We analyze the performance of RiskMetrics, a widely used methodology formeasuring market risk. Based on the assumption of normally distributed returns,the RiskMetrics model completely ignores the presence of fat tails in thedistribution function, which is an important feature of financial data.Nevertheless, it was commonly found that RiskMetrics performs satisfactorilywell, and therefore the technique has become widely used in the financialindustry. We find, however, that the success of RiskMetrics is the artifact ofthe choice of the risk measure. First, the outstanding performance ofvolatility estimates is basically due to the choice of a very short (one-periodahead) forecasting horizon. Second, the satisfactory performance in obtainingValue-at-Risk by simply multiplying volatility with a constant factor is mainlydue to the choice of the particular significance level.
展开▼
