st: Panel-robust Hausman Test

[Lukas Borkowski <Lukasbork@hotmail.com>]

Dear all,

I am using Stata 11.2 and I am trying to run a panel-robust Hausman Test as suggested by Cameron & Trivedi, Ch. 8. The code is

preserve
qui xtreg lngdp_wb ptot ppri psec pter $controls, re 
scalar theta = e(theta)
global yandx lngdp_wb ptot ppri psec pter $controls
sort id 
qui foreach var of varlist $yandx {
by id: egen mean`var' = mean(`var')
gen fe`var' = `var' - mean`var'
gen re`var' = `va' - theta*mean`var'
}

* Wooldridge's auxiliary regression for the panel-robust Hausman test
qui reg relngdp_wb reptot reppri repsec repter relnoecd_dist reoecd_comlang_ethno redisteq ///
relandrel relif remor_u5 rerat_bir relnpop_tot relnland reopenk remalfal94 ///
feptot feppri fepsec fepter felnoecd_dist feoecd_comlang_ethno fedisteq felandrel felif ///
femor_u5 ferat_bir felnpop_tot felnland feopenk femalfal94, vce(cluster id)

* Test of the null-hypothesis gamma==0
test feptot feppri fepsec fepter felnoecd_dist feoecd_comlang_ethno fedisteq felandrel felif ///
femor_u5 ferat_bir felnpop_tot felnland feopenk femalfal94
restore


Even though I have been successfully running this test a couple of times before, suddenly an error message appears after scalar theta = e(theta) telling me that there would be a type mismatch when generating theta. However, all variables are numeric and there cannot be any kind of type mismatch.

Does anyone have a solution to this rather stupid but time-consuming problem? Thanks!

#
Lukas Borkowski




*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/