EXPERT SYSTEMS IN BUSINESS LENDING

Business lending is changing in form and practice as technology and competition continue to change how banks deal with their business credit customers. One of the most important business loan trends today focuses on the increasing use of so-called expert systems-predictive equations that use selected data on a customer's business and financial history to classify a loan request as likely or unlikely to meet the bank's quality standards. Advanced statistical techniques, such as multiple discriminant analysis and logistic regression analysis, are used to sample from a bank's past business loan records a limited set of variables such as the business customers credit rating, recent profitability (e.g., ROE or ROA), degree of liquidity(e.g.,current or acid-test ratio) and leverage (e,g., ratio of debt to equity capital), coverage ratios (e.g., the customer's earning before interest and taxes relative to interest and principle payments on debt and leases), and other variables that reflect economic or industry conditions statistical tests are conducted until a set of variables is found that appears to separate at a statistically significant level " good" from "bad" business loans that the bank has made in the past. It is assumed that these same variables will be able to distinguish (at a statistically valid level) between "good" and "bad" future loan requests.