The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
December 15,1989 eCONOMIC COMMeNTORY Federal Reserve Bank of Cleveland Making Judgments About Mortgage Lending Patterns by Robert B. Avery In the mid-1970s, there was considerable public debate about whether financial institutions engaged in redliningthe practice of refusing to lend to, or limiting the number of home mortgage loans made in, poor or predominantly minority neighborhoods. Congressional concerns about such practices were a major factor in the passage of the Home Mortgage Disclosure Act (HMDA) in 1975 and the Community Reinvestment Act (CRA) in 1977. Concerns about redlining have resurfaced in the past few years as a result of increased CRA-related protests of bank applications and several investigations of alleged discrimination in mortgage lending patterns in several cities. 1 Partly as a result of these concerns, Congress recently enacted a law that substantially expands several features of the HMDA and CRA.2 The HMDA requires financial institutions to publicly disclose the number and dollar amount of their mortgage loans by geographic location. Due to recent changes in the law, institutions will be required to report the distribution of their mortgage applicants by sex, race, and income within each geographic area they serve. HMDA reporting requirements will also be extended to previously uncovered mortgage companies. The CRA establishes regulatory processes to encourage institutions to meet ISSN 0428·1276 the credit needs of their entire community, including low- and moderateincome areas, in a manner consistent with safe and sound practices. Evaluation of an institution's record in meeting such needs is a routine part of the bank examination process and is used as a factor in deciding the merits of applications for bank charters, deposit insurance, mergers, branches, and for other regulatory actions. CRA compliance ratings and a summary of the compliance examination now will be made public for all covered institutions. In addition, federal regulators issued a joint CRA statement on March 21, 1989, encouraging financial institutions to expand their management of, and attention to, CRA responsibilities, including procedures for information dissemination and recordkeeping.f Before any further extensions to the HMDA and CRA are undertaken, it would be desirable to evaluate the existing statistical evidence on redlining and to determine whether the weight of evidence is strong enough to reach definitive conclusions. This paper evaluates the general methodology used in some of the statistical studies. These research methods, which are employed to examine overall lending patterns across neighborhoods, are then compared with compliance procedures followed by regulators concerned with the performance of individual financial institutions. - Studies examining whether mortgage lenders discriminate against borrowers in minority and lower-income areas have traditionally analyzed the relationship between aggregate annual mortgage lending within a neighborhood and the neighborhood's characteristics. Regulatory-agency compliance examiners make judgments about the mortgage lending procedures adopted by individual lenders. The differences in these two methods of evaluation are not easily reconciled. • A History of Redlining Concerns The renewed interest in red lining and CRA-related issues follows a considerable lull in regulatory activity in the early I980s. For example, only one of the 2,337 consumer complaints received by the Federal Reserve Board in 1984 was CRA-related. Similarly, only three requests for regulatory action by the Federal Reserve were protested on CRA grounds in 1984, and more than 98 percent of the institutions examined by the The value of these studies depends criti- When studies show that lending pat- This finding seems to be constant both for the country as a whole. To address acceptance of mortgage-participation securities led to a substantial rise in FHA-insured and VA-guaranteed loans cally on the selection of control variables. If other relevant factors included in the analysis adequately control for mortgage demand and appropriate lend- terns differ among neighborhoods after demand and risk factors are accounted for, inappropriate lender actions are likely to be responsible. However, over time and across cities. For example, a study of mortgage lending in Cleveland between 1977 and 1979 concluded that commercial banks and this concern, a study was conducted at the Federal Reserve Board using HMDA data drawn from the entire country.' This study was designed as a and to an increased role of FNMA and FHLMC as guarantors of conventional mortgage loans4 The use of variablerate mortgages and private mortgage in- ing risk, for example, then it can be argued that the residual relationship between race and lending reflects the inappropriate actions of lenders. If some controversy exists about the interpretation of lending risk variables in these studies. It is clear that lenders have a right to be concerned about the S&Ls were considerably less likely to be the source of mortgage financing in integrated and all-minority neighborhoods than in predominantly white prototype of the kind of systematic, large-scale evaluation of CRA-related mortgage lending patterns that would be feasible with the HMDA data avail- surance also became