. United States General Accounting Office Report to Congressional Requesters GAO” November 1990 SECONDARY MORTGAGE MARKET Information on Underwriting and Home LOXE Atlanta GAO,‘RCED-91-Z United States GAO General Accounting Office Washington, D.C. 20548 Resources, Community, and Economic Development Division B-239564 November 28,199O The Honorable Alan J. Dixon, Chairman The Honorable Christopher S. Bond, Ranking Minority Member Subcommittee on Consumer and Regulatory Affairs Committee on Banking, Housing, and Urban Affairs United States Senate Your June 22, 1989, letter asked us to examine various aspects of home mortgage lending. Specifically, you asked us to examine whether (1) underwriting criteria of secondary market institutions and private mort- gage insurers contribute to racial discrimination and (2) statistical evi- dence of discrimination exists in the Atlanta, Georgia, area by home mortgage lenders that are not depository institutions. As you know, by purchasing home loans, the secondary mortgage market agencies spread financial risk and provide liquidity to primary lenders, thereby making additional credit available to qualified bor- rowers. Nondepository institutions include organizations such as mort- gage bankers as opposed to depository institutions such as banks, savings and loans, and credit unions. As discussed with your office, we were unable to fully address your areas of concern because of the absence of readily available data on lending activities of nondepository institutions and the extensive effort required to gather and analyze such data from original sources. Thus. we agreed to provide you with information on (1) underwriting guide- lines established by certain secondary mortgage market agencies to help them determine whether they should purchase, insure, or guarantee single-family home loans made by lenders and (2) statistical data on such loan activity by these agencies in the Atlanta, Georgia, metropol- itan area. L We reviewed and summarized underwriting guidelines and loan activity data of the Federal National Mortgage Association (Fannie Mae) and the Federal Home Loan Mortgage Corporation (Freddie Mac), which together market most of the dollar value of mortgages sold in the WC- ondary market. The Government National Mortgage Association (Gmnie Mae) also is a secondary market organization that guarantees sewrlt ies backed by mortgage loans insured by the Department of Housing and Urban Development (HUD) or guaranteed by the Department of 1’crt~rans Affairs (VA). Ginnie Mae has no underwriting guidelines of its olvn rl(jr Page 1 GAO/RCED91-2 Secondary Mortgage Market B-239664 does it have an automated system for identifying the details of the indi- vidual HUD or VA loans that it handles. Therefore, we also summarized HUD’S and VA’S loan approval guidelines and loan activity data to represent Ginnie Mae. The statistical data for Fannie Mae, Freddie Mac, and HUD represent the number and value of single-family home mortgage loans that these orga- nizations purchased or insured within 80 residential ZIP code areas in a five-county metropolitan Atlanta area during the period July 1, 1987, through June 30, 1989. Unlike Fannie Mae, Freddie Mac, and HUD, VA could not provide us with loan data at the ZIP code level. Therefore, we summarized the VA data separately by county, the only level of data available to us. Thus, the VA loan activity data represent the number and value of VA-guaranteed loans in the five metropolitan Atlanta counties. The loan activity data provide information on the extent and location of secondary market loan activity in the metropolitan Atlanta area. Iiow- ever, these data should not be used to derive conclusions on discrimina- tion in secondary market loan activity because data limitations prevented us from determining the reasons for the variations in the activity among ZIP code areas. A key limitation in the data is that it would be very difficult to determine the demand for loans-a primary factor in determining whether credit needs for housing have been met and a potentially significant reason for differences in loan activity among ZIP code areas-because such data are not regularly maintained by 1enders.l Other limitations include the use of multiple sources of data that we could not verify for accuracy, the absence of comparable housing markets within ZIP code areas, and the lack of information on the race of the actual buyer. (See app. I for details on data limitations.) Fannie Mae and Freddie Mac provide similar guidance to lenders from whom they purchase mortgages. Each agency sets out its standards in a Seller/Servicer Guide provided to lenders to assist them in making acceptable loans. While Ginnie Mae does not have any underwriting guidelines, it relies on the HUD- and VA-established guidelines that lenders use to determine which mortgages qualify for HUD insurance or a L-\ guarantee. ‘The Financial Institutions Reform, Recovery, and Enforcement Act of 1989 (Public, 1-1~ I’ ‘I 73 I, August 9, 1989, now requires lenders to retain information on loan applications. II)XIV..I( I ILLII!.made. and loans purchased starting on January 1,199O. Page 2 GAO/RCED91-2 Secondary Mongagr Market B-239564 All four agencies (Fannie Mae, Freddie Mac, HUD, and VA) require that in determining which loans to approve lenders consider a potential bor- rower’s ability and willingness to repay the debt and the property’s value. The agencies suggest that lenders consider factors such as the borrower’s housing expense-to-income ratio and track record in fulfilling previous debts. Property appraisals are required to ensure that the property’s market value is sufficient to provide adequate collateral for the mortgage loan. During the 2-year period from July 1, 1987, through June 30, 1989, Fannie Mae, Freddie Mac, HUD, and VA purchased, insured, or guaranteed 63,586 home loans in the metropolitan Atlanta area that we studied. Of the 57,227 loans purchased or insured by Fannie Mae, Freddie Mac, and HUD, 87 percent were for properties located in predominately white (defined as the range from 61-100 percent white) ZIP code areas. About 83 percent of these loans were for properties in ZIP code areas having average annual incomes of $35,000-$74,999. Median home prices were also highest in the predominately white, higher income areas. Of the 6,359 VA-guaranteed loans, 53 percent went to white individuals and 47 percent went to minority individuals. A secondary mortgage market buys and sells mortgage loans or securi- Background ties backed by mortgage loans. The secondary market agencies are not primary lenders and have no direct contact with borrowers. The agen- cies do not originate mortgage loans; rather, they purchase loans from lenders or guarantee securities based on the loans. Fannie Mae, Freddie Mac, and Ginnie Mae were created by the Congress for the purpose of sponsoring a secondary market for mortgages. Although under federal charter, Fannie Mae and Freddie Mac are pri- vate corporations. However, Ginnie Mae is a United States government corporation. From 1987 through the first half of 1989, the three agen- cies accounted for 57 percent of the total dollar volume of secondary mortgage market loan purchase activity.2 Fannie Mae and Freddie Mac together market most of the dollar value of conventional mortgages in the secondary market. They mainly purchase conventional mortgages from lenders such as mortgage banking compa- nies, thrifts, commercial banks, and others. The mortgages are packaged 2The source of these data, HUD, includes Giie Mae’s issuance of securities in the loan purchase data. However, technically, Ginnie Mae does not purchase mortgages. Page 3 GAO/RCED91-2 secondary Mortgage Market B239564 into securities and are sold to investors. The sale of such loans and secu- rities returns funds to the institution originating the loan, creating liquidity and allowing the originator to make additional loans or other- wise reuse the funds. Ginnie Mae is the secondary market organization that guarantees securi- ties backed by HuBinsured and VA-guaranteed home loans. Ginnie Mae does not purchase mortgage loans but guarantees the timely payment of principal and interest for privately issued securities. Ginnie Mae- approved banks, mortgage lenders, and other institutions issue securi- ties based on pools of HuDinsured and VA-guaranteed loans. Through the Ginnie Mae guarantee, these securities are backed by the full faith and credit of the United States. In the event the issuer defaults on the securi- ties, Ginnie Mae takes over principal and interest payments to the investor. Underwriting is the process of identifying potential risks associated with financial instruments and developing guidelines to assess the expected costs of covering those risks. Underwriting standards indi- rectly establish the qualifications that loans to individual borrowers must meet to be eligible for delivery to the secondary mortgage market. Lenders use the underwriting process to determine whether they will make individual loans or whether the loans will be held in their portfo- lios or later sold in a secondary market. The underwriting guidelines for Fannie Mae, Freddie Mac, HUD, and VA Underwriting are built around certain risk assessments. The risk assessments-which Guidelines lenders also consider in analyzing the degree of risk associated with making a home mortgage loan-include the following: l The borrower’s ability to repay the debt. l The borrower’s willingness to repay the debt. . The sufficiency of the property to secure the mortgage. To make the first two assessments, underwriting guidelines address fac- tors such as past credit history, current and projected income, and expenses. This information is used in making the lending decision. When a lender decides to make a loan, it sets loan terms, including an interest rate, collateral values, and other conditions consistent with the risks involved in the loan. Therefore, an individual with a good credit rating and sufficient collateral may receive more favorable terms than a bor- rower with a delinquent payment history or limited financial resources. Page 4 GAO/BCEJS91-2 Secondary Mortgage Market B-239564 In making the third assessment, lenders rely on property appraisals to assess the sufficiency of the property to secure the mortgage. In the 80 Atlanta ZIP code itreas we studied, Fannie Mae and Freddie Mac Loan Activity in purchased 45,700 mortgage loans, and HUD insured 11,527 mortgage Atlanta loans during the 2-year period ending June 30, 1989. In the five-county, Atlanta metropolitan area, VA guaranteed 6,359 loans during the same period. The number of homeowners in various income and race population clas- sifications differed substantially in most cases. Therefore, we generally presented loan activity data for Fannie Mae, Freddie Mac, and HUD in terms of the number of loans per 100 homeowners in an income or race category. Fannie Mae, Freddie Mac, and HUD purchased or insured almost twice as many home mortgage loans per 100 homeowners in predominately white (61-100 percent white) areas as in the predominately minority (defined as the range from O-40 percent white) areas. In the areas with 6 1 per- cent to 100 percent white population, Fannie Mae, Freddie Mac, and HUD purchased or insured 13.9 loans per 100 homeowners. In areas with 61 percent to 100 percent minority population, the agencies’ loan activity was 7.0 loans per 100 homeowners. Fannie Mae, Freddie Mac, and HUD purchased or insured the greatest number of loans per 100 homeowners in ZIP code areas with higher income levels. Loan activity per 100 homeowners was 1.8 times as great in areas with average income levels of $35,000 to $74,999 as it was in areas with average income levels of $7,500 to $34,999. Home prices were generally higher in predominately white and higher income areas. For example, in predominately minority (O-20 percent white) areas, the median home price was about $56,000, and in the pre- dominately white (81-100 percent white) areas, the median home price was $101,000. The