Home > Research > Publications & Outputs > Statistical modelling to predict corporate defa...

Electronic data

View graph of relations

Statistical modelling to predict corporate default for Brazilian companies in the context of Basel II using a new set of financial ratios

Research output: Working paper



This paper deals with statistical modelling to predict failure of Brazilian companies in the light of the Basel II definition of default using a new set of explanatory variables. A rearrangement in the official format of the Balance Sheet is put forward. From this rearrangement a framework of complementary non-conventional ratios is proposed. Initially, a model using 22 traditional ratios is constructed. Problems associated with multicollinearity were found in this model. Adding a group of 6 non-conventional ratios alongside traditional ratios improves the model substantially. The main findings in this study are: (a) logistic regression performs well in the context of Basel II, yielding a sound model applicable in the decision making process; (b) the complementary list of financial ratios plays a critical role in the model proposed; (c) the variables selected in the model show that when current assets and current liabilities are split into two sub-groups - financial and operational - they are more effective in explaining default than the traditional ratios associated with liquidity; and (d) those variables also indicate that high interest rates in Brazil adversely affect the performance of those companies which have a higher dependency on borrowing.