Is it possible to apply hurdle models (like the Craggit, probit and truncated models) to panel data, preferably with fixed effects to control for unobserved heterogeneity?
In Stata, the user-written command craggit
only allows to use pooled panel data, but not to control for unobserved heterogeneity…
In general, the hurdle model is quite simple to estimate: probit model on the binary outcome variable, truncated (or lognormal) on the (positive values of the) outcome variable.
Is it possible to just estimate a RE probit (or even FE logit) in a first step, and a fixed effects model in the second – and to actually call that a hurdle model?
Thanks.
Best Answer
You can estimate double hurdle models to panel data with random effects using the module called
dhreg
.The Stata module is described on the following article (paywall until 2017):
http://www.stata-journal.com/article.html?article=st0359
You can find and install the software by typing
findit dhreg
on Stata.