xthtaylor: Estimation with no time-invariant exogenous variables

[Alexandre Sonderegger]

Dear all,

Is it currently possible to estimate xthtaylor model with no time-invariant exogenous variables ?
If not possible, do you know any alternative Stata package (including community packages) to estimate
HT model which deals with this specification ?

Currently, I get error:

"There are no time-invariant exogenous variables in the model.
If you have those variables specified, they may have been removed because of collinearity.
r(198);"

Best regards,

Alexandre

[Sebastian Kripfganz]

I have suggested to StataCorp long time ago to amend xthtaylor such that it can be run in such cases. Unfortunately, I am still waiting for such an update to happen.

In your case, you can still trick the command to do what you want by generating and including a regressor for the constant (which is exogenous and time-invariant), and running the command with the noconstant option; the regression intercept is then the coefficient of that constant regressor:

webuse psidextract
gen byte cons = 1
xthtaylor lwage wks south smsa ms exp exp2 occ ind union fem blk ed cons, endog(exp exp2 occ ind union fem blk ed) constant(fem blk ed cons) nocons

[Alexandre Sonderegger]

Thank you for your suggestion, it works great.
It would be great that StataCorp implements this feature, users could then also use this specification without an intercept.