0
$\begingroup$

I have two equations to estimate on two different datasets.I want to test if the coefficient for x are the same. The two equations are $Y_{1t} = \alpha_{11}*X_{1t}+ \alpha_{12} * X_{2t} $ and $Y_{1t} = \alpha_{21}*X_{1t}+ \alpha_{22} * X_{2t} $. The two equations are estimated using two samples of the same dataset. In the first data set , $Z==1$, and the second, $Z==2$. But since the populations is a panel , and $Z$ changes over time. so the same individual could appear in both samples. I am estimating the two equations using fixed effects model.

I have two questions for the procedure.

  • How is the test statistics constructed for constraints $\alpha_{11}=\alpha_{12}$
  • Issue in implementation. Some has suggested using suest in stata. But I am running a panel regression. And suest does not work with xtfeg. I know there is a routine xtsur . But it constantly gives me an error message. And others seem to run into the same problem. So my question is What is the proper way to implement this in a statistical software (Stata, R ,etc).
Improve this question
$\endgroup$
7
  • $\begingroup$ You are running two fixed-effects panel regressions on data sets where there's no correlation except within each panel? $\endgroup$
    – dimitriy
    Commented Jan 8, 2015 at 16:29
  • $\begingroup$ I suspect there would be correlations. The two samples are separated based on their smoking status in each wave. So there could be individuals showing in both two waves, due to quitting and relapsing. $\endgroup$
    – Yan Song
    Commented Jan 8, 2015 at 16:35
  • $\begingroup$ You really need to clarify what you're doing and the structure of your data. In the mean time, take a look at the Blackwell SJ paper cited in the comments here. $\endgroup$
    – dimitriy
    Commented Jan 8, 2015 at 18:03
  • $\begingroup$ There also might a way to do this using gsem. $\endgroup$
    – dimitriy
    Commented Jan 8, 2015 at 18:14
  • $\begingroup$ I edited the questions for background. $\endgroup$
    – Yan Song
    Commented Jan 8, 2015 at 19:21

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.