Statalist


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

Re: st: unit-root tests: xt-fisher, Hadri, ipshin


From   Joachim Landström <joachim.landstrom@fek.uu.se>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: unit-root tests: xt-fisher, Hadri, ipshin
Date   Tue, 24 Nov 2009 22:30:15 +0100

I would advise you to go and look into the actual academic references before you use those unit root models and scrutinize their assumptions. Do they differ for what you have and how applicable are those models really for you?

Since your N is small and has a low N/T-ratio I would recommend to use a unit root model that has robust finite sample properties while being assymptotic in T.

/Joachim


Quoting Carlos Rodriguez <carlosrodriguez1993@gmail.com>:

So, even if some panels have unit roots but not all, would it be
fine if I run the regression in first-differences to address the issue
of unit roots? Or is this not advisable when not all panels have unit
roots.

Should I first-difference all control variables in additon to the main explanatory one and the dependent variable? What about the binary (1/0) indicators?

Thanks again
Carlos



On Thu, Oct 29, 2009 at 7:09 PM, Scott Merryman
<scott.merryman@gmail.com> wrote:
On Thu, Oct 29, 2009 at 2:39 PM, Carlos Rodriguez
<carlosrodriguez1993@gmail.com> wrote:
Dear Stata users,

I am a bit confused and I'd very much appreciate your advice and clarification.

I wish to test for unit roots in my time-series cross sectional data
(20 countries, 30 yearly observations for each), but the tests
available in Stata10 provide opposite answers so my question is:
what's the criteria I should follow for choosing among these tests?

According to the xtfisher and the ipshin tests, the null hypothesis of
unit roots is rejected.

In all the panels

But, if I use the Hadri test, instead, the null hypothesis of *no*
unit roots is rejected.

In all the panels.

That is in the Fischer test:
Ho: All panels contain unit roots
Ha: At least one panel is stationary

and in the Hadri LM test:
Ho: All panels are stationary
Ha: Some panels contain unit roots

So couldn't you conclude that some panels, but not all, contain a unit root?

Scott
*
*   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/




--
Joachim Landström


*
*   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