The Collateral Rule: An Empirical Analysis of the CDS Market

We study the empirical determinants of collateral requirements in the cleared credit default swap (CDS) market: how margins depend on portfolio risks and market conditions, and what the implications are for theoretical models of collateral equilibrium. We construct a novel data set containing CDS portfolios and margins posted by all participants to the main CDS clearinghouse, ICE Clear Credit, covering 60% of the U.S. market. We provide direct empirical evidence that margins are much more conservatively set than what a Value-at-Risk (VaR) rule would imply, and are unequally implemented across participants. We show that more extreme tail risk measures have a higher explanatory power for observed collateral requirements than VaR, consistent with endogenous collateral theories such as Fostel and Geanakoplos (2015) where extreme events dominate in determining collateral. The dependence of collateral requirements on extreme tail risks induces potential nonlinearities in margin spirals, dampening small shocks and amplifying large ones. We also confirm empirically the main channel through which collateral-feedback effects operate in many theoretical models of equilibriums with financial frictions, such as Brunnermeier and Pedersen (2009), highlighting the prominent role of aggregate volatility and funding costs.