Dear Sebastian, It is known that RE (random effect) estimator assumes that the individual-specific effect $\alpha_i$ is uncorrelated with $x_{it}$ but FE (fixed effect) estimator allows them to be correlated. My question is that, under what assumption can we use the correlated random effects models (https://www.stata-journal.com/articl...article=st0283)?
-
Login or Register
- Log in with
Comment