Issue with xtpmg

[Thuy Nguyen]

Dear esteemed members,

I am running a panel PMG and MG regression on the impacts of bank credit on growth. I used xtpmg in STATA 15 to run the regression.
STATA seems to work well for MG model:
xtpmg d.LGDPPC d.LBCPS d.LGEXP , lr(l.LGDPPC LBCPS LGEXP) ec(ECT) replace mg

------------------------------------------------------------------------------
Mean Group Estimation: Error Correction Form
(Estimate results saved as mg)
------------------------------------------------------------------------------
------------------------------------------------------------------------------
D.LGDPPC | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
ECT |
LBCPS | .5945616 .2240737 2.65 0.008 .1553852 1.033738
LGEXP | -.9632138 .8217571 -1.17 0.241 -2.573828 .6474006
-------------+----------------------------------------------------------------
SR |
ECT | -.0679927 .0288838 -2.35 0.019 -.1246038 -.0113815
|
LBCPS |
D1. | -.0638153 .031625 -2.02 0.044 -.1257992 -.0018315
|
LGEXP |
D1. | -.0905296 .0695371 -1.30 0.193 -.2268197 .0457605
|
_cons | .7099954 .2590646 2.74 0.006 .2022381 1.217753
------------------------------------------------------------------------------

However, when I ran the same test using pmg and let the program to run overnight but after such a long time, I still havn't got any results. Instead, STATA kept running and I have just got a long list of log likelihood for running-on iteractions like this:
Iteration 0: log likelihood = 1140.7413 (not concave)
Iteration 1: log likelihood = 1150.9154
Iteration 2: log likelihood = 1153.5223
Iteration 3: log likelihood = 1157.7877 (not concave)
Iteration 4: log likelihood = 1161.554 (not concave)
Iteration 5: log likelihood = 1161.554 (not concave)
Iteration 6: log likelihood = 1161.554 (not concave)
Iteration 7: log likelihood = 1161.554 (not concave)
Iteration 8: log likelihood = 1161.554 (not concave)
Iteration 9: log likelihood = 1161.554 (not concave)
Iteration 10: log likelihood = 1161.554 (not concave)
Iteration 11: log likelihood = 1161.554 (not concave)
Iteration 12: log likelihood = 1161.554 (not concave)
Iteration 13: log likelihood = 1161.554 (not concave)
....

And until this morning, the iteration reached more than 200 and is keeping running on now.

Could you please share your experiences or suggestions on these issues? I might have got some errors somewhere in specifying the model?

Many thanks for your advices,

Very best wishes,

Thuy Nguyen

[Elina Kim]

Dear esteemed members,

I am running a panel PMG and MG regression on the impacts of bank credit on growth. I used xtpmg in STATA 15 to run the regression.
STATA seems to work well for MG model:
xtpmg d.LGDPPC d.LBCPS d.LGEXP , lr(l.LGDPPC LBCPS LGEXP) ec(ECT) replace mg

------------------------------------------------------------------------------
Mean Group Estimation: Error Correction Form
(Estimate results saved as mg)
------------------------------------------------------------------------------
------------------------------------------------------------------------------
D.LGDPPC | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
ECT |
LBCPS | .5945616 .2240737 2.65 0.008 .1553852 1.033738
LGEXP | -.9632138 .8217571 -1.17 0.241 -2.573828 .6474006
-------------+----------------------------------------------------------------
SR |
ECT | -.0679927 .0288838 -2.35 0.019 -.1246038 -.0113815
|
LBCPS |
D1. | -.0638153 .031625 -2.02 0.044 -.1257992 -.0018315
|
LGEXP |
D1. | -.0905296 .0695371 -1.30 0.193 -.2268197 .0457605
|
_cons | .7099954 .2590646 2.74 0.006 .2022381 1.217753
------------------------------------------------------------------------------

However, when I ran the same test using pmg and let the program to run overnight but after such a long time, I still havn't got any results. Instead, STATA kept running and I have just got a long list of log likelihood for running-on iteractions like this:
Iteration 0: log likelihood = 1140.7413 (not concave)
Iteration 1: log likelihood = 1150.9154
Iteration 2: log likelihood = 1153.5223
Iteration 3: log likelihood = 1157.7877 (not concave)
Iteration 4: log likelihood = 1161.554 (not concave)
Iteration 5: log likelihood = 1161.554 (not concave)
Iteration 6: log likelihood = 1161.554 (not concave)
Iteration 7: log likelihood = 1161.554 (not concave)
Iteration 8: log likelihood = 1161.554 (not concave)
Iteration 9: log likelihood = 1161.554 (not concave)
Iteration 10: log likelihood = 1161.554 (not concave)
Iteration 11: log likelihood = 1161.554 (not concave)
Iteration 12: log likelihood = 1161.554 (not concave)
Iteration 13: log likelihood = 1161.554 (not concave)
....

And until this morning, the iteration reached more than 200 and is keeping running on now.

Could you please share your experiences or suggestions on these issues? I might have got some errors somewhere in specifying the model?

Many thanks for your advices,

Very best wishes,

Thuy Nguyen

Hi! Unfortunately I cannot help currently, I also don't know, I've faced same situation. Noticed that it occurs when I put dependent variable in its natural logorithm in the short run model part.

I'd like to know whether you have found solution to this issue. Would be nice to hear from you concerning this if anything relevant
. Thanks.