0

the OLS estimators of the short-run parameters are $T^{1/2}$ consistent. What is the$T^{1/2}$ consistent ?

1190
  • 848
  • 3
  • 8
  • 20
  • A closely related question was asked here: http://stats.stackexchange.com/questions/207264/root-n-consistent-estimator-but-root-n-doesnt-converge/207281#207281 – Christoph Hanck Oct 03 '16 at 07:54

2 Answers2

1

Let the model be, $y=X\beta+e$ where $e_i\sim i.i.d N(0,\sigma^2<\infty)$ and $i=1,2,\ldots,T$. Then square root consistency means,

$$ \sqrt T (\beta-\hat\beta)\overset{d}{\rightarrow} N(0,(X'X)^{-1} \sigma^2) $$

TPArrow
  • 2,155
  • 11
  • 22
  • Thank you for your explanations. I have two more questions. First one is that what is asymptotically singular ? and second one what is asymptotically perfectly collinear? Please give an explanation thank you again :) – 1190 Oct 02 '16 at 14:11
  • 1
    You should really ask this as new questions! Comments is not the place for new questions. And, please include some more context than in this question. – kjetil b halvorsen Oct 02 '16 at 14:38
  • 1
    As @kjetilbhalvorsen suggests, you should go for a new question with more details. – TPArrow Oct 02 '16 at 14:41
  • 1
    Actually, the answer is not correct. First the asymptotic variance has too many X and second lim is imprecise in terms of the convergence mode. – Christoph Hanck Oct 02 '16 at 14:54
  • 1
    @ChristophHanck Thanks, I fixed that. What do you mean by the second lim? – TPArrow Oct 02 '16 at 15:14
  • The situation that you are describing is convergence in distribution, for which we commonly use notation like $\to_d$, and not $\lim$, which concerns deterministic sequences. – Christoph Hanck Oct 03 '16 at 07:53
0

Let the model be, $y=X\beta+e$ where $e_i\sim i.i.d N(0,\sigma^2<\infty)$ and $i=1,2,\ldots,T$. Then by central limit theorem:

$$ \sqrt T (\beta-\hat\beta)\overset{d}{\rightarrow} N(0,(X'X)^{-1} \sigma^2) $$

This in turn implies that $$\sqrt T (\beta-\hat\beta) = O_P(1)$$ The above equation defines "$\hat{\beta}$ is $\sqrt{n}$-consistent".

What does that definition mean? For some statistic $Z_n$, $Z_n = O_P(1)$ i.e. $Z_n$ is "bounded in probability" means the following: for all $\varepsilon > 0$ there exists a constant $C_{\varepsilon}$ such that: $$ \mathbb{P}(|Z_n| > C_{\varepsilon}) \leq \varepsilon $$

For more details also see: root-n consistent estimator, but root-n doesn't converge?

blooraven
  • 169
  • 6