median home price ranged from $53,000 in the lower income ($7,500-$24,999) areas to $146,000 in the higher income ($50,000~$74,999) areas. VA does not include ZIPcodes in its loan guarantee data base; howvtbr. it did provide us with data on a county level, by race. For the five**ounty area, 53 percent of the loans VA guaranteed were for white indivltirl;ils and 47 percent were for minorities. In terms of dollar values, .55 ptmvnt Page 5 GAO/ltCJD91-2 Secondary MO- Market 5239564 went to white individuals and 45 percent to minorities. We have summa- rized the VA data separately in appendix III of this report. We received written comments on a draft of this report from Freddie Agency Comments Mac and Fannie Mae within the 30 calendar-day limit specified by law (see app. IV and V). We also received oral comments from HI!D and VA officials. Subsequently, we received written comments on a draft of this report from HUD. The comments were received too late for us to present and evaluate without delaying the report’s issuance; therefore, they have not been reproduced in the report. However, copies of HUD'S comments will be provided to the requesters. The essence of HUD'S comments are discussed below. HUD stressed the importance of ensuring that the loan activity data not be misinterpreted. We agree. In fact, concern that the data could be mis- interpreted is one of the reasons we have caveated these data and pointed out that these caveats preclude reaching conclusions on discrim- ination in the secondary mortgage market. HUD also said that the discussion on its performance in the various ZIP code areas of Atlanta should be presented separately from the discus- sion on Freddie Mac’s and Fannie Mae’s performance. HUD belie\,es that because it has a much lower maximum loan amount than those agencies, a very different geographic pattern may emerge for its loan activity data than the pattern resulting from combining its data with those of Freddie Mac and Fannie Mae. Our report combines the loan activity data of Freddie Mac, Fannie Mae, and HUD because of the confidential nature of the data Freddie Mac and Fannie Mae provided to us. Freddie Mac acknowledged that it had leadership responsibih t ies in sup- port of affordable housing opportunities and against discriminatory lending practices. It also said that our report correctly points orlt r hat its guidelines will not permit the consideration of race of a borrowcar in any aspect of the loan underwriting process. Also, according to Frt~ci(iit~Mac, the information contained in our report which shows that t htb II IIrnber of loans purchased by Freddie Mac per 100 homeowners increati+ \VI~h the percent of whites within an area’s population “is consistent w rt 11the well-documented pattern of discrimination reported in the .At I;I11I ,1Con- stitution’s ‘The Color of Money’ series.” This result, according ! I 1l‘rcddie Mac, is to be expected since the mortgage loans it purchase5 t‘r( ‘RI Page6 GAO/RCEDSl-2 Secondaq Hdmwc(r .Market B-239564 lending institutions would reflect the mortgage loans originated by those lending institutions that are making such loans directly to borrowers. Fannie Mae said that the major finding of our report, that secondary market purchasing activity declines with decreasing neighborhood income or increasing neighborhood minority composition, warrants con- cern from all sectors of the mortgage finance industry. It also said that while our report raises serious questions and concerns, it is important that our findings be placed in the proper perspective given the data limi- tations described in our report. In this regard, Fannie Mae reiterated some of the limitations in the loan activity data described in appendix I and how these limitations preclude any inferences concerning the causes of the observed purchasing patterns. Freddie Mac and Fannie Mae also stressed their commitment to ensure that all potential homebuyers have equal access to credit. Freddie Mac pointed out that its recently created Affordable Housing Initiatives Department will help it and the industry design homeownership and rental programs to address this issue. Fannie Mae pointed out that it had a long-standing commitment to the homebuying credit needs of low- and moderate-income households and residents of inner city neighborhoods. It also said that many of its activities that particularly benefit inner city and low-income neighborhoods are not covered in this report, such as its purchases of multifamily mortgages and mortgage revenue bonds. A VA official agreed with the factual information presented in this report on that agency. This report discusses (1) the scope and methodology for our revieit . including data limitations that prevent us from reaching conclusions about the causes of differences in secondary market loan activity among areas in metropolitan Atlanta, (2) underwriting guidelines established by the secondary mortgage market agencies, and (3) a summary of the secondary market loan activity in the Atlanta area by population groups and income levels. We performed our work between July 1989 and May 1990. At your request, we plan no further distribution of this report until :30 days from the date of this letter. At that time, we will send topics to the appropriate congressional committees, the Secretaries of HI'D and L:!. the Page 7 GAO/RCED91-2 Secondary Mortgage Market 0239564 Chief Executive Officers of Fannie Mae and Freddie Mac, and the Presi- dent of Ginnie Mae. We will also make copies available to others upon request. If I can be of further assistance to you, please contact me at (202) 275- 5525. Major contributors to this report are listed in appendix VI. John M. Ols, Jr. Director, Housing and Community Development Issues Page 6 GAO/ECED91-2 secondary Mon~gr Market Page 9 GAO/RCED91-2 Secondary Maflmuv Market Contents Letter 1 Appendix I 14 Scope and Data Limitations 16 Methodology Appendix II 20 Underwriting General Requirements 20 Borrower’s Ability to Repay the Debt 21 Guidelines Used by Borrower’s Willingness to Repay the Debt 26 Fannie Mae, Freddie Sufficiency of the Property Value to Cover the Mortgage 28 Mac, HUD, and VA Appendix III 34 Secondary Market Atlanta Demographics Loan Activity by Population Group 34 37 Loan Activity in the Loan Activity by Income Level 40 Metropolitan Atlanta Loan Activity by Income Level and Population Group 43 VA Loan Activity 45 Area - Appendix IV 47 Comments From the Federal Home Loan Mortgage Corporation Appendix V 49 Comments From the Federal National Mortgage Association Appendix VI Major Contributors to This Report Page 10 GAO/RCJCD-91-2 Secondary Mortgage Market Tables Table II. 1: General Loan Requirements for Fannie Mae, 21 Freddie Mac, HUD, and VA Table 11.2:Comparison of Income Ratios for Fannie Mae, 23 Freddie Mac, HUD, and VA Table 11.3:Residual Monthly Incomes by Region for Loan 23 Amounts of $69,999 and Below Table 11.4:Residual Monthly Incomes by Region for Loan 24 Amounts of $70,000 and Above Table 111.1:Number of ZIP Codes by Population Group 35 Table 111.2:Number of ZIP Codes by Average Income 35 Level Table 111.3:Loan Activity and Number of ZIP Codes by 44 Population Group and Average Income Level Table 111.4:Number of VA-Guaranteed Loans for White 46 and Minority Individuals in Five Metropolitan Atlanta Counties Table 111.5:Dollar Value of VA-Guaranteed Loans for 46 White and Minority Individuals in Five Metropolitan Atlanta Counties Figures Figure III. 1: Percentage of Total Homeowners by 36 Population Group Figure 111.2:Percentage of Homeowners by Income Level 37 Figure 111.3:Number of Loans per 100 Homeowners by 38 Population Group Figure 111.4:Average and Median Loan Amounts by 39 Population Group Figure 111.5:Average and Median Home Prices by 40 Population Group Figure 111.6:Number of Loans per 100 Homeowners by 41 Income Level Figure 111.7:Average and Median Loan Amounts by 42 Income Level Figure 111.8:Average and Median Home Price by Income 43 Level Figure 111.9:Number of Loans per 100 Homeowners by 45 Population Group and Income Level Page 11 GAO/ECJZD-91-2 Secondary Mortgage Market Contents Abbreviations DMIS Donnelley Marketing Information Services Fannie MCUZ Federal National Mortgage Association Freddie Mac Federal Home ban Mortgage Corporation GAO General Accounting Office Ginnie MW Government National Mortgage Association HUD Department of Housing and Urban Development LTV loan-to-value ratio VA Department of Veterans Affairs ZIP Zone Improvement Plan Page 12 GAO/XCED-91-2 Secondary Mortgage Market Page 13 GAo/BCED@l-2 Secondary Mortgage Market Appendix I Scopeand Methodology We interviewed officials of Fannie Mae, Freddie Mac, and Ginnie Mae at their headquarters locations, in Washington, D.C., and HUD and VA offi- cials in their Atlanta Regional Offices to identify the appropriate under- writing guidelines. We obtained and reviewed copies of each organization’s underwriting guidelines for home mortgage loans. To provide statistical data on secondary mortgage market activity in the Atlanta, Georgia, metropolitan area, our work focused on the mortgage activities of Fannie Mae, Freddie Mac, HUD, and VA. We defined the Atlanta metropolitan area as the five counties of Clayton, Cobb, Dekalb, Fulton, and Gwinnett. To provide insights on the demographic characteristics of each ZIP code area, we obtained statistical data on race and income from the Donnelley Marketing Information Services (DMIS).' The demographic data are esti- mates of 1989 conditions based upon 1980 Census information on popu- lation and household income. The data included estimates of the number of individuals by race (white, black, and other) and number of house- holds per ZIP code area, as well as the number of households within spe- cific income categories for each ZIP code area. This data provide general demographic information, but does not represent the characteristics of specific borrowers associated with the agencies’ loan activity data dis- cussed in the report. The demographic data were compiled according to 1989 residential ZIP code data for the five counties. We sorted the statistical data for Fannie Mae, Freddie Mac, and HUD on two demographic variables-number of individuals by race and average income-for each ZIP code. The statistical data represent the number and value of single-family home mortgage loans that these organizations purchased, insured, or guaranteed in 80 ZIP code areas within the five-county metropolitan Atlanta area during the period July 1, 1987, through June 30, 1989. Although the agencies identified more than 80 ZIPcodes in the five coun- ties, our consolidation of loan activity and demographic data from various sources resulted in 80 ZIP codes for use in our study. Overall, our study of data for the 80 ZIP code areas represents about 85 percent of both the number and the total dollar value of loan activity the agencies reported to us. ‘Donnelley Marketing Information Services, a company of the Dunn & Bradstreet Coqwra111an.pre vides selected demographic information from the 1980 Census of the United States, proJwtr4 to reflect current year (1989) estimates. The Donnelley Demographics data base contams rstlmatrs for various demographic characteristics and is generally used for market analysis. Page 14 GAO/RCJCD91-2 Secondary Mortgage Market Appendix I Scope and Methodology We compiled the statistical data by postal ZIP codes because that was the lowest common level that Fannie Mae, Freddie Mac, and HUD could accu- mulate to provide the data. However, the VA data were not available below the county level because VA does not include ZIP codes in its loan guarantee data base. Therefore, the VA data are presented separately, by county. Throughout the report, we rounded all percentages to the nearest whole number. Unlike Fannie Mae, Freddie Mac, and HUD, VA included the race of bor- rowers associated with loans guaranteed in the data provided to us. However, according to a VA Program Analysis Specialist, VA could not provide complete data on borrower income; therefore, we sorted the data only by race. Because a