Journal of Financial Market Infrastructures

Study of correlation impact on credit default swap margin using a GARCH–DCC-copula framework

David Li and Roy M. Cheruvelil

  • Correlation among different assets is an important factor in determining margin charges to members relating to the portfolios presented for clearing at CCPs.
  • There are limitations to the use of overly simplified correlation assumptions in CCP margin models.
  • The first motivation of this research is to apply and understand time-varying, conditional correlation for a representative margin model to isolate the effect of correlation risk. The second motivation is to understand how sensitive various portfolios are to correlation shocks and identify tools available to mitigate correlation risk.
  • The research indicates it may be prudent to account for correlation dynamics when calculating margin at CCPs.

We establish generalized autoregressive conditional heteroscedasticity–dynamic conditional correlation (GARCHDCC) and constant conditional correlation (CCC) copula model frameworks to study time-varying correlation among credit default swap (CDS) single names (SNs) and its impact on certain risk measures of CDS portfolios that consist of names from different sectors within the eurozone (EU) and North America (NA). Our purpose is to better understand the direction and magnitude of impacts on such risk measures due to correlation changes. This study covers 188 NA SNs and 145 EU SNs from January 2008 to August 2017. We find that correlations between CDS SNs go through different correlation regimes during this period. As a result, CDS portfolio risk measures in the form of value-at-risk or expected shortfall show sizable variation due to correlation regime shifts from historical means. Depending on the correlation level (high or low) and the portfolio type, risk measures could be either underestimated or overestimated. Both directional and balanced portfolios could experience a sizable underestimation of the margin depending on the direction in which the correlation deviates from historical means.

Sorry, our subscription options are not loading right now

Please try again later. Get in touch with our customer services team if this issue persists.

New to View our subscription options

If you already have an account, please sign in here.

You need to sign in to use this feature. If you don’t have a account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here