PART III: STUDY RESULTS AND INTERPRETATION.

1. Worst case lending patterns: The most common variations..

   The Banking Research Project identified 62 lending patterns representing worst case situations. These are instances in which a major lender has undertaken virtually no home purchase loan origination activity in minority neighborhoods, or has substantially underserved such neighborhoods. For a lending pattern to be designated "worst case," four criteria had to be met:

   1. The lender had to rank among the top 20 home purchase loan originators within the metro area;

   2. The lender had to rank among the lowest of the major home purchase loan originators within a given metro area, in terms of the percentage of a lender's total home purchase loan originations in the metro area made in minority neighborhoods, as determined by HMDA statistics;

   3. The lender had to originate home purchase loans throughout most of the metro area or at least large segments of it; and

   4. The lender had to exhibit a pattern of excluding or substantially underserving minority areas.

   It is the combination of the third and fourth criteria -- a geographic lending territory of broad scope and the failure to lend in (or substantially underserve) minority neighborhoods -- that this study considers the unique feature of the worst case lending patterns.

   The lending pattern maps provide a clear picture of each lender's effective lending territory, and the market share approach is particularly useful in assessing a lender's marketing strategy.

   The 62 worst case lending patterns involved only 49 lending institutions because a number of lenders accounted for worst case lending patterns in more than one of the 16 metro areas. Sears Mortgage Corporation was a worst case lender in 6 of the 16 metro areas -- Chicago, Atlanta, Los Angeles, St. Louis, Oakland, and Pittsburgh. This nationwide pattern of excluding minority neighborhoods by a large-scale mortgage lender, in and of itself, casts serious doubt on the effectiveness of Fair Lending law enforcement. Prudential Home Mortgage Company was a worst case lender in 3 of the 16 metro areas -- Atlanta, New York City and Oakland.

   Apart from Sears and Prudential, six other lenders were mapped as worst case lenders in two metro areas each:

   Chase Home Mortgage Corporation in Chicago and New York City;
   IMCO Realty Services in Dallas and Houston;
   CTX Mortgage Company in Dallas and Houston;
   Margaretten & Company in Philadelphia and Miami;
   American Home Funding in Buffalo and Baltimore; and
   B.F. Saul Mortgage Company in Washington, D.C., and Baltimore.

   What was particularly revealing about the maps was the three distinct categories of worst case lending patterns that emerged:

   1. Loan activity throughout a metro area but not in minority neighborhoods.

      Some major mortgage lenders maintain a high or average market share throughout the entire metro area, or at least most of it. This pattern does not hold in minority neighborhoods, however, where mortgage lenders show virtually no lending activity, or lending activity that is very limited in scope. Prime examples of this dynamic include the lending patterns of Standard Federal Bank in the Detroit metro area (Map 13); United Postal Savings Association in the Missouri portion of the St. Louis metro area (Map 14); and IMCO Realty Services in the Houston metro area (Map 15).

      Other notable examples include Sears Mortgage Corporation in the Chicago metro area (Map 1), GE Capital Mortgage Services in the Philadelphia metro area (Map 16), and Chase Lincoln First Bank in the Buffalo metro area (Map 17), and CTX Mortgage in the Dallas metro area (Map 18).

   2. No loan activity in minority neighborhoods and some white neighborhoods.

      Some major lenders are active throughout large portions of the metro area, but are inactive in various white neighborhoods as well as the minority community. In the Washington, D.C., metro area, B.F. Saul Mortgage Company (Map 20) was an active lender in many white neighborhoods, but shows no lending activity in the great majority of census tracts in which the minority population concentration is greater than 50 percent. In the Atlanta metro area, Sears Mortgage Corporation (Map 21) provides another example of this pattern.

      In the Chicago metro area, Chase Home Mortgage Corporation (Map 22) was generally inactive in both the large minority community and a number of the white neighborhoods that lie immediately to the west or northwest of the minority community. In the Oakland metro area, Wells Fargo (Map 23) overall mortgage lending declined from 1990 to 1991, but the lender was consistent in avoiding home purchase lending in minority neighborhoods during both years.

      In the New York metro area, Prudential Home Mortgage Company (Map 24) was an active lender in many of the white areas of Westchester County, Manhattan, Queens, and Staten Island, but showed little lending activity in many of the white neighborhoods of Brooklyn and the Bronx, and almost no activity in the metro areas's many minority neighborhoods. The lending pattern map for Chemical Bank in the New York metro area (Map 25) is roughly comparable to Prudential's lending pattern map. American Home Funding exhibits a similar lending pattern in the Baltimore metro area (Map 26), as does Bell Federal Savings and Loan in the Pittsburgh metro area (Map 27).

