Announcement

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

  • OLS vs. Panel regression

    Hello everyone,
    actually I got a comment from the journal, I did not understand what the difference between OLS and panel regression.
    the comment says ((The authors are using OLS as a baseline methodology, but your time period is from 2009 2015, I suggest that the authors could conduct panel regression as it accounts for individual specific heterogeneity.))

    please someone helps, I want to know the difference and how to apply Panel regression.

    thanks in advance
    Alkebsee

  • #2
    Alkebsee:
    it is really difficul to explain in a reply why reviewer is in all likelihood right in suggesting to switch from pooled OLS to panel data regression.
    You can grasp /(at least) the backbones of panel data regression studying one of the many handbooks on this topic (see, for instance, the pivotal Jeff Wooldridge 's contributions).
    My best recommendation is seeking help from a co-autthor (if any) and/or local expert in econometrics.
    Kind regards,
    Carlo
    (Stata 18.0 SE)

    Comment


    • #3
      Originally posted by Carlo Lazzaro View Post
      Alkebsee:
      it is really difficul to explain in a reply why reviewer is in all likelihood right in suggesting to switch from pooled OLS to panel data regression.
      You can grasp /(at least) the backbones of panel data regression studying one of the many handbooks on this topic (see, for instance, the pivotal Jeff Wooldridge 's contributions).
      My best recommendation is seeking help from a co-autthor (if any) and/or local expert in econometrics.
      thank you so much
      I got your point
      but is panel regression is fixed effect model ?

      Comment


      • #4
        Alkebsee:
        not quite.
        The most used specifications in panel data regression are fixed and random effects.
        Kind regards,
        Carlo
        (Stata 18.0 SE)

        Comment


        • #5
          Among the basic estimators, fixed effects is the most robust as it removes the heterogeneity entirely. RE removes a fraction of it; pooled OLS none of it. But you can try all three, with cluster-robust standard errors. It's usually informative to do so.

          Comment


          • #6
            Originally posted by Carlo Lazzaro View Post
            Alkebsee:
            not quite.
            The most used specifications in panel data regression are fixed and random effects.
            Ok gratitude

            Comment


            • #7
              Originally posted by Jeff Wooldridge View Post
              Among the basic estimators, fixed effects is the most robust as it removes the heterogeneity entirely. RE removes a fraction of it; pooled OLS none of it. But you can try all three, with cluster-robust standard errors. It's usually informative to do so.
              thank you so much i tried fixed-effect but the results are not consistent. I will try Cluster model.

              Comment


              • #8
                Usually if FE is very different from POLS or RE you have to go with the FE results. It depends on how different RE and FE are, and the outcome of the Hausman test. But we can't help you any more without seeing output.

                Comment


                • #9
                  Originally posted by Jeff Wooldridge View Post
                  Usually if FE is very different from POLS or RE you have to go with the FE results. It depends on how different RE and FE are, and the outcome of the Hausman test. But we can't help you any more without seeing output.
                  thank you so much. actaully i m far away from my data of interest. when i m home i will show you all results i will got.
                  thank you again

                  Comment


                  • #10
                    I hesitate to quibble with Professor Woolridge's comment, but the choice between fixed effects (within) and random or between estimators also depends on the theory and the problem. If the theory is about within-panel variation over time, then fixed effects is clearly the way to go. However, if the theory is about relatively stable cross-panel differences, then the fe estimates are not informative. The derivation of the conclusion that testing fixed vs random effects and going fixed if they differ usually assumes that the within and between effects should have the same parameters. If this is the case, then differences between within and between probably reflect estimation problems in the between. However, in many situations, the cross-panel theory and mechanisms differ from within-panel theory and mechanisms.

                    Comment


                    • #11
                      I'll quibble a bit with Phil's quibble. It's hard to think of cases where, if the FE estimates and RE estimates differed significantly on time-varying variables that I would place much value on the RE estimates. I don't think of it as a problem of "variation" -- that sounds like a technical issue that doesn't really have anything to do with establishing causality. If the goal is a descriptive regression then just use pooled OLS. That's arguably as descriptive as RE or the between regression. If you want to control for systematic, unmeasured differences across units (individuals, firms, schools, and so on) then FE is preferred. If the variables of interest don't vary enough over time to identify the effects then we might need a new problem or a new data set. Now, if FE is introducing unnecessary noise then RE can be preferred -- but that's an empirical matter.

                      Another way to state it: How can any theory reliable conclude that unobserved heterogeneity is uncorrelated with the observed covariates? How could I ever be sure that, say, managerial talent is unrelated to firm inputs? The only theory that implies POLS or RE is suitable is if we have a randomized intervention -- still quite rare in the social sciences.

                      Comment

                      Working...
                      X