Bai and Ng Unit Root test for Cross-sectionally dependent panel

For technical questions regarding estimation of single equations, systems, VARs, Factor analysis and State Space Models in EViews. General econometric questions and advice should go in the Econometric Discussions forum.

Moderators: EViews Gareth, EViews Moderator

Arca
Posts: 3
Joined: Fri Nov 20, 2020 1:02 pm

Bai and Ng Unit Root test for Cross-sectionally dependent panel

Postby Arca » Fri Jan 07, 2022 10:35 am

I am currently using Eviews 12 (Serial # 12K00061, Enterprise edition registered to Shiv Nadar University). My question is about Bai and Ng Panel Unit Root test.

At the bottom of page 724, the Eviews Users Guide II says, “Lastly, the final table is a pooled version of the individual ADF test statistics in the previous table. Here, we cannot reject the null hypothesis that all of the cross-sections are simultaneously co-integrated.” The p-value for this pooled statistic is 0.59592 and the null hypothesis is “No cointegration among all cross-sections.”

Then, at the bottom of page 728, it says, “Lastly, the pooled version of the latter tests has a p-value which effectively (sic) zero, and therefore we fail to reject the null hypothesis that all cross-sections are not cointegrated.” The p-value for this pooled statistic is 0.00000 and the null hypothesis is “No cointegration among all cross-sections.”

These are confusing, if not incorrect.

EViews Mirza
Posts: 80
Joined: Sat Apr 22, 2017 8:23 pm

Re: Bai and Ng Unit Root test for Cross-sectionally dependent panel

Postby EViews Mirza » Fri Jan 07, 2022 10:57 am

Hi Arca!

Thanks for pointing this out. We'll fix the language to more clearly (correctly) reflect the p-value.

edward_leman
Posts: 2
Joined: Tue Apr 19, 2022 6:41 am

Re: Bai and Ng Unit Root test for Cross-sectionally dependent panel

Postby edward_leman » Tue Apr 19, 2022 7:39 am

In addition to what Arca asked, should the null in the pooled test address cointegration or stationarity? i am confused since the whole test (in my understanding) is about unit root.
Attachments
evieww.PNG
evieww.PNG (10.81 KiB) Viewed 4487 times

EViews Mirza
Posts: 80
Joined: Sat Apr 22, 2017 8:23 pm

Re: Bai and Ng Unit Root test for Cross-sectionally dependent panel

Postby EViews Mirza » Tue Apr 19, 2022 9:29 am

Quoting the original source for the test: "A panic attack on unit roots and cointegration" by Bai and Ng (2004), the following is found on page 1140:

"A pooled test of the idiosyncratic errors can be seen as a panel test of no cointegration, as the null hypothesis that ρ_i = 1 for every i holds only if no
stationary combination of X_it can be formed."

Hope this helps.

edward_leman
Posts: 2
Joined: Tue Apr 19, 2022 6:41 am

Re: Bai and Ng Unit Root test for Cross-sectionally dependent panel

Postby edward_leman » Wed Apr 20, 2022 12:33 am

Thank you, this really helps.


Return to “Estimation”

Who is online

Users browsing this forum: No registered users and 17 guests