I am conducting a data analysis. I have a panel with individual firms with firm-specific and macroeconomic variables. I would like to run an OLS regression adjusted for firm clustering effects and accounting for time effect (I have 449 firms and 36 time periods, unbalanced panel)

The question arises: When I include time-dummies will the effect of the "overall" macroeconomic variable absorbed by the dummies, in other word: Will including time-dummies lead to multicollinearity between the time-dummies and the "macroeconomic" variables?

I am wondering because different studies use this approach, although it appears more logical to me, to exclude the macroeconomic variables when including time-dummies.

Did I fail to consider anything important here?


This is an old question so not sure how much good it does but I'll have a go. If the macro-economic variables alter at the same speed (or slower) as your time variable, it should create perfect collinearity with time dummy variables. Depending on the kind of software you use this will either be automatically dealt with or result in not concave regression (stata) or cause a warning or error (R). (my experience is mainly with negative binomial regressions)

The different studies that use this approach will probably have some specific settings that allow both to co-occur or have macro-economic variables that change at different speeds (at different moments in time) to avoid perfect collinarity. Often different softwares deal with similar problems in distinct ways so it's hard to give more concrete advice at this point. (and perhaps pointless given the age of the question)


| cite | improve this answer | |
  • $\begingroup$ Hello Simon, thank you for your answer. I haven't been looking at this post, so sorry for my late answer. $\endgroup$ – eternity1 Nov 26 '14 at 19:17

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.