Capturing initial margin in counterparty risk calculations

Abstract

This paper compares a range of alternative approaches to incorporate Initial Margins (IMs) in the modelling of counterparty credit risk exposures. IMs have risen in importance following the rise of Central Counterparties to clear OTC derivatives and the recent legislation requiring bilateral margining for uncleared derivatives between financial counterparties. IMs have become an essential model component that drives exposure, associated regulatory capital requirements and valuation adjustments such as Credit Valuation Adjustment (CVA) and Margin Valuation Adjustment (MVA). The influence of the modelling choices is explored by means of typical derivatives portfolios. For the actual estimation of a path-dependent (‘stochastic’) IM through time the use of quantile regression is suggested as an econometrically reliable approximation. Banks’ internal counterparty risk models will likely exhibit a basis vis-à-vis the actual IM mechanisms in practice (for example, owing to different risk factor representations and/or calibrations). In this context, the paper suggests that a simplified representation in the form of a ‘dynamic IM’ can approximate most of the quantities of interest to a reasonable degree.