Measuring Systemic Risk With A Dynamic Copula-Based Approach

Abstract

This study examines the extent to systemic risk embedded in the credit and equity markets using a conditional value-at-risk (CoVaR) measure. To implement CoVaR, we employ a copula-based approach by considering different perspectives of dependence structure - a time-invariant setup with a variety of copula functions and a time-variant setup with the generalized autoregressive score model. We select the data of credit default swap (CDS) spread and stock price of the five individual companies traded in the US market - American Express, BBVA, Goldman Sachs, Morgan Stanley, and Wells Fargo - from 2001 to 2013 by dividing the three time-slots (pre-crisis, during-crisis, and post-crisis). We conduct the marginal modeling, copula parameter estimation, then we compute CoVaR values for all the asset-pairs using the best-fit copula model. We find that Student t copula is the best-fit structure and CoVaRs during the crisis period were the highest in all the asset pairs.