      In many cases, this pattern of excluding some white neighborhoods along with virtually all of the minority community is the product of upscale marketing strategies that focus on upper income and upper/middle income neighborhoods. To depict this marketing strategy, lending pattern maps for Chase Home Mortgage Corporation in the Chicago metro area (Map 2) and NBD Mortgage Company in the Chicago metro area (Map 28) were created using the diagonal overlay to indicate the location of upper or upper/middle income neighborhoods -- i.e, census tracts in which the median family income was greater than 110 percent of the metro area median family income. As these maps clearly demonstrate, Chase's and NBD's marketing strategy is targeted to the residents of such upscale neighborhoods.

      A second lending pattern map for Bell Federal Savings and Loan in the Pittsburgh metro area (Map 29) provides another perspective on lending patterns shaped by upscale marketing strategies. In this map, the diagonal overlay indicates census tracts in which the median family income is less than 90 percent of the metro area median family income -- i.e., low and moderate income census tracts, expansively defined. The map shows that Bell Federal engages in little lending in the Pittsburgh metro area's low and moderate income census tracts. Since virtually all of Pittsburgh's minority neighborhoods are low and moderate income neighborhoods, this upscale lending strategy has a powerful discriminatory effect, as shown by the first lending pattern map for Bell Federal with the minority neighborhood overlay (Map 27).

   3. Loan activity that stops abruptly at the edge of minority neighborhoods.

      Other major lenders have a high market share in a key segment of the metro area, but their lending stops abruptly at the edge of minority neighborhoods that lie adjacent to their lending areas. For example, the lending pattern map for Mid America Federal Savings Bank in the Chicago metro area (Map 30) shows that the S&L is a dominant lender in its primary service area, a large western segment of the metro area. But this intense lending activity comes to a dramatic halt at the edge of the minority neighborhoods that lie directly to the east of Mid America's primary service area. The map also shows that, in contrast to its lack of lending activity in the adjacent minority neighborhoods, Mid America does lend to a moderate extent in white neighborhoods that lie to the north and to the south of its primary service area.

2. Worst case lending patterns shaped by minority population concentration and composition.

   Among worst case lenders, lending activity tends to fall off substantially in neighborhoods where minorities comprise 50 percent or more of the population. The majority of the worst case lending pattern maps have a diagonal overlay that defines minority neighborhoods in terms of a 50 percent or greater minority population concentration. The diagonal overlay was set at 50 percent or greater for 37 maps (of the 62 worst case lending patterns); at 75 percent or greater for 9 maps; and at 25 percent or greater for 12 maps.

   In the case of two lending patterns -- Decatur Federal Savings & Loan in Atlanta and Nationsbank Mortgage Corporation in Washington D.C. -- we created twin minority overlay maps to explore the richer nuances of how minority concentrations and composition can affect lending patterns. We generated one map using a 50 percent or greater minority population concentration overlay, and a companion map using a 25 percent or greater minority population concentration overlay. We also created two lending pattern maps for the Miami metro area, one that uses a 50 percent or greater minority population concentration overlay, and another which indicates census tracts in which Black persons comprised 50 percent or more of the population.

   Four significant findings about minority population concentration and composition emerge from the worst case lending maps:

   • The maps suggest that the levels of minority population concentration which result in sharp declines in mortgage lending often vary from one metro area to another. For example, in the Atlanta and Detroit metro areas, a number of the worst case lenders appear to avoid lending in neighborhoods in which the minority concentration level is 25 percent or greater. On the other hand, in the Los Angeles metro area, several of the worst case lenders do not appear to curtail lending activity until the minority concentration level rises to 75 percent or greater.

   • In some cases, mortgage lending cut-off points vary with differences in the composition of the minority population -- e.g., Black or Hispanic. The minority population composition maps are useful in identifying lending patterns that differ between Black and Hispanic neighborhoods. The lending pattern maps for the Miami metro area provide the best example of this.

   • In many cases, the cut-off point for mortgage lending in minority neighborhoods is significantly influenced by the income level of the neighborhood. The median family income maps can be used to assess this factor. This relationship is evident in a number of the lending pattern maps for the Los Angeles and the Atlanta metro areas.

   • In some metro areas, the cut-off point for lending in minority neighborhoods varies between individual lenders.

   Many revealing insights into mortgage lending patterns -- particularly the variations in cut-off points -- can be obtained by combining the color-coded lending pattern maps with minority population concentration, minority population composition maps and median family income maps. (All of these maps are listed in Appendix C.)

   For example, in the Atlanta metro area, Decatur Federal Savings & Loan and Griffin Federal Savings Bank make very few mortgage loans in neighborhoods where the minority population concentration was greater than 75 percent. Decatur engaged in limited lending activity in integrated neighborhoods -- i.e., census tracts in which the minority population concentration is in the 25 percent to 75 percent range. But Decatur's lending in such integrated neighborhoods was generally confined to middle-income neighborhoods.

