Nonlinear Relationships and Their Effect on the Bankruptcy Prediction

Abstract

This study uses a generalized additive model to analyze nonlinear relationships and their effect on the bankruptcy prediction in terms of a company’s probability of default. Specifically, this study examines which of the performance indicators that are contained in a company’s annual financial statements affect nonlinearly that company’s probability of bankruptcy. Furthermore, this study examines the specific form that such nonlinear relationships take and interprets their economic relevance to predicting bankruptcy. On the basis of comprehensive data on German companies, this study shows empirically that there are statistically and economically relevant nonlinear relationships that influence the bankruptcy prediction. These nonlinear relationships are observed both below and above specific thresholds with respect to a company’s adjusted equity ratio, asset structure ratio based on tangible assets, adjusted return on assets, sales, and age. Our findings show that, to increase the accuracy of bankruptcy forecasts, it is necessary to take into account nonlinear relationships in models of bankruptcy prediction.