spatial dynamic panel model (GMM-FD)?

[pande suputra]

i have been using stata command "spregdpd" to analyze spatial dynamic panel model (GMM-FD). I can't find how to do Arellano-Bond test for zero autocorrelation in first-differenced errors in this command. Any suggestion? or any stata command to do an estimation for spatial dynamic panel model (GMM-FD)? Thank You

[Sebastian Kripfganz]

I would recommend to use the official GMM estimation commands for dynamic panel models (xtabond, xtdpdsys, xtdpd), or user-written commands (xtabond2, xtseqreg). You need to create new variables for your spatial lags first which might be a bit tedious with panel data, but the rest is standard GMM.

[pande suputra]

Hi Sebastian, thanks for your reply. if i have to do so, the spatial autocorrelation test is from statis panel data model? create new variabel, it's mean i have to multiply the spatial matrik with Y right? i already did that, and the estimation result is different from previous model (spregdpd)

[Sebastian Kripfganz]

if i have to do so, the spatial autocorrelation test is from statis panel data model?

Sorry, I do not have an answer to this question.

create new variabel, it's mean i have to multiply the spatial matrik with Y right?

Yes, that is correct.

i already did that, and the estimation result is different from previous model (spregdpd)

It is difficult so say anything about that because I do not know how exactly the command spregdpd obtains its result. Sorry again.

[pande suputra]

it's okay. thank you very much for your reply, it's mean so much for me

[Rajarshi Chowdhury]

Hello Sebastian. Could you please tell me can I get the direct and indirect effects estimates of the variables by using standard dynamic panel estimation with xtabond2 after creating spatial lagged variables?

[Sebastian Kripfganz]

No, you cannot easily get those effects without a dedicated command for spatial models (or considerable own programming efforts).