   In the Detroit metro area, mortgage lending activity by a number of the worst case lenders dropped dramatically or was virtually nonexistent in neighborhoods where the minority population concentration level was 25 percent or more, especially along the northern boundary of the City of Detroit. This can be seen in the lending pattern maps for Standard Federal Bank, GMAC Mortgage Corporation, First Nationwide Bank and Republic Bancorp Mortgage. A similar pattern is seen in the Washington D.C. metro area, in which Nationsbank Mortgage Company made almost no home purchase loans in the large segments of the metro area where the minority population concentration is greater than 50 percent. The company also makes very few mortgage loans in the even larger portion of the metro area where the minority population concentration is greater than 25 percent.

   By contrast, while IMCO Realty Services in the Houston metro area made virtually no mortgage loans in minority neighborhoods where the minority population concentration was above 75 percent, it did engage in a moderate level of lending in neighborhoods where the minority population concentration fell within the 25 percent to 75 percent range.

   An in-depth review of the lending ratios and lending pattern maps produced for this report suggests that there is no clear-cut statistical rule for identifying lenders that have excluded minority neighborhoods from their effective lending territories. This is because the level of minority concentration at which a given lender begins to exclude minority neighborhoods will vary considerably from metro area to metro area, and from lender to lender within the same metro area. The approximate cut-off points can be as various as 75 percent minority; 50 percent minority; 25 percent minority; 75 percent Black; or 75 percent Hispanic. Where lenders are drawn from multiple metro areas, much of this variation reflects demographic and economic differences between metro areas. But even among lenders from the same metro area, there is considerable room for individual lender preferences and biases to play a role.

   In assessing all of these lending pattern maps, it is important to keep in mind that race-based discrimination of any sort in mortgage lending is unlawful. Federal law prohibits lenders from varying their mortgage lending activity depending on the level of minority concentration or the composition of the minority population. The Fair Lending laws require that all types of minority neighborhoods be given equal access to mortgage financing. Thus, a mortgage lender cannot justify racial redlining in neighborhoods where the minority concentration level is 75 percent or more by citing greater lending activity in neighborhoods where the minority concentration level ranges from 25 percent to 75 percent. Similarly, lending in Hispanic neighborhoods does not mitigate the exclusion of Black neighborhoods, and vice versa.

   It bears special mention that HMDA statistics, including those in this report, often do not tell the full story. HMDA statistics by themselves are often used to measure a mortgage lender's level of mortgage lending in minority neighborhoods as a group. But these numbers do not show the distribution of a lender's loans by minority concentration or composition within the minority community. This distribution may be very significant in assessing whether or not Fair Lending laws have been violated.

   Several of the maps illustrate the inherent limitations of HMDA statistics in evaluating lending patterns for Fair Lending enforcement purposes. Margaretten & Company in the Miami metro area is the best example. The HMDA statistics in Table 10 indicate that the percentage of Margaretten's total home purchase loan approvals going to Black neighborhoods was roughly the same as its percentage for the Miami mortgage market as a whole. A different story emerges, however, from the map depicting Margaretten's lending pattern in the Miami metro area (Map 19) and a Banking Research Project census-tract-by-census-tract review of the loan origination data underlying the map. The more detailed statistics, when combined with a geographical depiction, reveal that while Margaretten was an active lender in several outlying Black census tracts to the south of the central city area, it provided comparatively few loans in the large cluster of Black census tracts that extend north from the central city area.

3. Decatur Federal S&L's lending pattern in Atlanta, Georgia

   In analyzing Decatur's pattern of home mortgage loan originations, the Justice Department noted that of the total number of mortgage loans originated by Decatur in the Atlanta Region in 1988, only 3 percent (45 loans) were made in Black census tracts, while the comparable percentage was also 3 percent in both 1989 (35 loans) and 1990 (33 loans).[12] For purposes of this calculation, the Justice Department defined Black census tracts as tracts in which Black persons comprised 50 percent or more of the population. The Banking Research Project's analysis of Decatur's 1991 HMDA data indicates that during 1991 only 2.6 percent (29 loans) of Decatur's home purchase loan originations within the Atlanta metro area were made in Black census tracts, as defined by the Justice Department.

   The Banking Research Project has prepared a lending pattern map for Decatur in the Atlanta metro area based on its 1991 HMDA activity (Map 37). The map concept is comparable to that employed by the Justice Department -- except that minority census tracts are defined as those in which minorities comprised 50 percent or more of the population, rather than those in which Black persons comprised 50 percent or more of the population. The map suggests that Decatur's mortgage lending pattern for 1991 had not changed much from the pattern which the Justice Department had alleged in its complaint to constitute a violation of the Fair Lending laws.

