NR 96-45 April 19, 1996 OCC Summarizes Results of Fair Lending Pilot Testing WASHINGTON, D.C. -- The Office of the Comptroller of the Currency (OCC) today issued a report summarizing the results of its pilot program to assess the use of "testers" to identify fair lending violations at national banks. "Matched pair tests were conducted at eight national banks and no violations of fair lending laws were detected," said Bert Otto, Acting Deputy Comptroller for Compliance Management. "Based on the pilot experience, we've concluded that matched-pair testing can be a valuable tool to use in certain situations in the agency's overall fair lending program." The OCC launched its pilot testing program in July 1994 to assess whether national banks and their mortgage subsidiaries treated consumers of different races and nationalities similarly during the pre-application phase of the mortgage process. In addition to actual testing, the OCC examined and evaluated testing processes currently in use by other regulatory agencies, fair housing organizations, mystery shopping vendors and others who had experience with matched-pair testing. Going forward, the OCC will consider further use of testing on a case-by-case basis when an evaluation of information from examiners, consumers or the media indicates that a bank may be engaging in illegal discrimination in the pre-application stage. The agency also has encouraged national banks to conduct fair lending self-assessments. Self-assessments include both self-evaluations of information contained in loan files and self-tests at the pre-application stage of the loan process. Interim guidelines on self-assessments were sent to national banks in OCC Bulletin 95-51, dated September 15, 1995. "A significant benefit of our fair lending pilot testing program is that the agency now can assist national banks in developing their own self-testing programs," Otto said. The report summarizing the OCC's fair lending pilot testing program accompanies this news release. The document is also available on the agency's Internet web site, found at http:\\www.occ.treas.gov, or it may be obtained by fax from the OCC Information Line at (202) 479-0141 document number 79645. # # # The OCC charters, regulates and examines approximately 2,800 national banks and 70 federal branches and agencies of foreign banks in the U.S., accounting for more than half the nation's banking assets. Its mission is to ensure the safety and soundness of the national banking system. The Office of the Comptroller of the Currency Fair Lending Pilot Testing Program Summary Report April 1996 OCC's Pilot of Fair Lending Testing at National Banks Background In 1993, the OCC assembled an interdepartmental team, under the direction of the Compliance Management Department, to develop and implement an OCC mortgage loan testing program. A pilot program to test whether national banks and their mortgage subsidiaries treated mortgage loan applicants of different races and national origin similarly during the pre-application phase was launched during 1994, and all tests were completed by January 1995. Those tests did not disclose any violations of fair lending laws. In addition to actual testing, the team examined and evaluated testing processes currently in use by other regulators, fair housing organizations, mystery shopper vendors, and others who had experience with testing. To prepare for testing and to set the parameters for this pilot, the team reviewed testing materials from the Urban Institute, the Federal Reserve Board, the Department of Housing and Urban Development (HUD), the University of Wisconsin, the University of Michigan, the Department of Justice (DOJ), and others. In addition, the team received a large amount of information in response to a "Request for Information" that the OCC published in the Commerce Business Daily. The team also interviewed staffs of other regulators including HUD, the Federal Trade Commission (FTC), and DOJ; a nonprofit fair housing group; a nonprofit research group; and a for-profit firm. Major Objectives After gathering information and becoming more experienced with the testing process, the team identified the following as its major objectives: - To plan, develop, and implement a pilot testing program. - To select national banks or their wholly owned mortgage subsidiaries to be tested on a pilot basis to determine if they were prescreening applicants on a prohibited basis. - To gain experience in testing and to evaluate the effectiveness of different testing methods used by organizations offering testing services. - To draw conclusions about whether, how and when the OCC could use testing in the future. Planning the Pilot Program Based on the information gathered, the team determined that the OCC should hire outside testing organizations to conduct tests. Since outside organizations already had hired, trained, and debriefed testers, the team believed that using those existing structures, reviewed and modified by OCC, could save time in implementing a pilot program. The team determined that the pilot program should: - Test for pre-application discrimination based on the race/national origin of the applicant. Although other prohibited bases could have been used, the team believed that statistical analyses and public attention in the last few years have focused on the race and national origin of applicants as the major cause for possible discrimination in mortgage lending. - Select blacks and Hispanics as the target racial/ethnic groups. Blacks and Hispanics are major minority groups in the United States and have been the target of recent statistical analyses and public concern regarding discriminatory treatment. - Focus on home purchase loans, using in-person, paired testers. Paired testing sends people who have comparable application characteristics into each institution to inquire about obtaining credit. In these tests, one person in each pair was white, and the other was either black or Hispanic. Focusing on home purchase loans allowed the pilot