st: Confusion about XTPMG

[john ebireri <jebireri@yahoo.com>]

Hello,

I am having a problem with where i should include the lag structure when using -xtpmg-. I am using Stata 12.1 for windows. 

I noticed in some published articles that the Error Correction Model produces short run coefficients with lag structures and the the long run cointegrating vector producing long-run coefficients without lags.

In my case, what i have explained above is opposite: short-run coefficients without lags and long-run coefficients with lags. It might be wrong specification on my part, but ill provide a sample of the command i use and the regression output i receive. I have 8 variables but ill use just 3 to show the problem i encounter:
 
xtpmg d.mvagdp d.bcon d.bnkdev if year >=1995, lr(l.mvagdp l.bcon bnkdev ) ec(ec) replace pmg technique(dfp)

 
Iteration 0:   log likelihood = -985.04495  
Iteration 1:   log likelihood =  -984.5807  (backed up)
Iteration 2:   log likelihood =   -984.579  (backed up)
Iteration 3:   log likelihood = -984.51101  
Iteration 4:   log likelihood = -984.50197  
Iteration 5:   log likelihood = -984.50194  
 
Pooled Mean Group Regression
(Estimate results saved as pmg)
 
Panel Variable (i): id                                       Number of obs      = 372
Time Variable (t): year                                    Number of groups   = 31
Obs per group: min = 12
avg = 12.0
max = 12
Log Likelihood     = -984.5019
 
D.mvagdp    Coef.       Std. Err.        z       P>z         [95% Conf. Interval]
 
ec           
bcon 
L1.           .0184939   .0195238     0.95   0.344    -.0197719  .0567598
             
bnkdev    .0503223   .0282728     1.78   0.075    -.0050914  .105736
 
SR           
ec           -.8136236   .0513977   -15.83   0.000   -.9143612 -.7128859
             
bcon 
D1.          .0107836   .1003137     0.11   0.914    -.1858276 .2073949
             
bnkdev 
D1.           1.379236   1.134416     1.22   0.224    -.8441793 3.60265
             
_cons      . 9586204    .612323     1.57   0.117    -.2415107 2.158752
 
 
 
How do i get the lags to be in the ARDL model rather than the long run cointegrating vector.
 
Thanks.
 
John.

*
*   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/