A Tale of Two Tensions: Balancing Access to Credit and Credit Risk in Mortgage Underwriting

Over the years 2000–2007, mortgage market underwriting conditions eased in response to public policy demands for increased homeownership. This easing of acceptable credit risk in order to accommodate increased access to credit, when coupled with the unanticipated house price declines during the Great Recession, resulted in substantial increases in delinquencies and foreclosures. The response to this mortgage market crisis led to myriad changes in the industry, including tightened underwriting standards and new market regulations. The result is a growing concern that credit standards are now too tight, restricting the recovery of the housing market. Faced with this history, policy analysts, regulators and industry participants have been forced to consider how best to balance the tension inherent in managing mortgage credit risk without unduly restricting access to credit. Our research is unique in providing explicit consideration of this trade‐off in the context of mortgage underwriting. Using recent mortgage market data, we explore whether modern automated underwriting systems (AUS) can be used to extend credit to borrowers responsibly, with a particular focus on target populations that include minorities and those with low and moderate incomes. We find that modern AUS do offer a potentially valuable tool for balancing the tensions of extending credit at acceptable risks, either by using scorecards that mix through‐the‐cycle and stress scorecard approaches or by adjusting the cutpoint—more relaxed cutpoints allow for higher levels of default while providing more access, tighter cutpoints accept fewer borrowers while allowing less credit risk.     

Differential Access to and Pricing of Home Mortgages: 2004 through 2009

This article documents trends and drivers of the residential mortgage market during the years 2004 through 2009, specifically focusing on the access to and pricing of mortgages originated by African‐American and Hispanic borrowers, and by borrowers living in low‐income and minority communities. Our analysis relies on a rich set of proprietary data that allow more expanded insights than can be obtained from the Home Mortgage Disclosure Act (HMDA) alone. We show that access to mortgage credit increased between 2004 and 2006 for the borrowers we focus on in our study and declined dramatically thereafter. Trends in access to credit were driven primarily by the changing credit mix of mortgage applicants and secondarily by the replacement of the Federal Housing Administration (FHA) for subprime as the dominant mode of nonprime originations and tighter underwriting standards. Throughout our entire period of study, these borrowers also consistently paid higher prices for their mortgages; however, the extent of this differential varied considerably over time and across groups. These pricing trends were driven primarily by changes in the FHA and subprime shares as well as by the market's increasingly aggressive pricing of credit risk.     

Reverse Mortgages: Contracting and Crossover Risk

A pricing model is developed for a reverse mortgage contract where the borrower receives payments either as a lump sum or in an annuity while the loan balance accumulates as a claim against the house. No underwriting criteria on income are applied. One risk of default is that the borrower will remain in the house after the negatively amortizing loan balance exceeds the value of the house. An explicit pricing model of the reverse mortgage permits the evaluation of this default "crossover" option. Alternative methods involving life insurance contracts and securitization are compared as secondary market channels.     

An Analysis of the Ex Ante Probabilities of Mortgage Prepayment and Default

Observed mortgage prepayment and default rates have been far different than the ruthless option exercise rates predicted by contingent claims models of mortgage pricing. The discrepancies have been attributed to both the competing-risk nature of prepayment and default and to transactions costs. This paper tries a different means of reconciliation. We introduce a third stochastic process, household income, into the usual pricing model that includes only the spot interest rate and the house price. The presence of income allows considering consumption-theoretic determinants of termination. The role of mortgage underwriting rules in restricting optimal prepayment is also explicitly modeled. Numerical ex ante prepayment and default rates based on the theoretical model come much closer to historical experience.     

Choice of Mortgage Contracts: Evidence from the Survey of Consumer Finances

This study revisits the empirical question of the determinants of the choice between fixed‐ and adjustable‐rate mortgages using data from the Survey of Consumer Finances that overcome some of the data limitations in previous studies. The results from a logit model of mortgage choice indicate that pricing variables and affordability are important considerations. We also find that factors, such as mobility expectations, income volatility and attitudes toward financial risk largely influence mortgage choice, with more risk‐averse borrowers preferring fixed‐rate mortgages. For households that are less risk averse, the mortgage type choice decision is less sensitive to pricing variables and income volatility, and affordability factors are not significant. These findings provide empirical support that underscores the importance of attitudes toward risks in mortgage choice.     

Automated Underwriting and the Profitability of Mortgage Securitization

This paper develops a game-theoretic model of mortgage securitization, which is then used to examine a potential effect of automated underwriting. The paper's primary supposition is that automated underwriting lowers the costs to competitive mortgage originators and a monopolist securitizer of identifying mortgage applicants who are good credit risks. Faced with lower underwriting costs, originators will screen a larger number of mortgage applicants in the hopes of holding more good risks in their portfolios and passing through more bad risks to the securitizer. This mounting adverse-selection problem causes the securitizer's expected revenues to decline; this effect can outweigh the cost-saving benefit of automated underwriting, causing the securitizer's return on equity to fall.     

