Commercial Mortgage Default: A Comparison of Logit with Radial Basis Function Networks

This article explores the use of artificial neural networks in the modeling of foreclosure of commercial mortgages. The study employs a large set of individual loan histories previously used in the literature of proportional hazard models on loan default. Radial basis function networks are trained (estimated) using the same input variables as those used in the logistic. The objective is to demonstrate the use of networks in forecasting mortgage default and to compare their performance with that of the logistic benchmark in terms of prediction accuracy. Neural networks are shown to be superior to the logistic in terms of discriminating between good and bad loans. The study performs sensitivity analysis on the average loan and offers suggestions on further improving prediction of defaulting loans.     

Cultural Affinity and Lending Discrimination: The Impact of Underwriting Errors and Credit Risk Distribution on Applicant Denial Rates

We examine the impact of underwriting errors on marginal loan denial rates when these errors or average creditworthiness differ across racial groups. We find that the noise in evaluating applicant credit risk can have a differential impact on marginal borrowers across racial groups. Consequently, discrimination can go undetected in an examination of relative loan denial rates when it is present. We also show that discrimination can be detected where none exists. Furthermore, we argue that it is likely that the errors a bank makes in assessing applicant credit risk are a consequence of its skill or affinity, which, in turn, is shaped by the underwriting experiences in the bank's market. A resulting implication is that banks that develop affinities in serving different market segments may have substantially different denial rate experiences across racial groups. This observation can shed light on the puzzling result that minority-owned banks tend to perform poorly in studies of lending discrimination. We conclude that underwriting errors call into question the reliability of the fair-lending guidelines used to assess all banks.     

Banks as Interest Rate Managers

Most banks have a two-tier pricing system, offering accounts at market-related interest rates and at posted rates that are changed at discrete intervals. In this paper, I develop a model of bank interest rate management. I consider a bank with two classes of loans and deposits in its balance sheet: One pays a market rate of interest, the other a posted rate. Market rates are exogenous and evolve stochastically over time. Posted rates are altered intermittently by the bank itself. The bank faces imperfect arbitrage by its customers between posted and market rate funds. Under simple assumptions about the stochastic process governing the market rate, I derive optimal decision rules for the adjustment of the posted rate and determine conditions under which these rules are asymmetric. A key prediction of the model is a negative correlation between market loan rates and the gap; this is more consistent with the behavior of British banks than is the contrary prediction of more standard models.     

Institutional disparities in the pricing of adjustable rate mortgage loans

In recent years, commercial banks and savings and loan associations in South Florida have consistently offered initial adjustment period teasers, or subsidies, on their adjustable rate mortgage loans (ARMs). This study adopts the size of the initial subsidy as a proxy for a lender's willingness to offer ARM loans and develops an econometric model which relates the size of the teaser to a series of internal variables (other lending parameters), and external variables (financial market conditions).The results suggest that subsidization policies are not identical across institutions. Specifically, savings and loan associations seem to be less willing than commercial banks to accept the interest rate exposure inherent in ARM lending when future loan rates are constrained by adjustment limits. Consequently, the study argues that the character of a lender's existing assets influences its reactions to the risk/return properties of new assets.This paper has benefitted greatly from the comments of the Journal's reviewers. Responsibility for remaining errors rests with the author.     

The Impact of Risk-Based Premiums on FDIC-Insured Institutions

This paper examines the effect of a series of announcements leading to the approval of risk-based deposit insurance premiums on returns to stockholders of commercial banks. Utilizing risk-weighted capital ratios and measures of overall risk, we group banks according to one of the nine-tier insurance categories subsequently defined by the FDIC. During the period in which the new insurance system was considered and approved, we found that stockholders of well-capitalized, healthy banks experienced wealth changes significantly different from those experienced by less than well-capitalized, less than healthy banks. Although many argued the premium range in the initial insurance schedule was insufficient, the results show that this initial risk-basing marked an important change in the relative burdens imposed by FDIC insurance.     

Government Deposit Insurance and the Diamond-Dybvig Model

The apparent banking market failure modeled by Diamond and Dybvig [1983] rests on their inconsistently applying their sequential servicing constraint to private banks but not to their government deposit insurance agency. Without this inconsistency, banks can provide optimal risk-sharing without tax-based deposit insurance, even when the number of type 1 agents is stochastic, by employing a contingent bonus contract. The threat of disintermediation noted by Jacklin [1987] in the nonstochastic case is still present but can be blocked by contractual trading restrictions. This article complements Wallace [1988], who considers an alternative resolution of this inconsistency.     

