Factors affecting the probability of bankruptcy: a panel data approach

Abstract

This paper investigates the importance of firm-specific factors in determining or explaining bankruptcy. By studying Norwegian firms from the period 2005-2012, we are able to examine this using binary regression models. First, we identified potential financial measures we believed to be associated with business failure. Then we selected 15 of these that are potentially correlated with the occurrence of bankruptcy along with 3 firm-specific characteristics. These measures were incorporated into different econometric models. During analysis, the 15 financial measures were reduced to 5: Two profitability measures, two solidity measures and one liquidity measure. We conclude that fixed effects are present in the data. Controlling for them enables us to identify the impact of accounting ratios on the probability of a bankruptcy more efficiently. In the logistic regression only two profitability measures remain significant, yet when we construct a prediction model for business failure this model has an overall accuracy of 74 %. Thus, we are also confident that incorporating firm-specific effects in the model enables us to identify good measures of how accounting data affects the probability of bankruptcy.