large number of renters in a ZIPcode area could distort the loan activity data, we adjusted the population of households in each ZIP code area to reflect only the estimated number of homeowners. For each ZIPcode, this adjustment was the estimated number of households in 1989 less the percentage of renters from the 1980 Census data. Generally, we presented the loan activity data in terms of the number of loans per 100 homeowners in an income or race category. We used this measure to provide greater comparability in terms of the population of homeowners among the various categories. To identify the racial composition of the ZIPcode areas, we defined two population groups- white and minority. We classified the demographic data in terms of the percentage of white individuals by ZIPcode area and sorted the data into five population groups ranging from O-20 percent white to 81-100 percent white. However, these population classifications can also express the percentage of minority individuals in a ZIP code area. For example, a population group of O-20 percent white can also be referred to as 81-100 percent minority. To show the loan activity within ZP code areas having various income levels, we estimated average household income per ZIPcode area. All 80 ZIPcodes had average incomes between $7,600 and $74,999. We sorted the mortgage loan activity data by the four income levels per ZIP code area: l $7,500~$24,999; l $26,000~$34,999; . $35,000-$49,999; and Page 16 GAO/IlCED91-2 Secondary blow Market Appendix I Scope and Methodology l $50,000-$74,999. We obtained information concerning home prices from Dataman Infor- mation Services, Inc.2 Dataman obtained the information from warranty and security deeds recorded in Fulton, Dekalb, Gwinnett, Cobb, and Clayton County courthouses. This home sales data were the most com- plete and recent that we identified. The data represent the amount paid to the nearest thousand dollars for single-family dwelling properties in the 80 ZIP code areas studied during the period September 1, 1988, through September 30,1989. We merged the home sales and demo- graphic data and computed the average and median sales prices of homes for the selected population and income categories of ZIP code areas. The loan activity data provide information on the extent and location of Data Limitations secondary market activity in the Atlanta metropolitan area. However, we were unable to determine the reasons for the variations in loan activity among areas in Atlanta because of data limitations. Conse- quently, the data should not be used to derive conclusions on discrimina- tion in secondary market loan activity. Reliabili .ty of Data Not As discussed above, we obtained data on loan activity and demographic characteristics from different agencies as well as private industry Verified sources. Because of time constraints and possible problems obtaining access to data, we could not perform any data reliability assessments to determine the accuracy of the data provided to us by these organizations. The demographic characteristics for each ZIP code area do not necessa- rily represent current (1990) conditions in the Atlanta area. The data on population by race and income levels are estimated 1989 data based on 1980 Census data. We obtained this data from a private marketing infor- mation service and did not analyze its method of projecting the data. ‘Dataman Information Services, Inc., is a privately owned company that collects real I-GUIVAIMSmort- gage data and prepares detailed analyses of housing activity. Dataman collects such tnf~I~~AI11in as the purchase price, mortgage amount, loan type, transaction date, and lender name frl NT)H .c.~.u~y and security deed records at county and municipal courthouses nationwide on a diulv bw\ l‘h~ r& estate and mortgage data are generally used for market analysis. Page 16 GAO/WED-91-2 Secondary Monw Market Appendix I Scope and Methodology Lack of Comparabil .ity of ZIP codes were the lowest common geographic area for which Fannie Mae, Freddie Mac, and HUD could provide loan activity data. However, ZIP Code Areas the lack of comparability in ZIP code areas in terms of demand for housing, the number of renters versus homeowners, and the condition of the housing stock may affect any comparison of loan activity data. In fact, any other unique circumstance for a particular ZIP code may affect the loan activity patterns and may not be representative of the agencies’ purchasing or insuring tendencies. The data available to us did not provide a means of controlling for vari- ations in loan demand across the various ZIP codes. We did not measure title transfers, the level of new home construction, the number of existing homes for sale, or resident mobility, which could provide some indication of the housing market and loan demand in each ZIP code. Because loan demand influences the number of loan originations, a greater demand for mortgage loans in one area versus another could indicate a racial pattern that is not caused by discrimination. To ensure greater comparability in terms of homeowners, we adjusted the population of households in each ZP code area to reflect only the estimated number of homeowners. The number of mortgageable residen- tial properties in a ZIP code area with a large number of households living in multifamily rental structures may be much less than the overall household count. For each ZIP code, this adjustment was simply the esti- mated number of households in 1989 less the percentage of renters from the 1980 Census data. However, we do not know what changes may have occurred in the numbers of renters and homeowners in each ZIP code over the past 10 years or what effect these changes would have on the purchasing patterns of the secondary market agencies. Few ZIP Codes in Some The demographics of the Atlanta area produced uneven distribution of the 80 ZIP codes over the five race and four income categories we Race/ Income Categories defined. In some instances, as few as one or two ZIP codes fell into a particular category. (See table 111.3.)The uneven distribution of the ZIP codes over the various categories may, in part, explain the loan activity patterns. For example, an income and/or racial composition category may reflect unique characteristics, such as proximity to commercial activity, which may prevent lenders from originating loans in those areas. Therefore, we cannot determine whether the loan activity pat- terns are representative of the agencies’ purchasing or insuring tenden- cies or some other factors. Page 17 GAO/RcEDsl-2 Secondary Mortgage Market Appendix I Scope and Methodology Compliance With Basic The loan activity data available to us did not contain information on Underwriting Criteria Not why loan applications were approved. Similarly, we had no information on why loan applications were denied. The loan activity data available Assessed did not contain information on loan selection; that is, the process by which lenders determine whether a given loan application is within the bounds of acceptable risk. The data are limited to the overall number and value of loan activity by ZIP code area for each agency and does not provide any information on the individual loans involved. For example, we did not have information on any individual borrower’s income or credit worthiness or the value of properties involved. In addition, we had no way to determine the condition or comparability of the housing stock among the various ZIP code areas. Since home appraisal is a major factor in determining whether a loan is within the bounds of acceptable risk, the condition of the housing stock may account for a difference in the level of loan activity among ZIPUK+ areas. The only data we have that are related to the mortgage underwriting process is ZIP code average income. However, this measure is limited because (1) the income of potential mortgagers may be significantly dif- ferent from the ZIP code average and (2) average income may be skewed by a few extremely high or low household incomes within a ZIPcode area. Similarly, population groups are defined by the percentage of whit0 or minority individuals in a ZIP code area; however, we cannot say whether the loan activity or the average and median home prices in thcstl ;irttas are related specifically to white or minority individuals. Renter Income Included in The demographic data we obtained from DMIS included the numbvr of households in seven income categories for each of the 80 ZIPcodt~ ZIP Code Average Income Based on this data, we calculated an average income for each LII’ (.ode in our study. Because the average income figure we calculated is t);t\t4 on household income, the incomes of renters as well as homeownt~rs XV included. The ZIP code average income figure we calculated ma>’ t)(btiis- torted to the extent that renters’ and homeowners’ incomes tilt‘t’t~! Duplication of Loan Data Fannie Mae and Freddie Mac loan data may duplicate HITD’S Io;~H (kit ;t to a small extent. We wanted to provide a perspective on the OLc~r;~ll( ;~nnie Page 18 GAO/RCED-91-2 Seconda M11nI&UWMarket Appendix I Soope and Methodology Mae activity in the Atlanta area, but Ginnie Mae does not have auto- mated data to show details on the specific HUD or VA loans it backs. Since Ginnie Mae handles about 90 percent of the HLJD and VA loans, we obtained statistical data from HUD and VA on their mortgage loan activity. We recognize that some duplication exists because both Fannie Mae and Freddie Mac purchased a small percentage of HUD and VA loans. VA Data Not Available by VA data were not available below the county level because VA does not include ZJP codes in its loan guarantee data base. Although Ginnie Mae ZIP Code handles 90 percent of all HUDand VA loans, we could not include VA data in our summary of loan activity by ZIPcode area. Therefore, we have provided a separate summary of VA loan activity by race for each county. For the 2-year period, VA guaranteed only 10 percent of the agencies’ loan activity in the 80 ZIPcode areas discussed in this report. Home Price Data Not The home sales (price) data we used reflect the average home values in the 80 ZIPcode areas during the period September 1,1988, through Sep Compatible With Loan tember 30,1989, and cannot be compared to secondary market loan Activity Data activity. Dataman obtained information on home sales from the war- ranty deeds recorded in the respective county courthouses. The home sales data represent all single-family dwelling property transfers that occurred during the period September 1,1988, through September 30, 1989. This time period represented the most complete and current data available to us. Also, no relationship exists between the race and income categories in our statistical presentations and the home prices within these catego- ries. We do not know the race or the income levels of the actual buyers. Finally, because the home price data and loan value data are from two different sources, the data cannot be used together to draw conclusions concerning loan-to-value (Lrv) ratios. Pye 19 GAO/RCTCWl-2 secondary hlw brket Appendix II Underwriting Guidelines Used by Fannie Mae, Freddie Mac, HUD, and VA This appendix addresses the underwriting guidelines established for individual home loans as set by the secondary market agencies (Fannie Mae and Freddie Mac) and the government agencies (HUD and w). Fannie Mae and Freddie Mac provide similar guidance to lenders from whom they purchase mortgages. Each specifies its standards in a Seller/ Servicer Guide provided to lenders to assist them in underwriting acceptable loans. Ginnie Mae does not have any underwriting guidelines of its own but will assist in marketing loans insured by HUD or guaranteed by VA. HCD and VA have each established their own credit and property appraisal guidelines for determining which mortgages to insure or guarantee. HUD'S underwriting standards are included in handbooks 4155.1 REV-2 (credit) and 4150.1 REV-l (appraisal). VA’S underwriting criteria are included in circular 26-80-l 1, revised (credit) and manual M26-2 (appraisal). As of January 1, 1990, Fannie Mae and Freddie Mac have had maximum General Requirements loan purchase amounts of $187,450, in the continental United States. HUD'S maximum loan insurance amount is $67,500 but can be $124,875’ in areas with prevailing high housing costs. The amount of the guaranty for a VA loan depends on the amount of the loan. Since December 1989, the guaranty has been (1) 50 percent of the loan amount for loans of $45,000 or less or $22,500 for loans of greater than $45,000, but not more than $56,250; (2) the lesser of 40 percent or $36,000 for loans of more than $56,250 and not more than $144,000; and (3) the lesser of $46,000 or 25 percent for loans of greater than $144,000. Table II. 1 outlines the general loan requirements for mortgages on single-family dwellings for the four organizations. ‘Effective January 12. 