Validating Risk Estimation Models in Russian Market

Abstract In world practice in financial markets, the Value-at-Risk model (hereinafter VaR) has been recognized as a benchmark of measuring risks. Therefore, VaR is usually used to establish and monitor risk limits and is also used to calculate capital requirements in financial institutions. However, despite the widespread use of VaR, there is no single standard for testing this risk model. Backtests are based on the number of assumptions, and the accuracy of the VaR depends on these assumptions. In this work, using the Monte-Carlo experiment, it was empirically revealed which backtests are more effective when violating the assumption of independence and the same distribution of breakdowns of the parametric Value at Risk in the developing Russian financial market. The results in the form of power tests for different nominal sizes, Value at Risk levels and sample sizes are presented in a table at the end of this work. Keywords: Parametric VaR, NAGARCH, violation of assumptions, power of backtests.