lags in GMM

[Joe Joyce]

GMM estimations are best suited for data with "long N" and "short T," in part due to the proliferation of instruments as the length of the time period expands. There are ways to deal with this, such as the use of the "collapse" command. In addition, in xtabond2 the lags can be specified using the "laglimits" command. What sort of rules of thumb are used to specify these lags? The second and third seem to be good candidates, but are there other considerations to be aware of?
Thanks--Joe

[Sebastian Kripfganz]

There is unfortunately no general rule. You might find the discussion in the following paper by Jan Kiviet helpful:

• Kiviet, J. F. (2020). Microeconometric dynamic panel data methods: Model specification and selection issues. Econometrics and Statistics, 13, 16-45

More on the GMM estimation of dynamic panel models in Stata:

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