st: Question on dynamic panel estimation

["Palan, Stefan (stefan.palan@uni-graz.at)" <stefan.palan@uni-graz.at>]

Hi everybody,

I have data from an experiment with 4 treatments, each with 6 markets, of which each again has 15 periods (So in Treatment 1, I have markets 1, 2, 3, 4, 5 and 6, and 15 observations each. In Treatment 2, I have markets 7, 8, 9, 10, 11 and 12, each again containing 15 periods). I would like to determine whether there are differences in the intercept and slope of the market price between the treatments (which are fixed within markets). Markets are independent of each other, but periods within a market are not. In other words, I expect there to be autocorrelation in observations of consecutive periods (at least at lag 1, possibly at more).

Now I am not sure how to get the information I want. If I estimate using xtabond, I lose any information on intercepts, since it uses first differences, right? I can use xtmixed with a random intercept for each market and including the lagged price as an independent variable, but does that lead to correct standard errors/consistency/...?

I hope someone can help me...

Best regards,
Stefan.

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/