Standard errors in sbinaryLGMM

[ALLemmens]

Dear G. Piras,

I am working with the spldv package in R, thank you very much for this clear and usable implementation!

I have two questions about standard errors:

• In the documentation it says: "The variance-covariance matrix can be computed using the traditional White-corrected coefficient covariance matrix from the last two-stage least squares estimator of the linearlized model." Are these also the standard errors that get reported if I use getSummary on the binlgmm object? Or do I need to do the White-correction myself via for example coeftest?
• Is it possible to use clustered standard errors when using the Klier & McMillen logit model as implemented by spldv? If so, how do I do it?

Kind regards. A.L. Lemmens

[mauricio1986]

Hi A.L Lemmens

1. Yes, the LGMM estimator uses the white-corrected variance-covariance matrix as default.
2. We are not completely sure why you'd use clustered standard errors in a spatial model. In any case, sbinaryLGMM delivers a fit value, which is basically an lm object. From there, you can use for example https://www.ne.su.se/polopoly_fs/1.216115.1426234213!/menu/standard/file/clustering1.pdf

Best,

Mauricio

[ALLemmens]

Thank you, very helpful!