Announcement

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

  • Instrumental Variable with binary endogenous variable and interactions and fixed effects - please help!

    Hello everybody,


    I please need your help, since I don't know how to proceed with my research.

    I'm estimating an inverted U-shape relationship between my outcome variable and my regressor and in a second step a moderating effect on this u-curve with a binary variable. This is done with a fixed effects estimator an implemented in stata as follows:

    Code:
     xtreg y c.x#c.x##i.dummy controls i.year, fe vce(cluster ID)
    The controls include lagged variables.

    Now I have reason to think that the dummy variable is endogenous, and I have two instrumental variables created for it.

    But how do I implement an instrumental variable regression with a binary endogenous variable, and that many interaction terms? And is there any way to keep the fixed effects estimator?

    Your help would be highly highly appreciated!







  • #2
    Sabrina:
    have you taken a look at -xtivreg, fe-?
    Kind regards,
    Carlo
    (Stata 19.0)

    Comment


    • #3
      Hello Carlo,

      yes I have tried, but I don't actually understand how to set up the code with the interactions and moreover, I have read at some places that you should use another model like a probit model if the endogenous regressor is binary (e.g. etregress?). I am just confused in what is the best option and how to set up the regressions because of the interactions.

      Do you have any advise?

      Comment


      • #4
        I don't want to annoy anyone, but if anybody could help this would really really save me here, deadline approaching

        Comment


        • #5
          Sabrina:
          https://www.stata.com/bookstore/micr...metrics-stata/ (pages 192-194) suggests (with a worked out example) -treatreg- for dealing with binary endogenous regressors.
          Kind regards,
          Carlo
          (Stata 19.0)

          Comment


          • #6
            Hello Carlo, thank you for your suggestion. Yes, this is the command I was referring to above, I read the same book. It's now called etregress. I'm still unsure though of how to implement the interactions.

            Comment

            Working...
            X