Previsão de falência bancária: um modelo de risco proporcional

Abstract

The purpose of this paper is to construct a bank failure prediction model that provides an early warning system capable of identifying future problem institutions using financial data. Such a system would allow a more efficient allocation of the scarce bank examination resources, and improvement of the quality of regulatorís performance, given that it may identify a problem bank while it is still possible to implement corrective measures. This is important because since the adoption of the Real Plan more than 50 of the 271 existing banks have been declared officialy insolvent and were subsequently closed, acquired or received assistance to prevent closure. Here the Cox proportional hazards model is applied to the prediction of bank failures in Brazil. Its most significant advantage is the provision of information regarding the expected time to failures. The results strongly suggest that a proportional hazards model could be an effective early warning tool. The classification accuracy of the estimated model is quite high, and the model identifies a considerable proportion of failures in advance.