Bankruptcy Prediction with Financial Ratios - Examining Differences across Industries and Time

Abstract: The purpose of this study is to examine how well different financial ratios can predict bankruptcy across industries and time. The study also examine whether including industry differences in a prediction model can increase its accuracy. Bankruptcy prediction models were estimated using logistic regression for each year between 2006 and 2011, with and without interaction terms accounting for industry effects. These were analyzed and tested on a holdout sample for their classification abilities. 311,930 annual reports from non-bankrupt companies and 5,257 annual reports from bankrupt companies were analyzed, covering the time period 2006 to 2011. The study shows that the bankruptcy-prediction ability of different financial ratios varies between years. However, only in some cases, significant differences between industries were found. The overall classification ability was not significantly increased when including the industry effects but using some specified cut-off values, a marginal increase was found.