Lessons Learned: Statistical Techniques and Fair Lending (original) (raw)
Related papers
A Reconsideration of Discrimination in Mortgage Underwriting with Data from a National Mortgage Bank
Discrimination in Financial Services, 1997
This paper, analyzing over 12,000 conventional and FHA/VA loan applications to a national mortgage lender in the 1989±1990 period, argues that mortgage denials occur only in a minority of cases, where the borrower has not learned the lender's underwriting rules in advance. Widespread borrower foreknowledge of such rules is demonstrated by a discriminant ®nding that 9 of 10 borrowers``correctly'' choose whether to apply under FHA vs. conventional programs, based on ®nancial and equity characteristics. This contrasts with the far lower ability of econometric models to identify approval/denial outcomes. It is revealing that denials on the basis of credit problems, the only important information generally not available until post application, account for most racial/ ethnic differences and borrower education affects the probability of approval of government insured loans more than loan to value. Contrary to common assumptions, race differences in FHA/VA lending are at least as pronounced as in conventional lending; and outcomes for Asians, correctly measured, diverge as much from outcomes for whites, as do outcomes for Hispanics and African American.
Discrimination in the Credit and Housing Markets: Findings and Challenges
Handbook on the Economics of Discrimination, 2006
Economics all too seldom provides straightforward guidelines for designing and analyzing statistical materials on subjects of great social importance. Since the economic theory of discrimination does provide a simple approach, it is too bad that studies of whether banks discriminate in mortgage lending have not utilized these insights.-Gary Becker (1993, P. 18) Eventually, even the definition of discrimination comes to mean different things to blacks and whites.-Derrick Bell (1980, P. 658)
Anatomy of a Fair-Lending Exam: The Uses and Limitations of Statistics
SSRN Electronic Journal, 2000
In this paper, we consider the role of statistical analysis in fair-lending compliance examinations. We present a case study of an actual fair-lending examination of a large mortgage lender, demonstrating how statistical techniques can be a valuable tool in focusing examiner efforts to either uncover illegal discrimination or exonerate an institution so accused. Importantly, our case also highlights the limitations of such statistical techniques. The study suggests that statistical analysis combined with comparative file review offer a balanced and thorough approach to enforcement of fair-lending laws. * The views stated here are those of the authors, and do not necessarily reflect those of the Board of Governors of the Federal Reserve System. We thank Robert Avery, Raphael Bostic, Glenn Canner, and Anthony Yezer for helpful comments.
Policy Issues Concerning Racial and Ethnic Differences in Home Loan Rejection Rates
Policy-making institutions associated with the mortgage lending market are facing increased pressure to respond to studies demonstrating wide disparity in loan rejection rates between white and minority applicants. Because the disparity is commonly attributed to discrimination, calls are going out to regulators and policy makers to address the issue. Unfortunately, not all the causes of the disparity are easily detected or fully understood, so policy makers must compare options without complete information. Despite the potential for imposing substantial costs on the market by implementing ill-considered and overly restrictive interventions, government inaction could lead to large costs in efficiency and equity if there is widespread discrimination in the market.
SSRN Electronic Journal, 2000
Over the past twenty years, understanding of and business practice in mortgage markets has been influenced significantly by the application of statistical models. Mortgage underwriting was automated using statistical models of default and default loss, and statistical models of denial rates and loan pricing were used to test for discrimination in lending. Efforts to measure mortgage market discrimination and credit risk have been propelled by an increase in the loan-level data available through various resources. Unfortunately, as researchers strived to produce results from these data, critical statistical errors were overlooked and then repeated in what has become the "conventional approach" to measuring discrimination and credit risk. The purpose of this paper is to reexamine the fundamental assumptions integrated into this conventional model and provide insight into why the results are both biased and inaccurate. This study will argue that conventional statistical models of discrimination and mortgage credit lack a sound basis in economic theory and rely on unrealistic and demonstrably false assumptions. As a result of these shortcomings, discrimination tests tend to produce false-positive indications of discrimination where none exists, and tests for default risk fail to predict instances where default rates are likely to rise significantly. A common theme underlies this essay: the mortgage lending transaction is extremely complex and involves many dimensions. Applicants, loan officers, underwriters and secondary market participants make decisions based on simultaneous consideration of many factors about which both the applicant and the lender must come to some mutual agreement. Applicants choose among mortgage lenders, products and terms based on their personal circumstances, with higher risk applicants self-selecting into loan programs with higher mortgage rates and higher rejection and default rates. These higher rejection and default rates are due to their self-selection into particular loan programs, not to differential treatment by lenders. The problem with conventional statistical techniques for estimating mortgage discrimination and credit risk is that these methods assume that borrowers never consider the effects of their decisions on the mortgage transaction. You do not need to be an economist to understand that mortgage applicants behave strategically when choosing mortgage products.
Estimation and Evaluation of Loan Discrimination: An Informational Approach
Many recent studies have analyzed whether lending discrimination exists. In all previous studies, the researcher faces constraints with the available data or modeling problems. In this article, we use a new informational-based approach for evaluating loan discrimination. Given limited and noisy data, we develop a framework for estimating and evaluating discrimination in mortgage lending. This new informational-based approach performs well even when the data are limited or ill conditioned, or when the covariates are highly correlated. Because most data sets collected by bank examiners or banks suffer from some or all of these data problems, the more traditional estimation methods may fail to provide stable and efficient estimates.
Anticipated Discrimination in the Home Lending Market
Housing and Society, 1981
Racial discrimination in the housing market has been a topic ofresearch for at least three decades. The present study extends this line of research by exploring how a newly considered factor, anticipated discrimination in the home lending market, differs by race and affects the probability ofowning a home. Using data from a sample ofhouseholds in Memphis, Tennessee, we show that significant differences in anticipated discrimination by race exist and that, certeris paribus, these expectations help to explain racial differences in home ownership. Combining our results with the findings of two homeownership preference studies, the implication is that anticipated discrimination reduces the demand for homes by creating a situation where blacks are less likely to apply for home loans because they feel that credit will be denied.