Margin requirements based on a stochastic correlation model

Abstract

We demonstrate that margin requirements of central counterparties show a significantly different behavior when calculated with a portfoliowise treatment instead of taking the weighted sum of the margin requirements of the components without accounting for their correlation structures. This is shown via simulating trajectories of a joint stochastic volatility–stochastic correlation model. Results indicate that an unnecessarily large overmargin requirement is set by regulators when the applied risk measure is not calculated via a portfoliowise treatment. Finally, accounting for the correlation structure of the assets during the margining process would not lead to an overly prudent method, nor would it cause greater procyclicality