1990, the Congress raised the “high cost” limit to $124.875 for fi.sA year 1990. The limit will revert to $101,250 after September 30, 1990, unless the Congress tlstt,nds the fEcal year 1990 increase. Subsequently, the President signed a temporary resolution that Inamtains the $124,875 loan limit until October 31,199O. Page 20 GAO/RCEDSl-2 Secondary Mortgage Market Appendix II Underwriting Guidelinea Used by Fannie Mae, Freddie Mac, HUD, and VA Table 11.1:General Loan Requirements for Fannie Mae, Freddie Mac, HUD, and Maximum loan Maximum loan-to- Mortgage insurance VA Agency amount value ratios. required Fannie Mae $187,450 in 95 percent For LTV greater than continental United 80 percent States Freddie Mac $187,450 in 95 percent For LTV greater than continental United 80 percent States HUD $67,500; $124,875 in 97 percent of first HUD provided prevailing high $25,000; 95 percent housing cost area of remaining value b b VA VA provided BThe LTV ratio expresses the loan amount as a percentage of the value of the property. Maximum LTV ratios may differ in certain situations specified in the guidelines. %A crrcular 26-80-l 1 (rev. Dec. 2. 1987) does not contain the maximum loan amount or the maxrmum LTV ratio acceptable to VA. According to VA’s Chief of Loan Processing, Atlanta Regronal Office, VA does not have a maximum loan amount or a maximum loan-to-value ratio. In determining the borrower’s ability to repay the debt, Fannie Mae, Borrower’s Ability to Freddie Mac, HUD, and VA recommend that the lenders relate the bor- Repay the Debt rower’s income and liabilities to the proposed housing payment. In doing so the underwriter should assess factors such as borrower income, housing expense-to-income ratio, total debt-to-income ratio, and employment. Income Fannie Mae, Freddie Mac, HUD,and VA recommend that lenders make a determination regarding the adequacy, stability, and continuance of the borrower’s income. Each organization requires that the lenders verify 2 years’ previous earnings in making such a determination. In addition to earnings from the borrower’s primary employment, the guidelines permit inclusion of the following items in determining total income: Secondary income such as bonuses, commissions, overtime, and part- time or second job income. Certain military compensation such as income from the National Guard, flight or hazard pay, or quarters allowance. Retirement or social security income. Income from federal, state, or local assistance programs, if disclosed by the borrower. Child support or alimony. Page 21 GAO/BcEDol-2 Secondmy Mortgwje hhrket Appendix II Underwriting Guidelines Used by Fannie Mae, Freddie Mac, HUD, and VA l Other verifiable sources of income. Fannie Mae’s guidelines also permit income from seasonal employment if it has continued for the past 2 years and the borrower expects to be rehired for the next season. In determining the borrower’s stable monthly income and earning poten- tial, Fannie Mae’s and Freddie Mac’s guidelines suggest that lenders con- sider factors such as the borrower’s education, training, technical skills, occupation, and past employment history. On a case-by-case basis, Freddie Mac also considers the borrower’s age in determining stable monthly income. When including income from federal, state, or local assistance programs, alimony, or child support, various factors must be considered in deter- mining the likelihood of such income continuing. These factors and the organizations requiring them include . whether payments are received pursuant to a written agreement or court decree (Fannie Mae, Freddie Mac, HUD, VA); . the length of time the payments have been received or are expected to be received (Fannie Mae, Freddie Mac, HUD, VA); . the regularity of receipt (Fannie Mae, Freddie Mac, HUD, VA); . whether legal procedures are available to compel payment (Freddie Mac, HUD, VA); and l the credit worthiness of the payer, including the payer’s credit history when available to the seller under the Fair Credit Reporting Act or other applicable laws (Freddie Mac, HUD, VA). Income Ratios For loans to be eligible for sale to the secondary market, the loans and the borrowers are usually required to meet certain qualifying financial ratios to set limits on the risks involved. Generally, underwriting guide- lines may establish maximum ratio percentages and require the applica- tion of these ratios on a loan-to-loan basis. The secondary market agencies and government agencies use two overall ratios as guidelines to qualify homebuyers. Table II.2 summa- rizes the housing expense-to-income ratios and total debt payment-to- income ratios acceptable to the four underwriting organizations Page 22 GAO/RCED91-2 !Secondmy Mortgngr Market Appendix II Underwriting Guidelines Used by Fannie Mae, Freddie Mac, HUD, and VA Table 11.2:Comparison of Income Ratios for Fannie Mae, Freddie Mac, HUD, and Monthly housing expense- Total monthly debt VA Agency to-income ratio payment-to-income ratio Fannre Mae 28 percent 36 percent (For LTV less than or equal to 90 percent) 33 percent (For LTV greater than 90 percent) Freddie Mac 28 percent 36 percent HUD 29 percent 41 percent VA nonea 41 percent %ee tables II 3 and 4 VA’S underwriting guidelines are unique in that they provide for a residual income method of qualifying a borrower. To qualify a borrower under this method, housing expenses (including mortgage payments) and other monthly payments are subtracted from the borrower’s net effective income. Net effective income is gross income less federal income taxes. The remaining value is the residual monthly income for family support. VA provides a table of residual monthly incomes by region based on Department of Labor consumer expenditure surveys. VA provides the residual income table as a guide to qualify borrowers; how- ever, ~4 states that these figures should not automatically trigger approval or rejection of a loan. Tables II.3 and 4 show the residual monthly incomes for family support for loan amounts up to $69,999 and above $70,000. Table 11.3:Residual Monthly Incomes by Region for Loan Amounts of $69,999 and Family size0 Northeast Midwest South West Below 1 $348 $340 $340 $379 2 $583 $570 $570 $635 3 $702 $687 $687 $765 4 $791 $773 $773 $861 5 $821 $803 $803 $894 aFor famrlies wrth more than five members, add $70 for each additional member up to a family of seven Page 23 GAO/RCJD91-2 Secondary Mortgage Market Appendix II Underwriting Guidelines Used by Fannie Mae, Freddie Mac, HUD, and VA Table 11.4:Residual Monthly Incomes by Region for Loan Amounts of $70,000 and Family size0 Northeast Midwest South West Above 1 $401 $393 $393 $437 2 $673 $658 $658 $733 3 $810 $792 $792 ~--__$882 4 $913 $893 $893 $995 5 $946 $925 $925 $1.031 ‘For families with more than five members, add $75 for each additional member up to a family of seven, Each of the organizations provide for compensating factors which may allow the borrower to exceed the maximum income ratios or residual income figures discussed above. Examples of compensating factors pro- vided for in the various guidelines are a large down payment; the demonstrated ability of the borrower to devote a greater portion of income to housing expense; the borrower’s net worth being substantial enough to evidence an ability to repay the mortgage regardless of income; the likelihood of increased earnings based on education, job training, or time employed or practiced in a profession; evidence of an acceptable credit history or limited credit use; less than maximum mortgage term; and a health or welfare or community service organization provides funds for unusual services, house repairs, and the like. In considering the borrower’s income ratios, Freddie Mac guidelines indi- cate that more flexibility is appropriate for the monthly housing expense ratio than for the debt payment ratio. All guidelines require that the lender provide a written explanation that includes the compensating factors that justify the use of higher ratios. Employment As part of determining the stability of income, lenders must look at the borrower’s employment history. Fannie Mae, Freddie Mac, and f11I) require the lender to verify the borrower’s employment for the 2 >xtars preceding the loan application. The borrower must explain any frequent changes or gaps in employment for this time period. VA also requlrcs ver- ification of the borrower’s preceding 2 years of employment; hok+x>\.cr, VA provides an exception if this period consisted of active military duty. Page 24 GAO/RCEl%91-2 !Secondaxy Mot-t- Market Appendix II Underwriting Guidelines Used by Fannie hfae, Freddie Mac, HUD, and VA Fannie Mae guidelines indicate that a shorter employment history may be acceptable with adequate verification for a borrower who has recently graduated from school or was recently discharged from the mil- itary. VA also provides that recently discharged veterans or those with employment of short duration require special attention. Freddie Mac states only that the lender must consider the circumstances surrounding gaps in employment. HUD guidelines state that no arbitrary limits should be set for the length of time a borrower must have held a position in order to be eligible. However, for employment of short duration, HUD requires special consid- eration. HUD also indicates that in cases where employment on a tempo- rary basis is customary, employment stability will depend on the availability of opportunities for re-employment. Finally, HUD provides guidance indicating that temporary unemployment due to action of rec- ognized labor unions does not necessarily make the borrower ineligible for an insured mortgage. Both Fannie Mae’s and Freddie Mac’s guidelines indicate that frequent job changes do not necessarily indicate the lack of stable income. Ror- rowers who change jobs frequently to advance within the same line of work and who are successful in that work should be considered fator- ably. However, frequent job changes without advancement or changes from one line of work to another could lead to unstable income. Frc>ddie Mac suggests that if the borrower has maintained a stable income over the recent past, job hopping without advancement should not result in unfavorable consideration. Freddie Mac requires that borrowers with unstable employment histories have demonstrated financial strength and the ability to meet financial obligations when due. Fannie Mae guidelines also state that borrowers who have questionable employment histories must have strong offsetting financial strtlngths to be considered for maximum financing. Fannie Mae defines masimum financing as an amount that is within 5 percent of the highest IXI ratio allowed for a specific type of mortgage. For example, for those typt’s of mortgages allowed a 95- percent LTV ratio, any financing that tlsc~t~t& 90 percent of the property’s value would be considered maximum financing. Fannie Mae Requirements Only Fannie Mae guidelines specify additional lending consltkr;~r :(,115 when the LTV exceeds 80 percent. For these high LTV loans, the I(‘III~(T on High LTV Mortgages must pay particular attention to the borrower’s Page26 GAO/RCED-91-2 Secondary %hr(awr Market Appendix II Underwriting Guidelines Used