   The Justice Department complaint described Decatur's lending pattern in the following terms:

   Decatur Federal defined its lending area pursuant to the requirements of the Community Reinvestment Act (CRA). Decatur Federal circumscribed its CRA lending area in Fulton County to exclude most of the predominantly black neighborhoods of the City of Atlanta and South Fulton County by largely following the tracks of the Seaboard Coast/Southern Railway line and the Georgia Railroad, which historically separated black neighborhoods from white neighborhoods in South Fulton County. As a result, over 76 percent of the black population of Fulton county at that time, based on the 1970 Census, was excluded from Decatur Federal's CRA lending territory, while the predominantly white areas of North Fulton county were included.[13]

4. Lack of applications from minority neighborhoods.

   There are two basic reasons why a mortgage lender may fail to maintain roughly the same overall market share in minority neighborhoods as in white neighborhoods. First, the lender's market share of applications from minority neighborhoods may be significantly lower than its market share of applications from white neighborhoods. Second, the lender's approval rate for applications from minority neighborhoods may be significantly lower than the market's approval rate for such applications -- i.e. the approval rate for all home purchase loan applications received by all lenders from minority neighborhoods in the metro area.

   Table 3 examines the relative importance of these two factors in contributing to the worst case lending patterns in the 16 metro areas included in this study. Table 3 provides HMDA statistics on home purchase loan application activity and approval rates for one worst case lender from each of the 16 metro areas. In constructing Table 3, the Banking Research Project selected from each metro area the worst case lender with the highest or second highest total number of home purchase loan originations in 1991. The worst case lender with the highest loan origination total was chosen, unless that lender showed a greater level of activity in minority neighborhoods than the lender with the second highest total, in which case the second highest lender was chosen.

   Table 3 indicates that the selected worst case lenders had a much higher market share of total home purchase loan applications in white neighborhoods than in minority neighborhoods. For most of the selected worst case lenders, the disparity in their market share of home purchase loan applications between minority and white neighborhoods was parallel to the disparity in their market share of home purchase loan originations between minority and white neighborhoods. Table 3 also shows that most of the selected worst case lenders had approval rates for home purchase loan applications from minority neighborhoods that were roughly comparable to the approval rates for all applications received by all lenders from such neighborhoods -- i.e., the market approval rate in minority neighborhoods.[14] Taken together, the application market share data and the approval rate data presented in Table 3 provide convincing evidence that it is a lack of applications from minority neighborhoods, rather than a low loan approval rate, that is the immediate cause of most of the worst case lending patterns.

5. Presumption of no marketing effort.

   The lending pattern maps are important not just for what they reveal about lending activity (or the lack thereof) in minority neighborhoods. The maps also reveal that the effective lending territories of major lenders do not arise spontaneously. Rather, they are actively shaped by the marketing strategies of lending institutions. Most major lenders do not wait passively for customers to walk into their offices and request loan application forms. Instead, they actively initiate specific marketing strategies that target certain types of customers, often upscale persons, and particular geographic areas. The lending patterns that emerge are thus the end result of a series of choices by mortgage lenders, such as where to locate retail offices; who to hire as agents to solicit mortgage loan applications; which real estate brokers and mortgage brokers to cultivate for business relationships; and what advertising tactics to adopt. This holds true for most major lenders, not just the worst case lenders.

   As shown in Table 3, among most worst case lenders mapped by this study, the problem is not unusually high loan denial rates for applications received from minority neighborhoods. It is that so few applications are received from these areas in the first place. The most compelling explanation for this fact is that lenders have excluded minority neighborhoods from their marketing efforts. This conclusion is buttressed by a key finding of this study -- that lenders do, in fact, exercise significant control over the shape of their effective lending territories.

   The exclusion of minority neighborhoods from a lender's marketing efforts could occur in a variety of direct or indirect ways. A lender might, for example, cultivate working relationships with real estate brokers who serve white neighborhoods, but fail to build ties with real estate brokers who primarily serve minority neighborhoods. Or, in hiring real estate appraisers, a lender might decline to hire appraisers who have experience in appraising properties in minority neighborhoods. Both of these practices were cited by the Justice Department in its claim against Decatur S&L for discriminatory marketing practices.[15]

   Other explanations or presumptions are possible, but they also raise serious discrimination issues. For example, a scarcity of applications from minority areas may mean that lenders use pre-screening tactics to discourage potential applicants. Or, potential applicants may be steered away from the lender by real estate brokers because the lender employs extremely conservative lending criteria geared to upscale borrowers -- a lending policy that is subject to challenge under the "disparate impact" standard in civil rights law.

   Another explanation for the lack of applications from minority neighborhoods is that there is little or no demand for home purchase loans. This "insufficient demand hypothesis" has a number of formulations:

   1. "The volume of home purchase loan activity in minority neighborhoods is simply too small to generate more than a handful of applications."

      As discussed in Section F below, HMDA data demonstrates that there is an active home purchase loan market throughout most of the minority community. Moreover, as pointed out in Section G below, the major mortgage lenders who have made serious efforts to market home purchase loans in minority neighborhoods have found a strong level of demand.

   2. "Mortgage borrowers from minority neighborhoods usually seek types of financing that are not provided by major lenders."

