Once again: The Hausman Test!

[odovakar]

Previously on the forum, someone wrote:

"Dear all,

Hossein Abbasi-Nejad and Shapour Mohammadi (both from University of Tehran) write an EViews program to run the Hausman test. It is available at http://papers.ssrn.com/sol3/papers.cfm? ... _id=672641. This program is generic and easy to use. All you have to do is replace the panel/pool identifiers (which usually begin with '_') and variable names. To run the test, open your workfile, then run the program (File, Run). You will get the result instantly."

I have downloaded the paper and now have a somewhat stupid question: How do run such a program in Eviews in the first place? Am I supposed to just copy-paste from the text, then put all of the chunks of code together? What do I do next, i.e. how do I read this program into Eviews and correlate it with my panel? I would be very grateful if someone could explain this in a "for dummies" step-by-step procedure. And I don't mean statistical dummies in this case...

[good0428]

Sorry, May I ask you some question about the Hausman test by e-mail?my email: good0428@hotmail.com

[EViews Gareth]

That program performs the Hausman Test for Fixed Effects, and was written for EViews 4.

EViews 5 and 6 have the Hausman test for Fixed/Random Effects built in - there is no reason to run that program at all.



However, in general, to run a program in EViews, simply click on File->New->Program, then enter the text into that program and hit Run.

[odovakar]

Good0428: Sure, go ahead. My e-mail is: odovakar@spray.se

QMS Gareth: Thank you once more; I could try that. Yes, I know that Eviews 5.0 (in my case) has this function built-in, but even though I asked about it before at the forum, I still haven't been able to figure out how to conduct the test. In Stata it's a piece of a cake for a fixed panel, but in Eviews this seems to be a whole science. Since so many people have already been asking about this at the forum, I think the Eviews developers should try to introduce, if possible, a more user friendly function in the next version of the software.

In my last post on how to conduct a "regular" Hausman specification test for fixed panels in Eviews, I was told to first conduct it for the dynamic equivalent, which I did. The only problem is that the rest of the info wasn't true, i.e. there is no menu alternative whatsoever with which to do the estimation. Or was I supposed so sit down and calculate b-B etc. by hand? Please, I really beg you (especially since, as mentioned above, several people have been asking about this by now): Do you think that you or someone else could write a short step-by-step guide on how to conduct this kind of Hausman test, or maybe change the text in the Eviews manual, so that it would be more general and comprehensible?

The problem in my case is that I did the estimation in Eviews with weights, which to my knowledge are not possible to replicate in Stata. I must therefore, really, succeed in conducting a Hauman specification test in Eviews, or I'm in deep trouble.

[EViews Gareth]

It really isn't all that difficult to perform.

If you already have an equation that has fixed effects, then simply re-estimate the same equation with random effects (where you previously chose Fixed effects, now choose Random effects). Once you have done that, click on View->Fixed/Random Effects Testing->Correlated Random Effects - Hausman Test.


I do not believe you are correct in saying that many people are having difficulty with this - the Hausman endogeneity test is the test that most people require information about, not the Hausman random effects test.

[odovakar]

"Once you have done that, click on View->Fixed/Random Effects Testing->Correlated Random Effects - Hausman Test."

Well, in that case, no wonder that I am having trouble, because like I said, that option isn't even available to me! Are you sure that this option exists in Eviews 5.0?
I did exactly like you said the first time you wrote this, but there really is no such thing in my Eviews 5.0. Maybe I need an update or something?

[EViews Gareth]

I'm not sure its available in 5.0. But I am sure it is available in 5.1, and since 5.1 is a free upgrade from 5.0, you should probably upgrade.