I need to calculate the variance of the following product on the log scale, $$ Z \sim \alpha \beta X Y $$ where $\alpha$ and $\beta$ are both scalars and $X \sim LN(\mu, \sigma^2)$ and $Y \sim U(a, b)$. $X$ and $Y$ are independent of each other. What I'm interested in is, $$ Var(\log Z) $$ On the log scale the scalars will not affect the variance, so is this correct? $$ Var(\log Z) = Var(\log(XY)) = Var(\log(X) + \log(Y)) = Var(\log X) + Var(\log Y) $$ I know $\sigma^2$, which is $Var(\log X)$, and I can calculate $Var(\log Y)$ form here.

Is this approach valid?

  • 3
    $\begingroup$ You may need to assume that parameter $a>0$ (currently not stated). $$Var(\log Y) = 1-\frac{a b (\log (b)-\log (a))^2}{(b-a)^2}$$ $\endgroup$ – wolfies Jul 18 '17 at 16:35
  • 1
    $\begingroup$ You are correct, that's the way to go. $\endgroup$ – jbowman Jul 18 '17 at 17:09

Your Answer

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

Browse other questions tagged or ask your own question.