Statalist


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

Re: st: STATA command for reporting the estimate of the variance of the time-invariant fixed effect.


From   Austin Nichols <austinnichols@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: STATA command for reporting the estimate of the variance of the time-invariant fixed effect.
Date   Fri, 5 Jun 2009 12:23:43 -0400

Stephen Armah <armah.stephen@gmail.com>:
-xtreg, fe- reports the variance of u_i and a test that they are all 0
by default; read the [R] manual entry for more detail. Note that this
does not give "an indication of the presence of endogeneity."  If the
u_i are uncorrelated with X, there is no endogeneity arising from
omitting the FE when you run a pooled regression.

I would think a much bigger endogeneity problem is that aid may be
affected by political stability, and both may be related to GDP growth
through a number of causal channels running in various directions.
I.e. aid and political stability are not exogneous, even after
demeaning by country.

On Fri, Jun 5, 2009 at 9:08 AM, Stephen Armah <armah.stephen@gmail.com> wrote:
> Dear Statalist,
> I am trying to determine how political stability influences the effect
> of aid on growth  in Sub-Saharan Africa.  I have 31 countries and 6
> years with some missing data so the panel is not balanced.
>  I am aware that there is possible endogeneity due to country-specific
> effects.  I know I can eliminate these country-specif effects by
> first-differencing, forward orthogonal differencing or by applying
> time-demeaning (the fixed effects transformation).
>
>  My question is I want an indication of how large the country specific
> effects are and if they are causing significant endogeneity bias. Note
> that because  have few observations, traditional Durbin-Wu_Hausman
> tests cannot pick up endogenety even if it is present so I cannot use
> tests of endogeneity to check for country specific effect endogeneity.
>
>  However, according to Woolridge's book, I can use a command in STATA
> to tell me what the variance of the fixed effect is. This will give me
> an indication of the presence of endogeneity. Then I can use a formal
> Breusch-Pagan test to confirm heteroskedasticity. Does anyone know the
> STATA command for reporting the estimate of the variance of the
> time-invariant fixed effect during fixed effects estimation.
>
> Sincerely,
> Stephen E Armah
> *
> *   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   |   What's new   |   Site index