PEDRONI COINTEGRATION TEST

[FernandoT]

PEDRONI COINTEGRATION TEST

Can somebody please help me?

I have a question over Pedroni Cointegration Panel Test. I calculated those test with EVIEWS 6, but the results (I think) are erroneous. The Pedroni panel cointegration statistic is constructed from the residuals of the spurious regression, Pedroni makes a total of eleven statistics with varying degree of properties (size and power for different N and T). All Pedroni statistics can be standardized and asymptotically normally distributed.

(Chi(N,T)-u N^0.5)/v^0.5

Nevertheless, the results show that the statistics are not distributed as the normal. For example:

Pedroni Residual Cointegration Test
Series: IVM MM
……………………......…………….Statistic…………….Prob……………….Statistic…………….Prob.
Panel v-Statistic……………….4.219500 ………….0.0001………………4.119485………….0.0001
Panel rho-Statistic……………..-0.400152…………0.3682………………-2.543473…………0.0157
Panel PP-Statistic……………..0.671083…………..0.3185……………….-1.254923 ………..0.1815
Panel ADF-Statistic……………-0.216806………….0.3897……………….0.172158………….0.3931

Alternative hypothesis: individual AR coefs. (between-dimension)

.........………………………………….Statistic……………..Prob.
Group rho-Statistic…………….-1.776207………….0.0824
Group PP-Statistic………………-0.824320…………0.2840
Group ADF-Statistic……………..0.538943 ………….0.3450

The p-value (right tail normal distribution) of 4.2195 (Panel v-statistic) is not 0.0001, or the p-value (left tail normal distribution) of 0.671083 (Panel PP-statistic) is not 0.3185. Or the p-value (left) of 0.538943 (ADF statistic) is not 0.3450.

Are results correctly calculated?
How I have interpret this results?

Thanks

[EViews Gareth]

Before I answer this, can you confirm that your copy of EViews 6 is up-to-date? You can check the build date by looking at Help->About EViews.

It should be October 21 2008 (full version) or August 20 2008 (student version). Earlier versions of EViews 6 had some problems with the Pedroni test, I believe.

[FernandoT]

Before I answer this, can you confirm that your copy of EViews 6 is up-to-date? You can check the build date by looking at Help->About EViews.

It should be October 21 2008 (full version) or August 20 2008 (student version). Earlier versions of EViews 6 had some problems with the Pedroni test, I believe.

My EVIEWS copy is:

Enterprise Edition
Dec/17/2007 Build.

[EViews Gareth]

You should update to the latest copy:

http://eviews.com/download/download.html

February 2008 saw a large number of changes/corrections to the Pedroni test. It may be that your answers change.

[FernandoT]

Thanks I will do.

[FernandoT]

Thank you, I have the problem solved.