Dear Sir or Madam:
I am Shaojuan Liao, the 3rd year Ph.D. student from Econ Department,
Virginia Tech.
I don't know whether it is appropriate to ask you questions on the
command pgmm. But I don't know how to deal with the case where all X are
exogenous and all T time periods' X can be used as the instrument.
Problem 1:
I know when X are predetermined, such as
Z=[y1,X1,X2, 0, 0, 0, 0, 0, ............
0, 0, 0, y1, y2, X1, X2, X3, ............
.....................................................] with 2 period
loss
I can put X as part of gmm.inst, and specify the lag.gmm
If X are not specified,
Z=[y1, 0, 0, 0, 0, 0, 0, ................. dX3
0, y1,y2, 0, 0, 0, 0, ................. dX4.
0, 0, 0, y1,y2, y3,0, ................. dX5]with 2 period loss
I don't put X as part of gmm.inst nor speficy the lag.gmm of X
But if X are strictly exogenous, I don't know how I can implement this case
by the command.
Z=[y1,X1.....XT, 0, 0, 0, 0, 0, 0, .................
0, 0,......0, y1,y2, X1.....XT,0,..................
..............................................................] with 2
period loss
Problem 2:
What is more, if y0 exists but X0 does not exist, so that the instrument
matrix is
Z=[y0,X1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,............
0, 0, y0, y1, X1, X2, 0, 0, 0, 0, 0, 0, 0,............
0, 0, 0, 0, 0, 0, y0,y1,y2 ,X1, X2, X3,0.............] with 1
period loss
How can I use the pgmm command to do that?
Problem 3:
I can get the same result as the command pgmm when everything is alright.
But sometimes, in the second stage, to calculate V= the sum ofW'(du){du') W
for each individual, I even get the V singular and uninvertable. What do
you think of the reason for this?
Thanks so much for your help!
Best!
Shaojuan
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