by Fannie Mae, Freddie Mac, HUD, and VA l adequacy of reserves after closing, . ability to make a monthly housing payment that is larger than his or her previous mortgage or rental payment, l ability to accumulate savings and to demonstrate proper debt management, l demonstrated capability for increased earnings in future years (espe- cially for adjustable rate mortgages), and . ability to maintain an excellent credit history. If the LTV ratio is above 90 percent, the lender may use higher qualifying ratios only if the borrower satisfies one of the following requirements: l Has financial reserves that can be used to carry the mortgage debt: part of the savings must be in the form of liquid assets that equal at least 2 months of housing expense payments. l Has a demonstrated ability to devote a greater portion of income to housing expenses (but the housing expense for the mortgage the appli- cant is seeking should not exceed the borrower’s previous housing expense), an excellent payment history on any prior mortgage obliga- tion, and an excellent credit history. l Has a debt payment-to-income ratio (at the time of the application) of 30 percent or less, an excellent payment history on any prior mortgage obli- gation, and an excellent credit history. ties’ guidelines suggest that the lender should consider the borrower’s Willingness to Repay track record for meeting previous credit obligations. This analysis the Debt requires a review of the borrower’s manner of paying obligations and the ability to manage financial affairs. According to HUD guidelines, past credit performance serves as the most reliable guide in determining the credit attitudes that govern the individual’s future actions. If the borrower has a history of slow payments on existing or previous debt, Fannie Mae, Freddie Mac, and VA require that the borrower provide an explanation. Lenders must also pay careful consideration to past bankruptcies and foreclosures and other adverse credit actions. HUD guidelines indicate that the lender should not look for the isolated case of unsatisfactory or slow payment of an account but for a general pattern of credit behavior. A period in the past containing financial dif- ficulty does not necessarily make the risk unacceptable, if, subse- quently, a good payment record has been maintained. However, if the Page 26 GAO/RCED91-2 Secondary Mortgage Market Appendix II Underwriting Guidelines Used by Fannie Mae, Freddie Mac, HUD, and VA borrower has adequate income but consistently fails to repay creditors promptly, the reasons for this pattern must be carefully analyzed. Fannie Mae standards state that a borrower who has satisfactorily made payments on outstanding or previous credit obligations may be consid- ered favorably. However, a borrower who continually increases liabili- ties and periodically bails out through refinancing and debt consolidation is a marginal credit risk. Freddie Mac relies on the lender’s determination that the borrower is credit worthy. Freddie Mac requires the lender determine that the bor- rower’s credit reputation would be acceptable to mortgage lenders in the area. VA states that if the credit analysis develops any derogatory credit infor- mation and, despite such facts, it is determined that the borrower is a satisfactory credit risk, the lender must explain the basis for the decision. Fannie Mae, Freddie Mac, and VA require a term of at least 2 years after a bankruptcy proceeding against the borrower before the applicant can be considered for a loan. A shorter term is acceptable if the lender can prove that extraordinary circumstances caused the bankruptcy. Exam- ples of extraordinary circumstances provided in the various guidelines are those that are beyond the control of the borrower such as a serious, long-term illness not covered by insurance, death of a principal wage- earner, or loss of employment due to factory slowdowns, strikes, or reductions in force. HUD requires that at least 1 year has passed before considering the borrower for a mortgage. In all cases the lenders must determine that the borrower has re-established good credit. Generally, Fannie Mae and Freddie Mac will not purchase and HII) will not insure any loan for a borrower who has defaulted on a mortgage within the past 3 years. However, if the borrower has owned property that was subject to foreclosure proceedings within the past 3 years, the lender must document that the foreclosure resulted from extraordinary circumstances. VA does not make specific comments concerning previous mortgage defaults; however, VA does address prior VA loan experience. w states that such experience, especially if it is recent, may be so unfa\.or;ible that further credit is not warranted. VA does not state an acceptabltL time Page 27 GAO/WED-91-2 Secondary Mortgage Market Appendix II Underwriting Guidelines Used by Fannie Mae, Freddie Mac, HUD, and VA frame after which adverse prior mortgage experience would not be con- sidered in the lending decision. Fannie Mae, HUD, and VA indicate that the absence of credit history will not generally be viewed as an adverse factor in credit underwriting. Fannie Mae and VA require that efforts should be made to develop evi- dence of timely payment of obligations such as rent and utilities pay- ments Fannie Mae guidelines indicate that when adequate credit histories cannot be established in this manner, the lender should con- sider very conservative mortgage terms only and even those may not be appropriate without strong offsetting factors. Freddie Mac does not address the absence of borrower credit history. The primary purpose of conducting a property appraisal is to estimate Sufficiency of the the fair market value of the property that is the collateral securing the Property Value to mortgage loan. The appraiser’s role is to provide a defensible estimate of Cover the Mortgage property value and provide a complete, accurate description of the property and other related information to support the appraiser’s esti- mate of market value and related risks. Appraisals are important because in the event of a default, the collateral’s market value is what stands between the lender and a potential loss. The lender can recover the investment in the property without suffering a loss only if the prop- erty can be sold for an amount greater than the unpaid loan balance plus the cost of foreclosure proceedings. Fannie Mae’s and Freddie Mac’s appraisal guidelines are included in their Seller/Servicer guides. HUD'S appraisal guidelines are in Handbook 4150.1 Rev-l. VA'S standards are in Manual M26-2 (January 26, 1988). Each agency uses the “Uniform Residential Appraisal Report” form in conducting the appraisal. Fannie Mae and HUD appraisal guidelines focus on analyses of the neigh- borhood or location of the property, a physical inspection of the site and improvements, and the valuation of the property which are the main sections on the “Uniform Residential Appraisal Report.” Freddie Mac and VA address these areas; however, their guidelines generally do not provide specific procedures for performing the appraisal other than requiring completion of the “Uniform Residential Appraisal Report .” Each set of guidelines outlines three approaches for appraisers to llse in determining the market value of property -market or sales comparison, cost, and income. Page 29 GAO/ltCEDBl-2 Secondary Mm-tgagr Market Appendix II Underwriting Guidelines Used by Fannie Mae, Freddle Mae, HUD, and VA Neighborhood/Location Fannie Mae and HUD guidelines provide for a neighborhood/location Analysis: Fannie Mae and analysis to determine the value of property based on economic, social, government, or environmental forces that affect it. Fannie Mae defines a HUD neighborhood as an area with a group of properties that have comple- mentary land uses. HUD'S location analysis involves a comparison of sim- ilar locations without regard to the character or quality of the building improvements that exist on the site. In other words, a vacant site should receive the same location evaluation as an improved site in a similar location-one that includes properties having similar amenities and values to those in the subject location/neighborhood. As stated in Fannie Mae and HUD guidelines, examples of the forces that may influence the market value of a property in a particular neighbor- hood or location are l industrial, commercial, agricultural, and retail sales activity (HUD); l price and wage levels or the purchasing power of individuals (HUD); l employment opportunities (HUD); . supply and demand for living units (HUD, Fannie Mae); l taxation levels (HUD); l population change (HUD); l attractiveness of neighborhood buildings (HUD); . neighborhood character and character of neighborhood structures (HUD); l age of structures (HUD, Fannie Mae); adequacy of transportation (HUD); l degree of development and growth rate (Fannie Mae); l . property values (Fannie Mae); . changes in property from owner-occupied to tenant-occupied dwellings (Fannie Mae); and high vacancy rates (Fannie Mae). l Fannie Mae states that neither the racial composition nor the age of the neighborhood is a reliable appraisal factor. Fannie Mae states that it does not designate certain areas as being acceptable or unacceptable, but does recognize that “locational factors are fundamental to proper appraising and prudent underwriting and that there is nothing improper about underwriting on the basis of a realistic perception of risk in a given neighborhood.” Fannie Mae also states that the appraiser must be impartial and specific in describing the favorable or unfavorable factors in a neighborhood and should avoid the use of subjective terms such as “pride of ownership” or “neighborhoodin transition.” Page 29 GAO/WED-91-2 Secondary Mortgage Market Appendix II Underwriting Guidelines Used by Fannie Mae, Freddie Mac, HUD, and VA HUD provides specific statements pertaining to property located in low- income areas. For example, concerning the attractiveness of neighbor- hood buildings, HUD states that the appeal of a location is strengthened if the buildings in a neighborhood are attractive as a group and harmonize with one another and with their physical surroundings and that a pleasing variety that results in harmoniously blended properties without monotonous repetition is desirable. HUD further states that “it has been demonstrated that a pleasing variety in dwelling design need not be sacrificed in a neighborhood composed of low-cost housing.” HUD also states that “areas occupied by low-income families will ordinarily have easier and less expensive access to [community] facilities.” In assessing the neighborhood, HUD and Fannie Mae require the appraiser to give consideration to environmental changes such as neigh- borhood decline. HUD considers that the “infiltration of commercial, manufacturing, industrial enterprises and other nonconforming uses in residential sections, and the physical deterioration of buildings in these sections, are other obvious and common causes” of neighborhood decline. HUD guidelines state that “consideration must be given to the causes of decline in desirability and utility of residential districts in order to develop the greatest accuracy in valuation estimates.” Fannie Mae’s guidelines do not give examples of causes of neighborhood decline but state that appraisers must consider the cause of the property’s decline and its effect on the property’s marketability. According to Fannie Mae’s guidelines, properties in areas of declining value must be reviewed with great care. Fannie Mae standards state that a “lender must not consider the use of maximum financing in any instance in which property values are declining.” HUD states that any older existing community which is found unaccept- able because of certain features adversely affecting its location may be eligible for special funds under its section 223(e) program. The purpose of the section 223(e) program is to permit the use of HUD mortgage insur- ance in older, declining urban areas, in order to provide housing for