      As noted in Section G below, major mortgage lenders are able to use a wide variety of mortgage instruments, including conventional loans, to serve the home purchase financing needs of minority neighborhoods.

   3. "Home buyers in minority neighborhoods are often reluctant to apply to major mortgage lenders for mortgage loans."

      In selecting a lender, most home buyers are heavily influenced by the recommendations of real estate brokers and other real estate professionals. In this setting, the marketing practices of mortgage lenders, including their relationships with real estate brokers, have a greater impact on a home buyer's choice of a mortgage lender than does the home buyer's own shopping behavior.

   Considering all the factors, the most convincing explanation for worst case lending patterns that stem from a lack of applications is that little effort has been made to market mortgage loans in minority neighborhoods. Thus, these lending patterns represent prima facie evidence that minority neighborhoods have been improperly excluded from a mortgage lender's marketing efforts.

6. Substantial mortgage loan market in minority neighborhoods.

   Lenders also contend that while they do, in fact, pursue even-handed marketing policies in minority as well as white neighborhoods, they receive very few mortgage loan applications from minority neighborhoods because the mortgage market is so thin in these areas. To examine this issue, the Banking Research Project used 1990 HMDA data to compare the overall volume of home purchase loan activity in minority neighborhoods versus white neighborhoods for the 16 metro areas covered by this study.

   Table 4 shows the average number of home purchase loan originations per 1,000 1-4 family housing structures and condominium units for four different neighborhood categories: (a) high concentration minority neighborhoods -- census tract population greater than 75 percent minority; (b) minority neighborhoods more broadly defined -- census tract population greater than 50 percent minority; (c) white neighborhoods -- census tract population greater than 75 percent white; and (d) white neighborhoods that are moderate income neighborhoods. In terms of neighborhood income levels, minority neighborhoods are more similar to moderate income, white neighborhoods than to white neighborhoods in general. Table 4 indicates that for the 16 metro areas, the flow of home purchase loans in minority neighborhoods broadly defined was on average (unweighted) 54 percent of the flow in moderate income white neighborhoods and 44 percent of the flow in white neighborhoods in general.

   Although the flow of home purchase loans in minority neighborhoods on the whole is significantly less than the flow in white neighborhoods, the data demonstrate that there is, nonetheless, an active home mortgage loan market throughout most of the minority community. This suggests that the dearth of mortgage loan applications received by worst case lenders from minority neighborhoods cannot be explained by the hypothesis that there is virtually no mortgage loan market in the minority community. Moreover, it must be emphasized once again that the market share concept used in the lending pattern maps adjusts each lender's loan activity to reflect the size of the mortgage market at the neighborhood level.

7. Major lenders can serve home financing needs of minority neighborhoods.

   Lenders often contend that the type of mortgage loans frequently sought by applicants from minority neighborhoods represent financing needs that are not readily served by major lending institutions. In order to explore this issue, the Banking Research Project examined the HMDA database for the 16 metro areas included in the study to determine whether there were major mortgage lenders that actively served the minority community. In many of these metro areas, there were, indeed, major lenders affirmatively serving minority neighborhoods -- especially large depository institutions operating in their home markets or the closely related mortgage affiliates of such institutions. The Banking Research Project mapped the lending patterns for several of these "affirmative" lenders.

   The lending pattern map for Citibank in the New York City metro area (Map 35) indicates that Citibank is a very active mortgage lender in the great majority of the metro area's minority neighborhoods, especially in Brooklyn, Queens, the Bronx, and Westchester County. The lending pattern map for Great Western Savings Bank in the Los Angeles metro area (Map 36) shows a comparable pattern for Great Western in a majority of the minority neighborhoods of Los Angeles (Map 36). Similarly, the lending pattern map for Gulf States Mortgage Company in the Atlanta metro (Map 37) area indicates that Gulf States is an active mortgage major lender in the majority of Atlanta's minority neighborhoods.

   Additional examples of major lenders that affirmatively serve the minority neighborhoods of their metro areas -- although not quite as dramatically as Citibank in New York and Great Western in Los Angeles -- are Talman Home Mortgage Corporation in Chicago, Citibank Federal Savings Bank in Chicago, Maryland National Mortgage Corporation in Baltimore, and Marathon Mortgage Corporation in Detroit. These affirmative lenders, many of which are large depository institutions or mortgage affiliates that work closely with such institutions, demonstrate convincingly that the home purchase financing needs of minority neighborhoods can be actively served by major lenders.

   The lending records of Citibank in New York City and Great Western in Los Angeles provide an especially interesting perspective on the potential role of major lenders in the minority community. Many mortgage lenders assume that loans made in minority neighborhoods should be insured by the federal government under the FHA loan insurance or the VA loan guarantee programs. Major lenders who are not actively involved in FHA and VA lending may use this assumption as a rationale for not marketing in minority neighborhoods. Yet, 100 percent of the 2,386 home purchase loans originated by Citibank in the minority neighborhoods of the New York metro area during 1990-91 were conventional loans, not government-insured loans. Similarly, 100 percent of the 1490 home purchase loans originated by Great Western in the Los Angeles metro area during 1991 were conventional loans. This shows rather dramatically that conventional financing can play a major role in minority neighborhoods.

