Influence of incorporating industry dummies in XTABOND2 command

[Kevin Traen]

Dear stata users,

For my thesis, I make use of the XTABOND2 command.
My command is the following:

My variable of interest is Accreceivable. I tried to incorporate a dummy NACE2D (an industry dummy) as follows:

This generates the following output for the tests:

Could someone tell me why the Sargan and Hansen test have no p-value? This was not the case with the output of the first command above.

Kind regards

[Sebastian Kripfganz]

There is a bug in xtabond2 that computes incorrect degrees of freedoms for the overidentification test when there are omitted coefficients in the regression output (which often happens when you use dummy variables). Clearly, negative degrees of freedom do not make any sense. See slide 78 of my 2019 London Stata Conference.

One suggestion would be to use my xtdpdgmm command instead of xtabond2:

• Kripfganz, S. (2019). Generalized method of moments estimation of linear dynamic panel data models. Proceedings of the 2019 London Stata Conference.
• XTDPDGMM: new Stata command for GMM estimation of linear (dynamic) panel data models

Alternatively, you could adjust your xtabond2 syntax by manually only specifying dummy variables for non-omitted years and industries, i.e. make sure that none of the coefficients in the output are shown as "omitted" or "empty".

[Kevin Traen]

Thanks for your reply! The thing is that al the industries are omitted, so I wil make use of your command.
Thanks again!