The Information Content of Bank Examinations

We identify three types of information from bank examinations—auditing information from verifying the honesty and accuracy of the bank's books, regulatory discipline information about the treatment of the bank by regulators, and private information about bank condition. We estimate these information effects by comparing the cumulative abnormal market returns associated with examinations in which the CAMEL rating remained unchanged, improved, and worsened. All three information effects are found to be greater for banks entering the examination process with unsatisfactory ratings from prior examinations. The only consistently strong effect found is that examination downgrades appear to reveal unfavorable private information about bank condition. The evidence also suggests that the information may reach the market in part through loan quality data released in quarterly financial statements.     

Optimal Loan Interest Rate Contract Design

This paper analyzes optimal loan interest rate contracts under conditions of risky, symmetric information for one-period (static) and multi-period (dynamic) models. The optimal loan interest rate depends upon the volatility of, and co-variation among the market interest rate, borrower collateral, and borrower income, as well as the time horizon and the risk preferences of lenders and borrowers. For a risk-averse borrower with stochastic collateral, variable interest rate contracts are, in general, Pareto optimal. For plausible assumptions, the optimal loan interest rate for the multi-period model often exhibits muted responses to changes in market interest rates, making fixed rate loans a reasonable approximation for the optimal loan. Hence, in the absence of optimal contracts, long-term (short-term) borrowers tend to prefer fixed rate (variable) contracts.     

“Investor sentiment” and the closed-end fund puzzle: A 7 percent solution

Lee, Shleifer, and Thaler (1991) argue that the irrational noise trader model of DeLong, Shleifer, Summers, and Waldmann (1990) ... is consistent with the published evidence on closed-end fund prices. ... However, Lee, Shleifer, and Thaler provide no indication of how much of the variability of a closed-end fund's discounts and premiums is due to such investor sentiment. Using the signal extraction technique of French and Roll (1986) to measure noise, this article estimates that on average only 7 percent of the variance of a standardized measure of weekly changes in discounts and premiums can be attributed to noise-trading activity. Investor sentiment, therefore, seems to account for very little of a closed-end fund's discount and premium variability over time.     

On the Economics of Subprime Lending

US mortgage markets have evolved radically in recent years. An important part of the change has been the rise of the subprime market, characterized by loans with high default rates, dominance by specialized subprime lenders rather than full-service lenders, and little coverage by the secondary mortgage market. In this paper, we examine these and other stylized facts with standard tools used by financial economists to describe market structure in other contexts. We use three models to examine market structure: an option-based approach to mortgage pricing in which we argue that subprime options are different from prime options, causing different contracts and prices; and two models based on asymmetric information–one with asymmetry between borrowers and lenders, and one with the asymmetry between lenders and the secondary market. In both of the asymmetric-information models, investors set up incentives for borrowers or loan sellers to reveal information, primarily through costs of rejection.     

Why Are Earnings Kinky? An Examination of the Earnings Management Explanation

Prior research has documented a kink in the earnings distribution: too few firms report small losses, too many firms report small profits. We investigate whether boosting of discretionary accruals to report a small profit is a reasonable explanation for this kink. Overall, we are unable to confirm that boosting of discretionary accruals is the key driver of the kink. We caution the use of the ratio of small profit firms to small loss firms as a measure of earnings management. We investigate and discuss a number of alternative explanations for the kink.     

Do mortgage rates vary based on household default characteristics? Evidence on rate sorting and credit rationing

Credit screening models suggest that lenders vary loan rates and debt ceilings across applicants on the basis of credit risk. We argue that regulatory constraints such as Fair Lending Laws may preclude rate sorting while increasing lender use of debt ceilings to adjust for applicant credit risk. Using household data from the 1983 SCF, we find that mortgage rates do not vary with applicant credit risk whereas related studies find that debt ceilings vary with borrower risk attributes. Together, these findings support arguments that regulatory constraints reduce rate sorting while increasing the use of non-price terms in the mortgage contract.     

Changes in Non-Local Lending to Small Business

In this paper, I use recently collected Community Reinvestment Act loan data to examine how small business lending in local geographic areas (defined as markets) by lenders not physically located in those areas changed between 1996 and 2001. The results show that the importance of outside lending increased substantially over this period when measured in terms of the number of loans rather than the dollar volume of loans. The levels and rates of growth in out-of-market lending are more modest if the share of out-of-market lending is expressed in dollar volumes and almost insubstantial if organizations that engage in substantial credit card lending are excluded as out-of-market lenders. Using a fixed-effects model and an extensive panel data set, I find that the share of outside lending into local geographic markets is positively associated with local market concentration and the average wage of tellers in the market, consistent with the hypothesis that outside loans are to some degree substitutes for in-market loans.     

