The Impact of Initial Margin on Derivatives Pricing with an Application of Machine Learning

Abstract

In this paper we consider the impact of bilateral initial margin on derivatives pricing. We first introduce the background of bilateral initial margin. Then, we focus on how initial margin effects counterparty credit exposures, capital requirements and funding costs. Nowadays, besides risk-neutral valuation principles, these components are included when pricing derivatives through so-called valuation adjustments (xVAs). Based on this, we present an approach which incorporates initial margin into existing xVA frameworks. Then, we give an overview of the most common methods for calculating future initial margin. Moreover, we present a regression model which is based on machine learning methods. Finally, we show some numerical results for a cross currency swap, an fx option and a Bermudan swaption including a scenario analysis. Here, it turns out that it is much more important to consider margin agreement, regulatory or non-modellable parameters in detail than implementing a most accurate dynamic initial margin model.