Stress-Testing for Credit Portfolios in the Mexican Banking System

Abstract

The aim of this paper is to present a scheme to assess vulnerability to adverse macroeconomic shocks in aggregate loan portfolios of the Mexican banking system. The used database consists of monthly aggregate information of the radius between nonperforming loans and total loans, for the period 2000-2014, for a total of 65 financial institutions. Through the Seemingly Unrelated Regressions (SUR) methodology, we estimated an equations system to link the default probability and macroeconomic factors, the obtained system was used to estimate the probability distributions, unconditional and conditional upon the occurrence of particular shocks in relevant macroeconomic variables. The outcome indicates that the estimated loss from the different distributions under stress does not increase significantly, so, we conclude that the risk in this sector is moderate under the selected scenarios. The most important result of the work is the obtained relation between default probabilities and stressed macroeconomic factors, but these results are limited by the probability of ocurrence of these induced stress scenarios and the horizon of the estimated econometric model.