Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.

# st: suspicious p-value?

 From "Sabina Kummer" To statalist@hsphsun2.harvard.edu Subject st: suspicious p-value? Date Thu, 10 Jan 2013 22:43:33 +0100 (CET)

```Dear all,

I am currently working with error correction model (Johansen approach) in order to estimate the level of tax income. The explanatory variable is GDP and I use quarterly seasonally adjusted data (that are expressed in log).

The coefficient of the cointegration equation is highly significant, with a p-value of 0.0000 (lower than the 1% level). I would like to know whether the fact that the p-value is so low is a sign of a spurious modelisation or of a problem with data or of any other problem. Is that common to have p-value of 0.0000 in the cases of error correction model?

The specification is the following:  D(TAX)=C(1)+C(2)*(TAX(-1)-0.292587*GDP(-1)-0.005554*@TREND)+C(3)*D(GDP(-1))+C(4)*D(GDP(-2))+C(5)*D(TAX(-1))+C(6)*D(TAX(-2))

The cointegration equation is: TAX(-1)-0.292587*GDP(-1)-0.005554*@TREND

Here is the output of the estimation:

Coefficient	Std. Error	t-Statistic	Prob.

C(1)	5.223485	0.953629	5.477480	0.0000
C(2)	-0.956099	0.174718	-5.472241	0.0000
C(3)	0.498795	1.992768	0.250303	0.8030
C(4)	1.872874	2.003406	0.934845	0.3528
C(5)	0.122361	0.145785	0.839326	0.4039
C(6)	0.097690	0.113331	0.861987	0.3913

R-squared		0.428218	    Mean dependent var	0.007543
Adjusted R-squared	0.391566	    S.D. dependent var	0.147855
S.E. of regression	0.115330	    Akaike info criterion	-1.413284
Sum squared resid	1.037484	    Schwarz criterion	-1.239655
Log likelihood	65.35795	    Hannan-Quinn criter.	-1.343487
F-statistic		11.68315	    Durbin-Watson stat	2.026847
Prob(F-statistic)	0.000000

Kind regards,

Sabina

---- Fin du message transféré ----

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/
```