ISDA has published a paper that explains why stress scenarios that central counterparties (CCPs) use for default fund sizing cannot be used for forward-looking initial margin (IM) simulators.

Typically, stress scenarios used by CCPs consist of a single step, transitioning directly from business-as-usual to stressed market conditions. For default fund sizing or liquidity requirement determinations, these one-step scenarios are sufficient because the focus is on measuring absolute loss under stress.

However, for many CCPs, the IM response function depends on the path taken during market stress events. Their models may yield different IM requirements based on how stress unfolds over time. As a result, IM simulators require scenarios that replicate the day-by-day progression of stress events and use this time series as input for margin simulation.

Most systemically important CCPs, whose margin levels are particularly relevant for clearing users, employ some form of filtered historical value-at-risk (VaR) model to calculate their IM.

These models have three pathways through which the different trajectories of a shock can influence margin levels:

1. The number of stressed scenarios influencing the VaR calculation;
2. The impact on the volatility estimate used for historical filtering;
3. The compounding effect of overlapping returns for margin periods of risk longer than one day.

This paper examines these three mechanisms and argues that for CCPs with path-dependent models, given the complex relationship between stress scenarios, VaR models and potential additional tools, scenario simulation should proceed incrementally.