Credit Default Swap Spreads and Systemic Financial Risk

Abstract

This paper presents a novel method to measure the joint default risk of large financial institutions (systemic default risk) using information in bond and credit default swap (CDS) prices. Bond prices reflect individual default probabilities of the issuers. CDS contracts, which insure against such defaults, pay o only as long as the seller of protection itself is solvent. Therefore, CDS prices contain information about the probability of joint default of both the bond issuer and the protection seller. If we consider the entire set of CDS contracts written by each financial institution against the default of each other institution we can learn about all pairwise default probabilities across the financial network. This information, however, is not sucient to completely characterize the joint distribution function of defaults of these banks. In this paper, I show how this information can be optimally aggregated to construct bounds on the probability of systemic default events. This method enables me to measure systemic default risk without making any assumptions about the joint distribution function. Two main results emerge from the empirical application of this method to the recent financial crisis. First, I show that an increase in systemic risk in large global banks did not occur until after Bear Stearns’ collapse in March 2008. Second, some of the large observed spikes in CDS spreads and bond yield spreads during this period (for example, following Lehman Brothers’ default) correspond to spikes in idiosyncratic default risk rather than systemic risk.