### Re: Fama-MacBeth regression

Posted:

**Mon Jul 10, 2017 11:41 am**Page **11** of **11**

Posted: **Mon Jul 10, 2017 11:41 am**

Posted: **Tue Jul 11, 2017 1:24 pm**

you mean in every research we have this problem? because in all researches in this area, balance sheet data are annual but return data are monthly.

Posted: **Tue Jul 11, 2017 2:55 pm**

The question you asked is about frequency conversion/interpolation, not Fama-MacBeth.

Posted: **Tue Feb 27, 2018 12:20 am**

Hi,

why exactly do you provide the second regression, where you average the return over time and then doing one single regression?

What's "interesting"?

Can you avoid the assumption of no serial correlation, which you have in the first regression due to the assumption that the standard error of beta and pricing error estimates are treated as independent drawings of a common distribution in the second step?

Would be nice if you can provide me with some insight here.

why exactly do you provide the second regression, where you average the return over time and then doing one single regression?

What's "interesting"?

Can you avoid the assumption of no serial correlation, which you have in the first regression due to the assumption that the standard error of beta and pricing error estimates are treated as independent drawings of a common distribution in the second step?

Would be nice if you can provide me with some insight here.

Posted: **Tue Feb 27, 2018 12:58 pm**

Study some of the writing on Fama-MacBeth, e.g. the papers here.

Posted: **Tue Apr 10, 2018 5:46 am**

Hi,

I would like to know how I can attribute the saved bhat values to the original depended variables.

The add in saves bhat values for each independent variable c under r1, r2, ... ri instead of using the original name. I guess it uses the variables in the alphabetical order, which eviews also uses to display them in the workfile. Am I right?

To give an example: I use 25 portfolios and each portfolio has a certain number of returns over time. The Fama MacBeth add in makes a time series regression for each portfolio in the first step and saves the coefficients as bhat. There it lists the coefficients c1-ci for each portfolio, but the portfolios are not named similar to their original name, but as r1-r25. I need exactly to know, which r1-r25 I can attribute to which portfolio.

Kinds

I would like to know how I can attribute the saved bhat values to the original depended variables.

The add in saves bhat values for each independent variable c under r1, r2, ... ri instead of using the original name. I guess it uses the variables in the alphabetical order, which eviews also uses to display them in the workfile. Am I right?

To give an example: I use 25 portfolios and each portfolio has a certain number of returns over time. The Fama MacBeth add in makes a time series regression for each portfolio in the first step and saves the coefficients as bhat. There it lists the coefficients c1-ci for each portfolio, but the portfolios are not named similar to their original name, but as r1-r25. I need exactly to know, which r1-r25 I can attribute to which portfolio.

Kinds

Posted: **Thu May 10, 2018 6:29 pm**

Hi there,

I want to use Fama-MacBeth regression add-ins in Eviews and need some help on using the Fama-MacBeth add-in. I tried to search it in User Guide I and II but I am not getting any help related to this.

Please advise.

Thanks,Sara

I want to use Fama-MacBeth regression add-ins in Eviews and need some help on using the Fama-MacBeth add-in. I tried to search it in User Guide I and II but I am not getting any help related to this.

Please advise.

Thanks,Sara