Discriminant Analysis: Applications In Finance

Abstract

This paper begins by defining and briefly explaining what discriminant analysis is. After noting the advantages and drawbacks of this statistical technique which is very similar to the regression analysis, the paper summarizes a number of important studies reported in the finance area during the last twenty years, using this technique. Some of the interesting efforts involve prediction of corporate bankruptcies, identification of conglomerate targets, prediction of bond rating, etc. The majority of the studies have relied on discriminant analysis to classify firms into two distinct groups. Few studies (e.g., studies on bond ratings) have used multiple discriminant analysis to identify more than two groups. Varying levels of accurate classification rates are reported depending on how the matched group was selected and which statistical procedures were used to select the independent, explanatory variables.