Econ 321 Group Project EVIDENCE OF DISCRIMINATION IN MORTGAGE LENDING B Y H E L E N F. L A D D
Goals of Paper Show that discrimination models can prove that there is discrimination in mortgage lending Define Racial Discrimination Disagreements on how to measure discriminatory behavior Prejudice over profits Define possible areas of desperate treatment or adverse impact Service area chosen Advertising and Marketing Prescreening of Mortgage applicants Higher Interest Rates or less favorable Loan Rates
Discrimination Theories Taste Discrimination Put prejudice ahead of profits Boston Fed study Default rates (More Minorities default than Whites Statistical Discrimination Lenders find it cheaper to rate on race to estimate the applicants creditworthiness rather than the applicants history Types of Mortgages (FHA vs. conventional) Do mortgage lenders discriminate because of potential profits
Federal Reserve Bank of Boston Study T E S T I N G F O R T A S T E D I S C R I M I N A T I O N U S I N G R E G R E S S I O N M O D E L S
The Data Used All loan applications from minorities in the Boston area, plus randomly selected white applications for the year 1990 3000 applications, 700 from Black and Hispanic Researchers asked for 38 more points of data than the HMDA required. This encompassed all the information loan officers collected
Data (Cont.) Included Important variables not used in previous studies like: credit history, mortgage credit history, public records of default and bankruptcy history, level of employment instability.
Why this Data? The area was chosen mostly for convenience to the Fed Bank of Boston This is the most comprehensive data used in a study on lending discrimination and accounts for a lot of the shortfalls in other studies just using the HMDA data Only takes into account a small geographic area.
Results Before adjusting for control variables the loan denial rate for whites was 10%, while it was 28% for minorities. When personal property is controlled for the denial rate for whites was 10%, while the denial rate for minorities was 18% Results can only account for discrimination after the loan process has started, Lenders can still discriminate by choice of service area and by negotiating the amount of the loan down Could potentially understate the rate of discrimination
Loan Default Studies Look at default rates for mortgage loans given to black and white borrowers. Testing for taste discrimination, not statistical Berkovec et al. use FHA insured loans instead of conventional loans as in the Boston Fed Study in two studies of their own.
First Approach If blacks are discriminated against their default rate should be lower because lenders will set a higher bar for them. Single family mortgage loans between 1987 and 1989. Data shows blacks actually have a higher default rate.
Results Blacks have a default rate of 9.0%, whites have a default rate of 4.3%. This is statistically significant, even with statistical controls there is a 2% differential.
Problems Minorities are more likely to use FHA loans than conventional loans. Loan defaults measured by lender foreclosures. The data set does not allow for control of credit history.
Problems cont. Neither lender nor researcher can observe all characteristics related to creditworthiness. These unobserveables increase the likelihood of default, in opposition to discrimination. Only can account for taste based discrimination.
Second Approach Again test for taste based discrimination. Discrimination should be higher when lenders have greater market power. Look at the coefficient on interaction of race and concentration of the lending industry.
Results & Problems Based on a large sample size, this coefficient is not statistically significant. The unobservables are not a problem because they have nothing to do with market power. No taste discrimination, but still not sure about Statistical discrimination.
Why Are These Interesting? Lenders find them favorable because they are not discriminating They study the bad discrimination and show that it is not present.
Redlining GEOGRAPHIC-BASED EXACERBATE PROBLEMS OF POOR NEIGHBORHOODS COMMUNITY REINVESTMENT ACT OF 1977 HOME DISCLOSURE ACT (1989) BENSTON AND HORSKY (1991) DISCRIMINATION?
Additional Research and Policy Directions Question: Where should research and policy with regard to discrimination in mortgage lending go from here? -A U D I T S T U D I E S -C R E D I T S C O R I N G
Audit Studies Pair of identical people differing only in race Inquire about mortgage loans from specific lenders Advantages Provides clear evidence of discrimination Insight as to how loan applicants are treated during prescreening
Audit Studies 1989 Housing Discrimination Studies 2,000 audits were conducted of real estate brokers National sample of metropolitan areas Results Brokers were more willing to assist whites than minorities 13.3% of black auditors were assisted 24.4% of white auditors were assisted
Audit Concerns I: Larger tests like the one done in 1989 are needed, but can get very expensive. II: Auditors need to specify what house is being purchased III: Ethical and legal issues of using mortgage brokers as test subjects. IV: Lenders will try and verify auditors information V: Risk of lenders finding out they are being tested
Audit Conclusion Necessary in moving forward with the next step in regulation Provides clear evidence to discrimination, which detailed studies like the Boston Fed cannot point out Civil rights activists believe audit studies are the only way to provide irrefutable evidence Researchers should move forward with large scale audits
Credit Scoring Background George Galster 1996, Rebecca Carter 1996 Credit Scoring Models Predict riskiness of the loan Estimates relationship between borrower characteristics and the property Historically, evaluating applications for: Consumer loans Credit Cards Now being used in mortgage market Examples: Fannie Mae, Freddie Mac
Credit Scoring Benefits Reduce: Loan application processing costs Racial discrimination in mortgage lending process George Galster, 1996 Lenders never see applicants
Credit Scoring Concerns Models developed by private firms, making them proprietary Too broad in results, not enough information Models have built in standards» May have adverse impacts on minority borrowers» Riskiness of loan cannot be justified In this way, credit scoring might simply substitute discrimination in the form of adverse impacts for discrimination in the form of disparate treatment. (Ladd, 59)
Research Study Conclusions Controversy on answer to question stems from confusion on how to define discrimination By legal definition, studies indicate discrimination exists However limited data suggests: Lenders operate more on guesses from experience than concrete data More research needed to generate criteria for lenders to evaluate applicants Attention to individual differences Reduce lender pressure to discriminate