EViews.com

Good morning,

How can i run *cross-sectional* regressions using a GMM framework in Eviews? I have pooled my variables but the only methods I see available are LS and TSLS. How do I perform a GMM? I also would like to compute Standard Errors using the Newey and West method.

Thank you very much! :)

Selin

Slightly confused by your question. If you want to do cross-sectional estimation, why have you pooled your data? Do you actually want to do panel estimation?

Hey Gareth, thanks a lot for your prompt reply. I have time-series data for 100 stocks. For each stock there is a series like return, volume, etc for the same time period. I am running cross-sectional regressions for different groups of stocks, so I need to transform my series into *panel* data? (I am actually new at cross-sectional regressions in Eviews. I thought I should pool my series to run cross-sectional regressions easily)
If I transform my data into panel data, how do I perform GMM?

Thanks a lot again! :)

Selin

I'm confused by your use of the word "cross-sectional". Cross-sectional data doesn't have time periods. Do you want to estimate using a single time period?

Or do you want to estimate using both your time dimension and the cross-section dimension at the same time?

I have a *single* time period (my time-series are given for the same dates and for the exact same period). I want to estimate using the cross-section dimension.
(I form 10 portfolios with the 100 stocks in my data set. Each portfolio consists of ten stocks. For example, I want to estimate how volume affects returns using these ten portfolios with GMM method.)

Given the current state of your data, this might be a little tricky. Essentially you'll have to convert from a pool workfile into a panel workfile, and then create a cross-section workfile and copy the data from the panel workfile to the cross-section workfile for the given date.

http://forums.eviews.com/viewtopic.php?f=7&t=74

Thank you so much! I will try it now. :) Have a great day!