Is it correct that dynamic panel without individual fixed effect (but with group fixed effects), no serial correlation in error can be estimated consistently through pooled ols?

For example, suppose we have $$ y_{it} = \rho y_{it-1} + \beta x_{it}+v_g+\epsilon_{it} $$ where $E(x_{it}\epsilon_{it})=0$ and $E(y_{it-1}\epsilon_{it})=0$.

Do we need to use Arellano and Bond's GMM method in the above case?

  • $\begingroup$ Is $y_{it}$ on the right-hand side a lag of your outcome? $\endgroup$ Jul 25, 2020 at 16:46
  • $\begingroup$ @ThomasBilach It's a typo. I have corrected it. $\endgroup$
    – chuck
    Jul 25, 2020 at 16:47


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.