Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

AW: st: RE: can Arellano-Bover estimator be used when some regressors or the dependent variable contain a unit root?


From   "Dithmer, Jan" <jdithme@food-econ.uni-kiel.de>
To   "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>
Subject   AW: st: RE: can Arellano-Bover estimator be used when some regressors or the dependent variable contain a unit root?
Date   Fri, 29 Jun 2012 15:40:08 +0200

As nobody seems to have any idea on this topic, I'll try my best to help you.

First of all, I am not aware of any program accounting for non-stationarity in a dynamic panel model.
However, xtabond2 is mainly for the "small T, large N" case. In this case spurious regression caused by non-stationarity should be no problem.
If T is large and N small, then you may run into problems with spurious regression.
If T is large and N too, the problem should vanish as well.
However, as is noted in the paper by Roodman, non-stationarity of variables may lead to a weak-instrument problem with respect to the Arellano-Bond estimator
and the Blundell-Bond estimator may have bad small sample properties.

Thus, if you have a small T, large N panel, you may not have to worry much. If not, an alternative may be to estimate a panel VAR model.
However, I don't know if it is available in Stata...

Hope, this somehow helps a bit. Would be nice if someone could join the discussion.

Best, Jan 
 

-----Ursprüngliche Nachricht-----
Von: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] Im Auftrag von sabbas gidarokostas
Gesendet: Thursday, June 28, 2012 7:50 PM
An: statalist@hsphsun2.harvard.edu
Betreff: Re: st: RE: can Arellano-Bover estimator be used when some regressors or the dependent variable contain a unit root?

I was wondering if anyone in the Stata community has any answer to my question

thanks

On 6/20/12, sabbas gidarokostas <sabbasgidarokostas@googlemail.com> wrote:
> On 6/20/12, Dithmer, Jan <jdithme@food-econ.uni-kiel.de> wrote:
>> Dear Sabbas,
>>
>> I would send this question directly to the author of the program 
>> xtabond2.
>>
>> Maybe you can ask him to post the answer on statalist, as it may be 
>> interesting for others as well.
>>
>> Best, Jan
>>
>> -----Ursprüngliche Nachricht-----
>> Von: owner-statalist@hsphsun2.harvard.edu
>> [mailto:owner-statalist@hsphsun2.harvard.edu] Im Auftrag von sabbas 
>> gidarokostas
>> Gesendet: Tuesday, June 19, 2012 7:43 PM
>> An: statalist@hsphsun2.harvard.edu
>> Betreff: Re: st: RE: can Arellano-Bover estimator be used when some 
>> regressors or the dependent variable contain a unit root?
>>
>> thank you Mark for your reply.
>>
>> I looked at the command -xtabond2-, Do you think that -xtabond2 will 
>> "automatically solve the problem of non stationarity? Because if not, 
>> then
>> xtabond2 has not other option available  that could cure this problem.
>>
>> thank you
>>
>> On 6/19/12, Hintz, Mark <Mark_Hintz@mentor.com> wrote:
>>> I think you'll need to take the first difference first, then use the 
>>> differences in your model. The model uses the first differences to 
>>> find the moment conditions, but it's fundamentally estimating the 
>>> undifferenced model, which is nonstationary. I think you want the 
>>> estimated model to reflect a stationary process, so you'll need to 
>>> feed the stationary first-differenced series into the Arellano-Bover 
>>> estimator.
>>>
>>> -----Original Message-----
>>> From: owner-statalist@hsphsun2.harvard.edu
>>> [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of sabbas 
>>> gidarokostas
>>> Sent: Monday, June 18, 2012 5:02 PM
>>> To: statalist
>>> Subject: st: can Arellano-Bover estimator be used when some 
>>> regressors or the dependent variable contain a unit root?
>>>
>>> Dear all,
>>>
>>> I have a dynamic panel regression with fixed effects and using the 
>>> xtunitroot iip i found that some variables contain unit root. Is is 
>>> still valid to use Arellano-Bover estimator?
>>> I think yes because we take the first difference. Am i right?
>>>
>>> cheers
>>> *
>>> *   For searches and help try:
>>> *   http://www.stata.com/help.cgi?search
>>> *   http://www.stata.com/support/statalist/faq
>>> *   http://www.ats.ucla.edu/stat/stata/
>>>
>>> *
>>> *   For searches and help try:
>>> *   http://www.stata.com/help.cgi?search
>>> *   http://www.stata.com/support/statalist/faq
>>> *   http://www.ats.ucla.edu/stat/stata/
>>>
>> *
>> *   For searches and help try:
>> *   http://www.stata.com/help.cgi?search
>> *   http://www.stata.com/support/statalist/faq
>> *   http://www.ats.ucla.edu/stat/stata/
>>
>> *
>> *   For searches and help try:
>> *   http://www.stata.com/help.cgi?search
>> *   http://www.stata.com/support/statalist/faq
>> *   http://www.ats.ucla.edu/stat/stata/
>>
>

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index