st: RE: RE: AR2 test for autocorrelation

[John Ebireri <j.ebireri.1@research.gla.ac.uk>]

Thanks Nick. Ill restructure the question.

John.

________________________________________
From: owner-statalist@hsphsun2.harvard.edu [owner-statalist@hsphsun2.harvard.edu] On Behalf Of Nick Cox [n.j.cox@durham.ac.uk]
Sent: 04 November 2011 13:05
To: 'statalist@hsphsun2.harvard.edu'
Subject: st: RE: AR2 test for autocorrelation

You need help in formulating questions. See for example

http://blog.stata.com/2010/12/14/how-to-successfully-ask-a-question-on-statalist/

and much specific advice in the Statalist FAQ.

These really are difficult questions to answer. You provide no specific details whatsoever of exactly what you did and exactly what Stata did in reply. There were some internal changes to the ML engine in Stata 11 that may account for differences in output between 10 and 11. However, major differences in output often reflect models that don't fit the data well in any case.

Nick


John Ebireri

I have 2 major problems and need some help on this;

First of all, in using the Systems GMM estimation in a unbalanced panel of 120 observation with 20 countries, i noticed that the output does not include the AR(2) test for autocorrelation. Could this be as a result of the missing values which tends to be large as the number of observations in the ouput is reduced  to only 51 observations?

Secondly, i do not know if any one eperiences this but using the same data, model specification and ecometric technique (System GMM in this case) on two different Stata versions (Stata 10.2-work and Stata 11-home) generates different outputs (coefficient estimates). Is this a software problem?


*
*   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/