Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


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

Re: Re: Re: st: Problems with FE and Hausman test


From   Christopher Baum <[email protected]>
To   "[email protected]" <[email protected]>
Subject   Re: Re: Re: st: Problems with FE and Hausman test
Date   Sat, 22 Feb 2014 14:46:05 +0000

<>
On Feb 22, 2014, at 2:33 AM, statalist-digest <[email protected]> wrote:

> 
> Thank you, this makes perfect sense now! So I could run
> 
>> xtreg, PERCENTGRAD IV1 IV2 IV3 IV4 IV5 IV6 IV7 IV8 Timeis2007, Timeis2008, be
> 
> And yes, I acknowledge that the rest of the model takes for granted every other thing ("says that they are the ONLY factors (other than geography and time) that affect the outcome.") But that's part of my focus, the argument that solution is just administrative/organizational and not in terms of funding and other things. My guess is that administrative/organizational changes help, but that a large portion can be attrbiuted to other things. 

No, you cannot, because the between estimator is a cross-sectional estimator making use of time averages, and time dummies will be dropped.

The idea that one can run a regression and focus on certain objects of interest leads in general to a seriously misspecified, biased and inconsistent regression model that is useless for any policy analysis. Let me give you an example from a prior student of mine, who wanted to explain GDP growth in subSaharan African countries by regressing growth, in a panel model, on US aid received. She indeed wanted to focus on the question of whether US aid had an impact on these countries' economies, but specified a model in which that was the ONLY systematic causal factor beyond fixed country and time effects. That paper received a very low grade, as the underlying premise was nonsensical. For instance, it was easy to demonstrate that the prevalence of AIDS in those countries was a very powerful explanatory factor. So developing a 'focus' on what you might be wanting to study is fine, but you need to control for all other factors in computing that partial derivative. Fixed effects ca!
 n do some of that, but unless you want to make the argument that all other state-specific factors are time-invariant, that is a very seriously flawed model, and one that should be rejected by any thoughtful reviewer.

Kit

Kit Baum
Professor of Economics and Social Work, Boston College, Chestnut Hill MA, USA
DIW Research Professor, Department of Macroeconomics, DIW Berlin, Berlin, Germany
[email protected]  |  http://ideas.repec.org/e/pba1.html












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


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index