1: \begin{abstract}

2: This paper considers the maximum likelihood estimation of panel data

3: models with interactive effects.

4: Motivated by applications in economics and other social sciences, a

5: notable feature of the model is

6: that the explanatory variables are correlated with the unobserved effects.

7: The usual within-group estimator is inconsistent. Existing methods for

8: consistent estimation are

9: either designed for panel data with short time periods or are less

10: efficient. The maximum likelihood

11: estimator has desirable properties and is easy to implement, as

12: illustrated by the Monte Carlo

13: simulations. This paper develops the inferential theory for the maximum

14: likelihood estimator,

15: including consistency, rate of convergence and the limiting

16: distributions. We further extend

17: the model to include time-invariant regressors and common regressors

18: (cross-section invariant).

19: The regression coefficients for the time-invariant regressors are

20: time-varying, and the

21: coefficients for the common regressors are cross-sectionally varying.

22: \end{abstract}