2
$\begingroup$

I have some questions when dealing with Value-at-Risk (VaR) and Conditional Value-at-Risk (CVaR).

Is there any relationship between $VaR_{\alpha}(X)$ and $VaR_{\alpha}(- X)$, or $CVaR_{\alpha}(X)$ and $CVaR_{\alpha}(-X)$ ?

Here, $VaR$ and $CVaR$ are defined as:

$$VaR_{\alpha}(X) := \inf \left\{x\in \mathbb{R}| Pr(X >x)\leq \alpha \right\}, \alpha \in [0, 1]$$

$$CVaR_{\alpha}(X) := \frac{1}{\alpha}\int_{0}^{\alpha}VaR_{s}(X)ds$$

$\endgroup$
  • $\begingroup$ Down-voted, I'd like to know what have you tried? Did you write down the definitions? $\endgroup$ – Bob Jansen Sep 28 '17 at 14:07
  • $\begingroup$ I believe you should improve your question to get a better response. $\endgroup$ – Bob Jansen Sep 29 '17 at 5:45
  • $\begingroup$ So, given those formula's, would you say that there is a relationship? $\endgroup$ – Bob Jansen Sep 29 '17 at 8:35
  • $\begingroup$ Actually, I am not sure. Since $VaR_{\alpha}(-X) = \sup{x \in \mathbb{R}| F_{X}(x) \leq 1-\alpha}$}, I was wondering whether there could be some relation between $VaR_{\alpha}(-X)$ and $VaR_{\alpha}(X)$. It does not seem apparently. Maybe some relation holds by applying some trick. $\endgroup$ – Xinyuan Sep 29 '17 at 9:07
2
$\begingroup$

We consider the case where the distribution function $F$ of $X$ is strictly increasing. Then \begin{align*} VaR_{\alpha}(X) &= \inf\{x: P(X >x) \le \alpha \}\\ &=\inf\{x: F(x)\ge 1-\alpha \}\\ &=F^{-1}(1-\alpha). \end{align*} Moreover, we note that the distribution function $G$ of $-X$ is defined by \begin{align*} G(x) &= P(-X \le x) \\ &=1-F(-x), \end{align*} Then, \begin{align*} VaR_{\alpha}(-X) &= G^{-1}(1-\alpha)\\ &=-F^{-1}(\alpha)\\ &=-VaR_{1-\alpha}(X). \end{align*} Furthermore, \begin{align*} CVaR_{\alpha}(-X) &=\frac{1}{\alpha}\int_0^{\alpha}VaR_s(-X)ds\\ &=-\frac{1}{\alpha}\int_0^{\alpha}VaR_{1-s}(X)ds\\ &={\color{red}{-}}\frac{1}{\alpha}\int_{1-\alpha}^1 VaR_s(X)ds\\ &=-\frac{1}{\alpha}\left(\int_0^1 VaR_s(X)ds - \int_0^{1-\alpha} VaR_s(X)ds\right)\\ &=-\frac{1}{\alpha}\int_0^1F^{-1}(1-s)ds +\frac{1-\alpha}{\alpha}CVaR_{1-\alpha}(X)\\ &=-\frac{1}{\alpha}\int_0^1F^{-1}(s)ds +\frac{1-\alpha}{\alpha}CVaR_{1-\alpha}(X)\\ &=-\frac{1}{\alpha}\int_{-\infty}^{\infty} x\, dF(x) +\frac{1-\alpha}{\alpha}CVaR_{1-\alpha}(X)\\ &=-\frac{1}{\alpha} E(X) +\frac{1-\alpha}{\alpha}CVaR_{1-\alpha}(X). \end{align*}

$\endgroup$
  • $\begingroup$ Hi Gordon. Thank you again for your help. Your answer is really helpful. But I still doubt about a sign. Could you help me check it ? \begin{equation} \begin{split} CVaR_{\alpha}(-X) &= \frac{1}{\alpha}\int_{0}^{\alpha}VaR_{s}(-X)ds \\ &= -\frac{1}{\alpha}\int_{0}^{\alpha}VaR_{1-s}(X)ds\\ &= -\frac{1}{\alpha}\int_{1}^{1-\alpha}VaR_{t}(X)d(1-t)\\ &= -\frac{1}{\alpha}\int_{1-\alpha}^{1}VaR_{t}(X)dt)\\ &= -\frac{1}{\alpha}\left(\int_{1-\alpha}^{0}VaR_{t}(X)dt)+\int_{0}^{1}VaR_{t}(X)dt)\right)\\ \end{split} \end{equation} $\endgroup$ – Xinyuan Oct 11 '17 at 10:59
  • $\begingroup$ \begin{equation} \begin{split} &= -\frac{1}{\alpha}\left(\int_{0}^{1}VaR_{t}(X)dt)-\int_{0}^{1-\alpha}VaR_{t}(X)dt)\right)\\ &= -\frac{1}{\alpha}\left(\int_{0}^{1}F^{-1}(1-t)dt - (1-\alpha)CVaR_{1-\alpha}(X)\right)\\ &= \frac{1}{\alpha}\int_{1}^{0}F^{-1}(s)ds + \frac{1-\alpha}{\alpha}CVaR_{1-\alpha}(X)\\ &= -\frac{1}{\alpha}\int_{0}^{1}F^{-1}(s)ds + \frac{1-\alpha}{\alpha}CVaR_{1-\alpha}(X)\\ &=-\frac{1}{\alpha}E[X] + \frac{1-\alpha}{\alpha}CVaR_{1-\alpha}(X). \end{split} \end{equation} $\endgroup$ – Xinyuan Oct 11 '17 at 11:00
  • $\begingroup$ @Xinyuan: Thanks for pointing this out. See the correction. $\endgroup$ – Gordon Oct 11 '17 at 12:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.