Discrimination and Mortgage Lending in Boston: The Effects of Model Uncertainty

In 1992 the Federal Reserve Bank of Boston conducted an analysis that examined the effects of race on mortgage lending in the Boston Metropolitan Statistical Area. Collecting data on all the possibly relevant information used in the lending process, they find when controlling for a subset of this information that race has a statistically significant effect on the decision to reject a mortgage application. Other researchers, using the same dataset, have shown that analysis of alternative subsets of the variables significantly reduces the effects of race. While theory should guide variable selection, there is often no unique theory to explain social science. In such cases, uncertainty in model specification causes one to be uncertain as to the true effects of the variables of interest. This paper accounts for the effects of model uncertainty by using Bayesian model averaging and finds a reduced effect of race and weakened evidence concerning the statistical significance of the effect.     

Are Minorities or Minority Neighborhoods More Likely to Get Low Appraisals?

We empirically examine the role of appraisal in the residential mortgage lending process—in particular, the incidence, consequences, and determinants of appraisal below contract purchase price. Using the Boston Federal Reserve Study data set, we find that, as expected, low appraised value significantly increases the probability of mortgage loan application rejection. We find no evidence that low appraised value is related to census tract racial composition, an important finding given the history of the appraisal industry; however, low appraised value is related to proxies for neighborhood quality. Moreover, properties securing adjustable rate mortgages, condominiums, and properties purchased by African American buyers show an increased probability of low appraisal, though the race effect result is highly sensitive to model specification.     

Under-specified Models and Detection of Discrimination: A Case Study of Mortgage Lending

This study examines how omitted variables affect underwriting models the OCC estimates during fair lending examinations. The purpose is to assess the effects of omitted variable bias common to most studies of discrimination in mortgage lending. The results show omitted variables have an important impact on both the estimate of the effect of race and on the identification of outliers for review. Further, there appears to be no consistent patterns to the direction of these impacts. This suggests that it is inappropriate to make generalizations concerning the direction of bias based on assumptions about correlations between omitted variables and race.     

Borrower Self-Selection,Underwriting Costs,and Subprime Mortgage Credit Supply

In the U.S., households participate in two very different types of credit markets. Personal lending is characterized by continuous risk-based pricing in which lenders offer households a continuous distribution of borrowing possibilities based on estimates of their creditworthiness. This contrasts sharply with mortgage markets where lenders specialize in specific risk categories of borrowers and mortgage supply is stepwise linear. The contrast between continuous lending for personal loans and discrete lending by specialized lenders for mortgage credit has led to concerns regarding the efficiency and equity of mortgage lending. This paper sheds both theoretical and empirical light on the differences in the two credit markets. The theory section demonstrates why, in a perfectly competitive credit market where all lenders have the same underwriting technology, mortgage credit supply curves are stepwise linear and lenders specialize in prime or subprime lending. The empirical section then provides evidence that borrowers are being effectively sorted based on risk characteristics by the market.     

Mortgage Lending,Default, and the Community Reinvestment Act

The Community Reinvestment Act (CRA) encourages banks to lend to low- and moderate-income individuals. This paper estimates the effect of the CRA on mortgage lending, exploiting variation in the set of banks whose lending performance is assessed in a given neighborhood due to redefinitions of Metropolitan Statistical Areas in 2003. Incorporating a typical tract into one additional banks' assessment area increased mortgage lending there by approximately 2%. Lending to low-income borrowers was particularly affected. While income-conditional default risk was little changed, CRA-induced loans were riskier than average, due to their borrowers' lower incomes.     

Credit History and the Performance of Prime and Nonprime Mortgages

Although nonprime lending has experienced steady or even explosive growth over the last decade very little is known about the performance characteristics of these mortgages. Using data from national secondary market institutions, this paper estimates a competing risks proportional hazard model, which includes unobserved heterogeneity. The analysis examines the performance of 30-year fixed rate owner occupied home purchase mortgages from February 1995 to the end of 1999 and compares nonprime and prime loan default and prepayment behavior. Nonprime loans are identified by mortgage interest rates that are substantially higher than the prevailing prime rate. Results indicate that nonprime mortgages differ significantly from prime mortgages: they have different risk characteristics at origination; they default at elevated levels; and they respond differently to the incentives to prepay and default. For instance, nonprime mortgages are less responsive to how much the option to call the mortgage or refinance is in the money and this effect is magnified for mortgages with low credit scores. Tests also reveal that default rates are less responsive to homeowner equity when credit scores are included in the specification.     

Do Black-Owned Banks Discriminate against Black Borrowers?

Black-owned and white-owned banks are studied to test for lending discrimination based on applicant's race. Using single-equation models the rndings suggest that black-owned banks may utilize applicant race in the mortgage granting decision. The result holds in a naive HMDA model and in an enhanced HMDA model with bank-specirc, demographic, and neighborhood characteristics added.     