widespread. The net effect of these changes has been to standardize the mortgage application process and to make it much easier for other relevant factors are not adequately controlled for, then it cannot be argued that the residual relationship reflects purely supply factors. This is a potentially critical weakness of virtually all redlining research to date, since adequately capturing the demand for credit is extremely difficult. credit quality of each loan. It is also clear that some lenders believe property values in some neighborhoods are less stable than in others, and therefore may have used this information in making lending decisions. What is not clear, however, is exactly how or whether lenders can legally incorporate areas with otherwise similar characteristics." However, the study also found that minority and integrated areas were considerably more likely to able at that time. such neighborhood factors into their lending decisions. with the net flow of housing-related financing to minority and all-white neighborhoods being about the same. Changes were particularly apparent in home mortgage lending. The growing Federal Reserve in that year received satisfactory performance ratings for CRA compliance. The records of other CRA regulators are comparable to those of the Federal Reserve. firms that did not want to hold mortgages for their own portfolios to specialize in originating mortgages. Casual evidence suggests that mortgage markets Several factors could explain this rela- have become more competitive as the number of players has increased. tively tranquil period. Community development corporations and other similar organizations were established in the early 1980s with the intent of providing financing and banking assistance to special areas of need within low- and moderate-income neighborhoods. It is also clear that the CRA has had a significant educational effect. More banks now appear to have a greater awareness of minority and lowincome business opportunities and are better equipped to seek this business through enhanced advertising and marketing programs. Despite these apparent signs of success, it is difficult to tell if the redlining issue has been resolved. A large number of dramatic changes took place in the structure of the consumer lending market between 1975 and 1989 that make it very difficult to isolate the effects of the CRA. Many of the depository functions of banking institutions were deregulated over this period, and branching restrictions were substantially eased. Savings and loan associations (S&Ls) were given expanded lending powers, usury laws were lifted, and nonbank lenders, such as General Motors and General Electric, became major players in the consumer market. • Statistical Evidence Most of the existing empirical research on redlining has focused on the analysis of geographic patterns of mortgage lending.i' The most common type of study has used HMDA data (or publ ic deed-title-transfer data) combined with census information to examine the relationship between aggregate annual mortgage lending within a neighborhood and the neighborhood's characteristics. With few exceptions, these studies have not used information about individual mortgage applicants or lenders. The prototypical aggregate lending study examines whether, after controlling for other relevant factors, the racial (or income) characteristics of a neighborhood are related to the amount and type of mortgage lending. Typically, only lending by banks and S&Ls required to report under the HMDA has been included, although some studies have supplemented this information with FHA and VA loans originated by mortgage bankers. In virtually all of the studies, data have been aggregated across classes of institutions. Thus, inferences can be drawn only about the behavior of institutions as a whole, not about individual lenders. be served by mortgage bankers offering FHA and VA loans and by home improvement loans from all lenders. Overall, these factors offset one another, Data for a representative year (1981) were gathered for the 318 standard metropolitan statistical areas (SMSAs) for which financial institutions were required to report their lending activity. This sample was reduced to 100 by removing all smaller SMSAs and those The types of control variables used in HMDA-based studies have varied considerably. In some studies, only the number of residents or the number of What do these studies reveal? Almost uniformly, studies with little or no con- A recent study of mortgage lending in with insufficient minority population to make a redlining study feasible. HMDA data on mortgage and home improvement loans were summed over owner-occupied houses in a neighborhood has been used as a control for mortgage demand. Yet, household mobility and property turnover differ trol for demand factors have found that predominantly white and high-income neighborhoods receive more mortgage loans per home than minority and Atlanta by the Atlanta Constitution reported similar findings.' Using a matched sample of middle-income white and minority neighborhoods, the author reporting institutions and arrayed by census tract within each SMSA9 Data represented more than 3,000 financial institutions, approximately two-thirds considerably across neighborhoods and over time. For example, the amount of new construction, the age of the