Model-based approach for scenario design: stress test severity and banks' resiliency

Published in: Quantitative Finance

This paper proposes a methodology to analyze stress test exercises conducted by the European Banking Authority (EBA) in the Italian banking sector. Specifically, it aims to assess (1) whether the severity of adverse scenarios has increased over time and (2) if the resilience of the Italian banking sector has improved.

The analysis is conducted using the Italian Banking Sector Evaluation (IBASE) model, a large Bayesian vector autoregression (LBVAR) model that includes macro-financial and bank balance sheet variables. In the model bank, macro-financial variables are closely interconnected. In particular, bank variables affect macro-financial variables, a feature absent in traditional stress testing models. Despite the limited scale of our model, any potential feedback among variables is allowed, and we show that it is a valid tool for scenario design in macroprudential stress testing.

Severity is measured by calculating the probability of realization of adverse scenarios for individual variables based on conditional forecast simulations from the model.

The 2018 stress test was the most severe overall, particularly for GDP, Euribor rates, and stock prices. Moreover, we find that the profile of GDP in the last exercise is comparable in terms of severity with previous crisis episodes (Subprime and Sovereign debt crisis), while the drop in stock prices is more severe than in past exercises.

We also find that the joint realization probability of the EBA adverse scenarios is extremely low; as such it is impossible to give an overall assessment of the severity of the exercises.

This is also due to the positive correlation between short-term rates and GDP that we find in the data when looking at past crisis episodes.

We, therefore, propose a novel approach. We develop a novel contribution, which is an aggregate measure of severity, defined as the weighted average of the severity of each variable‡, where the weights are given by the contribution of the shock of each variable to the variance of the bank aggregates. Starting from severity evaluation, we find that the severity of EBA’s regulatory stress tests has increased over time, suggesting that European regulators are ‘raising the bar’ on banks’ ability to survive periods of economic stress.

Counterfactual stress tests are constructed to evaluate resilience by rescaling the 2016 and 2018 scenarios to match the 2014 severity levels. Under these new artificial scenarios, we can evaluate the response of the banking system at different points in time to now comparable stress scenarios. Comparing responses shows that by 2018, banks could allow for smaller loan declines while still experiencing a larger (but still manageable) Tier 1 ratio reduction. We take this result as evidence that the resiliency of the banking sector to adverse stress scenarios has increased: banks can accommodate a lower decline in loans, i.e., maintain a higher exposure to risk in their balance sheets while keeping capital adequacy levels under control.

In summary, this paper develops a model-based approach to stress test analysis that considers dynamic bank balance sheets and potential feedback effects. It finds evidence that EBA stress tests have become more severe over time and that the Italian banking sector has become more resilient to adverse shocks through risk management practices. The methodology provides a framework to evaluate stress test scenarios and banking sector resilience systematically.

Read the full article.

Authors at the Department of Management

Giuseppe Lusignani - Full Professor

Academic disciplines: Financial Markets and Institutions

Teaching areas: Banking and Finance, Financial Derivatives, Investments, Risk Management.

Research fields:  Financial Crisis, Financial Intermediaries, Banking Regulation, Risk Management, Capital Adequacy.

Full Professor of Economics of Financial Intermediaries at the University of Bologna. Member of the Prometeia Associazione Scientific Committee, Honorary member, and Chairman of the Ethics and Disciplinary Committee of AIFIRM, the Italian Financial Risk Managers Association. Editorial board director of the journals Banca Impresa Società (Il Mulino) and Bancaria (Bancaria Editrice).