The Impulsive Approach to Procyclicality. Measuring the Reactiveness of Risk-based Initial Margin Models to Changes in Market Conditions Using Impulse Response Functions

Abstract

In recent years, many derivatives market participants received large margin calls in episodes of elevated market volatility such as the onset of the Covid-19 global pandemic and the illegal Russian invasion of Ukraine. The lack of some market participants’ preparedness to meet margin increases resulted in liquidity stress and reinvigorated the policy debate about how reactive margin should be to changes in market conditions. This debate has been hampered by the lack of a generally accepted way of measuring the reactiveness of the models used to calculate initial margin. The first contribution of this paper is to provide such a measure. We consider a step function in volatility, and examine the responses of various initial margin models to paths of risk factor returns consistent with this impulse, introducing the impulse response function as a convenient means of presenting this reaction. The results presented demonstrate that a model’s impulse response is a robust and useful measure of its reactiveness. This approach could be used both to measure initial margin model reactiveness, or procyclicality as it is often termed, and to capture the uncertainty in this measurement. It also provides significant, novel insights into the behaviour of some economically important margin models. In particular, the tendency of some filtered historical simulation value at risk models to over-react to sharp stepwise increases in volatility is demonstrated and the reasons for it are explored. The behaviour of two widely-used anti-procyclicality tools, the buffer and the use of a stressed period, are also analysed: the latter is found to be more successful at mitigating procyclicality than the former. The paper concludes with a discussion of the policy implications of the results presented.

Keywords: Anti-procyclicality, impulse response function, initial margin model, margin model response, procyclicality, volatility estimation

JEL Classification: G13, C52, C12