housing stock, property tax rates, con- lower-income neighborhoods. concluded that banks and S&Ls ex- of which were commercial banks, supplemented with data gathered from the 1980 census. dominium conversion rates, and the size and value of typical homes are all likely to vary with income and to affect mobility and mortgage demand. Attempting to account for these factors, the more thorough studies have used information on the number of housing transfers within a neighborhood. Many HMDA-based studies have also tried to control for various lending risks that may vary geographically. These risks are extremely hard to measure, and it is by no means clear how they should be regarded. Research studies typically have used the condition of the housing stock, foreclosure rates, and vacancy rates as representative of neighborhood risk to help explain lending patterns across neighborhoods. The uniformity of these conclusions starts to break down, however, when more complex control variables are introduced or when the activity of different types of lenders is examined. The gap between lending patterns in minority and white neighborhoods narrows considerably when controls for income and property transfers are introduced, although it does not always disappear entirely. However, even when overall differences in lending activity disappear after demand factors are controlled for, studies have generally found persistent differences in the type of lender servicing the neighborhoods. tended four to five times as many new mortgage loans per single-family housing unit to the predominantly white areas than to comparable minority areas. However, this disparity is cut in half when property transfers are controlled for. Although they did not perform a systematic analysis of lending patterns by nonbank lenders, the Constitution reports evidence that mortgage bankers are considerably more active in minority areas than in white neighborhoods, tending to corroborate the results found for Cleveland. • Systematic Evidence The above studies were virtually all ad hoc, initiated by individual researchers or community groups to investigate problems within a particular city. Although some evidence appears to have emerged about these individual cities, the studies fail to provide an evaluation Separate analyses were performed for each SMSA in the study. The number and dollar volume of mortgage and home improvement loans allocated to census tracts containing less than 10 percent, 10 to 40 percent, 40 to 80 percent, and more than 80 percent minority populations were compared. These comparisons were made controlling for various neighborhood characteristics represented by the census data. • A History of Redlining Concerns The renewed interest in red lining and CRA-related issues follows a considerable lull in regulatory activity in the early I980s. For example, only one of the 2,337 consumer complaints received by the Federal Reserve Board in 1984 was CRA-related. Similarly, only three requests for regulatory action by the Federal Reserve were protested on CRA grounds in 1984, and more than 98 percent of the institutions examined by the The value of these studies depends criti- When studies show that lending pat- This finding seems to be constant both for the country as a whole. To address acceptance of mortgage-participation securities led to a substantial rise in FHA-insured and VA-guaranteed loans cally on the selection of control variables. If other relevant factors included in the analysis adequately control for mortgage demand and appropriate lend- terns differ among neighborhoods after demand and risk factors are accounted for, inappropriate lender actions are likely to be responsible. However, over time and across cities. For example, a study of mortgage lending in Cleveland between 1977 and 1979 concluded that commercial banks and this concern, a study was conducted at the Federal Reserve Board using HMDA data drawn from the entire country.' This study was designed as a and to an increased role of FNMA and FHLMC as guarantors of conventional mortgage loans4 The use of variablerate mortgages and private mortgage in- ing risk, for example, then it can be argued that the residual relationship between race and lending reflects the inappropriate actions of lenders. If some controversy exists about the interpretation of lending risk variables in these studies. It is clear that lenders have a right to be concerned about the S&Ls were considerably less likely to be the source of mortgage financing in integrated and all-minority neighborhoods than in predominantly white prototype of the kind of systematic, large-scale evaluation of CRA-related mortgage lending patterns that would be feasible with the HMDA data avail- surance also became widespread. The net effect of these changes has been to standardize the mortgage application process and to make it much easier for other relevant factors are not adequately controlled for, then it cannot be argued that the residual relationship reflects purely supply factors. This is a potentially critical weakness of virtually all redlining research to date, since adequately capturing the demand for credit is extremely difficult. credit quality of each loan. It is also clear that some lenders believe property values in some neighborhoods are less stable than in others, and therefore may have used this information