The Atlanta Journal-Constitution study of lending patterns tracked home-purchase and home-improvement loans made by every bank, savings and loan association, and large credit union in metro Atlanta from 1981 through 1986.
The newspapers used the federal Freedom of Information Act to obtain the information on computer tape from the federal government, which compiles it from lenders' reports. These institutions are required to report the location of each loan by census tract under the federal Home Mortgage Disclosure Act. The reports were compiled by the Federal Financial Institutions Examination Council, a federal agency.
The newspaper matched the lending data with demographic data from the 1980 U.S. census, updated with 1987 information from the Atlanta Regional Commission.
The study also analyzed federal reports prepared for bank examiners, real estate sales records, market studies and other data, and included interviews with bankers, homebuyers, regulators, real estate agents, appraisers, housing specialists, legal experts and neighborhood leaders.
All articles were researched and written by Bill Dedman, a Journal-Constitution staff writer. The project was supervised by Hyde Post, assistant city editor for special projects, and Dwight Morris, assistant managing editor for special projects. Articles were copy-edited by Sharon Bailey.
All statistical analysis was directed by the Journal-Constitution staff using methodology largely developed by Calvin Bradford and Charles Finn of the Hubert H. Humphrey Institute of Public Affairs at the University of Minnesota. These researchers, who together have nearly 20 years of experience in the field of community economic development, also performed computer work and assisted in statistical analysis. Research assistance was provided by Stan Fitterman, a graduate student in city planning at Georgia Tech.
The Journal-Constitution also adapted some methods and ranking systems used by researchers at Johns Hopkins University and Temple University.
The study was controlled to ensure that each census tract in the study was homogeneous with regard to race, and that only tracts of similar income were compared.
Regarding race, a neighborhood was not considered to be white or non-white unless 80 percent of its residents were in that group; all other areas were considered integrated.
Regarding income, the 332 census tracts in the metro area were separated into income groups, based on each tract's median household income as compared with the metropolitan area's median income. These groups were (1) tracts with median incomes above 122 percent of the area median, which were excluded because no non-white tracts had income above that level; (2) tracts below 70 percent of the area median, which were excluded because many of these households would not be able to afford homeownership; and (3) the remaining "middle-income" tracts, which were further separated into three groups of "lower-middle," "middle-middle" and "upper-middle" tracts. (The metro area's median income at the time of the 1980 census was $18,355.)
In addition, certain neighborhoods were excluded to ensure comparability and so that any statistical bias actually would underestimate the differences between lending to white and black areas.
For example, an area that experienced a great deal of new construction naturally would receive more home-purchase loans than a stable neighborhood. Because most high-growth areas in metro Atlanta are predominantly white, including them would have overstated the discrepancy between lending to whites and blacks. Overall, Fulton and DeKalb counties, where all the predominantly black census tracts in the metro area are located, grew by 12 percent in single-family housing from the 1980 census to 1987, according to the Atlanta Regional Commission. To be conservative, any census tract in the metro area that grew by 10 percent or more during that time was excluded from the study.
The study also excluded six tracts that had shown a loss in single-family housing (most of them around Hartsfield International Airport); 22 tracts with fewer than 500 owner-occupied housing units to begin with; and, because their rates of growth or decline could not be determined, 33 tracts outside the seven-county Atlanta Regional Commission area.
After these deletions, the study focused on 64 middle-income tracts: 39 white, 14 black and 11 integrated.
All lending comparisons were based on an estimate from census data of the number of one-to four-family structures eligible for mortgages in each census tract. This estimate was achieved by adding the number of households in free-standing single-family homes or condominiums, half the number of households in duplexes, and the number of households in three- and four-unit structures divided by 3.5.
The Home Mortgage Disclosure Act data that form the basis of the study have limitations.
First, the law requires financial institutions to disclose to the public only the location of actual borrowers. No figures on the number of applicants or the percentage approved are available by race or location. Information about rejection rates of blacks and whites is available to federal regulators, but not to the public.
The Journal-Constitution asked each institution to volunteer information about applicants. Only two, Georgia Federal Bank and Fulton Federal Savings and Loan, the two largest savings institutions in the state, provided the information.
Second, the study could not include every home loan in the metro area, because not all lenders are required by federal law to report their loans. Only banks, savings and loans, and large credit unions are covered by the law.
Therefore the figures leave out loans made by separate mortgage companies owned by banks, unaffiliated mortgage companies, life insurance companies and individuals, which altogether make more than half the home-purchase loans. Congress recently expanded the law to require reporting, beginning with this year's loans, by these bank-owned mortgage companies.
Because most banks did not supply these data from subsidiaries and because banks traditionally lend less money on real estate than savings and loans, institutions were not ranked by their volume of lending to minority and lower-income neighborhoods. One ranking was based on comparisons of lending to black and white neighborhoods of comparable income, and the other on the portion of a bank's loans that go to black or lower-income neighborhoods.
To offset the lack of data on applications and on the subsidiary companies, the Journal-Constitution analyzed real estate records for 23 neighborhoods, including 16 of the 64 middle-income neighborhoods, for 1986. These records, compiled by Real Estate Data Inc. and made available by the DeKalb Board of Realtors, identify the seller, buyer, lender, type of loan, loan amount and other information.
In each case these real estate records showed at least as great a disparity between white and black lending as the federal loan data.
Although the study could not include all home loans, it did include 82,610 home-purchase loans and 26,721 home-improvement loans for a total of $6,241,374,000 in lending by banks and savings and loans over six years.
The people who worked on this series:
"The Color of Money" was researched and written over a period of five months by Journal-Constitution staff writer Bill Dedman. A native of Chattanooga, Tenn., Dedman, 27, joined the Journal-Constitution in 1987.
The project was supervised by Hyde Post, assistant city editor for special projects, and copyedited by Sharon Bailey. Dwight Morris, assistant managing editor for special projects, supervised the analysis of lending data.
The statistical analysis used methodology largely developed by Calvin Bradford and Charles Finn of the Hubert H. Humphrey Institute of Public Affairs at the University of Minnesota. These researchers together have nearly 20 years of experience in the field of community economic development. Research assistance was provided by Stan Fitterman, a graduate student in city planning at Georgia Tech.
The Journal-Constitution also adapted some methods and ranking systems used by researchers at Johns Hopkins University and Temple University.Go to the next article or back to the Color of Money index or Power Reporting
Reprinted with permission from The Atlanta Journal and The Atlanta Constitution. Further reproduction, retransmission or distribution of these materials without the prior written consent of The Atlanta Journal and The Atlanta Constitution, and any copyright holder identified in the material's copyright notice, is prohibited.
Please send comments and story ideas to Bill Dedman, Bill@PowerReporting.com
Home page: Power Reporting