Explaining and predicting bank failure using duration models: the case of Argentina after the Mexican crisis

This paper studies the role played by several bank specific financial indicators in determining the process of bank failure in Argentina after the Mexican crisis known as the "tequila effect". Due to the relative scarcity of previous studies, this paper priorizes the use of semiparametric and non-parametric methods which allow us to measure the effect of bank specific financial explanatory variables in the process of bank failure together with duration dependence effects without the need of arbitrary and possibly unrealistic assumptions. The dynamic of bank failures can be fairly characterized by observable factors, which discards the possibility that it had been governed by contagion processes solely. The non-monotonocity of the implicit hazard rate suggests that there were contagion effects, and that they had a strong influence in the first 200 days of the crisis.