low- and moderate-income families and to contribute to the upgrading or sta- bilization of such areas. Special funds have been appropriated by the Congress for this program since the insurance of mortgages in such areas constitutes a higher risk than other localities. HUD states that the Chief Appraiser in each field office should become acquainted with and be aware of such neighborhoods so as to assure that the special high- risk funds are used for properties in such areas. Page 30 GAO/RCEBBl-2 Secondary Mortgag(r Market Appendix II Underwriting Guidelines Used by Fannie Mae, Freddie Mac, HUD, and VA According to HUD, this is not to be confused with “redlining.” To redline is to withhold home loan funds or insurance from neighborhoods consid- ered poor economic risks. HUD states that it does not withhold insurance, but rather designates the insurance fund program which must be used in connection with the insuring of loans in these areas. Finally, HUD states that gentrification may reverse the rate of decline of an older neighborhood assuming it is not subject to heavy commercial or industrial encroachment. Gentrification occurs when people move into older or declining neighborhoods and restore the homes. Both Fannie Mae and HUD use an analysis of similar neighborhoods in determining the strength of the subject neighborhood-the neighbor- hood containing the property an applicant is trying to buy. Fannie Mae provides guidance for a neighborhood rating system in which the appraiser rates the various aspects of a neighborhood by comparing them to the same aspects of similar neighborhoods. A similar neighbor- hood is one that includes properties having similar amenities and values to those in the subject neighborhood. In performing this analysis, appraisers must rate the principal items in a neighborhood that are generally considered important by people when they purchase a home- convenience to employment, shopping, and schools; adequacy of public transportation or utilities; police and fire protection; general appearance of the properties; and property compati- bility. An average rating should indicate that the characteristics of sub- ject neighborhoods are equal to those that represent the norm for the market area and that are considered acceptable in competing neighbor- hoods. The use of neighborhood ratings should not preclude appraiser comments on the neighborhood conditions. Fannie Mae provides the fol- lowing example. If a neighborhood is characterized by a lack of mainte- nance or absence of local government services (which may be typlc*al for similar neighborhoods and, therefore, warrant an “average” rating 1.the appraiser still must describe such neighborhood conditions on the appraisal form. HUD requires a similar location analysis. HUD guidelines state that ;m acceptable location must be related to the needs of the prospect 1LILI MW- pants and to the alternatives available to them in other similar ICH;II Ions. Page 31 GAO/RCRDBl-2 Secondary Wongaur Harket Appendix II Underwriting Guidelinea Used by Fannie Mse, Freddie Mac, HUD, and VA Site Analysis Fannie Mae states that in order folr a property to qualify for maximum Requirements: Fannie Mae financing, the site should be of the size, shape, and topography that gen- erally conforms and is acceptable in the market area. HUD does not spe- and HUD cifically address a “site analysis” but includes similar guidelines such as those listed below in its location analysis discussed earlier. Fannie Mae suggests that the appraiser comment on the following factors that affect the site: . zoning classification and compliance; l a determination of the highest and best use of the land; l the acceptability of the utilities and streets; l the topography, shape, size, and drainage of the lot; and l whether property improvements are located in a flood hazard area. Appraisal Guidelines: Freddie Mac requires the lender to obtain an appraisal report for each Freddie Mac mortgage. The appraisal report must be completed in a manner that sup- ports the appraiser’s estimate of market value and presents to the reader a visual picture of the neighborhood, site, and improvements. The appraiser is encouraged to use the “Comments” section of the appraisal report or attach addenda to make this presentation. The appraiser must also evaluate the stability and marketability of the mort- gaged premises compared with other properties in the mortgaged prem- ises’ price range. Freddie Mac generally does not provide specific guidance for conducting the appraisal. Freddie Mac’s guidelines state that as a matter of corporate policy, Freddie Mac will not purchase any loan made that is supported by an appraisal report that makes reference to race or the racial composition of the neighborhood. Appraisal Guidelines: VA While VA generally does not provide specific procedures for conducting the appraisal other than requiring the use of the “Uniform Residential Appraisal Report,” the guidelines require the appraiser to use acctapt- able appraisal techniques and standards. VA'S guidelines also addwss other areas such as the following: . designation of appraisers, l procedures for requesting a determination of reasonable valutb. . procedures for assigning appraisers, l procedures for reviewing the appraisal report, l procedures for preparing certificates of reasonable value, l procedures for determining appraiser’s fees, and Page 32 GAO/RCED914 Secondary %l~cw-~mmtv Uarket Appendix II Underwriting Guidelines Used by Fannie Mae, Freddie Mac, HUD, and VA l VA'Sminimum property requirements for proposed and existing construction. Approaches to The four agencies suggest that appraisers use either the market or sales Determining Market Value comparison, cost, or income approaches to determine the value of a property. Fannie Mae and HUD place more emphasis on the market approach. Fannie Mae will not accept appraisals that rely solely on either the income or the cost approach. Freddie Mac does not specify that one method is preferred over another, but states that it does not rely heavily on the cost approach. VA guidelines indicate that the market approach will be used; however, in certain cases the cost or income approaches should be used as appropriate. The market or sales comparison approach uses the market price in determining the value of the property. The market price is the price at which a property may be currently bought or sold. Appraisers must determine the relationship between the market value and estimated market price through an analysis of all circumstances affecting the property and the transaction. Appraisers estimate the market value of a property by analyzing prices paid for similar properties. Appraisers con- sider the major characteristics of the similar properties and determine whether they add to or subtract from the value of the property. The cost approach involves valuing property as the sum of building repro- duction costs less depreciation plus the value of the land. The income approach involves looking at the actual return on investment from the subject property. Page 33 GAO/RCEDBl-2 Secoridary Mortgagr Market Appendix III Secondary Market Loan Activity in the Metropolitan Atlanta Area The summary data include loans purchased by Fannie Mae and Freddie Mac and loans insured by HUD for 80 residential ZIP code areas in a five- county metropolitan Atlanta area for the 2-year period ending June 30, 1989. As discussed in appendix I, we were unable to determine the rea- sons for the variations in loan activity among areas in Atlanta because of data limitations. Consequently, the data in this appendix should not be used to derive conclusions on discrimination in secondary market loan activity. Also, in the following discussion of loan data by popula- tion group and average income levels, we refer to the data for Fannie Mae, Freddie Mac, and HUD as secondary market loan activity. As previ- ously stated, VA could not provide loan data at the ZIP code level; there- fore, we presented the loan activity data for VA separately by race and county. The secondary mortgage market loan activity data were greatest in pre- dominately white (61-100 percent white) ZIP code areas and in ZIP code areas having average incomes between $35,000 and $74,999. Fannie Mae accounted for the majority of the loan activity (40 percent) while Freddie Mac and HUD accounted for 32 percent and 18 percent, respec- tively, during this period. The average loan amount purchased by Fannie Mae and Freddie Mac and insured by HUD was $77,093 and the median loan amount was $70,531. VA accounted for the smallest percentage (10 percent) of overall loan activity of the four organizations. Minority individuals received the greatest percentage of VA-guaranteed loans in Dekalb and Fulton C’oun- ties. However, in Cobb, Clayton, and Gwinnett Counties, white individ- uals received the greater percentage of VA loans. During the period September 1, 1988, through September 30, 1989. the average home price for the five-county Atlanta area was $12 1,772. The median home price was $93,000. An understanding of the demographic make-up of the metropolitan Atlanta Demographics Atlanta area is important to the assessment of.secondary market loan activity over race and income variables. Four of the five metropolitan Atlanta counties, Clayton, Cobb. [k~lialb, and Gwinnett, are predominately white (86, 92,68, and 97 wrc’tbnr white, respectively). Fulton County, which contains most of t htx VIt y of Atlanta, is 51 percent minority. The city of Atlanta, itself. is ti7 pbrc,ent minority. Page 34 GAO/RCED-91-2 !bcondarJ Morrgwr Market Appendix III Secondary Market Loan Activity in the Metropolitan Atlanta Area As discussed in appendix I of this report, we classified 80 ZIP code areas in the five-county metropolitan Atlanta area into five population groups. Most of the ZIP code areas were 81-100 percent white. (See table 111.1.) Table 111.1:Number of ZIP Codes by Population Group Population group Number of ZIP codes O-20 percent white 8 21-40 percent white 6 -.-- .- -___ 41-60 percent white 6 61-80 percent white 8 81-100 percent white 52 Total 80 We also classified the 80 ZIP code areas into four average income catego- ries. Most ZIP code areas had average incomes of $35,000-$49,999. (See table 111.2) Table 111.2:Number of ZIP Codes by Average Income Level Average income level Number of ZIP codes $7,500.