1
$\begingroup$

Does the distribution of $XY$ depend on $\theta$, when $X\sim\text{Exp}(\theta)$, $Y\sim\text{Exp}(1/\theta)$ and $X$ independent with $Y$?

I understand that from Wiki Parametrization 1 we have $XY$ does not depend on $\theta$ and intuitively this is correct, but I don't know how to show it. Here is what I've done.

\begin{align*} \mathbb{P}(XY\leq t)&=\int_{0}^{\infty}\mathbb{P}(XY\leq t|Y=s)f_{Y}(s)ds\\ &=\int_{0}^{\infty}\mathbb{P}(X\leq t/s)f_{Y}(s)ds \text{ -- by X independent with Y}\\ &=\int_{0}^{\infty}\left[1-e^{\theta t/s}\right]\frac{1}{\theta}e^{-\frac{s}{\theta}}ds\\ &=\frac{1}{\theta}\int_{0}^{\infty}\left[1-e^{\theta t/s}\right]e^{-\frac{s}{\theta}}ds\\ &=\text{a function of }\theta \end{align*}

I don't know why the distribution of $XY$ is independent with $\theta$.

$\endgroup$
2
  • 1
    $\begingroup$ It is because $X=\frac1{\theta} U$ and $Y=\theta V$ where $U,V\sim \text{Exp}(1)$. $\endgroup$ – StubbornAtom Jan 9 at 19:40
  • $\begingroup$ @StubbornAtom ha, thanks a lot. I was trying to find a general way of finding an ancillary statistic and missed this smart solution $\endgroup$ – Tan Jan 9 at 19:47
2
$\begingroup$

The distribution of $X\theta$ does not depend on $\theta$; it is Exp(1).

The distribution of $Y/\theta$ does not depend on $\theta$; it is also Exp(1).

So the distribution of $(X\theta)(Y/\theta)$ does not depend on $\theta$. But that is just the distribution of $XY$

$\endgroup$

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.