in making lending decisions. What is not clear, however, is exactly how or whether lenders can legally incorporate areas with otherwise similar characteristics." However, the study also found that minority and integrated areas were considerably more likely to able at that time. such neighborhood factors into their lending decisions. with the net flow of housing-related financing to minority and all-white neighborhoods being about the same. Changes were particularly apparent in home mortgage lending. The growing Federal Reserve in that year received satisfactory performance ratings for CRA compliance. The records of other CRA regulators are comparable to those of the Federal Reserve. firms that did not want to hold mortgages for their own portfolios to specialize in originating mortgages. Casual evidence suggests that mortgage markets Several factors could explain this rela- have become more competitive as the number of players has increased. tively tranquil period. Community development corporations and other similar organizations were established in the early 1980s with the intent of providing financing and banking assistance to special areas of need within low- and moderate-income neighborhoods. It is also clear that the CRA has had a significant educational effect. More banks now appear to have a greater awareness of minority and lowincome business opportunities and are better equipped to seek this business through enhanced advertising and marketing programs. Despite these apparent signs of success, it is difficult to tell if the redlining issue has been resolved. A large number of dramatic changes took place in the structure of the consumer lending market between 1975 and 1989 that make it very difficult to isolate the effects of the CRA. Many of the depository functions of banking institutions were deregulated over this period, and branching restrictions were substantially eased. Savings and loan associations (S&Ls) were given expanded lending powers, usury laws were lifted, and nonbank lenders, such as General Motors and General Electric, became major players in the consumer market. • Statistical Evidence Most of the existing empirical research on redlining has focused on the analysis of geographic patterns of mortgage lending.i' The most common type of study has used HMDA data (or publ ic deed-title-transfer data) combined with census information to examine the relationship between aggregate annual mortgage lending within a neighborhood and the neighborhood's characteristics. With few exceptions, these studies have not used information about individual mortgage applicants or lenders. The prototypical aggregate lending study examines whether, after controlling for other relevant factors, the racial (or income) characteristics of a neighborhood are related to the amount and type of mortgage lending. Typically, only lending by banks and S&Ls required to report under the HMDA has been included, although some studies have supplemented this information with FHA and VA loans originated by mortgage bankers. In virtually all of the studies, data have been aggregated across classes of institutions. Thus, inferences can be drawn only about the behavior of institutions as a whole, not about individual lenders. be served by mortgage bankers offering FHA and VA loans and by home improvement loans from all lenders. Overall, these factors offset one another, Data for a representative year (1981) were gathered for the 318 standard metropolitan statistical areas (SMSAs) for which financial institutions were required to report their lending activity. This sample was reduced to 100 by removing all smaller SMSAs and those The types of control variables used in HMDA-based studies have varied considerably. In some studies, only the number of residents or the number of What do these studies reveal? Almost uniformly, studies with little or no con- A recent study of mortgage lending in with insufficient minority population to make a redlining study feasible. HMDA data on mortgage and home improvement loans were summed over owner-occupied houses in a neighborhood has been used as a control for mortgage demand. Yet, household mobility and property turnover differ trol for demand factors have found that predominantly white and high-income neighborhoods receive more mortgage loans per home than minority and Atlanta by the Atlanta Constitution reported similar findings.' Using a matched sample of middle-income white and minority neighborhoods, the author reporting institutions and arrayed by census tract within each SMSA9 Data represented more than 3,000 financial institutions, approximately two-thirds considerably across neighborhoods and over time. For example, the amount of new construction, the age of the housing stock, property tax rates, con- lower-income neighborhoods. concluded that banks and S&Ls ex- of which were commercial banks, supplemented with data gathered from the 1980 census. dominium conversion rates, and the size and value of typical homes are all likely to vary with income and to affect mobility and mortgage demand. Attempting to account for these factors, the more thorough studies have used information on the number of housing transfers within a neighborhood. Many HMDA-based studies have also tried to control for various lending risks that may vary geographically. These risks are extremely hard to