skip navigation

Saturday, December 20, 2014

[ – ] Text Size [ + ]  |  Print Page

Cascade: No. 56, Winter 2004

Foreclosures: Factors That Influence Default and Loss Rates

Homeownership can bring great joy and the potential for financial gain. If not dealt with properly, it can also lead to heartache and dire monetary consequences. Owning a home entails maintaining it and fulfilling financial obligations. While not doing the former might lower the home’s market value, failure to do the latter could result in foreclosure. The increase in the number of mortgage foreclosures is not only a concern of homeowners but also the lenders who supply their mortgages. The rise in the number of foreclosures across the nation has raised our awareness of the devastation that foreclosure creates and challenged our understanding of the underlying causes.

While academics and practitioners have studied foreclosures, they have focused primarily on default behavior and the factors that contribute to it. These studies have given lenders in both the conventional and FHA sectors valuable information for assessing risk when evaluating prospective loans. Moreover, FHA mortgage scoring relies solely on default probability for determining risk.

In a recent study for the U.S. Department of Housing and Urban Development’s Office of Policy Development and Research,1 Robert F. Cotterman points out that while knowledge of default probabilities is useful, previous studies have given little if any attention to the dollar value of losses and its determinants. He argues that underwriting guidelines and loss mitigation strategies should also consider loss severity along with default probabilities. Cotterman addresses this imbalance by developing estimates of the effect of location factors and borrower characteristics on loss severity.

Why is it important to consider loss severity when evaluating risk? Cotterman notes that one reason is the possibility that factors influence the probability of default differently than they influence loss severity. A better understanding could lead to improved assessments of portfolio risk.

Another rationale is that loss severity is related to race differently than the probability of default. This is of particular interest since minorities are subject to relatively more factors that lead to default than non-minorities. As a result, mortgage scoring systems based solely on default probabilities tend to assign less favorable scores to minorities. Those concerned with this racial disparity in scoring outcomes suggest that when minorities default, they tend, on average, to generate smaller dollar losses. Consequently, a mortgage scoring system based on dollar losses instead of solely on default might improve the fate of minorities.

Data and Methodology

Cotterman uses data on FHA-insured loans from 1992, 1994, and 1996 to investigate the factors that influence both default probabilities and dollar loss rates, as well as the manner in which such influence arises. He uses different levels of analysis, ranging from simple statistical summaries and descriptive regressions to more sophisticated statistical analysis. Cotterman sought to provide some insight into what underlies loan loss, as well as to allow what is learned to be used in underwriting practice. Thus, he restricted the characteristics to be studied to those that could be obtained at the time of loan applications. He also recognized that defaults may occur over the full term of a loan. Nonetheless, Cotterman limited the defaults to those occurring within the first three years of a loan because of data limitations. Plus default is generally much more heavily concentrated in the early years of a loan.

Before discussing his results, Cotterman clearly defines key terms used in his analysis. For example, the loss rate of a defaulted loan is the number of dollars lost per dollar lent. He further distinguishes between two types of loss rates: conditional and unconditional. The former is calculated over only those loans that default, while the latter is calculated over all loans: those that default and those that do not. Moreover, the unconditional loss rate is equal to the conditional loss rate multiplied by the default rate.

Cotterman offers the following example to illustrate this relationship. If “the default rate were 5 percent and the loss rate among those defaulting (i.e., the conditional loss rate) were 60 percent, then the unconditional loss rate would be 3 percent (0.05 x 0.6 = 0.03). That is, even though losses are, on average, 60 cents on the dollar among defaulting loans, losses are on average only 3 cents on the dollar among all loans.”

Cotterman indicates his preference for using the unconditional loss rate for its usefulness in valuing a portfolio of loans and in deciding whether to underwrite a loan as compared with simply knowing the expected default rate.

Since Cotterman’s specification of an unconditional loss rate contains the default rate as a component, any attempt to determine the factors underlying the former must consider the specification and estimation of the latter. Thus, he estimates a model of default alongside a model of conditional loss rates. The explanatory variables Cotterman uses in his default model fall into five categories (with some examples for each): credit characteristics—a proxy for a borrower’s capacity to pay (front-end ratio, credit score, and number of reserve monthly payments after closing); characteristics of the loan (loan-to-value ratio and note rate); characteristics of the area housing market and home relative to the area market (house price growth and relative house prices); race- and income-related characteristics of the individual and neighborhood (applicant’s race and monthly income; and other characteristics (tract income and judicial foreclosure state).2

Cotterman identifies several components that contribute to the loss rate, including the unpaid principal balance less sales price received in property disposition, forgone interest, holding cost, sales expense, and foreclosure, acquisition, and conveyance (FAC) costs. He finds that the components vary in their contribution to the loss rate: unpaid balance less sales price contributes the most; foregone interest, FAC, and sales expense contribute about the same; and holding cost contributes the least.

Findings and Implications

Cotterman's analysis yields some revealing findings. While some findings might seem intuitive, they underscore the importance of thoughtful analysis to confirm conventional wisdom. Overall his results point to the important role that differences in the timing of default-related events-the time from origination until default, the time spent in foreclosure processing, and the time spent in property disposition-play in determining loss rates. Loans that take longer to default tend to lower dollar loss rates, while loss rates tend to rise with the amount of time spent in foreclosure and property disposition. Moreover, various key characteristics of the borrower, the lender, and the market affect loss rates differently. For example, increases in the front-end ratio, loan-to-value ratio, the note rate, and borrower incomes tend to increase loss rates. Increases in credit scores, mortgage payments held in reserve, loan amounts, house price growth, relative house prices, and tract incomes tend to lower loss rates. His results also suggest that blacks, Hispanics, and those in underserved areas and judicial foreclosure states tend to have higher loss rates.

The interplay between default rates and loss rates was apparent in Cotterman's analysis of data on applicants in 1992 and 1996. He found that even though the loss rates per default fell while the default probabilities rose, the loss rates per loan rose across these cohorts because the default probabilities were dominant. His estimations, though quite tentative, indicate that using expected unconditional loss rates as a basis for underwriting rather than default probabilities would be unlikely to improve the prospects of black applicants. Using such an underwriting scheme to rank risks (rather than using only estimated default probabilities) would lower blacks' representation in the low-risk category and raise it in the high-risk category. In general, Cotterman suggests that a mortgage scoring system based on both the probability of default and dollar losses yields a more comprehensive measure of risk.

  • 1 "Analysis of FHA Single-Family Default and Loss Rates." Unicon Research Corporation: Santa Monica, CA, March 2004. Cotterman is vice president of Unicon. The study can be seen at www.huduser.org/publications/pdf/
    fhasinglefamilydefaultlossrates.pdf
    .
  • 2 A judicial foreclosure state is one in which mortgages are used for the purchase of real property. Once a lender in one of these states proves that a loan is in default and has exhausted all attempts to resolve the matter with the homeowner, the lender can seek relief in the courts, i.e., a judicial foreclosure.