We analyze around 200 different financial time series, i.e. components of Dow Jones, Nasdaq, FTSE and Nikkei with seven different VaR approaches. We differentiate our analysis according to characteristics that can be observed. Our analysis shows that in high risk situations in which the time series show high volatility risk and high fat tail risk the current Basle II guidelines fail in the attempt to cushion against large losses by higher capital requirements. One of the factors causing this problem is that the builtin positive incentive of the penalty factor resulting from the Basle II backtesting is set too weak. Therefore, we propose adjustments regarding the Basle II penalty factor that take different risk situations into account and lead to higher capital buffers for forecast models with a systematic risk underestimation.
Risk evaluation Value-at-risk Basle II backtesting GARCH