Systemic Risk and the Underlying Statistical Assumptions of SRISK

• Activity Number: 173
• Type: Topic Contributed
• Date/Time: Monday, August 10, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Section on Risk Analysis
• Author(s): Andrew Wilcox* and Peter Bloomfield
• Companies: North Carolina State University and North Carolina State University
• Keywords: Systemic Risk ; SRISK ; DCC-GARCH ; SIFIs ; Financial Crisis

Abstract

In the wake of the Financial Crisis of 2007-2009 government and academic researchers have focused on defining and measuring systemic risk. This research has the potential create more effective regulation regarding Systematically Important Financial Institutions (SIFIs). A leading measure of systemic risk is the SRISK index introduced by Brownlees and Engle (2011). SRISK uses a Long Run Marginal Expected Shortfall (LRMES) calculation to estimate the capital shortage a firm would be faced with conditional on a substantial market decline. In this work we show that the use of of a non-Gaussian working likelihood in their estimation procedure can produce significant changes in LRMES. Specifically longer tailed distributions have the largest effect with differences as large as 20% for certain firms and dates. Additionally we provide a bootstrapping procedure that allows us to quantify the full amount of error in estimating LRMES. These findings allow for a better understanding of SRISK and can help guide decisions regarding its use in a regulatory framework.