Systemic Risk of Commercial Banks: A Markov-Switching Quantile Autoregression Approach

This paper extends the Conditional Value-at-Risk approach of Adrian and Brunnermeier (2011) by allowing systemic risk structures subject to economic regime shifts, which are governed by a discrete, latent Markov process. This proposed Markov-Switching Conditional Value-at-Risk is more suitable to Supervisory Stress Scenario required by FederalReserve Bank in conducting Comprehensive Capital Analysis and Review, since it is ca-pable of identifying the risk states in which the estimated risk levels are characterized. Applying MSCoVaR to stress-testing the U.S. largest commercial banks, this paper finds that the CoVaR approach underestimates systemic risk contributions of individual banks by around 131 basis points of asset loss on average. In addition, this paper constructs Banking Systemic Risk Index by value-weighted individual risk contributions for specifically monitoring the systemic risk of the banking system as a whole.