Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Question: How can I test whether individual variables are endogenous (with unobserved heterogeneity) using the Hausman test?

    Hello,

    I am using the Hausman-Taylor IV estimator (xthtaylor). The command requires me to specify which independent variables are endogenous in the model, i.e., those which are correlated with the unobserved heterogeneity. Previous studies explain they have been able to identify whether each independent variable is endogenous through using fixed effects models with the Hausman test. I'm aware that the Hausman test can be used to identify whether a fixed effects model or random effects model should be used, but how can it be used to identify the endogeneity of each independent variable?

    (I am using Stata/BE 17.0)

  • #2
    Patrick:
    I would take a look at the comprehensive helpfile that comes with the community-contributed module -xtoverid-.
    Last edited by Carlo Lazzaro; 27 Mar 2023, 09:28.
    Kind regards,
    Carlo
    (Stata 18.0 SE)

    Comment


    • #3
      Thanks for recommending -xtoverid-.

      I've figured out how to obtain the Sargen-Hansen statistic, but how do I (if possible) see the difference between the fixed-effect and random-effect coefficients on a variably-by-variable basis, and their respective standard errors? I was hoping to use the standard errors to obtain t-statistics in order to find out which variables are endogenous with the unobserved heterogeneity.

      Comment

      Working...
      X