Multivariate Multiple Regression

[dada cox]

Hi Stata-users

i´m currently writing at my thesis and have a problem i dont really know how to solve; maybe there is someone who knows how to do that ;-)

For my thesis i´m analyzing stock returns of banks operating abroad - have to measure the exposure of banks in eastern european countries (cross border).
I have about 10 banks with daily stock returns from 2006-2015 for every country, these stock returns then have to be regressed with 10 daily local industry-indices for every country in which they are operating. Finally i have about 15 rows with almost 3000 lines for each of these countries.

In the paper that serves as template they analyze around 12000 banks worldwide and regress each of them with these local industry-indices and provide an overall-like regression and correlation result at the end which should serve somehow as a summary for all these banks.

Have been googling for such a long time but unfortunately without any useful result. The only thing I found is that this kind of statistics must be called multivariate multiple regression... or something like that; and i also think that the use of dummy variables is not very appropriate in this case because of the data structure.

Is it possible to do something like that with stata and if so, how??

Maybe someone knows

Thanks a lot!!!!

[Marcos Almeida]

- messagHello "dada cox",

Welcome to the Stata Forum. Is this your real name? If not, please re-register just by clicking on the "contact us" button.

Also, I kindly ask you to read the FAQ: there, you may find great suggestions on how to present a query and how to improve the chances of getting advice from the Forum members.

Going directly to your query, you asked whether it is possible to perform a multivariate regression in Stata.

Please take a look at this text (http://www.ats.ucla.edu/stat/stata/dae/mvreg.htm), and see examples with - manova - and with - mvreg - and see if they apply to your needs.

Last but no least, even though your study design is not yet clear (at least to me), it seems there is a panel data structure, pointing to a different analysis.

Best,

Marcos

P.S: I assume you are not referring to multiple linear regression (note: after reading Cox's - in fact, to be clear, Nick's, not Dada's - message, I realized the mistyping and changed the "P.S," from "multivariate" to "multiple").

[Nick Cox]

Marcos gives good advice. We could be related, but I too have to ask whether Dada Cox is your real name. More importantly, I am shocked that you seem to be expected to write a statistically-based thesis without statistical background. If so, you should take some courses first, or hassle your supervisor/advisor/committee for support.

To your point, multivariate regression implies multiple responses; multiple regression just multiple predictors. In my experience almost all mentions of "multivariate regression" confuse it with "multiple regression". If you don't know the difference, then the advice has to be to understand multiple regression first.