$24,999 9 $25,000~$34,999 16 $35.000-$49,999 45 $50,000-$74,999 - 10 Total 80 We estimated that for 1989 the total population of homeowners in the 80 ZIPcode areas was 459,462. (See app. I for estimation methodology.) Most of the total population of homeowners (78 percent) live in the 60 predominately white ZIP code areas (61-100 percent white). Only 16 per- cent of the total population of homeowners live in the 14 predominately minority (O-40 percent white) ZIP code areas and about 6 percent live in the 6 integrated ZIP code areas (41-60 percent white). (See fig. III. 1.) Page 35 GAO/RCED91-2 Secondary Mortgage Market Appendix III Secondary Market Loan Activity in the Metropolitan Atlanta Area Figure 111.1:Percentage of Total Homeowners by Population Group I 81-100% White Source: Demographic data provided by DMIS and the 1980 Census The greatest percentage of the total population of homeowners (58 per- cent) live in ZIP code areas having average incomes of $35,000-$49,999. The second largest percentage of homeowners (about 21 percent) live in ZIP code areas having average incomes of $25,000-$34,999. Finally, the smallest percentage of total homeowners live in ZIP code areas having average incomes of $50,000-$74,999 and $7,500-$24,999 (about 15 per- cent and 6 percent, respectively). (See fig. 111.2.) Page 36 GAO/ECED91-2 Secondary Mortgage Market Appendix III Secondary Market LOan Activity in the Metropolitan Atlanta Area Figure 111.2:Percentage of Homeowners by Income Level $3!5,000-$49,999 Source. DemographIc data prowded by DMIS and the 1980 Census Loan Activity by one variable-the racial composition of the ZIP code areas, the data Population Group show that such activity was greater in predominately white (6 1- 100 per- cent white) ZIP code areas. The average and median loan amounts and home prices also increased as the percentage of white population increased in a ZIP code area. For the 80 ZIP code areas in the five metropolitan Atlanta counties. the number of loans per 100 homeowners was higher in the predominately white ZIP code areas than in the predominately minority or integrated ZIP code areas. (See fig. 111.3.)The number of loans per 100 homeowners ranged from 6.2 in the predominately minority (O-20 percent white) ZIP code areas to 8.3 in the relatively integrated (41-60 percent white minority) ZIP code areas. However, the 61-80 percent white ZIP code areas received 17.2 loans per 100 homeowners, and the 81-100 percent white ZIP code areas received 13.7 loans per 100 homeowners. Page 37 GAO/RCED-91-2 Secondary Monga~ Uarket Appendix Ill Secondary Market Loan Activity in the Metropolitan Atlanta Area Figure 111.3:Number of Loans per 100 Homeowners by Population Group 25 Laanspu1ooHolnoownor8 Populatia+l Grorp by Peranl whita Source: Loan data prowded by Fanme Mae, Freddie Mac, and HUD DemographIc data provtded by DMIS and the 1980 Census. Average and Median Loan The average and median loan amounts purchased or insured by Fannie Mae, Freddie Mac, and HUD increased as the percentage of white popula- Amounts by Population tion increased. (See fig. 111.4.)For example, the average loan amount Group increased about 76 percent (from $46,168 to $81,179) from the predomi- nately minority population groups (O-20 percent white) to the predomi- nately white population groups (81-100 percent white). The median loan amount increased 103 percent (from $38,763 to $78,762) over the same range. Page 38 GAO/RCED91-2 Secondary Mwtgaar(r Market Appendix III Secondary Market Loan Activity in the Metropolitan Atlanta Area Figure 111.4:Average and Median Loan Amounts by Population Group 90 Loan Amount In Thouaanda B Awrage Loan Amount -1-1 Medii Loan Amount Source: Loan amounts provided by Fanme Mae, Freddie Mac, and HUD. DemographIc data prowded by DMIS and the 1980 Census. Average and Median Home The average home price increased from $67,161 in the predominately minority (O-20 percent white) ZIP code areas to $132,485 in the 81-100 Prices by Population percent white ZIP code areas. The median home prices increased from Group $56,000 to $101,000 over the same range. (See fig. 111.5.) Page 39 GAO/RCED91-2 Secondary Mortgage Market Appendix Ill Secondary Market Loan Activity in the Metropolitan Atlanta Area Figure 111.5:Average and Median Home Prices by Population Group 140 DdlaminThouunds 130 120 110 100 60 60 70 60 50 ..-- &20% whit. 2140% whit. 41- WhHa 61-w% White El-100% Wh Papubtian Gnwp by Pwcanl Whit@ - Aim-age Hwne price ---- Median Home Price Source. Home price data provided by Dataman Inc. Demographic data provided by DMIS and tne 1980 Census When considering the patterns of secondary market loan activity over Loan Activity by only the ZIP code average income variable, the data show that such Income Level activity was greater in ZIP code areas having higher average incomes ($35,000-$74,999) than in those having lower average incomes (57,.500- $34,999). Average and median loan amounts and home prices also increased over the range from lower to higher average incomes. Total loan activity per 100 homeowners in ZIP code areas with the two highest income categories ($35,000-$49,999 and $50,000-$74.499$) was 1.8 times the total loan activity in those ZIP code areas with the tkvo lowest average income levels ($7,500~$24,999 and $25,000-53~.9!)9$. The ZIP codes in the $35,000-$49,999 and $50,000-$74,999 incomc groups each received 14.2 loans per 100 homeowners while ZIPcotir~sin the $7,500-$24,999 and $25,000~$34,999 income groups receivchcl3 8 and 8.2 loans per 100 homeowners, respectively. (See fig. 111.6.) Page 40 GAO/RCEB91-2 Secondary Mort~(agv Market Appendix III Secondary Market Loan Activity in the Metropolitan Atlanta Area Figure 111.6:Number of Loans per 100 Homeowners by Income Level 20 Loan8pu100Homoownon Source: Loan data prowded by Fannie Mae, Freddie Mac, and HUD. Demographic data prowded by DMIS and the 1980 Census. Average and Median Loan The average amount of the loans purchased by Fannie Mae and Freddie Mac and insured by HUD increased about 113 percent and the median Amount by Income Level loan amount increased by 161 percent as the average income of the LIP code areas increased. For example, the average loan amount was $44,746 in the ZIP code areas with average income levels of $7,500 to $24,999, and $95,274 in the areas with average incomes of $50.000 to $74,999. The median loan amount was $36,393 in the lowest income area and $95,138 in the highest income area. (See fig. 111.7.) Page 41 GAO/RCEB91-2 Secondary MorQww %rkN Appendix Ill Secondary Market Loan Activity in the Metropolitan Atlanta Area Figure 111.7:Average and Median Loan Amounts by Income Level 100 LomAmountlnlhoumnb - AverageLanAmount 1-1, MsdianLoanArnom Source: Loan amounts prowded by Fannie Mae, Freddie Mac, and HUD. Demographic data provided by DMIS and the 1980 Census. Average and Median Home The average home price increased from $67,119 in ZIP code areas with Price by Income Level the lowest income level ($7,500~$24,999) to $176,425 for those arcas with the highest incomes ($50,000-$74,999). The median home pnce increased from $53,000 to $146,000 over the same range. (See fig. 111.8.) Page 42 GAO/RCED91-2 !3econdaq Y~n~(u(r Market Appendix III Secondary Market Loan Activity in the Metropolitan Atlanta Area Figure 111.8:Average and Median Home Price by Income Level 190 Dollaninlhomada rm / 150 130 110 90 m !I0 lncomo Lovol - Average Home Price -1-1 MedianHornePrica Source: Home pnce data prowded by Dataman Inc. Demographic data provided by DMIS and the 1980 Census. Previously, we showed the patterns of secondary market loan activity Loan Activity by over the isolated variables of racial composition and average income of Income Level and the ZIP code areas. Here, we have presented the loan activity across the Population Group population groups at four average income levels. In general, loan activity fluctuated over population groups having the same average income. However, for ZIP code areas having an average income of $25,000-$34,999, the number of loa.ns per 100 homeowners increased as the percent of white population increased. Loan activity per 100 homeowners varied with no specific pattern over the population groups having average incomes of $7,500~$24,999, $35,000-$49,999, and $50,000-$74,999. Also, the number of ZIP wdc areas in the population groups/income levels varied from 0 to 37. ( SW table 111.3.) Page 43 GAO/RCEDSl-2 Secondary Mortgagr .Market Appendix IIl !hcondary Market Loan Activity in the Metropolitan Atlanta Area Table 111.3:Loan Activity and Number of ZIP Code8 by Population Group and Number of Loan activity per Average Income Level Population group income/level ZIP codes 100 homeowners O-20 Percent White: $7,500~$24,999 5 56 $25,000-$34,999 3 6.8 $35,000~$49,999 0 . $50,000-$74,999 0 . 21-40 Percent White: $7,500~$24,999 3 46 $25,000~$34,999 2 72 $35,000~$49,999 1 13.8 $50.000-$74.999 0 . 41-80 Percent White: $7,500~$24,999 1 - 18.4 $25,000~$34,999 3 79 $35,ooo-$49,999 2 8.3 $50,000-$74,999 0 . 81-80 Percent White: $7,500-$24,999 0 $25,000~$34,999 3 93 $35,000-$49,999 5 21.5 $5o,ooo-$74,999 0 . 81-100 Percent White: $7,500~$24,999 0 . $25,000~$34,999 5 96 $35,000-$49,999 37 142 $50.000-$74.999 10 142 Source: Loan activity data provided by Fannie Mae, Freddie Mac, and HUD. Number of ZIP codes from DMIS. Not all of the population groups contained each of the income levels. For example, only one income level, $25,000-$34,999, is common to each of the five population groups. For this income category, the number of loans per 100 homeowners increased as the percent of white population increased. Homeowners in the predominately white (81-100 ptlrcvnt white) ZIP code areas received 41 percent (or 2.8) more loans per 100 homeowners than in the predominately minority (O-20 percent iv bite) ZIP code areas at this income level. In addition, seven race/incomcl (xtrgo- ries for predominantly minority, middle- and upper-income mu ()f Atlanta show no activity, because our study did not contain an!. LIPcode areas that fell within these race/income categories. Consequthnr 1) this does not necessarily mean that secondary market agencies arcaIII IT Page 44 GAO/RcEDsl-2 Secondaq M~wtmgr ,Markct Appendix Ill Secondary Market Loan Activity in the Metropolitan Atlanta Area buying loans in these areas of Atlanta. (See fig. 111.9.The data presented in table III.3 form the basis for fig. 111.9.) Figure 111.9:Number of Loans per 100 Homeowners by Population Group 26 AvuagoLowmpu1ooHomoowmn and Income Level 20 1s 10 5 0 o%-2o%whilo 21lbauwhlto 41%60% whit. 619HO% Whtto BlXl#)w whna Porcallt whltr 1 1 $7,50&s24,992 $25,oow34,999 @5.ooo-s49,sss m s5o,ooo$74,9QQ Source: Loan data prowded by Fanme Mae, Freddie Mac, and HUD. Demographic data prowdea by DMIS and the 1980 Census. Minority individuals received more VA-guaranteed loans than did u’hlte VA Loan Activity individuals in Dekalb and Fulton Counties. In Clayton, Cobb, and Cwin- nett Counties, white individuals received more loans. The dollar ~.aluc of VA loans guaranteed follows a similar pattern. VA guaranteed its greatest number of loans in Dekalb and Cobb Co~lntres, respectively. In Cobb County i’O percent of VA’S total loan volume n’a~nt to white individuals. However, in Dekalb County minority indivltiu;lls received 64 percent of VA’S total loan volume. Minority individu;ll\ Page 45 GAO/RCED-91-2 Secondary Moml(ayr Warket Appendix III Secondary Market Loan Activity in the Metropolitan Atlanta Area received the greatest percentage of loan volume in Fulton County (71 percent) and the least in Gwinnett County (25 percent). (See table 111.4.) Of the five metropolitan Atlanta counties, only Fulton County has a pre- dominately minority population (51 percent minority). Table 111.4:Number of VA-Guaranteed Loans for White and Minority Individuals Loans for Individuals in Five Metropolitan Atlanta Counties Total White Minority Countv loans Number Percent Number Percent Clavton 1.183 605 51.1 578 48.9 Cobb 1,488 1,044 70.2 444 29.8 Dekalb 1.541 550 35.7 991 64.3 Fulton 933 272 29.2 661 70.8 Gwmnett 1,214 908 74.8 306 25.2 Total 6.359 3.379 53.1 2.980 46.9 Source. Loan data by race provided by VA VA also guaranteed its greatest dollar volume of loans in Cobb and Dekalb Counties. In Cobb County 70 percent of VA’S total dollar volume went to white ‘individuals. However, in Dekalb County minority individ- uals received 64 percent of VA’S total loan volume. Minority individuals received the greatest percentage of VA’S total dollar volume in Fulton County (65 percent). (See table 111.5.) Table 111.5:Dollar Value of VA- Guaranteed Loans for White and Minority Dollars in thousands Individuals in Five Metropolitan Atlanta Loans for Individuals Counties Total White Minority dollar Dollar Dollar County value value Percent value Percent Clavton $80,026 $40,374 50.5 $39,652 49.5 Cobb 122,118 86,000 70.4 36,117 29.6 Dekalb 116,827 43,850 37.5 72,978 62.5 Fulton 69,015 24,059 34.9 44956 65.1 Gwinnett 97,890 73,512 75.1 24,378 24.9 Total $485,876 $267,795 55.1 $218,081 44.9 Average $97,175 $53,559 $43,616 Source: Loan data by race provided by VA Page 46 GAO/RcEDsl-2 Secondary Mortgage Market Appendix IV . Cements From the Federal Home Loan Mortgage Corporation 1 September 6. 