Content provided in partnership with

A comprehensive model for managing credit risk on home mortgage portfolios

Decision Sciences, Spring 1996 by Smith, L Douglas, Sanchez, Susan M, Lawrence, Edward C

ABSTRACT

Managing credit risk in financial institutions requires the ability to forecast aggregate losses on existing loans, predict the length of time that loans will be on the books before prepayment or default, analyze the expected performance of particular segments in the existing portfolio, and project payment patterns of new loans. Described in this paper are tools created for these functions in a large California financial institution. A forecasting model with Markovian structure and nonstationary transition probabilities is used to model the life of a mortgage. Logistic and regression models are used to estimate severity of losses. These models are integrated into a system that allows analysts and managers to depict the expected performance of individual loans and portfolio segments under different economic scenarios. With this information, analysts and managers can establish appropriate loss reserves, suggest pricing differentials to compensate for risk, and make strategic lending decisions.

More Articles of Interest: Mortgage Delinquency Migration: An Application of Maximum Entropy Econometrics; Executive summary: best practices in mortgage default risk measurement and...; Portfolio forecasting tools: what you need to know; The competing risks framework for mortgages: modeling the interaction of...; The State of the Art in Credit Portfolio Modeling

Subject Areas: Decision Support Systems, Financial Models, Forecasting, Logit Modeling, Markov Processes/Chains, and Risk Analysis.

INTRODUCTION

There were over $440 billion in residential mortgages held by U.S. banks in June, 1993. This important segment accounts for about 27% of the institutions' loan portfolios. Portfolio managers at financial institutions are responsible for ensuring that credit is granted judiciously, and for maintaining adequate reserves to cover future losses that may occur on outstanding mortgages. To perform this function, managers must work closely with loan officers to establish bank policies and standards with regard to the credit worthiness of borrowers, required collateral, and loan terms. In addition, portfolio managers must continuously assess the likely impact of changing economic conditions on risk and return characteristics of the portfolio.

During the last two decades, rapid price appreciation in many real estate markets created substantial equity for homeowners. Under these favorable conditions, mortgage defaults were relatively low, and the accumulation of equity cushioned lenders against large losses. However, with the collapse of the real estate markets in the Southwest in the mid-1980s, it became clear that managing loan portfolios of all types requires new analytical tools to assess risk in changing environments. More recently, the weak real estate markets in New England and California reinforced the pressing need for better information and controls. WIth better analytical tools, financial institutions can establish appropriate reserves against credit losses, price their products to reflect the underlying risks (e.g., charging premiums on "jumbo" loans), identify market segments from which they would like to attract new business, evaluate the benefits of mortgage insurance policies with different limitations, and compare the risks and returns associated with portfolio segments that may be bought or sold in secondary markets.

To support decisions of this nature, managers require systems to estimate the expected performance of each existing loan in the mortgage portfolio over its remaining life. They need to be able to estimate how the rates of loss and prepayment will change through time and be affected by changes in key economic variables. Ideally, managers could simulate the expected lifetime performance of a new loan under different economic scenarios (showing, for example, how it may vary with product type, financial leverage, and the use of mortgage insurance). The purpose of this paper is to present a comprehensive model that forms the core of a system to support such analyses and decisions in a major California financial institution.

RELATED RESEARCH

Broome and Nourse [1], Cooperstein, Redburn, and Meyers [3], Gardner and Mills [5], Vandell [11], and Weagley [12] have identified borrower characteristics and economic factors that are useful for predicting default by holders of fixed-rate, adjustable-rate (ARM), and graduated mortgages. Emerging as primary explanatory variables are: the homeowners' equity position, the property age, the change in home prices, the borrowers' monthly income to expenses, interest rates and unemployment. Of these, the equity level (usually measured by the loan-to-value ratio) is clearly dominant. Lawrence, Smith, and Rhoades [8], studying loans for manufactured homes, found that borrower delinquency patterns are important indicators of the likelihood of default.

Campbell and Dietrich [2] dealt with prepayment in addition to default and showed the importance of considering the effects of changing interest rates and other contemporaneous economic influences. Green and Shoven [6], using a proportional hazard model for California mortgages from 1975 to 1982, confirmed that prepayment behavior on fixed-rate obligations is highly sensitive to changes in interest rates after origination. Zom and Lea [14] studied the default and prepayment tendencies of borrowers using the Canadian rollover mortgage. Their results show that increases in interest rates raised the probability of default, and that partial prepayment is dependent on the rates of return for other investments. Cunningham and Capone [4] summarize the mortgage characteristics, property characteristics, borrower characteristics, and economic variables that have been reported by other authors to be useful in predicting default and prepayment on mortgages. Using multinomial logit models with variables for each of these factors, they compare the termination experiences of adjustable-rate versus fixed-rate mortgages in periods with volatile interest rates and housing prices. Consistent with industry notions, their simulations indicate that ARMs have higher default risks and lower prepayment probabilities than fixedrate obligations. In addition, the termination risks of ARMs are directly related to the size of rate caps and the frequency of adjustment. Richard and Roll [9] and Kang and Zenios [7] constructed prepayment models for mortgage backed securities (MBS) that rely on a pool of mortgages to collateralize bonds sold to investors. These models included four elements to describe the tendency to prepay a mortgage: the interest rate on the mortgage relative to prevailing rates as a measure of the economic incentive to refinance, a factor for seasonal variations, a "seasoning factor" for the early years when prepayments rates are higher, and a "burnout" effect to recognize that prepayment activity decreases for older loans. Combined, the last two variables accommodate the nonlinear effects of aging on propensity to prepay. Zipkin [13] suggests a Markov chain to represent prepayment rates under stochastic interest rates in the evaluation of cash flows from fixed-rate mortgages. Acknowledging "the standard modeling tradeoff between simplicity and realism," he proposes the analytical approach using Markov chains as an efficient alternative to "elaborate stochastic simulation" in evaluating mortgage-backed securities [13, p. 684]. The Markovian approach was implemented on a personal computer, in contrast to the use of parallel processing computers for stochastic simulations.

A comprehensive model for managing credit risk on home mortgage portfolios

Decision Sciences, Spring 1996 by Smith, L Douglas, Sanchez, Susan M, Lawrence, Edward C

Find Research Guides for:

Find Featured Titles for: News

Most Popular Articles in Business

Most Popular Publications in Business