Failed delivery and daily Treasury bill returns

If the seller of a Treasury bill does not provide timely and correct delivery instructions to the clearing bank, the bank does not deliver the security. Furthermore, the seller is not paid until this failed delivery is rectified. Since the purchase price is not changed, these fails generate interest-free loans from the seller to the buyer. This article studies the effect of failed delivery on Treasury bill prices. We find that investors bid prices to a premium to reflect the possibility of obtaining the interest-free loans that fails represent. This premium is a function of the opportunity cost of the fail. We also find that the bid-ask spread varies directly with the length of the fail. We rule out that our results are due to liquidity premiums, or to a general weekly pattern in short-term interest rates or the bid-ask spread.     

Ethnicity, Endogeneity, and Housing Tenure Choice

Home ownership rates for New Zealanders of European descent aged twenty five to fifty four are 16 percentage points higher than for Maori in the same age group. This article explores the relative attainment of home ownership of the two ethnic groups by estimating a series of tenure choice models and decomposing the difference in rates into endowment and residual effects. Particular attention is given to the endogeneity of current income and wealth relative to the tenure choice decision and to the methods for decomposing group mean differences. The article also applies more appropriate methods for estimating incomes and wealth than have been used in most previous studies of tenure choice. The study concludes that only a small proportion of differences in home ownership rates is explained by household endowments. It is shown that controlling for the endogeneity of income and wealth has a substantial impact on the tenure choice and decomposition results.     

Corporate Real Estate and Stock Market Performance

An interesting question in corporate real estate literature is whether real estate can improve the stock market performance of property-intensive non-real estate firms. Using a data set comprising 75 non-real estate corporations that own at least 20 percent properties, this paper empirically assesses and compares the pair-wise return, total risk, systematic risk and Jensen abnormal return performance of composite (with real estate) and hypothetical business (without real estate) firms. We employed Morgan Stanley Capital International world equity index instead of a local market index to provide some insights into the performance of the local market relative to the global market during the 1997–2001 volatile periods experienced by many Asian countries. Our results suggest the inclusion of real estate in a corporate portfolio appears to be associated with lower return, higher total risk, higher systematic risk and poorer abnormal return performance. It is therefore likely that non-real estate firms own properties for other reasons in addition to seeking improvement in their stock market performance. Further research is needed to explore the main factors contributing to corporate real estate ownership by non-real estate firms.     

Bank lending ‘manias’ in theory and history

A popular view of banking crises sees them as consequences of prior bank lending manias. Such manias are supposed to be especially likely in legally unrestricted banking systems, where banks can issue notes and are not subject to statutory reserve requirements. Here it is argued that the bank lending mania hypothesis (1) exaggerates the role of subjective factors, including bankers' confidence or optimism, as a stimulus to bank lending, and (2) is not supported by evidence from past, legally unrestricted banking systems.     

The “Rent-To-Own” mortgage and prepayment risk

The most important risk factor in the mortgage and mortgage-backed security market has been prepayment risk. Various innovations have arisen to deal with it but none hedge it fully. The Rent-To-Own (RTO) mortgage discussed here is a mortgage instrument that reduces or even reverses prepayment risk. It does so by creating an incentive structure within the framework of the mortgage contract that penalizes prepayment when interest rates are low and rewards it when interest rates are high. This is the opposite of standard mortgages. The RTO incentive structure is based on a unique buyout feature. Borrowers who want to buy out the financial interest of the lender may do so whenever they want, but the buyout price is a negative function of the market interest rates prevailing currently, that is, at the time of the buyout. Hence the lower these rates, the higher the buyout price. Other advantages of the RTO mortgage are also described.     

The Timing and Persistence of CRA Compliance Ratings

We evaluate supervisory practices in enforcing the Community Reinvestment Act (CRA) by examining whether or not supervisors consider observable, bank-specific characteristics in (1) scheduling CRA compliance examinations and (2) determining whether, and for how long, a given CRA rating persists. Failure to confirm such a relationship would be consistent with criticism that the evaluation criteria are so vague that supervisors can essentially assign any rating they want for compliance purposes. Analysis of a sample of several thousand commercial banks, observed over a relatively stable regulatory regime, indicates that both examination scheduling and the persistence of examination ratings are associated with residential loan levels, a presumed cornerstone of the CRA, as well as other financial, regulatory and market factors. We conclude that CRA enforcement during this period reflected, at least in part, objective evaluation criteria.     

A RECONSIDERATION OF DISCRIMINATION IN MORTGAGE UNDERWRITING WITH DATA FROM A NATIONAL MORTGAGE BANK

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 finding that 9 of 10 borrowers correctly choose whether to apply under FHA vs. conventional programs, based on financial 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 a re 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.