8. Injury to minority persons and minority neighborhoods.

   Some lenders may be tempted to claim that their failure to extend their marketing activities into the minority community does not produce the concrete injury that occurs when loan applications are denied or aggressive prescreening discourages potential applicants. Yet in broad economic terms, the refusal of some lenders to trade in the minority community reduces competition and lending options in these neighborhoods. This results in higher prices for mortgage borrowers and narrower options than are available in the white community. Considerable anecdotal evidence indicates that borrowers in minority neighborhoods often pay a premium for loan products. Such premiums are often charged by lenders who specialize in serving the minority community. Large lenders, by contrast, who must price their mortgage loan products in the competitive mainstream mortgage market, generally charge lower rates and impose fewer points on mortgage loan products than many lenders who target minority neighborhoods. The more large mortgage lenders exclude minority neighborhoods from their marketing areas, the greater the market opportunity for specialized lenders to extract a premium from borrowers in these neighborhoods.

9. Exclusionary lending criteria contribute to worst case lending patterns.
   1. Impact of exculsionary lending criteria.

      A number of the worst case lenders appear to have restrictive lending criteria. Even if such criteria are applied even-handedly to both white and minority loan applicants, they will have a disproportionate adverse impact on applicants who seek home financing in minority neighborhoods. Such disparate impacts can arise as the result of systematic differences in applicant income levels or other credit-related characteristics between white and minority neighborhoods or differences in housing stock characteristics between white and minority neighborhoods.

      Restrictive lending policies or practices can have such a disparate impact on minorities. These policies and practices include:

      1. Minimum loan amount thresholds that are set above the comparatively low home purchase loan amounts often sought in minority neighborhoods;

      2. High down-payment requirements;

      3. Very low ceilings on monthly-payment-to-income ratios;

      4. Refusal to engage in FHA insured lending or to develop flexible conventional financing alternatives;

      5. Refusal to lend on 2-4 family structures in metro areas where minorities are disproportionately dependent on this type of housing structures; and

      6. Use of appraisal standards that tend to undervalue 1-4 family structures in minority neighborhoods. (This last example may involve direct discrimination as well as discrimination via disparate impact.)

      Restrictive lending criteria and practices, such as those listed above, are likely to dramatically shrink the pool of qualified loan applicants in most minority neighborhoods. They also deter the submission of loan applications or referrals from real estate brokers in such neighborhoods. From a major lender's point of view, such restrictive lending criteria may limit the anticipated number of loan customers from minority neighborhoods to such an extent that there appears to be little business justification for marketing loans in these areas. However, such conservative lending criteria, in the context of the worst case lending patterns mapped for this study, are clearly subject to challenge under the "disparate impact" or "effects test" concept of lending discrimination.

      The HMDA database itself provides interesting evidence on the adverse impact that conservative lending criteria can have on the flow of home purchase loans to minority neighborhoods. The HMDA data files created by the Banking Research Project reveal the percentage of each lender's home purchase loan originations within a metro area that have been made to low and moderate income borrowers, and also a percentage of such home purchase loans originated made by all lenders within the metro area.[16] Quite often, the major home purchase loan originators who ranked lowest in terms of the percentage of their loans made to low and moderate income borrowers were also identified as worst case lenders.

      For example, in the New York City metro area, 5.08 percent of all home purchase loan originations were made to low and moderate income borrowers. Among the top 10 mortgage lenders in the metro area, the three lenders with the lowest percentage of loan originations to low and moderate income persons were Prudential Home Mortgage Company at 0.57 percent, Chase Home Mortgage Corporation at 1.03 percent, and Chemical Bank at 1.09 percent. All three were responsible for worst case lending patterns.

      Similarly, in Washington D.C., Chicago, and Los Angeles, the major home purchase loan originators with the lowest percentage of their home purchase loan originations going to low and moderate income borrowers were also identified as worst case lenders. These lenders were Nationsbank Mortgage Corporation and B.F. Saul Mortgage Company in the Washington D.C. metro area; NBD Mortgage Company in the Chicago metro area; and Citibank Federal Savings Bank in the Los Angeles metro area.

   2. Fair Lending law incorporates the "effects test" concept of discrimination.

      The Fair Housing Act and the Equal Credit Opportunity Act do not just apply to credit policies and practices that involve disparate treatment of minorities (intentional discrimination). They also prohibit policies and practices which, although neutral on their face, have a "discriminatory effect" or "disparate impact" on minorities. Under this discriminatory effect or disparate impact approach to mortgage lending discrimination, a showing that credit policies and practices have a disparate impact on minorities can be used to establish a prima facie case of unlawful discrimination. Once a prima facie has been established, the burden shifts to the defendant to justify the challenged policies and practices as a business necessity.[17]

      The application of the "effects test" concept of discrimination to Fair Lending law was made clear by the U.S. Senate Banking Committee in its report accompany the amendment of the Equal Credit Opportunity Act in 1976. The Committee explicitly states that Congress intended to incorporate within the legislative ban on credit discrimination the "effects test" concept developed by the U.S. Supreme Court in the employment discrimination area.

