The Impact of Business Groups on Bankruptcy Prediction Modeling

The bankruptcy prediction literature generally ignores corporate ownership and assumes companies are independent economic entities. In Continental Europe this latter assumption does not hold, due to the importance of business groups. Using a sample of mostly non-quoted Belgian medium and large sized companies, we show that the predictive power of several accounting ratios that are commonly used in bankruptcy prediction models (e.g. performance, leverage, liquidity and efficiency) is different for group member companies as compared to stand-alone companies. By exploiting these differences in relative importance, model fit can be improved without adding any new information. Performance can be increased further by directly adjusting for group related factors, e.g. by including a measure of financial health of the group as a whole. Finally, it is shown that group adjustments can also improve the fit of some well-known existing prediction models.