With standard setters considering further guidance on initial margin in centrally cleared markets, the World Federation of Exchanges (WFE) published joint research with David Murphy, visiting professor at the Department of Law of the London School of Economics (LSE) introducing a novel measure of initial margin model reactiveness (or procyclicality) that, unlike other measures, does not rely on particular risk factor paths and quantifies the uncertainty in the measurements.
The WFE Research Working Paper, “The Impulsive Approach to Procyclicality”, arrives soon after the recent BCBS-CPMI-IOSCO Review of Transparency and Responsiveness of Initial Margin in Centrally Cleared Markets, which seeks to establish a standardized measure of margin responsiveness across the international clearing community.
The responsiveness of initial margin models, both for centrally and non-centrally cleared trades, has been part of the discussions about how to best manage liquidity pressures in the financial system. However, the debate about how reactive margin should be to changes in market conditions has been hampered by the lack of a generally accepted and statistically sound way of measuring such reactiveness.
Common measures of model procyclicality, including the method being proposed by the Joint Working Group on Margin (JWGM) set up by the Basel Committee on Banking Supervision, are strongly dependent on the individual scenarios and, as a consequence, cannot be used to compare margin models across different market conditions and are unlikely to provide a sufficient understanding of how models (and anti-procyclicality tools) could behave when there is an unanticipated change in market conditions. They also ignore the uncertainty surrounding the measurements.
The methodology proposed in this research overcomes these limitations, measuring margin responsiveness to shifts in volatility and market dynamics by:
- Utilizing an impulse response function (IRF) in a Monte Carlo simulation setting,
- Capturing the variability (uncertainty) surrounding margin model reactiveness,
- Providing a tool to assess and compare models that does not depend on a particular scenario.
The results presented demonstrate that a model’s impulse response is a robust and useful measure of its reactiveness. Using this measure, the analysis also sheds light on different aspects of procyclicality management:
- There is a trade-off between model reactiveness and the uncertainty of future model behavior,
- The use of a stress period and the addition of a buffer, two commonly used anti-procyclicality (APC) tools, do not significantly reduce the likelihood of a model over- or under-reacting. In particular, the buffer performs poorly.
- The impact of the choice of core margin model far exceeds the impact of the APC tools analyzed,
- While the filtered Value at Risk (VaR) models analyzed offer the benefit of higher speeds of reaction, they tend to over-react to sharp, stepwise increases in volatility.
These results support the adoption of an outcome-based approach to procyclicality, recognizing that there is no single correct level of procyclicality, but only acceptable choices given a specific situation (including risk factor dynamics, portfolio characteristics, participants’ funding liquidity arrangements), and the desired trade-off between reactiveness, potential extent of over-reaction, and margin accuracy.
This research also highlights the need for an adaptation of the current methodology proposed by BCBS-CPMI-IOSCO before it is put into effect: either by capturing the degree of uncertainty in the measurement, or by including an appropriate warning of the standard setter’s proposed method’s limitations.
Pedro Gurrola-Perez, head of Research at the World Federation of Exchanges, said in a statement: “As the policy landscape looks set for change in this area, the research we have published today contributes to the sum of knowledge on this topic, sheds light on the shortcomings of the current approaches and proposals to measure model reactiveness (shortcomings which have often obscured the debate around procyclicality), and provides an alternative method for consideration which brings clarity to the discussion and provides the concepts of a model being over- or under-reactive with a statistically robust footing.”
David Murphy said in a statement: “As someone who has been studying the procyclicality of initial margin for some years, I am very happy to have been involved with this work. The importance of the measurement of procyclicality is increasingly recognized. For the first time, this paper presents a measure that allows different initial margin models to be robustly compared.”