Prediction of Bankruptcy Using Financial Ratios, Information Measures, National Economic Data and Texas Economic Data

The main purpose of this study is to develop a bankruptcy prediction model for the small business firm. Data was collected from the Dallas Small Business Administration (SBA), making this study specific to its decision makers. Existing research has produced models which predominately use financial ratios and information measures either independently or combined, and a few research models have used economic trends. This study varies from past studies in that it includes regional economic variables from the states of Texas. A sample of three-year data for 138 firms included fifteen bankrupt firms. This proportion of bankrupt/nonbankrupt firms approximates the proportion of repayed/defaulted loans in the SBA. Stepwise regression, set at the .15 level of significance, reduced a total of fifty-three variables to nine. These nine variables were then used to test twelve predictive models. All twelve models tested improved the SBA repayment rate and only two of the twelve would have caused the SBA to deny loans to applicants who eventually repaid. The study determined the model that included financial ratios, information measures, and Texas economic variables as best. It was also demonstrated that some of the variables used in this model could be eliminated without decreasing the predictive power of the model. The best of twelve models improved the SBA default rate by 40 percent without denying a loan to any applicant that eventually repaid.