[R] Options for running Hausman-Taylor with robust standard errors

Alex I alexierino at hotmail.com
Fri Nov 13 10:34:48 CET 2015


This has been eating at me for a very long time. From everything I understand, the plm package does not yet allow robust standard errors to be calculated for the Hausman-Taylor using the coeftest command you would for other estimators (i.e. pooled, within, random). This is of course the error that plm shoots out when you try to run it in the package, which you can see below:


Error in vcovHC.plm(ht, type = "HC0") : 
  Model has to be either random, within or pooling model


It just seems like such a strange oversight given the popularity of HTM in econometrics and the almost granted necessity of robust standard errors.


What are my options, perhaps using other simple packages in R, to calculate a Hausman-Taylor using robust standard errors for my panel data regression? I’m not really sure if sharing data or much code will help in this case.


Alex
	[[alternative HTML version deleted]]



More information about the R-help mailing list