Could Efficiency Analysis Help in Predicting Bank Failure? The Case of the 2001 Turkish Crisis

Inefficient, or weak banks, are continuously pointed out as threatening financial stability as they are more likely to fail than efficient ones and thus to weaken the whole banking sector (BIS, 2002). Consequently, banks' efficiency analysis, a measure of bank's management efficiency, should be a useful tool for policy makers to predict bank failure. This paper tests the usefulness of efficiency analysis in predicting bank failure by using the 2001 Turkish banking crisis as a case study. It implements in a first step both parametric, stochastic frontier analysis (SFA), and non-parametric, data envelopment analysis (DEA), to predict Turkish banks' efficiency scores over 1996 to 2001. During this period, 19 out of 55 were taken over by the Turkish Saving and Deposit Insurance Fund (SDIF). In a second step, efficiency scores are tested against the standard CAMELS model to check their capacity in predicting banks' probability to fail, i.e., SDIF taken-over. The findings indicate that banks taken-over were less cost efficient than others. Moreover, banks which were lending to connected parties were also less cost efficient than others. On the methodology side, our results using a parametric approach are found to be more significant than the non-parametric ones.