Page 1 of 1

Fama-MacBeth regression

Posted: Sat Jun 16, 2018 1:34 pm
by Florina Burdet
I am an undergraduate student and I really need your help with a problem, I just hope somebody will reply me soon enough so I can finish my work.
My problem is about Fama-MacBeth methodology, that I don't really understand in all it's aspects. My teacher asked me to apply this methodology, but not as usual on portfolios, but on assets. I have 36 assets, more exactly, part of the CAC40 index which I'm supposed to use it as a proxy for the market portfolio.
My very very important problem is that I don't really understand how to form a panel in Excel (as my teacher told me) and then to introduce it in EViews and run just the cross-section regression of F-MB. the period that I've considered in my study is 2004-2017 and I don't really know how I'm suposed to apply F-MB just for 36 assets and for just 14 years. I will attach you my data in EXCEL so you can se what I'm talking about. I really need your help with this, Eviews community. :(
CAC40.xlsx
(99.13 KiB) Downloaded 415 times


In EXCEL I've done the following computations: rolling beta, rolling returns, I've applied simple regressions for every asset with CAC40 to find the intercept and the residuals and then the standard deviation of these residuals that I thought I needed to use them in F-MB model as being Sm. In the results obtained by F-MB, the did the determination coefficient between the return of every portfolio the formed and the return of market portfolio M, but am I supposed to compute this R2 for assets instead of portfolios, but using the same formula?

The idea is that if you could help me to create that panel of data in EViews, I think I'm able to manage all the other operations to obtain the results about returns and the behavior of the market. I'd be really grateful as at this moment I'm really helpless with this situation.

Re: Fama-MacBeth regression

Posted: Sat Jun 16, 2018 2:36 pm
by startz
I can see why you're confused. The Excel file seems to have 36 series plus a date. There don't appear to be any explanatory variables. Nothing looks like a panel.

Re: Fama-MacBeth regression

Posted: Sat Jun 16, 2018 8:58 pm
by Florina Burdet
Thank you very much for your reply.
I don't really know what to do from this point on. I've done a kind of panel in Excel, at my teacher's suggestion, but I'm really not sure how I should upload this to EViews and then apply the Fama-MacBeth add-in on this data.
panel.xlsx
(493.93 KiB) Downloaded 417 times

I've computed the returns Ri for every asset in every month from my testing period (2007-2017), in rolling with estimation period 2004-2006, the same with beta. Then I've computed beta i squared and then the residuals obtained from regressions between every asset and CAC40. Could you please tell me if this is the right procedure? and how should I apply Fama-MacBeth regression on this panel, if correct?

My problem is that I need to do the panel, then I can easily apply all sort of regressions.
Please help.