Consider a random sample from exponential distribution with mean $\frac{1}{\theta}$. I have to prove that $nX_{(1)}$ is not consistent for $\frac{1}{\theta}$ . A sufficient condition for consistency is

$$\lim_{n \to \infty}E(\hat\theta)=\frac{1}{\theta}\quad ;\quad \lim_{n \to \infty}\operatorname{Var}(\hat\theta)=0$$

Since $X_{(1)}$ follow $\exp(n\theta)$, the second condition, was seen to be violated as its variance $\frac{1}{\theta^{2}}$ doesn't tend to zero. Is this enough to prove that $nX_{(1)}$ is inconsistent?

Can i say that since $X_{(1)}$ converge in probability to $\frac{1}{n\theta}$, $nX_{(1)}$ converge in probability to $\frac{1}{\theta}$. But this does not prove what i want.

Im confused about the right method to be used.


The sufficient condition you use is not strong enough to conclude anything. Proceed from definition.

To restate convergence in probability somewhat roughly, $T_n$ is said to be a consistent estimator of the parametric function $g(\theta)$ if for every positive $\varepsilon$, $P_{\theta}\left[|T_n-g(\theta)|<\varepsilon\right]\to 1$ as $n\to\infty$ for all $\theta$ (or equivalently $P_{\theta}\left[|T_n-g(\theta)|>\varepsilon\right]\to 0$).

Since $nX_{(1)}$ is an exponential variable with mean $1/\theta$, you will find that the probability $P_{\theta}\left[\left|nX_{(1)}-\frac{1}{\theta}\right|<\varepsilon\right]$ does not even depend on $n$. It does not converge to $1$. Hence proved.


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.