Predicting financial distress

Abstract

Research on financial distress has been carried out for many years. Various models have been used to explain the probability of a firm’s propensity to be distressed. Given the lack of agreement on the best model to study financial distress, the paper will attempt to compare the rolling-logit model with the logit regression model. The aims of this study are to find out: 1) which variables of financial ratios, industry relative ratios, and firms’ sensitivity to macroeconomic variables, are to be included as determinant variables in financial distress model; and 2) compare each model predicting ability and performance. The research is descriptive verification while the method used is a case study using cross-sectional pooled data. The sample is manufacturing companies listed in IDX period 2000-2015. The distressed company was defined as a firm that has negative book equity value in the observation period 2015. The data analysis used is descriptive analysis, Mann-Whitney U test, backward stepwise regression, logit regression, rolling-logit regression, and jackknife validation test. The findings indicate: 1) determinant variables to predict the probability of firm’s financial distress were EBIT to Sales, EBIT to total assets, current assets to total assets, net worth to sales, sales to total assets, cash to sales, cash to total assets, inventory to sales, quick assets to sales, firms’ sensitivity to M2 and real exchange rates, previous bankruptcy probability; 2) rolling-logit regression model as general exhibit higher predicting ability compared to logit regression