1990 Mr. John M. 01s Director, Housing and Community Development Issues U.S. General Accounting Office Warrhington, DC 20546 Dear Mr. 01s: Thank you for the opportunity to comment ou the GAO Report, kstr mv in the u.” As a major source of fundr for home loana nationwide. the Federal Home Loan Mortgage Corporation (“Freddie Mac”) acknowledges its leadership responsibilities in support of affordable housing opportunities and against discriminatory landing practicer. We are pleased that this report draws a clear distinction bet-en primary market lending and secondary market activity. As the report reflects. Freddie Mac doea not make mortgage loans directly to borrowers; rather, it purchases mortgages from lending institutions. The finding in the report that the number of loans purchased by rraddie Mac per 100 homeomercl increares with the percent of whites within an area’s population is consistent with the well-documented pattern of discrimination as reported in the Atlanta Constitution’s “The Color of Money” series. As a remit, ona would expect secondary market purchases to reflect primary market originations. We are also pleased that GAO’s report highlights attempts to assure that racial discrimination does not occur in the underwriting decision, as reflected in Freddie Mac’s underwriting guidelines. As GAO correctly points out, Freddie Mac guidelines will not permit the consideration of race of a borrower in any aspect of the loan underwriting process. In closing, I should mention that as part of our effort to easura that all potential homebuyers have equal access to credit, we recently created ao Affordable Aouaing Initiatives Department to help us, and the entire industry, design homeomerahip and rental programs to address this issue. Affordable housing pr0gram.s nay not directly address racial discrimination in lending practicer. However, we do hope that these programs -- which will combine Page 47 GAO/RCEKMlS Secondary Mortg~gr Yarkrt Appendix N Comments From the Federal Home Loan Mortgage Corporation John H. 01s September 6, 1990 Page 2 financial risk-sharing with more flexible undmwriting criteria -- will have the added benefit of improving homeomerehip opportunities for tboee who have traditionally had more difficulty in obtaininq credit. Sincerely. cszI%Adw Leland C. Brendsel Chairman and Chief Executive Officer LCB:tepr019lN Page 48 GAO/RCED91-2 Secondary Mortgage ,MMarket Appendix V comrrients From the Federal National Mortgage Association 3SlO Wimwia Avenue. NW William R. Malone Wnhingtan. DC 24015289g Ssniw Vie0 Reaiknt- mz 762 7130 Policy and Public ARkn x‘I4 4 FannieMae September 10, 1990 John M. 018, Jr. Director, Housing and Community Development Iesues United States General Accounting Office 441 G Street, NW Room 4073 Washington, DC 20540 Dear Wr. 016: Thank you for the opportunity to review and comment on your report to Senator Dixon concerning secondary market underwriting ntandardr, and mortgage purchaee patterns in Atlanta. I also appreciate the professionaliem exhibited by you and your staff and the receptiveness to input you exhibited throughout the conduct of this study. The major finding of the study, that secondary market purchasingfaecuritization activity declines with decreasing neighborhood income or increasing neighborhood minority composition, warrants concern from all mectora of the mortgage finance industry. At Fannie Mae, we bold a long-standing commitment to the homebuying credit needs of low- and moderate-income houaeholde and residents of inner city neighborhoods. This commitment has translated into a broad- bamed strategy to meet the credit needs OS those with limited incomes. Wany of our activities that particularly benefit inner-city and low-income neighborhoods are not covered in the study, much as our purchases of multifamily mortgagee and mortgage revenue bonds. These activities play an important role in providing decent, safe, and affordable housing for low- and moderate-income and minority households and residents of inner city neighborhoods. Pannie Wae ~EI continuing ita development of creative products and programs that are designed to address the credit needs of low- and moderate-income and minority communities. We are committed to the objective of providing an equitable distribution of home mortgage credit to such communities in Atlanta and in all metropolitan areas across the nation. As noted above, the GAO study raises serious questions and concerna. However, given the limitations acknowledged by GAO, it is important that the findings be placed in the proper perspective. As the report's conclusion points out, severe data limitations preclude any inferences concerning the causes Pmnk Mu 'lb USA'sHousing Partner Page49 GAO/RCEBBl-2SecondaryMortgageMarket Appendix V Commente From the Federal National Mortgage Association of the observed purchasing patterns. The study does not attempt to employ, nor will the data permit, a statistical analysis of the apparent relationship between secondary market purchasing activity and neighborhood racial composition or income. Though the numbers and graphs presented in the study euggeet a relationship between these factors, the data problems and the lack of statistical analysis prevent a conclusive finding to this end. As the report itself euggeete, the apparent relationship between race and loans per 100 homeowners could be a result of the hidden effects of factors not included in the study. The report mentions several of these missing factors, such ae individual borrower's (as opposed to neighborhood average) income and creditworthiness and the condition of neighborhood propertiee. Without including these factors and without conducting some form of statistical analysis to isolate the effect of the race variable, it is not possible to establish that a relationship does in fact exist between neighborhood racial composition and secondary market purchasing activity. If, for example, prospective home purchasers in minority neighborhoods have lower incomes, poorer credit histories, and lees wealth than prospective purchasers in white neighbor- hoods, then the minority neighborhoods will experience lower lending volumes than the white neighborhoods. Because a loan cannot be purchased or eecuritieed until it is made by a lender, a tendency for lower lending volumes in minority neighborhoods would translate into lower secondary market activities in these neighborhoods. A similar argument could be constructed regarding the apparent relationship between neighborhood average income and secondary market purchasing activity. The income measure used in the study is too imprecise, and the omitted factors (e.g., household wealth) potentially affecting lending activity are too numerous, to definitively conclude that secondary market activity declines with neighborhood income. Though numerous data limitations are discussed at some length in the study, these problems are severe enough to warrant reiteration. First, only 80 of the approximately 200 zip codes for which data were provided were included in the study. Though substantial difficulties inevitably arise in assembling a data set from several disparate sources, it would have been useful in assessing the study to know which ZIP codes were included and which were omitted. Without such information, it is difficult to know how the findings might have differed if alternative ZIP codes were analyzed. 2 Page 50 GAO/RCEDslI Secondary Mowge Market Appendix VI Major’Contributors to This Report Edward Kratzer, Assistant Director Resources, Robert Procaccini, Assistant Director Community, and Patrick Valentine, Assignment Manager Patrick Doerning, Senior Operational Research Assistant Economic Mitchell Karpman, Senior Operational Research AssiStant Development Division, Washington, D.C. Jerry Coffey, Evaluator-in-Charge Atlanta Regional Fannie Bivins, Site Supervisor Office Deborah Baker, Evaluator James Landers, Programmer Analyst (995194) Page 53 GAo/ltcEDw3 second8Iy MOM Market Appendi.xV CommentsFromtbeFederalNational MortgageAeeociation The studyts discussion of insufficient observations in some race/income categories is also worthy of repeating. This problem is particularly eevere when the data are analyzed by income and race simultaneously. The finding that secondary market activity decreases with increaningminority composition for the 925,000 to $34,999 income group is very difficult to eubstantiate becau8e there are only two or three neighborhoode in most of the racial categories within this income grouping. An examination of Figure III.9 exemplifies how problematic and potentially misleading this sparsity of observations can be. This figure shows zero loan8 per 100 homeowners for ZIP codes with O-20 percent white population and average incomes of $34,000-$74,999. A casual look at this figure might suggest that the necondary market is not buying any loans in the predominantly minority, middle- and upper-income neighborhoods of Atlanta. What is not apparent from Figure III.9 is that there are no ZIP codes includsd in the study which fall within these race/income categories. A failure to cross-reference Figure III.9 with Table III.3 may lead to the mistaken conclusion that the lack of loans is caused by ths failure of financial institutions to provide an adequate supply of credit. Another major concern which is expressed in the study is that the data set does not provide sufficient information by which to assess neighborhood variations in loan demand. Because loan demand is a critical determinant of the volume of loan originations and, ultimately, of loan purchasen, this data limitation prevents consideration of a major possible explanatory factor. Without information on loan demand across neighborhoods, it io not possible to determine whether low- income and minority neighborhoods received fewer loans because credit was not available or because there was less demand for loans in these neighborhoods. This data limitation prevents evaluation of a central issue: whether or not the need for mortgage credit is being fulfilled in low-income and minority communities. The study notes a number of other important limitations, which are outlined briefly below: 0 The data were assembled from several sources, with no attempts at verification of data validity. 0 The number of homeowners per ZIP code, which is used to calculate the number of loans per 100 homeowners, is calculated based on an 1989 estimate of total households and the 1980 Census count of renter households. It is possible that these figures do not accurately reflect the current residential makeup of the ZIP codes used in the 3 Page51 GAO/IUXD9l-2!ikcondaryMortgage Market Appendix V Comments From the Federal National Mortgage As.sociation study. Any errors in the estimates of owner households per ZIP code translate into inaccurate lending rates. 0 The income measure used in the study has several problems. First, average ZIP code income may be guitc different from the incomes of prospective mortgagors in a ZIP code. Because it is the income of the prospective mortgagor that is important in a lender's assee8ment of a loan application, this difference could substantially distort the income analysis in the study. Another problem with the income measure is that ZIP code average income can be skewed by a few unusually high or low household incomes and therefore may not accurately reflect the typical income for an area. 0 In the study, all FHA loans are assumed to be purchased by the Government National Mortgage Association, when in fact a small proportion are purchased by Fannie l4ae and Freddie Mac. This assumption leads to double-counting of a small number of FHA loans. The numerous difficulties mentioned above are indicative of the complexities of the problem which you have endeavored to study, and of the lack of readily available and reliable data. Over the coming months, new market information such as enhanced Home Mortgage Disclosure Act and 1990 Census data will be available. This information will enable businesses and regulators to better identify and address housing market discrimination. The efficient and effective allocation of credit to minority and low-income neighborhoods and communities deserves careful attention not only by GAO, but also by the housing industry as a whole. Sincerely, Page 52 GAO/RCEDSl-2 Secondary Murt+ge Market
Secondary Mortgage Market: Information on Underwriting and Home Loans in the Atlanta Area
Published by the Government Accountability Office on 1990-11-28.
Below is a raw (and likely hideous) rendition of the original report. (PDF)