On Reverse Stress Testing for Worst Case Scenarios: An Application to Credit Risk Modeling of Tunisian Economic Sectors

In Tunisia substantial economic and financial vulnerability are mainly caused by the civil unrest and the subsequent regime change after the Tunisian revolution in 2011. In order to quantify this fragility, in this paper, we use reverse stress testing (RST) based on combination of relevant risk factors that lead to the worst case in which the bank becomes unviable and insolvent. Given the financial stress testing scenarios shortcomings (i.e. plausibility, subjectivity), the use of RST methodology is explained by the fact that it identify the probability of realization of such scenarios. We apply this new methodology on Tunisian banks which are the core for financial stability; more specifically, we focus on credit risk RST. We choose the upper bound for value at risk (VaR) in order to identify worst VaR at probability of realization a. Our proposed framework relies on the logistic regression to identify stress probability and the linear regression of financial stress index output gap versus macroeconomic risk factors in order to search for scenarios at 1%, 3% and 5% levels of plausibility. Empirical results show that Tunisian banks have likely to reach the WVaR in 2012 at 5% level. Besides the more extreme the scenarios which are considered the less plausible they become. Our reverse FST is a complement, and never a substitute for risk manager and what matters most is the mindset of those employing it.