The Cap Rate Spread: A New Metric for Commercial Underwriting

This study introduces the cap rate spread as a novel metric for underwriting commercial mortgages. Cap rate spread is the difference between the cap rate and the fixed coupon rate. The spread predicts performance risk in a sample of 24,951 commercial mortgage‐backed securities loans during 1993–2011. We demonstrate that the cap rate spread includes crucial information about performance risk. The results arise from the role of the cap rate spread in generating positive or negative leveraged returns to equity in situations where additional equity is required. Incorporating simplistic cap rate spread requirements in commercial underwriting is expected to reduce loan performance risk.     

Price Dispersion in Mortgage Markets

Using transaction‐level data on Canadian mortgage contracts, we document an increase in the average discount negotiated off the posted price and in rate dispersion. Our aim is to identify the beneficiaries of discounting and to test whether dispersion is caused by price discrimination. The standard explanation for dispersion in credit markets is risk‐based pricing. Our contracts are guaranteed by government‐backed insurance, so risk cannot be the main factor. We find that lenders set prices that reflect consumer bargaining leverage, not just costs. The presence of dispersion implies a lack of competition, but our results show this to be consumer specific.     

Commercial Mortgage Defaults: Proportional Hazards Estimation Using Individual Loan Histories

This paper examines the theory of commercial mortgage default and tests it using a data set of 2,899 loan histories provided by a major multi-line insurance company. A default model is estimated which relates subsequent default incidence and timing to contemporaneous loan term, borrower, property and economic/market conditions. Maximum likelihood estimation is used to estimate a hazard function predicting conditional probability of default over time. Results confirm many expected default relationships, in particular the dominance of loan terms and property value trends over time in affecting default. The effectiveness of the model in discriminating between "good" and "bad" loans is explored. Implications for underwriting practice and credit risk diversification are noted. Finally, suggestions are made for extending these results in pricing applications.     

Semivariance of Property Value Estimates as a Determinant of Default Risk

Standard practice in the residential mortgage underwriting industry is to estimate collateral values via independent appraisals conducted by third parties. This paper empirically examines the role of property value ( i.e. , appraisal) uncertainty as a determinant of default on residential mortgage loans. Based upon an analysis of 1,428 residential loans drawn from the portfolio of a national mortgage lender, we find evidence that semivariance in property value uncertainty is related to default risk. Specifically, subject properties that are valued above the sales price of recently sold "similar and proximate" properties show evidence of greater default risk. Interestingly, a variance (range) measure of property value uncertainty is not significantly related to default risk.     

Does Credit Scoring Produce a Disparate Impact?

The widespread use of credit scoring in the underwriting and pricing of mortgage and consumer credit has raised concerns that the use of these scores may unfairly disadvantage minority populations. A specific concern has been that the independent variables that comprise these models may have a disparate impact on these demographic groups. By “disparate impact” we mean that a variable's predictive power might arise not from its ability to predict future performance within any demographic group, but rather from acting as a surrogate for group membership. Using a unique source of data that combines a nationally representative sample of credit bureau records with demographic information from the Social Security Administration and a demographic information company, we examine the extent to which credit history scores may have such a disparate impact. Our examination yields no evidence of disparate impact by race (or ethnicity) or gender. However, we do find evidence of some limited disparate impact by age, in which the use of variables related to an individual's length of credit history appear to lower the credit scores of older individuals and increase them for the young.     

Handing Over the Keys: A Perspective on Mortgage Default Research

This paper is the text of the 1992 Presidential Address for the American Real Estate and Urban Economics Association. A comparative evaluation of mortgage default research finds that both the residential and commercial markets evolved from informal underwriting rules, to formalized (though unvalidated) ratios and rules of thumb, to early risk ratings based upon empirical evidence, to gener-alizable econometric models of default, to option-based pricing models. The commercial market lagged the residential market by about 10 to 20 years at first but is now only about five years behind. The survey finds that research and progress in understanding mortgage credit risk has been precipitated by a public policy need or mandate, data availability, and adequate technology. The absence of any one of these factors has hindered progress in the past. Finally, six emerging issues in default research are identified and discussed: (1) the degree of "ruth-lessness" with which default is exercised, (2) loan recourse, (3) the magnitude and timing of revenues and losses associated with default, (4) loan modification, (5) default in a portfolio context, and (6) leasehold default. Progress in these areas will enhance the efficiency of both the residential and commercial markets.     

Hedging Forward Mortgage Loan Commitments: The Option of Futures and a Future for Options

The purpose of this paper is to analyze theoretically the problem of how mortgage lending institutions assume the interest rate risks inherent in the granting of fixed-rate mortgage loan commitments.
Two approaches to hedge this risk are analyzed. First, the use of the GNMA futures market is evaluated from the standpoint of how it might be used to hedge against mortgage commitment risks. Secondly, the use of an appropriate pricing model—the Black-Scholes option pricing model—is offered as a proxy for establishing the market value of a fixed-rate mortgage commitment. This model is extended and empirically estimated for several hypothetical environments.
The paper demonstrates a basic flaw in the GNMA futures market as a hedge against mortgage commitments. Once this is established, the use of the options pricing approach is offered as a more rational approach for hedging these risks.     

