Comparative Analysis of Value at Risk Estimates Based on Procyclicality

The initial margin requirements for a portfolio of financial instruments are calculated through risk-based models, which commonly lose in their accuracy during periods of market stress. Usually, margin requirements for the same instruments take lower values in periods before market stress and support higher unjustified values in crisis and post-crisis periods for a long time. In other words, risk models are commonly procyclical. It can become a reason of unpleasant consequences for counterparties. In this work we make a comparative analysis between existing models of Value at Risk estimation and on the base of the investigation of the procyclicality effect suggest our own way of market risks modelling. The accuracy and efficiency of the proposed method are estimated trough backtesting and certain procyclicality measures that are presented in this paper.