In a paper I submitted to a journal, I estimated a fixed effects panel model with regional fixed effects. The reviewer of my paper in his comments said my estimation was bad because I "simply used OLS with regional effects". I'm confused.

I thought the fixed effects estimator is different to the OLS estimator, and that his statement is wrong. Am I wrong?

My specification is:

crimei,t = δ + θi + αt + βaccessibilityi,t + εi,t

i= district (n=600)
t= year (2003-2014)
δ= constant
θ= district fixed effect
α= year dummy

  • 1
    $\begingroup$ They're the same. Maybe the referee has a problem with the linear model, rather than your specification. Was it an econ journal? $\endgroup$ – generic_user Sep 24 '18 at 11:18
  • 1
    $\begingroup$ What is the level of observation of the data? Is it region*time? Or are your data at a finer level than regions on the non-time dimension? $\endgroup$ – Roland Sep 24 '18 at 12:09
  • 1
    $\begingroup$ See stats.stackexchange.com/questions/174243/… $\endgroup$ – Christoph Hanck Sep 24 '18 at 12:27
  • 1
    $\begingroup$ This is probably more an econometrics question than a stats question. We probably can't comment unless we know more about your paper. But often times when I see econ papers presented at conferences, I see people using linear models in predictive settings, where they are either overfit or underfit. Other times I see people using linear models to derive ceteris paribus marginal effects, where ceteris paribus marginal effects don't make any sense given the nature of the problem. I don't know if either of these criticisms apply to your paper. Some referees are just dumb or lazy. $\endgroup$ – generic_user Sep 24 '18 at 15:12
  • 2
    $\begingroup$ As an economist, my guess would be that your referee's comment is not about the nature of the estimation technique, but more about your identification strategy. Controlling for the fixed heterogeneity of districts might not exhaust the endogeneity issues of the relationship between accessibility and crime. Just a guess, as I don't know your paper. But I agree with @generic_user that the criticism is probably more targeted towards the econ than the stats. $\endgroup$ – Roland Sep 24 '18 at 17:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.