measure, and it is by no means clear how they should be regarded. Research studies typically have used the condition of the housing stock, foreclosure rates, and vacancy rates as representative of neighborhood risk to help explain lending patterns across neighborhoods. The uniformity of these conclusions starts to break down, however, when more complex control variables are introduced or when the activity of different types of lenders is examined. The gap between lending patterns in minority and white neighborhoods narrows considerably when controls for income and property transfers are introduced, although it does not always disappear entirely. However, even when overall differences in lending activity disappear after demand factors are controlled for, studies have generally found persistent differences in the type of lender servicing the neighborhoods. tended four to five times as many new mortgage loans per single-family housing unit to the predominantly white areas than to comparable minority areas. However, this disparity is cut in half when property transfers are controlled for. Although they did not perform a systematic analysis of lending patterns by nonbank lenders, the Constitution reports evidence that mortgage bankers are considerably more active in minority areas than in white neighborhoods, tending to corroborate the results found for Cleveland. • Systematic Evidence The above studies were virtually all ad hoc, initiated by individual researchers or community groups to investigate problems within a particular city. Although some evidence appears to have emerged about these individual cities, the studies fail to provide an evaluation Separate analyses were performed for each SMSA in the study. The number and dollar volume of mortgage and home improvement loans allocated to census tracts containing less than 10 percent, 10 to 40 percent, 40 to 80 percent, and more than 80 percent minority populations were compared. These comparisons were made controlling for various neighborhood characteristics represented by the census data. Control variables included the number data used for the national analysis: (I) of owner-occupied houses, the growth in the number of housing units since 1970, median household income, the median value of owner-occupied hous- there was no information on housing transfers; (2) no loan-risk or foreclosure data were used; and (3) information on most mortgage banker lend- ing, the median age of the housing stock, and the percentage of housing stock older than 40 years. Although imperfect, these variables were included ing was not used. • Compliance The CRA directs regulators to "encourage" financial institutions to meet the credit needs of their entire community. However, the act provides no guidance as to how credit needs are to The effect of these omissions is quite apparent when comparing the conclusions of this study with those of the be determined. Moreover, only those institutions seeking regulatory actions, such as mergers, are subject to any penalties for not meeting their to attempt to control for differences in the demand for home loans across neighborhoods, as well as for the differing risks of lending in various neighborhoods. On the surface, results from this study look similar to those of earlier redlining studies. Controlling only for the number of owner-occupied housing units, census tracts with under 10percent minority population had an average of almost twice the number of Cleveland study reported earlier and with another comprehensive analysis for the Boston area.lo Compared with earlier findings that incorporated these factors, The Federal Reserve Board study would have overestimated the differential in lending by banks and S&Ls between minority and other areas. It also would have underestimated total lending in minority areas by all lenders. community's needs. These vague directives have led regulatory oversight to focus primarily on procedure. Institutions are required to state publicly how they themselves define their community and the services they provide. Compliance examiners then determine whether an institution's procedures appear to treat all neighborhoods falling within its self- new mortgage loans per housing unit as all-minority areas. However, these results changed markedly when other variables were controlled for and when examined at the larger SMSA level. These omissions are not easily remedied. Title-transfer information is defined community on an equitable basis. Historically, examiners have not collected in a systematic way and is costly to obtain. In the Cleveland study, it was necessary to take address data from the county transfer files and paid particular attention to marketing and loan processing procedures in determining whether neighborhoods are treated equitably. Tracts with over 80-percent minority population were projected to have fewer mortgages per unit than pre- use maps and "geo-coding" programs to assign transfers to census tracts. Foreclosure data had to be put together in a similar way. In some cities, these dominantly white tracts in 65 of the 100 SMSAs when other factors were controlled for. These differences were statistically significant in only nine data may be extremely difficult to obtain in a usable format. Evidence reported by the regulatory agencies suggests that virtually all financial institutions are in