Fama-MacBeth regression

[jeric]

Hello,

Is it possible to get the coefficient values from the first step (time-series regression) of Fama-Macbeth regression?

Thanks in advance.

[EViews Rebecca]

I've added a line to the Fama-MacBeth add-in to produce the coefficients from the timeseries regressions. The updated program is now available for download and more details are in the documentation.

[metrix]

Dear Rebecca,
thank you for the update, i hope that the next version of update can do Fama French (1993) regression? :)

Nice day.

[EViews Rebecca]

Don't wait for another update. Now that the add-in is producing the time series coefficients, it should be easier for you to follow Fama-French 1993 methodology.

[metrix]

the bhat containing the coeficients , Now I understood.
thank you.

[mmc23]

hi Rebecca
I have a simple suggestion :
can you show the regression output of the first step in equation object in the spool object results, just to show Sdr-error and t-stat.....
and very useful add-in. Thanks for sharing.
Kind Regards.

[EViews Rebecca]

It should be very simple for you to modify the add-in to put the ts regression output into a spool. For example, in the add-in code replace the fragment beginning with "' calculate betas from ts regressions of portfolio/asset returns on market returns" with:

Code: Select all

' calculate betas from ts regressions of portfolio/asset returns on market returns %subbeta_summary = @getnextname("beta_summary") spool {%subbeta_summary} matrix(!num_p, !num_f) {%subbhat} for !i=1 to !num_p equation {%subbetaeq}.ls {%subrets}(!i) c {%subfacs} for !j=1 to !num_f {%subbhat}(!i, !j) = @coefs(!j + 1) next {%subbeta_summary}.append {%subbetaeq} close {%subbetaeq} next delete {%subbetaeq}

[mmc23]

thanks for your help, thanks a lot.

[ecofin]

hi Rebecca
it's very useful and quick to add a 'Sample dialog' to this Add-in, would you like to help me for the code.

best regards

[EViews Rebecca]

Dialogs in EViews generally don't allow you to set samples, so I would suggest simply using the workfile sample.

[ecofin]

thanks for your response.

[metrix]

Dear Rebecca,
This Add-in use OLS for cross-sectional average regression, '' In applying standard OLS formulas to a cross-sectional regression, we assume that the right-hand variables beta are fixed. The beta in the cross-sectional regression are not fixed, of course, but are estimated in the time-series regression. This turns out to matter, even asymptotically...'' (John H. Cochrane, ASSET PRICING, chapter 12 pp 239-240).
How can use others methods for the estimation like GMM or EGLS for the cross sectional regression out put of Fama-MacBeth methodologie. can you add two methods EGLS and GMM with the OLS in the Add-in (to make the choice with the three methods OLS or GMM and EGLS for estimating).
the reference: John H. Cochrane, ASSET PRICING, Revised edition, Princeton University Press, Chapter 12 pp 229-252.

it's just that the add-in must take all special cases.

nice day for you.
Kind Regards. Metrix

[EViews Gareth]

It would be easier for you to modify the code to change the estimation method yourself

[metrix]

I am a beginner, I'm not an expert in programming like you Mr Gareth, would you like to help me to have some ideas about the code.

thanks in advance.

[EViews Gareth]

Looks like the subroutine FMB is the part that does the calculations. The line:

Code: Select all

rowplace(g, @transpose(@inverse(@transpose(design) * design) * @transpose(design) * retvec), j)

is the one that performs OLS and puts the row of coefficients into the matrix g. Just change that line to do whatever calculation you want, instead of OLS.