Does the Value-at-Risk legal framework lead to inaccurate and procyclical risk estimations? Empirical Evidence from the EU countries.

Abstract

This analysis tests whether the quantitative requirements of the law (Basel and Committee of European Securities Regulators) regarding the Value at Risk (VaR) framework may lead to inaccurate and procyclical VaR estimations. We apply two of the most popular VaR models, the Historical (HVaR) and the Exponential Weighted Moving Average (EWMA VaR) models, to a wide sample of 13 European Indices during the period 2002-2019. The empirical evidence confirms our assumptions that the legal framework in many cases leads to inaccurate and procyclical VaR estimations. Moreover, we show that the limitation on the required data inputs does not really contribute to a more stable financial environment. Further, we show that the current framework does not examine the procyclicality issue. The evidence in this study shows that the current legal framework needs some reforms: (a) the guideline on the minimum number of data inputs for the VaR estimations should be removed, taking provided that the accuracy of the applied VaR model is often evaluated, and (b) the current backtesting procedure does not examine whether VaR estimations are representative of the financial conditions. An additional backtesting procedure at a lower that the 99% confidence level could resolve this issue.