compliance with current CRA procedures. On the It does not appear that the recent extension of HMDA reporting requirements surface, this seems inconsistent with the evidence found in the redlining studies that the racial (or income) composition of neighborhoods appears to neighborhoods. When dollars of mortgages per dollars of housing value were examined, all-minority areas were projected to have fewer dollars than all- by Congress will fully remedy these kinds of problems faced by researchers. Congress did extend HMDA coverage to independent mortgage be related to aggregate lender activity in some areas. Again, however, compliance is focused on the procedures of individual institutions, whereas the companies, which addresses a serious weakness in the current reporting system. II Institutions will now also be re- statistical studies focus on aggregate market conditions. white areas in 52 of the 100 SMSAs. quired to file information on the dis- How do these findings compare with the studies for individual cities? Three major weaknesses were evident in the tribution of mortgage applicants by sex, race, and income within each census tract. In addition, lenders must report whether the application was ap- cases, however, and in four SMSAs, predominantly minority areas were found to receive proportionately more mortgage credit than similar white proved. While perhaps useful in individual compliance cases, however, this additional information is unlikely to say much about neighborhood risk or demand. Some institutions could be found to have poor CRA programs where, hoods even in cases in the aggregate, all neighbor- were adequately served. On the Since the CRA provides sideration a tually all financial evaluation of lending credit needs would also of safety and soundness, fair and comprehensive neighborhood other hand, the fact that an institution have to take into account makes tached to lending. viously, lending variety of sources fewer loans in some areas will not necessarily record. imply a poor CRA If institutional all applicants procedures treat fairly, but it happens some minority ty turnovers that areas have fewer proper- or fewer qual ified appli- cants, resulting legitimate As indicated and at times can be a would not contribute By necessity, these judgments to an institution's market. is not very active borhoods. loan might not be of the CRA even such a practice fewer mortgage that in the mortgage for example, found in violation though own An institution might result in loans to some neigh- A lender could, also have different home-improvement loan record. it is entirely Thus, possible that individual institutions could all be in compliance with CRA procedures, yet have some neighborhoods ing substantially from financial receiv- fewer mortgage institutions loans than other neighborhoods. definitions. Consideration in market one believed may have to HMDA to provide agement cannot account practices focused be easily modified aggregate The information conditions. to assess how truly serves the credit needs of its community, whether one institution job than another, sive to collect. tations is difficult could guidelines, yet collec- all fol- tively be part of an environment which some neighborhood application evaluation efforts institutions is and difficult. lenders comprise involved housing studies behavior. to ensure only a subset It is clear that research should continue. wide pattem is probably thought, evidence experience loan disparity mortgage Economic Review, Federal Reserve Bank of the Federal Reserve Bank of Cleveland. The tion, op. cit. author extends special thanks to Glenn Can- Boston, September/October 1989, pp. 3-30; The of lenders-as not be suffi- that all neighborhoods comparable amounts ner and Mark Snidermanfor "Mortgage Redlining: A Multicity Cross Sec- ments and suggestions. tion Analysis," Working Paper, Board of 1984. Federal Reserve Bank of Cleveland or of the 9. The HMDA requires each institution to Board of Governors of the Federal Reserve report its total annual mortgage and home im- System. provement loans by census tract. Census 4,000 people. The 100 SMSAs used in the Federal Deposit Insurance Corporation. See study average more than 210 tracts apiece. ly homogeneous population, typically about 10. See Glenn B. Canner, "Red lining and Mortgage Lending Practices," in Research 4. These acronyms describe, respectively, in Urban Economics: A Research Annual, J. the Federal Housing Administration, the Vernon Henderson, ed., Greenwich, Conn.: JAI Press, 1981, pp. 67-10 I. 5. See Glenn B. Canner, "Redlining: Re- The views stated herein are those of the author and not necessarily those of the Federal Home Loan Bank Board, and the Home Loan Mortgage Corporation. helpful com- Governors of the Federal Reserve System, tracts are areas designed to contain a relative- al Mortgage Association, and the Federal of the Cornell University and a visiting scholar at 8. Robert B. Avery and Glenn B. Canner, 3. These regulators are the Federal Reserve Board, the Comptroller of the Currency, the the Federal Reserve Bulletin, vol. 75, May 1989, p. 351. Robert B. AvelY is an associate professor at 11. In 1981, mortgage bankers were estimated to originate an average of 29 percent of the new mortgages in the 100 SMSAs search and Federal Legislative Response," used in the study. Their share ranged from a