Abstract
This paper explores the prediction of bankruptcy of Greek retail and wholesale trade companies and, in particular, the relation between the forecasting ability of the Logit model and the degree of homogeneity of the samples of bankrupt and healthy companies. A sample of 119 bankrupt companies was matched with an equal sample of healthy companies for the period 2003-2014, based on year, sector, and sub-sector, which was formed by random selection. Using the method of factor analysis, seven financial ratios were selected, which are the independent variables of the model. Applying the Logit model, the results showed a significant explanatory capability of the model in the trade sector as a whole as well as in the wholesale trade sub-sectors, which increases as the homogeneity of samples of bankrupt and healthy companies increases.In particular, the predictive capability of the model that we used improved by 14.3% regarding the classification of bankrupt firms when the same methodology was applied from the broader sector to the sub-sector. Moreover, the independent variable of capital structure has the highest stability and contributes substantially to the discriminant validity of the Logit model. Key words: Bankruptcy, corporate bankruptcy, sectoral forecasting models, financial and accounting ratios.
Highlights
The limited investigation of the effects of industry features on the prediction of firm bankruptcy may often lead forecast models to unreliable results
The results showed the superiority of the logarithmic probability model (Logit) model among statistical techniques and the more effective forecasting ability of support vector machines among artificial intelligence models
The sample consisted of 56 bankrupt companies which were matched to 56 healthy firms; the results showed an overall predictive precision of 74.1% for the multivariate discriminant analysis model and of 67.2% for the logarithmic probability model
Summary
The limited investigation of the effects of industry features on the prediction of firm bankruptcy may often lead forecast models to unreliable results. The degree of homogeneity of samples depends on many factors and in particular accounting and financial characteristics, such as the type of inventories, their sensitivity and obsolescence, the level of competition, the different accounting methods (Chava and Jarrow, 2004), the credit policy followed, financing factors, allocation of working capital components, the legislative framework, the import or export-orientation of the firm, the size of the firm, etc For these reasons, in the last decade, in particular, surveys of business failure forecast have focused on sector-specific samples. This study contributes to the literature in several important ways It provides unique empirical evidence, as the increasing homogeneity of firm samples (as is the case in sector and sub-sector samples) leads to a better forecasting ability of the models.It highlights the differences between the effect of the same financial ratios of firms belonging to different sectors and subsectors of economic activity on bankruptcy prediction. The analysis we present can be applied in future research to assess the effect of sample homogeneity on other sectors such as the manufacturing industry.this study provides some important approachesto failure prediction that should be helpful to policymakers, investors, banks, regulators,and financial analysts with regard to the forecasting and preventing of economic distress
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
