Computing MVA via Regression and Principal Component Analysis

Abstract

Financial institutions are currently facing the challenge of incorporating future bilateral initial margin into their counterparty credit risk exposure simulations - for accurately computing their xVAs, their limit utilizations or their regulatory CCR capital requirements. In this article, we present a method for approximating future bilateral initial margin amounts. Our method uses path dependent sensitivities with respect to model state variables, and it uses the pseudo-inverse of the model-to-market Jacobian. We assess our method’s performance for interest rate swaps and for European swaptions, in the case where the risk-neutral dynamics of interest rates is given by the Hull-White one factor model.