      The prohibitions against discrimination on the basis of race, color, religion or national origin are unqualified. In determining the existence of discrimination on these grounds, as well as on the other grounds discussed below, courts or agencies are free to look at the effects of a creditor's practices as well as the creditor's motives or conduct in individual transactions. Thus judicial constructions of anti-discrimination legislation in the employment field, in cases such as Griggs v. Duke Power Company, 401 U.S. 424 (1971), and Albemarle Paper Company v. Moody, (U.S. Supreme Court, June 25, 1975), are intended to serve as guides in the application of this Act, especially with respect to the allocations of burdens of proof.[18]

      Similarly, the Fair Housing Act has been widely interpreted by the federal courts to prohibit housing practices that have a disparate impact on minorities unless the defendant can demonstrate a business necessity for such policies.[19] According to Professor Robert G. Schwemm, a leading authority on the Fair Housing Act:

      By 1988, therefore, a strong consensus had developed among the circuits that the proper meaning of Title VIII included a discriminatory effect standard. Only the First, Tenth, and D.C. Circuits have not been heard from on this issue. Not a single court of appeals currently espouses the view that the effect theory is inappropriate for Title VIII cases.[20]

      Thus, the effects test concept provides an important basis for challenging marketing policies of mortgage lender's that focus exclusively on upscale neighborhoods. Where such policies have the effect of excluding minority neighborhoods -- as they often do -- they can be challenged under credit discrimination law. And, the exclusion of minority neighborhoods from a lender's marketing area can also be viewed as a form of direct or intentional discrimination.

10. Consolidation of home purchase loans made by affiliates.

    Under HMDA reporting rules, corporations that engage in mortgage lending in a metro area through two or more different subsidiaries report separate HMDA data for each subsidiary. In some instances, banking organizations that operate through multiple subsidiaries will have as many as three or four different HMDA reporters within the same metro area. An important regulatory issue is whether, for Fair Lending enforcement purposes, the HMDA data reported by a corporation's various subsidiary lenders for the same metro area should be examined separately for each subsidiary or consolidated for all subsidiaries.

    This question carries great significance in cases where there is a major divergence among a corporation's subsidiaries in their treatment of minority neighborhoods or minority loan applicants. For example, in the New York City metro area, Chase Home Mortgage Corporation (Map 31) made comparatively few home purchase loan originations in the metro area's extensive minority neighborhoods. On the other hand, beginning in 1991, Chase Manhattan Bank -- an affiliate of Chase Home Mortgage Corporation -- became an active lender in some of the minority neighborhoods of Brooklyn (Map 32). Similarly, Nationsbank Mortgage Corporation engages in very little home purchase loan origination activity in the minority neighborhoods of the Washington DC metro area (Map 33). Yet, its affiliate, Nationsbank DC, was active as a mortgage lender in a number of the minority census tracts of the District of Columbia during 1991 (Map 34).

    In some banking organizations, the mortgage lending activity of a subsidiary mortgage company may be operationally entwined with that of a subsidiary bank. In such cases, it might be appropriate to consolidate the HMDA data of the two lenders for Fair Lending enforcement purposes. However, many large bank-related mortgage companies operate on a nationwide or regional basis, and are active mortgage lenders in a number of metro areas in which their bank affiliates are not present. Where such mortgage companies conduct their mortgage lending operations independent of the activities of their bank affiliates, or on a geographic scope that extends beyond the deposit base of their bank affiliates, they should be judged on their own behavior for Fair Lending enforcement purposes. When they pursue policies that exclude minority neighborhoods, they should not be allowed to obscure this pattern by consolidating their HMDA data with that of their bank affiliates. If their marketing or lending policies are discriminatory, these policies should be subject to direct enforcement action without regard to the policies of an affiliated bank.

    Allowing the mortgage lending activities of bank affiliates to "offset" discriminatory lending patterns by large bank-related mortgage companies will create confusion concerning Fair Lending compliance requirements. It would also send a very mixed signal to large-scale mortgage originators by implying that there is a double Fair Lending compliance standard: a lower standard in metro areas where discriminatory lending patterns can be offset by that of any bank affiliates, and a higher standard in metro areas where no such offset is possible. Where these mortgage companies operate with highly centralized lending policies and procedures -- as they often do -- such a double standard would be operationally difficult to implement. Under any circumstances, it would create confusion concerning the requirements of Fair Lending compliance.

