Failure prediction of retail firms through use of financial ratios

Paul Thomas McGurr, Purdue University

Abstract

Financial ratios have long been used for business analysis. Researchers have formulated business failure prediction models utilizing financial ratios. However, few failure prediction studies have focused on specific industries. This study focuses on use of financial ratios to discriminate between failed and nonfailed firms in the retail industry. Using a matched sample of 66 failed and 66 nonfailed retail firms obtained from the COMPUSTAT database of publicly traded companies, multiple discriminant analysis was utilized to develop a retail prediction model which accurately classified 78% of the sample firms as failed or nonfailed. Hotelling's T$\sp2$ test determined that the mean vector of the discriminant function for failed firms differed from that of nonfailed firms. The model was further validated using a jackknife procedure and split sample analysis. Failed or nonfailed classification accuracy was found to be significantly better than chance. The retail prediction model's classification accuracy was compared to that of Zavgren's logistic model developed using manufacturing companies, and Deakin's multiple discriminant analysis developed using firms from several industries. Because the models were applied to the same set of firms, results were analyzed using McNemar's test. The retail prediction model was significantly more accurate than Zavgren's model but not significantly more accurate than Deakin's. Both the retail prediction model and Deakin's model had return on assets as their most significant variable. Return on assets accurately classified retail firms as failed or nonfailed in a univariate model. Although the retail prediction model, Deakin's model and the return on assets model all showed overall predictive accuracy, both Deakin's model and the return on assets model made more Type I errors classifying failed firms as nonfailed. The retail prediction model had greater accuracy identifying retail failure but less accuracy identifying nonfailure.

Degree

Ph.D.

Advisors

DeVaney, Purdue University.

Subject Area

Business community|Accounting|Finance

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS