Academic journal article Journal of Money, Credit & Banking

A Multinomial Logit Analysis of Problem Loan Resolution Choices in Banking

Academic journal article Journal of Money, Credit & Banking

A Multinomial Logit Analysis of Problem Loan Resolution Choices in Banking

Article excerpt

Management of problem Loans is one of the most critical aspects of the commercial banking business. The importance of the task is further heightened by increased loan defaults due to a decline in the economy. In 1991, for example, loan write-offs at all U.S. banks averaged 0.6 percent of outstanding credit. In the same period, another 2.1 percent of the outstanding credit was classified as nonaccrual and past due.(1) Surprisingly, there is a dearth of empirical research in the literature on problem loan resolutions, perhaps due to the lack of easily accessible data. Given the sensitive nature of credit files and the existence of strict federal confidentiality laws, banks have been generally reluctant to supply detailed data for empirical research on problem loan resolutions.

This paper presents the first empirical work on the topic using actual problem loan files from fifty-two banks in twenty-five states. The banks were chosen randomly from a population of small to medium size banks with less than $1 billion in assets. Collectively, participating banks supplied 155 files with loan origination dates from the period 1978-1989. The maximum number of loan files provided by a single bank was five with the average bank submitting three files.

Based on theoretical papers by Herring (1989) and Geppert and Karels (1992), we design our empirical work around a multinomial logit model that defines problem loan resolution choices as a function of joint borrower and lender decisions. A bank will choose a workout option if its expected value is greater than the outcome under a no-workout plan. For the borrower, if the reputational penalty due to a default is less than the opportunity cost of the next best alternative, the borrower will have an incentive to default. If the reverse holds, then the borrower will be better off with a workout.

We also examine factors that make a lender label a resolution as either a success or a failure through a binomial logit model. While the lender's perspective may be biased, the analysis may provide useful information regarding the decision process.

The remainder of the paper is organized as follows. Section 1 provides a review of the related literature. Section 2 examines the problem-loan decision process. Section 3 describes data collection. Section 4 presents multinomial logit models that explain the resolution choices. Section 5 presents the empirical results of the logit models. Section 6 evaluates the determinants of a successful resolution from the perspective of the lender. Section 7 concludes the paper.

1. PREVIOUS RESEARCH

While to the best of our knowledge there is no previous empirical work related to the problem loan resolution choices, the following papers provide some direct insights by examining the borrower characteristics in delinquent loans. In an analysis of "good" and "bad" loans, Rakes (1973) isolated the borrower attributes that contributed to slow payment. He recognized that many borrowers were habitually late in remitting their mortgage payments while never so late to be in default. He concluded that the characteristics of late-paying borrowers were quite different from those who defaulted on their loans. Ang, Bowling, and Chua (1979) also studied late-paying borrowers in the context of automobile loans issued by a large California bank. They found statistically significant results to differentiate late-paid loans from other loans.

In commercial lending, Orgler (1970) developed a credit-scoring model based on bank examiners' data to aid bank officers and regulators to classify existing loans into good and bad groups. Cowen and Page (1979) conducted similar research using minority small business borrowers and identified borrower attributes that determined successful and unsuccessful loans. Altman (1980) examined the charge-off and recovery experience of forty-nine banks with four hundred problem loans for the 1969-76 period. Through a questionnaire mailed to banks, he collected the data on the loan balance twelve months prior to charge-off, the balanced charge-off, and the amount recovered after the charge-off. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.