11. Consolidation of loan originations and loan purchases.

    Many mortgage lenders purchase home mortgage loans originated by other lenders as well as originating their own mortgage loans. In evaluating Fair Lending compliance, there are serious questions about consolidating a lender's loan origination and loan purchases into one category.

    To determine whether any of the 62 worst case lenders were, in fact, purchasing a substantial number of home purchase loans secured by properties located in minority neighborhoods -- thereby masking failures to originate their own loans to such neighborhoods -- the Banking Research Project reviewed HMDA data files on both purchased loans and loan originations for the 16 metro areas.

    The results of this review (presented in Table 5) indicate that in 29 of the 62 worst case lending patterns, lenders were not purchasing home purchase loans in the metro area, or were purchasing only an inconsequential number of such loans. Specifically, in 25 cases no loans were purchased, while in 4 cases less than 3 loans were purchased. Among the 33 cases where the worst case lender was purchasing 3 or more home purchase loans, it was unusual for such purchases to include loans from minority neighborhoods in sufficient number to substantially offset the adverse pattern of the lender's home purchase loan origination activity. In only 5 instances did consolidation of the lender's loan purchases and loan originations raise the percentage of the lender's loans going to minority neighborhoods by more than 1 percentage point.

    Regardless of these statistics, there are good reasons, from a Fair Lending enforcement perspective, why purchased loans should not be consolidated with loan originations in evaluating the lending patterns of individual lenders -- or at least good reasons to establish a presumption against such consolidation. Mortgage loans purchased by a lender may have different rates, fees, and terms than loans originated by the lender. Where purchased loans have costs that are higher and terms more disadvantageous to the borrower, Fair Lending enforcement would be compromised if data on purchased loans and loan origination were consolidated. Giving mortgage lenders equal credit for purchased loans and loan origination -- despite possible disparities of rates, fees and terms of loans -- would perversely work to perpetuate the dual housing finance market: a low-cost market in white neighborhoods and a high-cost market in minority neighborhoods.

    In fact, the very existence of a dual housing finance market constitutes prima facie evidence of a "structural" violation of the Fair lending laws. The Fair Housing Act prohibits mortgage lenders from charging higher prices or imposing more onerous loan terms based on the racial composition of the neighborhood. When a mortgage borrower in a minority neighborhood pays more for mortgage credit or receives more onerous terms than mortgage borrowers in white neighborhoods, there should be a presumption of unlawful discrimination, unless it can be demonstrated that the higher mortgage costs are warranted by higher default risk or substandard condition of the dwelling.

12. Mortgage companies dominate the list of worst case lenders.

    As indicated above, mortgage lenders reporting under HMDA can be grouped into seven general categories of mortgage lenders:

    1. Independent mortgage companies, such as Sears and Prudential, which are not controlled by a depository institution or a bank holding company;

    2. Mortgage companies that are non-bank subsidiaries of bank holding companies, such as Chase Home Mortgage Corporation;

    3. Mortgage companies that are direct subsidiaries of commercial banks;

    4. Mortgage companies that are direct subsidiary of savings institutions;

    5. Commercial banks;

    6. Savings institutions; and

    7. Credit unions.

    Table 6 shows the distribution, by type of lender, of the 62 worst case lending patterns and the 49 lenders responsible for these patterns.

    Independent mortgage companies accounted for 23 of the 62 worst case lending patterns (37 percent). Many of the 13 independent mortgage companies responsible for these 23 worst case lending patterns are very large mortgage originators active in a great number of metro areas across the nation. In the metro areas where they were mapped, they were among the largest home purchase loan originators. They are also active in many other metro areas where they operate as middle-tier lenders, often ranking somewhere between 20th and 50th in terms of home purchase loan originations. Quite often as middle-tier lenders, they exhibit the same focus on upscale, white neighborhoods and avoidance of minority neighborhoods that is found in more dramatic fashion in their lending patterns in the metro areas where they were mapped.

    Mortgage company subsidiaries of depository institutions and of bank holding companies -- commonly referred to as bank-related mortgage companies -- accounted for 17 of the worst case lending patterns (27 percent). Among this group, mortgage company subsidiaries of bank holding companies accounted for 7 worst case lending patterns; mortgage company subsidiaries of savings institutions accounted for 6 such patterns; and mortgage company subsidiaries of commercial banks accounted for 4 such patterns.

    Savings institutions accounted for 14 worst case lending patterns (23 percent). Many of the savings institutions responsible for worst case lending patterns are locally-based depository institutions that have slowly built up a large share of the mortgage market within their home metro area. Unlike the independent mortgage companies, the majority of these savings institutions are not major mortgage lenders in a wide range of different metro areas.

    Commercial banks in their direct mortgage lending activities accounted for only 8 of the worst case lending patterns (13 percent). To some extent, this may reflect the fact that an increasing number of commercial banks are shifting their mortgage lending activities to mortgage company affiliates or mortgage company subsidiaries.