Board of Governors of the Federal Reserve high of 68 percent to a low of 2 percent. System, Staff Study No. 121, October 1982. of lending. not as great as is commonly when important demand factors lender activity and lending is related risk to aggregate in many areas. The nor agreed are neither upon. to change through increased institution there is for this, however, well understood Attempts such that the racial composition neighborhoods reasons cient to ensure the nation- Although of mortgage on and 7. "The Color of Money," Atlanta Constitu- Veterans' Administration, the Federal Nation- even good citizenship necessary as it is-may and debate that in the determination • Conclusion issue is important be- Since regulated on the part of all regulated the redlining have The act as good citizens pattems, in Boston, 1982-1987," New England enacted on August 9, 1989. procedures, in their communities. parties Cleveland, Summer 1981. 2. The Financial Institutions Reform, Recovery, and Enforcement Act of 1989 was on in- and the CRA legislation lenders Economic Review, Federal Reserve Bank of "Geographic Patterns of Mortgage Lending Lending in C1eveland-1987," is that have focused most statistical regulated neighbor- of these data to the of individual likely to be costly circumstances hind them, are designed of aggregate across Karl E. Case, and Constance R. Dunham, May 1-4, 1988; and "Race and Mortgage in needs ap- - Robert B. Avery and Thomas M. Buynak, "Mortgage Redlining: Some New Evidence," "The Color of Money," Atlanta Constitution, for these apparently institutions' regulations, 6. Cuyahoga Plan of Ohio Inc., July 1989. dealt with aggregate data could be used demand compliance Footnotes 1. See, for example, Katharine L. Bradbury, of difficult, this outcome reliance are likely to be controversial, and expensive. The lack of a consensus causes redlining approach without on financial regulations and conditions argues should further view about the associated strongly with that such an not be undertaken study. BULK RATE U.S. Postage Paid Cleveland,OH Permit No. 385 or does a better and expen- And, regardless much information firms to take into market needed well an institution on man- in individual institutions whereas Even if available good measures risk-adjusted hoods, emphasis. and census as mortgage system that individual dividual and dif- that currently are taken into account, A compliance possible low regulatory contradictory market for example, have a very strong rules. It is entirely pear not to be met. as it is in other kinds of Institutions ferences are that loan transactions. be given to cost differentials made relative follow procedures with current The likely reason to a poor CRA rating. market, lend- in fewer acceptances, this circumstance self-defined pre- in the mortgage are in compliance • that vir- institutions and marketing regulatory the risk at- risk can arise from a factor ing decision, On the other hand, it appears for the con- is collected, of how interpre- Federal Reserve Bank of Cleveland Research Department P.O. Box 6387 Cleveland, OH 44101 of the data may still vary widely. Address Correction Requested: Please send corrected mailing label to FAULH880101000 000 001 CMTY LEE 0 FAULHABER CLEVELAND 1 the above address. RESEARCH Material may be reprinted provided that the source is credited. Please send copies of reprinted materials to the editor. 9 • tV £1 6 B 330 AH'VU81l HDB 3S3B Some institutions could be found to have poor CRA programs where, hoods even in cases in the aggregate, all neighbor- were adequately served. On the Since the CRA provides sideration a tually all financial evaluation of lending credit needs would also of safety and soundness, fair and comprehensive neighborhood other hand, the fact that an institution have to take into account makes tached to lending. viously, lending variety of sources fewer loans in some areas will not necessarily record. imply a poor CRA If institutional all applicants procedures treat fairly, but it happens some minority ty turnovers that areas have fewer proper- or fewer qual ified appli- cants, resulting legitimate As indicated and at times can be a would not contribute By necessity, these judgments to an institution's market. is not very active borhoods. loan might not be of the CRA even such a practice fewer mortgage that in the mortgage for example, found in violation though own An institution might result in loans to some neigh- A lender could, also have different home-improvement loan record. it is entirely Thus, possible that individual institutions could all be in compliance with CRA procedures, yet have some neighborhoods ing substantially from financial receiv- fewer mortgage institutions loans than other neighborhoods. definitions. Consideration in market one believed may have to HMDA to provide agement cannot account practices focused be easily modified aggregate The information conditions. to assess how truly serves the credit needs of its community, whether one institution job than another, sive to collect. tations is difficult could guidelines, yet collec- all fol- tively be part of an environment which some neighborhood application evaluation efforts institutions is and difficult. lenders comprise involved housing studies behavior. to ensure only a subset It is clear that research should continue. wide pattem is probably