Statalist The Stata Listserver

[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: RE: check results of xtabond2

From   "Urban, Dieter" <>
To   <>
Subject   st: RE: check results of xtabond2
Date   Thu, 18 May 2006 13:43:32 +0200

Maybe you have used the two-step efficient GMM estimator of
Blundell-Bond? This estimator is known to have a large small-sample
bias, where small samples can actually be quite large. (Recall that GMM
estimators are generally consistent in large samples but many of them
are biased in small samples.) Try and set your number of observations to
1 million and check again. Or use the one-step estimator or the
Windmeijer correction if this is indeed the problem. 

Dieter Urban

-----Original Message-----
[] On Behalf Of Vivian
Sent: Thursday, May 18, 2006 6:48 AM
Subject: st: check results of xtabond2 

I was wondering if anyone has verified the validity of the results of
xtabond2. I have simulated a data and performed xtabond and xtabond2.
The results of xtabond is very close to my specified vector of
parameters in the simulated data. Unfortunately, this is not the case
for xtabond2. The resulting parameters were simply far from my
specification and also with the results of xtabond. As far as I know,
Blundell and Bond estimator is better than the Arellano and BOnd
estimator. It seems that there is something wrong with xtabond2. I know
that this is not part of Stata's functions and that Stata is not liable
for it. I hope that somebody could verify the results of xtabond2
especially that there are a lot of Stata users using this procedure. If
somebody is interested, I could send my data and the results. 

Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around
*   For searches and help try:

*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index