I have no problem running a regression with dynamic fixed effect model, but having error messages such as "initial values not feasible" when "dfe" is replaced with "pmg" for pooled mean group estimation. Data is balanced panel with N=49 and T=47.

xtpmg d.var1 d.V7 d.V9 d.var15, lr(L.var1 V7 V9 var15) ec(ec) pmg
initial values not feasible
r(1400);

The results with FE are below.

xtpmg d.var1 d.V7 d.V9 d.var15, lr(L.var1 V7 V9 var15) ec(ec) dfe

full option not meaningful with dfe
ignoring option and continuing...
------------------------------------------------------------------------------
Dynamic Fixed Effects Regression: Estimated Error Correction Form
(Estimate results saved as DFE)
------------------------------------------------------------------------------
------------------------------------------------------------------------------
| Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
ec |
V7 | 1.455627 .318333 4.57 0.000 .8317059 2.079548
V9 | -.3265571 .7070167 -0.46 0.644 -1.712284 1.05917
var15 | -.028163 .0408433 -0.69 0.490 -.1082145 .0518884
-------------+----------------------------------------------------------------
SR |
ec | -.6995318 .0349926 -19.99 0.000 -.768116 -.6309476
|
V7 |
D1. | .4446153 .1973381 2.25 0.024 .0578396 .8313909
|
V9 |
D1. | 5.66951 1.254664 4.52 0.000 3.210415 8.128606
|
var15 |
D1. | .1621954 .0314758 5.15 0.000 .1005039 .2238868
|
_cons | -18.10949 12.19552 -1.48 0.138 -42.01227 5.793285
------------------------------------------------------------------------------


What could be a cause of this error?

Thank you.

Yoojung