How confident are we of margin model procyclicality measurements?

Measuring the responsiveness of a market risk model is relevant whenever the focus is on evaluating if a model is over- or under-reacting to changes in market conditions. Such is the case, for example, in the discussion about procyclical effects of the initial margin models used both in the centrally cleared and non-centrally cleared worlds to estimate the potential future exposure of portfolios. By definition, these models are sensitive to changes in market risk and, as a consequence, when market risk increases, initial margin requirements will tend to increase. To mitigate procyclical effects, CCPs have put in place different procyclicality mitigation tools across their risk management arrangements. However, after the Covid-19 stress, there have been renewed discussions about further monitoring, measuring, and mitigation of models' procyclicality. This paper contributes to the discussion by bringing the attention to the fact that the standard measures of model responsiveness are random variables and, as such, are subject to uncertainty. Therefore, any decision or policy making that is based on these measures requires, to be robust, to take into consideration the impact that uncertainty will have on expected outcomes. To estimate such impact, the analysis examines the case of some typical margin models, both empirically and within a Monte Carlo simulation setting. The results show there is a significant amount of uncertainty when measuring responsiveness, which raises questions about the effectiveness of model-focused, hard-rule approaches to procyclicality