Predicting the Risk of Bank Deterioration: A Case Study of the Economic and Monetary Community of Central Africa

Preventing bank failures is one of the fundamental concerns of decision-makers and it justifies the existence of banking supervision authorities. That is why bank supervisors have put in place measures aimed at enabling them to detect banks' difficulties early enough. The aim of the present study was to put forward an autoregressive logistic model of bank deterioration that could be used in countries in the CEMAC area as an early warning system (EWS). As explanatory variables, the model uses the financial variables that make up credit rating systems, the efficiency scores obtained using Data Envelopment Analysis as a proxy of management quality, shareholding, and the GDP growth rate; and as economic environment variables, it uses the inflation rate and the real effective exchange rate. The model also takes into account country effects. Although the simulations carried out in this study revealed that the model was less effective when used on out-of-sample data, the study still came up with a new typology of banks. For example, it found that a bank would be considered solid if the model had predicted it to be solid and its observed state was indeed solid at the predicted date. Such a typology makes it possible to use the results of the model by combining the model's prediction with the evolution of the situation of the bank in question. A refinement of this new typology can enable a better orientation of the bank supervision authority's prompt corrective action so as to prioritize the actions to be taken and to adapt them to the specific difficulties of each bank.