Hausman test after xtivreg - REIV

[Maye Ehab]

Dear all,

I am trying to conduct the endogeneity test after implementing xtivreg command. However, the Hausman test used is not working. So, I tried suest command suggested by Stata but it also did not work.
So, my question is how to conduct the endogeneity test after REIV taking into consideration that I use bootstrapping for the standard errors.

Here are the steps that I tried:

sort Findid year
xtset Findid year 

*Without IV 
xi: xtreg crhrsday $X5 if female==1, re vce(robust) 
estimates store hoursdayfwotiv

*With IV
xi: xtivreg crhrsday $X4 (crtrvtmp= $IV2) if female==1, re vce(bootstrap, reps(500) seed(1)) first small
estimates store hoursperdayfemales

hausman hoursdayfwotiv hoursperdayfemales , sigmaless force
*Hausman did not work. Yielded a negative chi square and State suggested the use of suest. 

suest hoursdayfwotiv hoursperdayfemales

I also tried the below codes. However, it does not take into consideration the bootstrapping.

reg crtrvtmp $X4 $IV2
predict v1hat, resid
xtreg crhrsday crtrvtmp v1hat $X4 $IV2, re vce(robust)
test v1hat

Please advise me.

Thanks,
Maye

[Carlo Lazzaro]

Maye:
what about considering the user-written programme -xtoverid-?

[Maye Ehab]

Thank you Carlo for the advice. I am actually using -xtoverid- to test the strength of my instrument.
Can you guide me to a reading that help me understand the output of this command? I tried the help file but it is not exhaustive of all the output provided.

Many thanks,
Maye

[Carlo Lazzaro]

Maye:
unfortunately, possibly due to its completness .-xtoverid- did not come with an article published on the Stata journal, as it is frequent for most of the user-written programmes.
Hence, the only source of information I know (and use) is the related -help file-.