开发者:上海品职教育科技有限公司 隐私政策详情

应用版本:4.2.11(IOS)|3.2.5(安卓)APP下载

Pina · 2021年01月06日

问一道题:NO.PZ201512181000007102

* 问题详情,请 查看题干

问题如下:

The number of Flusk’s VaR breaches most likely resulted from:

选项:

A.

using a standard normal distribution in the VaR model.

B.

using a 95% confidence interval instead of a 99% confidence interval.

C.

lower market volatility during the last year compared with the lookback period.

解释:

C is correct. Flusk experienced zero daily VaR breaches over the last year yet incurred a substantial loss. A limitation of VaR is its vulnerability to different volatility regimes. A portfolio might remain under its VaR limit every day but lose an amount approaching this limit each day. If market volatility during the last year is lower than in the lookback period, the portfolio could accumulate a substantial loss without technically breaching the VaR constraint.

A is incorrect because VaR was calculated using historical simulation, so the distribution used was based on actual historical changes in the key risk factors experienced during the lookback period. Thus, the distribution is not characterized using estimates of the mean return, the standard deviation, or the correlations among the risk factors in the portfolio. In contrast, the parametric method of estimating VaR generally assumes that the distribution of returns for the risk factors is normal.

B is incorrect because a specification with a higher confidence level will produce a higher VaR. If a 99% confidence interval was used to calculate historical VaR, the VaR would be larger (larger expected minimum loss). During the last year, none of Flusk’s losses were substantial enough to breach the 5% VaR number (95% confidence interval); therefore, if McKee used a 1% VaR (99% confidence interval), the number of VaR breaches would not change.

老师好 这题为什么不能选A? 因为var假设是正态分布,但真实损失的分布是左偏,所以VAR 的一定情况下的损失就估计大了,所以真实损失就一直没hit 到VAR 值? 谢谢。

2 个答案

星星_品职助教 · 2021年01月06日

同学你好,

这道题的VaR基于的是historical method,即真实历史数据和这些数据对对应的真实的分布,并不是正态分布。

Pina · 2021年01月06日

老师好,是不是我说的情况下, VAR 应该是估计小了 不是估计大了? 因为var假设是正态分布,但真实损失的分布是左偏,所以VAR 的一定情况下的损失就估计了 所以不能选A?谢谢。