said that they were concerned about such costs and that the additional expenses associated with data collection and reporting would, in part, be passed on to borrowers. According to the officials, most of the costs associated with a reporting requirement would involve developing the information technology necessary to capture and report the data, including system integration, software development, and employee training. Moreover, the officials said that, as with HMDA data, verifying, any reported data would also entail costs, including expenses associated with conducting internal audits. The regulatory agency responsible for assembling, verifying, and reporting the data to the public would also accrue costs for these activities.13 Some researchers and representatives from consumer groups we contacted said that they did not think that the costs associated with required collection and reporting of data on personal characteristics of nonmortgage loan applicants would be significant because many lenders already collect and report data on personal characteristics under HMDA. But representatives from banks and banking organizations, along with one researcher, said that lending information systems and personnel were not 13 According to FRB officials, it will cost the agency approximately $3.5 million to process the 2008 HMDA data. Page 14 GAO-08-1023T integrated in many mortgage and nonmortgage organizations. For this reason, they reiterated that a data collection and reporting requirement would involve additional system integration and employee training costs, among others. Limiting a Data Collection and Reporting Requirement to Specific Types of Nonmortgage Loans Would Also Have Benefits and Costs One potential option to mitigate the costs associated with a requirement that regulated lenders collect and report data on the personal characteristics of those seeking nonmortgage loans would be to limit the requirement to certain types of loans, such as small business and/or automobile loans. Similar to mortgage loan applications, small business and automobile loan applications are often made on a face-to-face basis, which could enhance the ability of lenders to help verify the race, gender, or other personal characteristics of the applicants. In contrast, lenders’ capacity to record data on personal characteristics for other types of nonmortgage applicants, such as applicants for credit card loans, may be limited by the fact that credit card loan applications and credit decisions are typically done by mail or over the Internet. However, researchers, federal bank regulatory staff responsible for fair lending oversight, banking officials, and representatives from some consumer groups we contacted cautioned that there were still significant complexities and potential costs associated with a data collection and reporting requirement that was limited to small business lending. Unlike mortgage and automobile lending, which have relatively uniform underwriting criteria, these officials said that small business loan underwriting is heterogeneous and more complex. For example, the types of financing that small businesses typically seek can vary widely, ranging from revolving lines of credit to term loans, and the risk of the collateral pledged against these loans may also vary widely (i.e., from relatively secure real estate to inventory).14 As discussed previously, studies of possible discrimination in small business lending that use SSBF data consider a variety of other indicators of creditworthiness, such as applicants’ credit scores, personal wealth, and history of bankruptcy. Without information on key underwriting variables, the officials said, research based on the reported data could be subject to significant controversy and potential misinterpretation, much like research based on HMDA data, which lacks information on these variables. At the same time, 14 We note, though, that small business owners may also use their personal residences as collateral to secure business loans. Page 15 GAO-08-1023T costs for the necessary technology, employee training, and data verification would likely increase as the range of data that lenders were required to collect and report increases. One option to potentially enhance federal oversight of the fair lending laws, while mitigating lender cost concerns, would be to require lenders to collect data on personal characteristics for small business loan applicants, and perhaps other types of nonmortgage lending like automobile lending, and make the data available to regulators but not require public reporting of such data or any other information. This approach could facilitate federal bank regulators’ ability to prioritize fair lending examinations for regulated lenders because the agencies currently do not have ready access to data on personal characteristics for nonmortgage loan applicants. It could also limit lender costs because they would not have to collect, publicly report, and verify data on a range of underwriting variables because regulators already have access to this information. However, due to the lack of a public data reporting requirement, such an option would not enhance the capacity of researchers, Congress, and the public to better understand the possibility of discrimination in nonmortgage lending. In closing, assessing the potential for discrimination in nonmortgage lending is an important and complex issue. While current data sources, primarily FRB’s SSBF and SCF provide important insights into possible discrimination in certain types of lending, they both have limitations that may impede the ability of researchers, regulators, Congress, and the public to further assess lender compliance with the fair lending laws. It is also not yet clear how FRB’s decision to discontinue the SSBF and incorporate elements of the survey into an expanded SCF beginning in 2010 will impact the already limited amount of information about possible discrimination in nonmortgage lending. Therefore, from a public policy perspective, now may be the time to consider whether the benefits of additional data for research and regulatory purposes outweigh the costs of collecting the data, as well as the trade-offs of various options to enhance available data, from a purely voluntary program to a data collection and reporting requirement, and decide whether such a requirement is warranted. Mr. Chairman, this concludes my prepared statement. I would be pleased to respond to any questions you or other Members of the Subcommittee may have. Page 16 GAO-08-1023T GAO Contact and Staff Acknowledgments (250405) For further information about this testimony, please contact Orice M. Williams on (202) 512-8678, or at williamso@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this statement. Individuals making key contributions to this testimony include Wesley M. Phillips, Assistant Director; Benjamin Bolitzer; Emily Chalmers; Kimberly Cutright; John Forrester; Simin Ho; Omyra Ramsingh; Robert Pollard; Carl Ramirez; and Ethan Wozniak. Page 17 GAO-08-1023T This is a work of the U.S. government and is not subject to copyright protection in the United States. The published product may be reproduced and distributed in its entirety without further permission from GAO. However, because this work may contain copyrighted images or other material, permission from the copyright holder may be necessary if you wish to reproduce this material separately. GAO’s Mission The Government Accountability Office, the audit, evaluation, and investigative arm of Congress, exists to support Congress in meeting its constitutional responsibilities and to help improve the performance and accountability of the federal government for the American people. GAO examines the use of public funds; evaluates federal programs and policies; and provides analyses, recommendations, and other assistance to help Congress make informed oversight, policy, and funding decisions. GAO’s commitment to good government is reflected in its core values of accountability, integrity, and reliability. Obtaining Copies of GAO Reports and Testimony The fastest and easiest way to obtain copies of GAO documents at no cost is through GAO’s Web site (www.gao.gov). Each weekday, GAO posts newly released reports, testimony, and correspondence on its Web site. To have GAO e-mail you a list of newly posted products every afternoon, go to www.gao.gov and select “E-mail Updates.” Order by Mail or Phone The first copy of each printed report is free. Additional copies are $2 each. A check or money order should be made out to the Superintendent of Documents. GAO also accepts VISA and Mastercard. Orders for 100 or more copies mailed to a single address are discounted 25 percent. Orders should be sent to: U.S. Government Accountability Office 441 G Street NW, Room LM Washington, DC 20548 To order by Phone: Voice: TDD: Fax: (202) 512-6000 (202) 512-2537 (202) 512-6061 Contact: To Report Fraud, Waste, and Abuse in Federal Programs Web site: www.gao.gov/fraudnet/fraudnet.htm E-mail: fraudnet@gao.gov Automated answering system: (800) 424-5454 or (202) 512-7470 Congressional Relations Ralph Dawn, Managing Director, dawnr@gao.gov, (202) 512-4400 U.S. Government Accountability Office, 441 G Street NW, Room 7125 Washington, DC 20548 Public Affairs Chuck Young, Managing Director, youngc1@gao.gov, (202) 512-4800 U.S. Government Accountability Office, 441 G Street NW, Room 7149 Washington, DC 20548 PRINTED ON RECYCLED PAPER
@FedFRASER