Bankruptcy Prediction Ahead of Global Recession: Discriminant Analysis Applied on Romanian Companies in Timiş County

The purpose of this paper is to evaluate the potential of financial ratio analysis performed by employing public data on predicting bankruptcy during the economic crisis period. The population subjected to our study was composed of the 26,980 Romanian companies from Timiş County that submitted financial reports for 2007 to the fiscal authorities. Based on the financial data that was published from these reports by the Romanian Ministry of Public Finance, twelve financial ratios for each of the 26,980 companies have been computed. The twelve ratios were chosen by taking into consideration the recommendations of the literature, as well as the availability of financial data. We were aware of the fact that other sources of information might improve the prediction of bankruptcy, but, as the access of the external stakeholders to such information sources is limited, we decided to search for bankruptcy predictors only within the financial data published online by the Ministry of Finance, which is easily available to everybody. The statistical analysis of the correlation between the values of each financial ratio and the frequency of the bankruptcy event led to the retention of five ratios as possible explanatory variables in a bankruptcy prediction model. The initial analysis also led to the reduction of the target population to 4,327 companies. By means of discriminant analysis, we proposed a model capable of predicting bankruptcy for the target population with an out-of-sample accuracy of 69.3%. Our findings show that the financial statements from one year prior to the beginning of the economic crisis in Romania reflect the weaknesses that make the companies susceptible to bankruptcy. We believe our model to be of practical use, as it is able to accurately discriminate between bankrupt and non-bankrupt firms over a five-year period, by only employing synthetic publicly available financial data.