panel data - xtreg, fe

[lizziebee]

Dear statalist, I do have a question concerning fixed effects models. I have unbalanced panel with 6 million observations in 5 years - id is the panel variable and year the time variable. LHS variable is health care expenditures, RHS variables include dummies for sociodemographic characteristics. Is there any possibility to include additionally zip code fixed effects in the model, without generating dummies? Because if this is not possible, I am confronted with 2000 dummies for the zip codes... Maybe someone can help me – that would be great! Thanks in advane! Lisa Wallner

[Brendan Halpin]

Are you irrevocably wedded to xtreg, fe? Your data has a multi-level structure, so xtmixed seem appropriate, with ID nested within location:

xtmixed expenditure rhsvars || zip: || id: