极值理论(EVT)在汇率受险价值(VaR)计算中的应用

The application of extreme value theory on computing value-at-risk of exchange rates

本文讨论根据极值理论 (EVT)计算金融市场风险重要量度——受险价值 (Va R)的一种新方法 ,给出金融资产组合收益或损失尾部分布的二阶展开式的参数估计形式 ,并以此为基础提出用确定临界值并估计Va R的“两次子样试算法”,最后用 1971- 1998年的日元 /美元汇率 6 70 0多个历史数据验证在极端条件下用EVT估计 Va R具有很高的准确性 .

As an integrated method to measure finance risks, value at risk(VaR) has been investigated widely in recent years. This paper introduces a new model to compute VaR - extreme value theory (EVT). It gives parametric estimations of the tail distribution by means of second order expansion. A percedure and program named “two step subsample bootstrap” was developed to find the start point of extreme and to estimate VaR. The computation result of JPY/USD foreign exchange rate presented at last proved that the EVT method has a good character in estimating VaR.