# Two-way model estimation

I have a model with 2-way fixed effects and panel data.

$$y_{it}=\alpha_i+\delta_t+\beta X_{it}+ \theta D.$$

The coefficient of interest is $\theta$. (D is time dummy). I can estimate the model with RE estimator but the Hausman test indicates the estimator is inconsistent. Is it possible to estimate the model with fixed effects and how to implement it in R?

• Your use of $\alpha$ for two different things is a bad idea; consider $y_{11}$ -- now $\alpha_1$ represents two different things. – Glen_b Jan 4 '16 at 3:05