Regression with panel data: How to display and store the estimated coefficients?

[Meiqi Wong]

Dear all,

I need to examine the effect between stock return R and change of exchange rate FX over a rolling 3 year-window for my thesis, also known as β_1i in the following regression:

Rit = β0i + β_1i*FXt + β_2i*RMRFt + β_3i*SMBt + β_4i*HMLt + εit

I’m working with monthly panel datasets with a time range from 2005-2014, each dataset contains data of 3 years and the same sample of firms i (e.g. dataset 1 from 2005-2007, dataset 2 from 2006-2008, etc.).

After I declared my data to be panel data, I have entered the following command for a GLS random-effects regression (of course separately for each dataset):

xtreg R FX RMRF SMB HML, re

Now I've been struggling on how to retrieve the estimates for coefficient β₁i_, which I need for further calculation and regression (2SLS). Is there a way to display and store the estimated coefficients for each firm?

I’m grateful for any advice, if you need more information please let me know. Thank you!


Kind regards,

Meiqi

[Attaullah Shah]

There must be an id for each firm, assuming that the variable is named as id, then you might use foreach loop such that

levelsof id, local(firmid)

foreach i of local firmid{

     xtreg R FX RMRF SMB HML, re if id==`i'

     outreg2 using regression_results.xls, append ctitle(firm`i')
}

You can use other methods to extract coefficient values instead of using outreg2, see http://www.stata.com/statalist/archi.../msg00892.html

[Meiqi Wong]

It works out, thank you very much for your instruction!

Best regards,
Meiqi