program to use Home Mortgage Disclosure Act (HMDA) data to select institutions to be tested and readily available mortgage underwriting ratios to construct tester profiles. Using in-person interviews avoided the ambiguities associated with telephone contacts. - Construct fictional qualifications that show testers to be marginally qualified for credit, with minority testers slightly more creditworthy than non-minority ones. - Audiotape tests, where permitted by law. - Select target lenders on the basis of potential discrimination based on characteristics of the applicant, rather than characteristics of the neighborhood. - Identify a variety of geographic areas in which illegal discrimination could exist. - Use different professional testing organizations, including both fair housing organizations and marketing research firms (sometimes called "mystery shopper" firms) to perform pilot testing for the OCC. Four organizations one fair housing organization and three market research firms conducted tests. Fair housing organizations and market research organizations have fundamentally different approaches to testing. Fair housing organizations generally recruit testers from local, nonprofit organizations, such as universities and the League of Women Voters. They generally conduct fewer tests and the testers are aware that they are testing for illegal discrimination. They also train their testers more extensively in bank lending processes. Market research organizations, on the other hand, recruit testers from local marketing research firms. They conduct numerous tests as bases for statistical analysis and use testers that are generally unaware of the purpose of the test. The organizations were located in different portions of the United States, which allowed the OCC to achieve wide geographic coverage while keeping costs low by selecting a testing organization located near the lender to be tested. - Use no testing company that had a business relationship with an institution selected for testing. Selecting Institutions for Testing Institutions were selected for inclusion in the pilot program primarily based on HMDA and census data and supervisory information. The team began by identifying metropolitan statistical areas (MSAs) that had sufficiently large minority populations that a low number of minority applications would raise the possibility of unequal treatment. It then identified HMDA reporters who received more than a certain minimum number of home purchase loan applications and eliminated any MSA that did not have any such institutions. The team then further reduced the potential MSAs by eliminating census tracts in which the minority population was so small that the testers might be exposed, i.e., that the lender might be suspicious that they were not part of the local community. For national banks (or wholly owned mortgage subsidiaries of national banks) in the remaining MSAs, the team determined the percentage of minority applications each received and compared their performance to that of other HMDA reporters. The goal was to identify any bank that received significantly lower percentages of minority applications than other institutions in the same market. Once the team identified the national banks (and subsidiaries) with the lowest percentages, it reviewed other information about the bank; specifically, the OCC's internal Supervisory Monitoring System (SMS), Community Reinvestment Act (CRA) reports, and consumer complaints. Then the team looked at the geographic distribution of the offices of the potential candidates for testing and eliminated any banks that did not have offices located near minority populated census tracts. The team made final selections to avoid concentrating resources in just one part of the country. The result was a group of eight institutions seven national banks and one wholly owned mortgage company subsidiary of a national bank that were tested. Five institutions were tested for pre-application discrimination against Hispanics and three were tested for pre-application discrimination against African Americans. Lessons Learned from the Pilot Program - Agency involvement in the testing process is critical because different test structures, testing companies, testers, and local conditions can affect what testers find. One testing company in the pilot devised virtually identical fictional qualifications for all testers. Without agency involvement to amend those qualifications, the institution would quickly have detected that it was being tested. In one test, there were significant differences in the way individual loan officers dealt with the testers, including the amount of time they spent with the testers, the amount of information they shared with the testers, and in the criteria for obtaining a loan that they explained to the testers. However, each loan officer treated all testers he or she dealt with similarly. The differences in treatment, therefore, were attributable to the loan officers themselves not the race of the testers. In another test, few testers, either minority or non-minority, encountered loan officers that addressed the testers' creditworthiness. Only one of 26 testers was asked about savings, investments, or credit history. In contrast, 85 percent of testers said the loan officer described the next step to take. The loan officers evidently viewed these interviews as very preliminary. One testing company went beyond the parameters of the test and attempted to ascertain "whether effort was made to make a loan to an applicant who presents that opportunity." In doing so the company found significant differences in treatment, although not necessarily related to access to credit. Follow-up review by the agency found no evidence of different and less favorable treatment of minorities. OCC staff determined that indications of disparate treatment were weak and primarily involved unverifiable testers' subjective perceptions, such as how "friendly" the loan officer was to the tester. - The OCC needs to impose high standards