I think I'm depressed : Some question about Dynamic Panel Data on System GMM and LSDVC

[Jaja Jma]

I'm new to Stata. I was estimating dynamic panel data with N = 10 and T = 18 using one-step GMM (I have passed the instrument validity, consistency, unbiased, and multicol test) and got a suggestion to use LSDVC. Is my estimate incorrect? I tried to estimate with lsdvc but due to my limited understanding of English, I don't understand how to estimate it so the lsdvc estimation that I did seems to have an error. After I input the data, do the encode, then xtset, should I do a first-difference on the variables? (my variables have a stationarity problem) Then, what is the correct stata command to estimate it? I estimated it with lsdvc d_y d_x1 ... d_xn Year, initial(ab) vce(cluster id). What tests do I need to pass to say that my model is good? I'm currently trying hard to understand how to use Stata while learning English better rn.

[Carlo Lazzaro]

Jaja:
welcome to this forum.
As a first step, see the -D.- operator available from -help tsvarlist-.

[Sebastian Kripfganz]

With N=10, GMM estimators for dynamic panel models are not very reliable. I recommend to have a look at my xtdpdbc command, which implements an improved version of the LSDVC method:

• Kripfganz, S., and J. Breitung (2022). Bias-corrected estimation of linear dynamic panel data models. Proceedings of the 2022 London Stata Conference.

[George Ford]

I don't think you'd difference in an LSDV model, since you've got a lag of the DV. with xtdpdbc, which looks like a solid option, you can specify AR terms.

[Jaja Jma]

Thank you very much for your informative insight, Sir.
At least now I understand better what I have to learn

[George Ford]

Spend some time getting the feel for the LSDV model. Anything you can find (including Youtube videos). If that's the way you are going, then it will pay off in the end.