Olga I am no statistician but I thought that for xtfisher, the null
hypothesis is the existence of a unit root; a significant test statistic
means that you can reject the null hypothesis, and thus that there is no
unit root. Your discussion below suggests that you understand the
interpretation of the xtfisher test to be the opposite.
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Olga
Gorbachev Melloni
Sent: Sunday, September 10, 2006 11:18 AM
To: [email protected]
Subject: st: xtfisher vs pescadf results
Dear Stata users,
I am confused. I am trying to run a panel unit root test and I am
getting
different results from xtfisher and pescadf. Why is that? I am attaching
an
output below that compares the two.
Can somebody point me to the potential mistake? I need to run panel unit
root test that allows for cross sectional dependence, thus pescadf would
be
my choice, but I am not understanding the output correctly it seems.
thank you,
___________________________________________________________________
webuse grunfeld,clear
pescadf invest, trend lag(1)
Pesaran's CADF test for invest
Cross-sectional average in first period extracted and extreme t-values
truncated
Deterministics chosen: constant & trend
t-bar test, N,T = (10,20) Obs = 180
Augmented by 1 lags (average)
t-bar cv10 cv5 cv1 Z[t-bar] P-value
-2.203 -2.740 -2.880 -3.150 0.241 0.595
____________________________________________________________________
xtfisher invest, trend lag(1)
Fisher Test for panel unit root using an augmented Dickey-Fuller test (1
lags)
Ho: unit root
chi2(20) = 55.2784
Prob > chi2 = 0.0000
*** based on this output I would accept unit root under xtfisher but
reject
under pescadf
____________________________________________________________________
pescadf invest, lag(1)
Pesaran's CADF test for invest
Cross-sectional average in first period extracted and extreme t-values
truncated
Deterministics chosen: constant
t-bar test, N,T = (10,20) Obs = 180
Augmented by 1 lags (average)
t-bar cv10 cv5 cv1 Z[t-bar] P-value
-2.242 -2.210 -2.340 -2.600 -1.572 0.058
____________________________________________________________________
xtfisher invest, lag(1)
Fisher Test for panel unit root using an augmented Dickey-Fuller test (1
lags)
Ho: unit root
chi2(20) = 13.8730
Prob > chi2 = 0.8369
*** based on this output I would accept unit root under pescadf but
reject
under xtfisher
whats going on? I used also the nodemean with pescadf, but the results
were
the same.
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