Simultaneity bias in mortgage lending: A test of simultaneous equations models on bank-specific data

This study uses simultaneous equations models and single-equation models to test for simultaneity bias in mortgage refinance data compiled by a regional bank. The purpose of the study is to assess the claim that single-equation models of the lending decision produce biased and inconsistent parameter estimates of endogenous mortgage terms. Bank-specific data are analyzed to avoid bias resulting from uncontrolled policy, training, or underwriting differences across banks. Importantly, the data contain all variables the regional bank identified as important factors in explaining its loan disposition results. After controlling for applicants' debt, income, credit history, and requested loan term, I find that the race coefficient in single-equation models is biased upward, while the loan-to-value ratio coefficient is biased downward, although both biases are insignificant. Overall, the results suggest that simultaneous equations models are preferable to single-equation models in tests for discrimination, and can be used to determine the extent of race coefficient and loan-to-value ratio coefficient bias in single-equation models.     

Beyond the Transaction: Banks and Mortgage Default of Low‐Income Homebuyers

We evaluate the effects of the lending institution and soft information on mortgage loan performance for low‐income homebuyers. We find that even after controlling for the propensity of a borrower to get a loan from a local bank based on observable characteristics, those who receive a loan from a local bank branch are significantly less likely to become delinquent or default than other bank or nonbank borrowers, consistent with an unobserved information effect. These effects are most pronounced for loans originated to borrowers with marginal credit, where soft information may have a stronger effect. These findings support previous research on information‐driven lending, and provide additional explanation for observed differences in mortgage loan performance between bank and nonbank lenders.     

Banks’ lending behavior and monetary policy: evidence from Sweden

The main purpose of this paper is to investigate the aggregate data about bank loans which may hide significant information about the monetary transmission mechanism. This study, by disaggregating bank loans data and using the relevant interest rates in Sweden, investigates the behaviour of banks after a monetary policy tightening. By using an unrestricted VAR model and impulse response analysis, our results show that a shock on the policy rate affects the main components of the banks’ loan portfolios differently. Initially, banks do not reduce lending to firms and households and they present a sluggish reaction concerning the relevant interest rates. On the contrary, they reduce lending to mortgage credit institutions significantly since real estate lending can be considered as a risky long-term investment. Moreover mortgage credit institutions reduce lending for housing purposes to non-bank public. This reduction is mainly driven by flexible rate loans and loans secured on tenant owned apartments. Consequently, theses actions have a significant effect on real economic activity, by amplifying the initial shock from the tightening monetary policy. The latter result provides evidence of the bank lending channel in Sweden working via mortgage lending and could be very important for policy makers.     

Risk versus Demographics in Subprime Mortgage Lending: Evidence from Three Connecticut Cities

This paper analyzes patterns in subprime residential mortgage lending using 2006 Home Mortgage Act Disclosure data for the cities of Bridgeport, New Haven and Waterbury, Connecticut. The analysis applies models presented in earlier research and has the objective of assessing the relative importance of demographic versus risk factors in subprime mortgage lending decisions. Regression equations are estimated for census tracts and individual borrowers and include demographic variables and property risk measures. The results find race and ethnicity to be significant determinants of subprime lending in the borrower equations that include the full set of risk measures. Neighborhood educational levels are found to have an inverse and often significant association with subprime mortgage loans. Property risk measures present mixed results regarding their significance in subprime lending, suggesting that risk may have played less of a role in loan originations in 2006 than it did in earlier studies.     

Economic Consequences of Transparency Regulation: Evidence from Bank Mortgage Lending

We examine the economic consequences of a rule designed to improve consumers' understanding of mortgage information. The 2015 TILA-RESPA Integrated Disclosures rule (TRID) simplifies the mortgage disclosures provided to consumers. As a consequence, TRID-affected mortgages become a less attractive investment opportunity to banks. Our main results document that mortgage applications affected by TRID are less likely to be approved following the rule's effective date. We find evidence consistent with both a decrease in consumers' information processing costs and an increase in banks' secondary market frictions, providing insight into the potential channels through which this reduction in mortgage credit operates. We also find that banks partially compensate for reduced mortgage lending by increasing small business lending, and that fintechs absorb mortgage demand in areas with reduced mortgage lending by banks. Our study documents real actions that firms take in response to disclosure transparency regulation and contributes to the literature on the economic consequences of such regulation.     

A Disequilibrium Model of Savings and Loan Associations

This paper discusses the consistent specification and estimation of asset demand equations in a disequilibrium model of financial markets. We estimate the effective asset demands of savings and loan associations, allowing for rationing in the mortgage market. These disequilibrium estimates are not very different from the estimates of notional demands with no rationing assumed. Savings and loans seem to be least affected by excess demand situations in that they are apparently not reluctant to raise mortgage rates and/or to ration borrowers.     

