Robert O. Edmister, An Empirical Test of Financial Ratio Analysis for Small Business Failure Prediction, The Journal of Financial and Quantitative Analysis, Vol. 7, No. 2, Supplement: Outlook for the Securities Industry (Mar., 1972), pp. 1477-1493

ratios with a specific occurrence or condition has not been quantitatively defined until very recently.In recent years several articles have reported scien tific ratio analysis research.William H. Beaver's study of firms during 1954-64 found that some ratios predicted failure 2 up to five years in advance.Edvard Altman introduced multi ple discriminant analysis as a statistical technique in finan cial ratio research and found five ratios which reliably pre-3 dieted bankruptcy up to two years prior to its occurrence.In his 1969 unpublished dissertation, David Ewert isolated 4 three ratios as good predictors of trade credit quality.The need for empirical verification of priori beliefs is being recognized by scholars, and scientific research of ratio analy sis is being undertaken.This dissertation will report the results of a study attempting to improve the empirical foundation for a theory of ratio analysis and to answer the question, is ratio analysis of small business financial state ments useful in predicting the failure of a small business? 2

This paper develops a credit-analysis model encompassing the accuracy of analytical methods, quality of applicants, cost of acquisition and analysis, profit from good loans, and losses from bad loans. Information generally available to the lending institution and subjective estimates can then be used to select from among alternative credit-granting systems the system with the greatest expected net present value. Each institution is thus able to find the credit granting system most appropriate for its particular market and analytical abilities.The model's profit maximizing objective and broad scope make it useful for setting credit department standards of performance. Costs can be compared with theoretical values of performance computed from loss rates, acceptance rates, and market information. The conditional probabilities, the chances of making the correct decision, can also be estimated for use in comparing methods of analysis or individual analysts. Unlike the loss rate, the conditional probability is an independent, unbiased measure of a method's accuracy.The example presented dealt with consumer installment loans, but the formulation is applicable to direct lending of any type. It provides the means for comparing loans with differing initial costs as well as widely varying risk classes and maturities. Financial institutions making direct loans add substantial values to capital supplied by the money and capital markets. The model is a theoretical formulation of the relationship between the cost and output of credit analysis.

The problem of monitoring the ongoing receivables collection experience of an enterprise which sells on credit is, in essence, the problem of identification. The concern is an accurate appraisal of customer account payment patterns — in particular, a determination of whether and to what extent those patterns vary over time. Successful execution by the credit manager of his responsibilities for policy formulation, collection enforcement, and forecasting necessarily depends heavily on the availability to him of a reliable reporting mechanism.