3
$\begingroup$

I'm asked to verify if the following estimator of $\mu$: $$\hat{\mu}=\frac{1}{n+1}\sum_{t=1}^{n}y_t$$ converges in probability by computing its $p \lim$ (note that $E[y_t]=\mu$). I've made the following computations:

\begin{align} p \lim(\hat{\mu})&=p\lim \left ( \frac{1}{n+1}\sum_{t=1}^{n}y_t \right) \\ &=p\lim \left ( \frac{1}{n+1} \frac{n}{n}\sum_{t=1}^{n}y_t \right ) \\ &=p\lim \left ( \frac{n}{n+1} \frac{1}{n}\sum_{t=1}^{n}y_t \right ) \\ &=\frac{n}{n+1}\ p\lim \left ( \frac{1}{n}\sum_{t=1}^{n}y_t \right ) \\ &=\frac{n}{n+1}\ p\lim \left ( E[y_t] \right ) \\ &=\frac{n}{n+1}\ \mu \end{align}

Can I say that for $n \to +\infty$ the fraction $n/(n+1)$ is equal to 1 and thus $p \lim(\hat{\mu})=\mu$ or must I leave the ratio in front of $\mu$?

$\endgroup$
1
  • $\begingroup$ In the first three lines of your calculations, "$n$" was a bound variable: you could have replaced it by any variable name you please without changing the meaning. At the fourth line, where $n/(n+1)$ appears outside the limit, "$n$" suddenly appears unbound--and thereby is meaningless. That line is invalid. $\endgroup$
    – whuber
    Oct 19, 2015 at 16:03

1 Answer 1

3
$\begingroup$

You can say that the limit of the fraction $\frac{n}{n+1}$ is $1$ and use the Law of Large Numbers for the sample mean, yes. It's a mistake however to take the limits sequentially. They have to be taken at the same time. Here for instance, you can use the fact that if $A_n \xrightarrow{P} A$ and $B_n \xrightarrow{P} B$ then $A_n B_n \xrightarrow{P} AB$. And you are done.

$\endgroup$

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.