concerning the quality of information produced by testing because of its enforcement responsibilities. To ensure that testing produced information that was sufficiently clear, reliable, and precise to determine whether violations had occurred and to support an enforcement proceeding, the OCC insisted on data standards more rigorous than a research or customer service project would need. Some standards that arose from the pilot test are: Testing companies must focus on and document lender conduct that facilitates or limits access to credit. In debriefing testers, questions should not call for interpretation or subjective responses. Although in other testing situations such observations are valued for the information they provide on customer service, from an enforcement perspective, they are not sufficiently clear, reliable, or comparable to serve as a basis for determining whether discrimination occurred. Tester debriefing questions cannot rely on simple yes/no answers. They must seek more of the facts presented by the lender ("What, if any, fees must be paid?") and ask about substantive differences in treatment related to access to credit. The debriefing questions cannot assume that every tester followed the prescribed scenario. They must ask the tester to report the facts he or she described to the lender. This allows the OCC to be more confident that the testers present themselves to the bank as similarly qualified. Testing companies must understand the importance of the documentation of the test. In one case a testing company official's notations on questionnaires filled out by testers might have compromised the value of the test report as evidence. - Using testers presents a variety of difficulties. The possibility that they might be asked to testify made some potential testers reluctant to participate. Testers sometimes find the fictional scenarios complicated and confusing, and therefore, have a hard time staying with the "script." Testers and testing companies do not always understand the OCC's concept of "marginal" qualifications. - Testing companies must understand what they are testing. Even when it was delivering the final report, one testing company had basic misconceptions about the activity it had just completed. Such misconceptions may not compromise the test; however, they present the disturbing possibility that a testing company may misinterpret results, and reinforce the importance of agency oversight. Conclusions and Recommendations for Future OCC Testing Matched-pair testing can be a valuable tool in the OCC's overall fair lending program. Examples of situations in which the OCC might consider conducting pre-application testing are: - In a compliance examination, examiners discover that the institution reports receiving virtually no mortgage loan applications from a geographic area of the city that is predominantly minority. Other lenders report receiving significant numbers of applications to purchase property located in this geographic area. - Examiners are approached during an examination and told that one of the loan officers quotes significantly higher interest rates to minorities inquiring about consumer credit than are quoted to whites making similar inquiries. - The OCC receives several unrelated complaints from members of the same minority group against the same bank alleging that they were discouraged from applying for no apparent reason or were met with such resistance when they inquired about consumer loans that they decided to apply elsewhere. - During a CRA examination, minority community contacts report that it is difficult to obtain mortgage loans from a bank because the bank requires a 30 percent down payment. Yet, examiners conducting a concurrent fair lending examination report few mortgage loans to minorities and numerous loans to whites with less than 30 percent down payments. These examples are not inclusive. The OCC will consider testing at an institution when information from examiners, consumers, or the news media indicates the institution may be engaging in illegal discrimination in the pre-application stage. The OCC will not target institutions solely on the basis of HMDA and census data and will not seek to conduct any specific number of tests each year. When the OCC does identify an institution for testing: - An OCC fair lending specialist will manage all testing, including on-site coordination with the vendor before, during, and after each test. That coordinator will observe or review each step of the process, including identifying and contacting the vendor, estimating the number of tests to be conducted, developing the applicant profiles and scenarios, training and debriefing the testers, analyzing the test reports, monitoring for indications that the bank has detected the testing, and reaching any conclusion about whether discrimination occurred. - Tests will be refined, particularly with respect to applicant profiles and the identification of lender loan terms and policies. - More tests will be completed at a decision center (e.g., bank, branch, mortgage company), even if it means testing fewer decision centers of the same institution. To avoid uncertainties in evaluating the testing data, at least four pairs of testers will be sent into a single office of an institution. This approach will also increase the likelihood that some minority and non-minority testers will speak to the same individuals. - OCC testing will be conducted by private fair housing organizations when possible. In the pilot program, testers from fair housing organizations were more professional and thorough and, typically, had participated in a number of tests previously. Testers from fair housing organizations offer the additional advantage of knowing they are testing for discrimination, though not the specific prohibited basis being targeted, and generally are willing to provide testimony, if necessary. - Tests will be audiotaped where permitted by state law for better documentation and to help deflect any allegations of tester bias.