Ivan,
I am not sure what you did with the xtreg command, but the model
looks to me as spatial competition, you have 20 regions that compite
for the people. In your case I will take a look of that literature before
to move on something on heteroskedasticity of the current model.
For the same reason above, you could re-post the question changing
the subject to spatial, with that you have more chances to receive
feedback from other users that work with that field.
Rodrigo.
On 9/12/07, Ivan Etzo <[email protected]> wrote:
> Dear all,
>
> I'm working with a panel dataset with yearly regional migration flows for
> every pair of regions, that is 20x19x7 (origin regions-destination
> regions-years).
>
> The indepvar is the yearly migration flow from the ith region (origin) to
> the jth region (destination). The covariates are per capita GDP (origin and
> dest region) unemp rate (origin and dest region) population (origin and
> destinat) and distance. so in total 7 covariates. All variables are in log.
>
> I'm in the early stage of my analysis and new with Stata. So far I estimated
> the model with the xtreg command and I got satisfying results, both in terms
> of sign of coeff and their significance.
> However I'm not sure all of this is enough (or correct). For example per
> capita GDP and unemployment rate are likely to be correlated and there might
> be heteroskedasticity..
>
> My question is: do you think I'm using the proper model for such data set or
> maybe I need to look for more advanced techniques? Should I use a dynamic
> model (arellano bond), or maybe the xtgls command (for heteroskedasticity),
> or xtivreg (for endogeneity)?
>
> Thank you to all of you
>
> Ivan
>
> *
> * For searches and help try:
> * http://www.stata.com/support/faqs/res/findit.html
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
>
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
