Making margin models less procyclical

Last month, the Bank of England (BOE) published a Financial Stability Paper entitled “An investigation into the procyclicality of risk-based initial margin models”. After the Global Financial Crisis, there has been growing concern that procyclical margin requirements (margins are higher in times of market stress and lower in calm markets) induce complacency in good times and panic in bad times. There is therefore a desire to reduce procyclicality, but this is difficult to do without sacrificing the risk sensitivity of the margin system.

The BOE paper uses historical and simulated data to compare various margin models on their risk sensitivity and their procyclicality. Though they do not state this as a conclusion, their comparison does show that the exponentially weighted moving average (EWMA) model with a floor (minimum margin) is one of the better performing models on both risk sensitivity and procyclicality. This is gratifying in that India uses a system of this kind.

However, the study leaves me quite dissatisfied. First procyclicality is measured in terms of elevated realized volatility. Market stress in my view is better measured by implied volatility (for example, the VIX) and by measures of funding liquidity. Second, the four models that the paper compares are all standard pre-crisis models. Even when they use simulated data from a regime switching model, they do not consider margin model based on regime switching. Nor do they consider models based on fat tailed distributions. There are no models that adjust margins slowly to reduce liquidity stresses in the system. Finally, they do little to quantify the tradeoff between risk sensitivity and procyclicality – how much risk sensitivity do we have to give up to achieve a desired reduction in procyclicality.

Posted at 2:16 pm IST on Thu, 19 Jun 2014         permanent link

Comments