thought, evidence experience loan disparity mortgage Economic Review, Federal Reserve Bank of the Federal Reserve Bank of Cleveland. The tion, op. cit. author extends special thanks to Glenn Can- Boston, September/October 1989, pp. 3-30; The of lenders-as not be suffi- that all neighborhoods comparable amounts ner and Mark Snidermanfor "Mortgage Redlining: A Multicity Cross Sec- ments and suggestions. tion Analysis," Working Paper, Board of 1984. Federal Reserve Bank of Cleveland or of the 9. The HMDA requires each institution to Board of Governors of the Federal Reserve report its total annual mortgage and home im- System. provement loans by census tract. Census 4,000 people. The 100 SMSAs used in the Federal Deposit Insurance Corporation. See study average more than 210 tracts apiece. ly homogeneous population, typically about 10. See Glenn B. Canner, "Red lining and Mortgage Lending Practices," in Research 4. These acronyms describe, respectively, in Urban Economics: A Research Annual, J. the Federal Housing Administration, the Vernon Henderson, ed., Greenwich, Conn.: JAI Press, 1981, pp. 67-10 I. 5. See Glenn B. Canner, "Redlining: Re- The views stated herein are those of the author and not necessarily those of the Federal Home Loan Bank Board, and the Home Loan Mortgage Corporation. helpful com- Governors of the Federal Reserve System, tracts are areas designed to contain a relative- al Mortgage Association, and the Federal of the Cornell University and a visiting scholar at 8. Robert B. Avery and Glenn B. Canner, 3. These regulators are the Federal Reserve Board, the Comptroller of the Currency, the the Federal Reserve Bulletin, vol. 75, May 1989, p. 351. Robert B. AvelY is an associate professor at 11. In 1981, mortgage bankers were estimated to originate an average of 29 percent of the new mortgages in the 100 SMSAs search and Federal Legislative Response," used in the study. Their share ranged from a Board of Governors of the Federal Reserve high of 68 percent to a low of 2 percent. System, Staff Study No. 121, October 1982. of lending. not as great as is commonly when important demand factors lender activity and lending is related risk to aggregate in many areas. The nor agreed are neither upon. to change through increased institution there is for this, however, well understood Attempts such that the racial composition neighborhoods reasons cient to ensure the nation- Although of mortgage on and 7. "The Color of Money," Atlanta Constitu- Veterans' Administration, the Federal Nation- even good citizenship necessary as it is-may and debate that in the determination • Conclusion issue is important be- Since regulated on the part of all regulated the redlining have The act as good citizens pattems, in Boston, 1982-1987," New England enacted on August 9, 1989. procedures, in their communities. parties Cleveland, Summer 1981. 2. The Financial Institutions Reform, Recovery, and Enforcement Act of 1989 was on in- and the CRA legislation lenders Economic Review, Federal Reserve Bank of "Geographic Patterns of Mortgage Lending Lending in C1eveland-1987," is that have focused most statistical regulated neighbor- of these data to the of individual likely to be costly circumstances hind them, are designed of aggregate across Karl E. Case, and Constance R. Dunham, May 1-4, 1988; and "Race and Mortgage in needs ap- - Robert B. Avery and Thomas M. Buynak, "Mortgage Redlining: Some New Evidence," "The Color of Money," Atlanta Constitution, for these apparently institutions' regulations, 6. Cuyahoga Plan of Ohio Inc., July 1989. dealt with aggregate data could be used demand compliance Footnotes 1. See, for example, Katharine L. Bradbury, of difficult, this outcome reliance are likely to be controversial, and expensive. The lack of a consensus causes redlining approach without on financial regulations and conditions argues should further view about the associated strongly with that such an not be undertaken study. BULK RATE U.S. Postage Paid Cleveland,OH Permit No. 385 or does a better and expen- And, regardless much information firms to take into market needed well an institution on man- in individual institutions whereas Even if available good measures risk-adjusted hoods, emphasis. and census as mortgage system that individual dividual and dif- that currently are taken into account, A compliance possible low regulatory contradictory market for example, have a very strong rules. It is entirely pear not to be met. as it is in other kinds of Institutions ferences are that loan transactions. be given to cost differentials made relative follow procedures with current The likely reason to a poor CRA rating. market, lend- in fewer acceptances, this circumstance self-defined pre- in the mortgage are in compliance • that vir- institutions and marketing regulatory the risk at- risk can arise from a factor ing decision, On the other hand, it appears for the con- is collected, of how interpre- Federal Reserve Bank of Cleveland Research Department P.O. Box 6387 Cleveland, OH 44101 of the data may still vary widely. Address Correction Requested: Please send corrected mailing label to FAULH880101000 000 001 CMTY LEE 0 FAULHABER CLEVELAND 1 the above address. RESEARCH Material may be reprinted provided that the source is credited. Please send copies of reprinted materials to the editor. 9 • tV £1 6 B 330 AH'VU81l HDB 3S3B