Announcement

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

  • Hausman test has suggested Random effects model but I want to use Fixed effects

    Hello everyone,

    I am using EU dataset for 27 countries for 5 different years. My goal is to determine what factors are influencing part-time farming in the EU. There are two types of part-time farming. I am running three different regressions. One for all types combined and two for two different types of part-time farming. Basically, the aim is to know what affects each type of part-time farming.

    I am interested in using the fixed-effect model for all 3 regressions. But Hausman test suggests a random effect model for the 1st regression and the fixed-effect model for the rest of the two.
    However, I want to use fixed effed model for all the 3 regressions.
    Can you guys please guide me, how can I use fixed effect model for the 1st one with a scientific explanation? and if I cannot use fixed effects model, how can I explain the difference of model in these regressions?

    Thanks

  • #2
    The only thing you lose by estimating a random effects model with the fixed effects estimator is efficiency. So it is fine to use the fixed effects estimator even if the Hausman points to random effects, but not the other way around. Bill Gould has a nice Stata FAQ on this.
    https://www.stata.com/support/faqs/s...effects-model/

    Comment


    • #3
      Originally posted by Andrew Musau View Post
      The only thing you lose by estimating a random effects model with the fixed effects estimator is efficiency. So it is fine to use the fixed effects estimator even if the Hausman points to random effects, but not the other way around. Bill Gould has a nice Stata FAQ on this.
      https://www.stata.com/support/faqs/s...effects-model/
      Thanks Andrew for your answer. Could you please guide me, how would you mention this in the research article?

      Comment


      • #4
        I would not discuss it as it is something that is well known. By efficiency, we mean that you will not be able to estimate the coefficients of time invariant variables with fixed effects as well as the intercept, but I guess you already know this. If a referee challenges you, then you just refer them to any standard graduate level panel data econometrics textbook.

        Comment


        • #5
          Muhammad:
          as an aside to Andrew's excellent replies, you may want to take a look at the community-contributed module -mundlak- (just type -search mundlak- to spot and install it).
          Kind regards,
          Carlo
          (Stata 19.0)

          Comment


          • #6
            Nobody can stop you from using fixed effects if you want to. You explain in your article that you think that controlling for country fixed effects is important because of this and that reason, and this is it.

            Comment

            Working...
            X