Panel Data - Unit specific regression coefficients vs constant coefficients

[Andrés Álvarez]

Dear Statalist,

I am currently, using Stata 12.1, working on a panel data set of the 48 contiguous states over 17 years. Hence, N, the number of units is larger than T, the number of recorded instances for each group.

I have used both the xtreg command in the fixed effects and random effects version and am content with the results.

I would like to, however, in a next step, try to estimate the same two specifications, but with unit specific, time invariant regressors, that is:

Fixed effects: y_it = constant_i + X_it*beta_i + e_it (*)
Random effects: y_it = constant + X_it*beta_i + v_i + e_it (**)

From reading the longitudinal/panel data reference manual (Release 10) I feel the "xtrc" is not what I am looking for since it computes a common beta and then adds a stochastic term for each group.

Upon using another book (1), I thought the definition of the seemingly unrelated regression model came close to equation (*). However, it is stated there that this specification may only be used if T exceeds N, which is not the case for me.

I feel a bit lost, not knowing how to procedd, so, as so often, I rely upon the kindness of strangers. A kindness much appreciated. Best regards,

Andrés Álvarez


(1) - An Introduction to Modern Econometrics Using Stata, Christopher F. Baum, Stata Press

[Klaus Pforr]

Dear Andres,

the random slope or random coefficient model (**) is described in Cameron/Trivedi (2009) in chapter 9.5.5 and in Rabe-Hesketh/Skrondal (2012) in chapters 4 and 5.5.4. Those are implemented in Stata 12 in the multi-level mixed effects commands or glamm or the new gsem package.

The individual slope model (*) is described in Wooldridge (2010) in chapter 11.7.2. There is an implemetation by Volker Ludwig (xtfeis http://www.stata.com/meeting/germany...y10_ludwig.pdf).

Best

Klaus

[Andrés Álvarez]

Dear Klaus,

I will research those option. The pdf looks very intriguing. Thank you ever so much.

Best regards,

Andrés

[paulvonhippel]

I am interested in Volker Ludwig's implementation, too. But all I see in the pdf are slides from a presentation. Where is the ado file?

[francis tandoh]

HELLO FRIENDS AM WORKING ON A PANEL TRESHOLD COINTEGRATION, I DONT KNOW IF THE CODE EXIST FOR THIS IN STATA AND IF IT DOES CAN ANY ONE SHOW ME HOW IT IS DONE. THANKS

[Carlo Lazzaro]

Francis:
- please, do not chime in a thread seeking advice for a different issue, but start out a new one undrr a different title;
- a generally agreed netiquette recommends not to shout one's message using upper-case letters, By they way, they might hamper much more than help the likelihood of getting useful replies.

Kind regards,
Carlo