Executive Summary
Credit unions are turning to automated credit scoring models as pressures for speedy lending decisions intensify. The models efficiently reduce multiple lending variables to simple numbers that indicate, with varying accuracy, the probability of timely repayment. Credit scoring is the most complex control tool in many managers' lending repertoire. Development of a credit scoring system often requires the services of an outside vendor with extensive command of econometric technique.
The authors of this report, demystify credit scoring and explain its benefits, limitations, and how it can be integrated into the overall lending process. They have chosen to summarize their analyses of sophisticated econometric techniques in a straight-forward, non-mathematical style.
What is this research about?
This report serves as a primer on credit scoring, explaining different approaches to constructing scorecards. It reviews three different types of scoring systems, and explains how to use scorecards most effectively in an underwriting system, and discusses the use of judgmental underwriting systems. It also provides a condensed users guide to credit scoring systems.
It shows how to evaluate the data used to construct a scorecard and explains the importance of the data used. It reviews the advantages and disadvantages of three different types of scoring systems: 1) customized, 2) generic, and 3) semi-customized.
The authors explain how to use scorecards most effectively in an underwriting system, and they discuss the use of judgmental underwriting systems, behavioral models, and expert systems, as they relate to underwriting systems.
What are the credit union implications?
The role of a credit scoring system in the overall lending process is changing. The most common criticisms of credit scoring are its rigid treatment of applicants and inability to accommodate "special cases," since the scorecard is a mechanical gauge of likelihood of repayment. Credit unions should review the benefits and pitfalls of the two most common scorecards, customized and generic, while considering a new type of scoring system—semi-customized— which might be more amenable to the credit union condition.
This report is sponsored by the McIntire School of Commerce, University of Virginia.