Mortgage Brokers,Origination Fees,Price Transparency and Competition

This article examines the dynamics between mortgage broker competition, origination fees and price transparency. A reverse first‐price sealed‐bid auction model is used to motivate broker pricing behavior. Confirming the model predictions, our empirical analysis shows that increased mortgage brokerage competition at the Metropolitan Statistical Area level leads to lower fees. The findings are robust to different measures of fees as well as different measures of competition. We also provide evidence that broker competition reduces mortgage origination fees on retail (nonbrokered) loans as well. In addition, our results indicate that pricing complexity is an important determinant of fees, and increased broker competition is associated with a higher probability of a loan being priced with transparency. Our results suggest that mortgage brokers increase competition and lower fees in the mortgage market.     

Lenders’ pricing strategy: Do neighborhood risks matter?

This article explores the different pricing strategies of lenders who originate both government-sponsored enterprise (GSE) and non-GSE loans. We find that conditional on loan and borrower characteristics and some observable local economic factors, mortgage rates on GSE loans vary significantly across regions. However, we observe no sizable regional variation in loan amounts or default risk. By contrast, the mortgage rates on non-GSE loans depend almost entirely on borrowers and loan characteristics. In addition, we find that spatial variations in GSE mortgage rates are highly responsive to regional prepayment risk. Our results are robust to various controls for neighborhood characteristics, including regional-level bank competition, borrower accessibility to mortgages, and household income levels. Overall, the findings offer a novel insight into how lenders adjust pricing strategies in response to a changing lending environment. The results provide implications relating to the present and imminent dangers of housing bubbles and the intensified refinancing wave following the COVID-19 pandemic.     

Housing Finance in a Stochastic Economy: Contract Pricing and Choice

An empirical analysis of macroeconomic time series from the mortgage, housing, capital and labor markets is based on life-cycle consumption and mortgage option pricing considerations. Vector autoregression techniques characterize the long-run equilibrium and short-run dynamics of the mortgage market as it relates to the other sectoral markets. A simultaneous-equations model characterizes the partial equilibrium in the differentiated products market for fixed- and adjustablerate mortgage contracts. The empirical results reveal the impacts that market conditions have on mortgage volumes and prices, and they generally support the implications of the consumption and pricing theories.     

Are Investors the Bad Guys? Tenure and Neighborhood Stability in Chelsea,Massachusetts

In this article, we examine the role of investors and occupant‐owners in an urban context during the recent housing crisis. We focus on Chelsea, Massachusetts, because it is a dense city, dominated by multifamily housing structures with high rates of foreclosure for which we have particularly good data. We distinguish between occupant‐owners and investors using local data, and we find that many investors are misclassified as occupant‐owners in the Home Mortgage Disclosure Act data. Then, employing a competing risks framework to study ownerships during the period 1998 through mid‐2010, we find that local investors, who tend to invest more in relation to purchase prices and sell more quickly, experienced approximately 1.8 times the mortgage foreclosure risk of occupant‐owners, conditional on financing. Nonlocal investors have no statistically significant difference in foreclosure risk from occupant‐owners. Nonetheless, those owners with subprime purchase mortgages (most of whom are occupant‐owners) faced the highest foreclosure risk when house prices fell.     

Valuing Retail Shopping Center Lease Contracts

In this article we set up a real option model of retail shopping center leases. The model incorporates the effects of stochastic sales externalities and the possibility of tenant default, and in the presence of these effects, we derive and solve a partial differential equation that can be used to price a lease transaction. The model then sums across all tenants to determine the value of the shopping center. The model generates a number of new predictions, including why a Jorgensonian user cost of capital may overestimate shopping center values, why the general industry practice is to ignore percentage rent payments and tenant default risk in commercial mortgage underwriting and why shopping center owners may not act opportunistically as most observers seem to think they do.     

Adjustable-Rate Mortgages, Economic Fluctuations, and Lender Portfolio Change

In this study we examine the effects of economic fluctuations on the repayment behavior of a portfolio of adjustable-rate mortgages (ARMs). Because the U.S. experience with ARMs is quite recent, we have used data on a form of ARM used in Canada, the rollover mortgage. The results of our analysis suggest that use of ARMs similar to the rollover mortgage may reduce but not eliminate interest-rate risk for lenders, as borrowers, albeit constrained, prepay above-market-rate loans. In addition, we find that the periodic payment change inherent in the rollover mortgage does not lead to higher default rates and, therefore, credit risk.     

Evaluating Statistical Models of Mortgage Lending Discrimination: A Bank-Specific Analysis

We present our efforts to develop bank-specific models to test for the presence of mortgage lending discrimination. We discuss the potential for selection and simultaneity biases and delineate the conditions under which a single-equation model is appropriate. The results from three national banks demonstrate that, by incorporating the specific underwriting guidelines of each bank, our alternative approach significantly improves the ability of the model to explain the outcomes of the mortgage lending decision process when compared to a single generic specification applied across all banks. Our results also demonstrate the difficulties encountered in attempting to incorporate the specifics of a bank's underwriting criteria and the remaining potential for omitted-variables problems.