GMM estimation problem and f-statistic

[Luisfer Indetec]

HI to everyone!
I´m working on gmm model.
I have stata 14.

My data has 32 individuals and 13 years each of them.
The problem is that when I estimate the GMM with the following rutine:

xtabond2 pibe usa tasa tipo pet , gmm(l.pibe,collapse) iv(tasa tipo) ///
noleveleq nodiffsargan twostep orthogonal small

I do not get the F-stat value.

I use this method because theres endogeneity in the exchange rate (tasa )and the monetary policy (tipo) . I decided to use endogenous instruments, for that reason the model is a Differece GMM.
I also collapse the estimation because there are more instruments than groups, and I looked for robust estimation. The problem appears when I use a one step and two steps estimation, I know the two step is more efficient.

Why do not I have the F-stats output? Is it part of the estimation outcome?

[Asraf Zaman]

I have encountered the same problem. Have you got any solution?
Thanks

[Sebastian Kripfganz]

It is not clear which F-statistic is wanted. Additional statistics can be obtained with the ivreg2 command after the xtdpdgmm command (as a replacement for xtabond2). Here is an example:

webuse abdata
xtdpdgmm n L.n w k, gmm(L.n, lag(0 .) collapse model(fodev)) gmm(w k, lag(1 .) collapse model(fodev)) vce(robust) twostep small nocons
predict iv*, iv
ivreg2 n (L.n w k = iv*), gmm2s cluster(id)

For details, please see my 2019 London Stata Conference, in particular slides 39 and following:

• Kripfganz, S. (2019). Generalized method of moments estimation of linear dynamic panel data models. Proceedings of the 2019 London Stata Conference.

Regarding the opening post, there are several other problems. First, the lagged dependent variable needs to be explicitly specified in the list of regressors. Second, the iv(tasa tipo) option does not create valid instruments when these variables are endogenous with respect to the idiosyncratic error component.

[Fazlul Sumon]

hello,

i have encountered the same problem. I have N=62 and T=11. When i run two step difference GMM i did not find F statistics. Moreover, i have got the following:

Warning: Two-step estimated covariance matrix of moments is singular.
Using a generalized inverse to calculate optimal weighting matrix for two-step estimation

Dynamic panel-data estimation, two-step difference GMM
------------------------------------------------------------------------------
Group variable: country Number of obs = 558
Time variable : year Number of groups = 62
Number of instruments = 24 Obs per group: min = 9
F(0, 62) = . avg = 9.00
Prob > F = . max = 9

why i got that? what is the solution

[Shaizy Khan]

Hello Everyone!

I am not getting the F statistics value on performing difference and system GMM. What can be the possible reason?
Please help!

Thank you!