Current Wealth,Housing Purchase,and Private Housing Loan Demand in Japan

Japanese households accumulate wealth for down payments at a high rate. Therefore, current wealth plays an important role in home acquisition as well as public loans whose direct mortgage lending is a strong support for home purchasers. We estimate the wealth effect on private mortgage debt as well as housing consumption by applying a model where mortgage-debt demand is derived from house-purchase decisions and is determined jointly with housing consumption. We use a simultaneous equation Tobit estimation method. Wealth effects on private mortgage debt, likelihood of borrowing, and housing consumption are not elastic. On the other hand, a change in housing consumption affects the likelihood of borrowing elastically much more than the private mortgage amount of borrowers. Housing and private mortgage markets fluctuate very closely with the number of participants in the mortgage market. Therefore, the number of housing starts is linked strongly to the private mortgage market.     

Disparities in Mortgage Lending,Bank Performance,Economic Influence,and Regulatory Oversight

This study investigates factors affecting changes in the disparity of home mortgage denial rates between white and minority loan applicants in the U.S. during the period 1991–1997. We develop a two-stage least-squares regression model that incorporates applicant-level characteristics, neighborhood characteristics, regional economic data, and bank-specific data as explanatory variables. Some have argued that mortgage lenders were under increasing pressure from industry regulators to extend additional credit to minorities and low-income groups during the period under study. The model includes each institution's periodic CRA rating as a proxy for regulatory influence. An alternative explanation is that market forces, such as improvements in economic conditions and in bank financial condition and performance, affected default loss estimates and credit standards in a way that disproportionally benefited minority and low-income applicants. The empirical findings are consistent with the latter hypothesis. We conclude that policy makers should consider the impact of market factors when assessing the allocation of mortgage credit in a particular demographic market. The findings also underscore the importance of controlling for lender assessments of credit risk when evaluating compliance with CRA and fair lending statutes.     

Introduction: New directions in Information and Screening in Real Estate Finance

Conclusion The evidence presented in this special issue supports the view that preexisting information held by mortgage lenders plays an important role in mortgage approvals. This argues for mortgage lending programs that make efficient use of lender information, and it supports the importance of local financial intermediaries for lending. It also suggests that mortgage finance is an important element in turning transitory shocks into persistent ones that shape macroeconomic and regional business cycles.This Journal has from its inception, with the publication of a seminal article on information and incentives on mortgage contract terms by Dunn and Spart (1988), pushed forward the frontier of knowledge on information issues in real estate finance. This special issue presents empirical evidence on the importance of this aspect of mortgages.     

Regulation of Subprime Mortgage Products: An Analysis of North Carolina's Predatory Lending Law

This paper estimates the effect of North Carolina's high-cost mortgage law on the subprime mortgage market in that state. The results indicate that creditors sharply restricted lending to higher risk consumers in North Carolina following passage of the law. Creditors did not restrict lending in neighboring states or to lower risk consumers in North Carolina. These results suggest that the restriction in North Carolina was due to rationing in response to higher costs imposed by the law. The findings of this study are of importance beyond North Carolina. Other states and municipalities have proposed or passed similar or more restrictive laws. These laws risk taking back some of the gains in credit availability that lower income and higher risk consumers gained in the 1990s.     

Political influence and banks: Evidence from mortgage lending

We show that banks expand mortgage lending in the home states of Senate Banking Committee chairs, and the effect is more pronounced in counties where the incumbent senator faces a competitive re-election race. Banks strategically target politically active borrowers. Consequently, banks’ profitability increases after favoring the incumbent politicians’ constituents, but they suffer a deterioration in mortgage asset quality in the long run. Our findings imply that political power could distort private capital allocation beyond conventional political contribution channels.     

A Comparison of Unexplained Racial Disparities in Bank-Level and Market-Level Models of Mortgage Lending

Statistical models of mortgage lending have been used by both academics and regulators to assess the importance of racial discrimination in lending decisions. Models estimated by bank regulators are specified at a bank level, allowing regulators to focus on possible disparate treatment discrimination against minorities. In contrast, the academic literature has tended to estimate models combining data across many banks. We argue that the market-level approach uses a mis-specified model whose estimates do not clearly measure any well-defined concept of discrimination. Using data from eight banks, we find important differences between bank-level and market-level models, with market-level models producing larger estimated racial effects than bank-level models.     

The Economics of Low-Income Mortgage Lending

The presumption that mortgage markets for low-income borrowers and neighborhoods are underserved by lenders has led to a variety of increased government interventions on the supply side of the housing market. Although many studies of low-income lending at the neighborhood level have been published, none is from the firm's perspective. We adopt such a framework to test the twin propositions that the low-income mortgage market is no different from the non-low-income mortgage market and that the low-income mortgage market is underserved.We examine empirically whether the operating costs including credit losses, revenues, and profits of savings and loan institutions engaged in more low-income lending differ systematically from those that do less low-income lending. We find that firms engaged in more low-income mortgage lending have higher costs than those engaged in less low-income lending, which is consistent with higher credit risk for low-income loans. Nevertheless, these firms are no more profitable than those that do less low-income lending, which is inconsistent with a market for low-income mortgage lending that is currently underserved.