Announcement

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

  • Can Hausman test be disregarded depending on the context (Interpreting random effects results)?

    Hi everyone, I have a panel data (2017-2021) set where I am trying to assess whether CEO traits (IV) influence a firm's environmental innovation score (DV).
    I have tested whether I have a panel-wise effect (which I do), and I have conducted the Hausman test to decide whether I should use a fixed effects regression or a random effects one.

    When I run the Hausman test it clearly indicates I should use a Random effects regression, since Prob > chi2 = 0.7223. Mundlak approach also suggests against fixed effects.
    However, in the context of what I am trying to investigate, it seems to me that a fixed effects regression would make much more intuitive sense as I am trying to investigate characteristics that influence a firm from within.

    I also read this: "Whenever there is a clear idea that individual characteristics of each entity or group affect the regressors, use fixed effects. For example, macroeconomic data collected for most countries overtime. There might be a good reason to believe that countries’ economic performance may be affected by their own internal characteristics: type of government, political environment, cultural characteristics, type of public policies, etc. " (Torres-Reyna, 2007). I thought this sounds very applicable to what I am trying to do.

    My questions are therefore:
    1. Should I stay with the random effects regression or can I, in the context of my study, choose to use the fixed effects instead, despite the Hausman test result?
    2. If I should use the random effects regression, how do I interpret the results, particularly in regards to the coefficient?
    According to what I read the coefficients from the resulting regression would indicate "change in the output (y) when the predictors change one unit over time and across entities (average effect)" (Torres-Reyna, 2007). However, I reckon it does not make sense in my example as I am not looking for across firms (entities) .

    PS: I did see there was a thread here (see link) about a similar question, but I believe it was diverted as it turned out the person in question did not have panel wise effects to start with:


    Hope my question is clear and thank you in advance for any advice!
    Hi all, I am running both a random-effects model and a fixed-effect model, to see whether a firm's ESG score has an effect on return. xtreg Return
    Last edited by Teresa Dilling; 23 Nov 2023, 17:35.

  • #2
    I believe the guidance on the Hausman test is inadequate, both in advising when to use the test and in teaching the correct interpretation of its results. In any case, fixed effects is always consistent, so proceed with it without considering anything else. There was a nice discussion in a recent thread on exactly this topic, see https://www.statalist.org/forums/for...itive-definite.

    Comment


    • #3
      Dear Andrew, thank you so much for you reply!

      Comment

      Working...
      X