Let $X \sim \mathcal N(\mu, \sigma^2)$ which denotes some random costs. By making the investment $k > 0$ one can change costs to $Y(k) \sim \mathcal N(e^{-k}\mu,e^{-2k}\sigma^2)$ such that costs changes from $x$ to $y(k) + k$. I have from a paper the following relationship, but don't know how to verify it. Can someone hint me to what happens in between? \begin{align} \mathbb E\left[e^{Y(k) + k}\right] = \exp\left(e^{-k}\mu + \frac{1}{2}e^{-2k}\sigma^2 + k\right) \end{align}
Asked
Active
Viewed 1,170 times
0
-
what is the MGF of normal distribution? – MAN-MADE Sep 20 '17 at 12:15
-
Because $Y$ takes the mean and sd of $X$ changed through $k$. – clueless Sep 20 '17 at 12:19
-
Thanks. I added a line. There is not more in it. – clueless Sep 20 '17 at 12:29
-
Once $\mu$ and $\sigma$ are assumed known, the information about $X$ is irrelevant, unless you are not telling the whole story. – bluemaster Sep 20 '17 at 12:29
-
2clueless: Here is a clue: try to avoid accepting an answer a mere 4 minutes after it is posted. – Did Sep 20 '17 at 13:00
1 Answers
2
\begin{align} & \operatorname{E} (e^{Y(k) +k }) = (2\pi )^{-\frac{1}{2}}e^k \sigma^{-1}\int_{\mathbb{R}}e^{x+k}e^{\frac{(x-e^{-k}\mu)^2}{2e^{-2k}\sigma^2}} \, dx \\[10pt] = {} & (2\pi )^{-\frac{1}{2}}e^{2k} \sigma^{-1}\int_{\mathbb{R}}e^{-e^k \mu +\frac{1}{2}e^{-2k}\sigma^2}e^{\frac{(x-(e^{-k}\mu -e^{-2k}\sigma^2))^2 }{2e^{-2k}\sigma^2}} \, dx \\[10pt] = {} & e^k \cdot e^{-e^k \mu +\frac{1}{2}e^{-2k}\sigma^2}=e^{-e^k \mu +\frac{1}{2} e^{-2k} \sigma^2 +k} \end{align}
-
Going back to the integral is certainly not the easiest way (nor the most illuminating one). – Did Sep 20 '17 at 12:59