Interpreting pvarsoc command (optimal lag)

[Maria Connor]

I want to use a dynamic panel data model and I am trying to figure out what will be the optimal lag to use. I am using the command pvarsoc and these are the results I have got:

Selection order criteria
Sample: 2011 - 2019 No. of obs = 1654
No. of panels = 233
Ave. no. of T = 7.099

+---------------------------------------------------------------------------------------------------
| lag | CD | J | J pvalue | MBIC | MAIC | MQIC
|----------------------------------------------------------------------------------------------------
| 1 | .9999375 121.1734 .0020412 -471.7027 -38.82659 -199.3
| 2 | .9999236 76.4745 .1364925 -397.8264 -51.5255 -179.9042
| 3 | .9999165 55.38916 .2159705 -300.3365 -40.61084 -136.8949
| 4 | .9999074 23.97426 .8453458 -213.1762 -40.02574 -104.2151
| 5 | .9998875 15.50961 .4876777 -103.0656 -16.49039 -48.58508
| 6 | .9967625 . . . . . |
+-------------------------------------------------------------------------------------------------

Could anyone provide any sources to interpret this chart? It would really be appreciated.

[Jenner Kerr]

Anyone got an idea about how to interpret this? I've got a similar question so would be interested in the answer. FWIW you should use code de-limiters though

[Nick Cox]

See https://www.statalist.org/forums/for...cs-assumptions

[Sebastian Kripfganz]

The model selection criteria are those proposed by Andrews and Lu (2001, Journal of Econometrics). I also discussed their use in the context of short-T dynamic panel data models in my 2019 London Stata Conference presentation; see the section on model selection:

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