Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.

# Re: Re: st: Regression with different firms

 From "felix kreppel" To statalist@hsphsun2.harvard.edu Subject Re: Re: st: Regression with different firms Date Fri, 10 Aug 2012 16:01:38 +0200

```Ok I have the solution: I used the Fama-Macbeth regression commmand xtfmb (net search xtfmb). The procedure is as follows: In the first step, for each single time period a cross-sectional regression is performed. Then, in the second step,  the final coefficient estimates are obtained as the average of the first step coefficient estimates.

I just tried to do this with my dataset with regression equation:

xtfmb market_return smb hml wml but

But when he gives me the regression output hey says variable coefficient = 0 and standard error omitted?

-------- Original-Nachricht --------
> Datum: Fri, 10 Aug 2012 14:27:55 +0200
> Von: "felix kreppel" <felix.kreppel@gmx.de>
> An: statalist@hsphsun2.harvard.edu
> Betreff: Re: Re: st: Regression with different firms

>
> My original empirical analysis works as follows:
>
> I am estimating a 4-Factor Model (with 4 factors: SMB, market_return, HML,
> WML which are the same for all firms) augmented by a fifth explanatory
> variable (which influence I want to evaluate) which is calculated as the
> average weekly standard deviation of excess return 12 months prior to month t
> for each firm:
>
> return_i_t=a*market_return_t+b*SMB_t+c*HML_t+d*WML_t+e*std_i_t
>
> where t indicates the month and i indicates the firm over a sample period
> of 25 years.
>
> What I did so far to solve my regression problem was to average all firm
> returns to an equally weighted index and also averaged all the previous
> volatilities to an equally weighted index and then estimated the following
> regression
>
> return_t=a*market_return_t+b*SMB_t+c*HML_t+d*WML_t+e*std_t
>
> with the command: newey return market_return smb hml wml std, lag(4) to
> address the serial correlation in the error terms.
>
> I do not know, however, if this approach works. Especially averaging all
> the previous standard deviations to one independent variable.
>
> Isn't there a possibility to run a regression for each single firm (say
> for each year) and then average coefficients, significane levels and standard
> errors together over the whole time period?
>
>
>
>
> > -------- Original-Nachricht --------
> > Datum: Fri, 10 Aug 2012 10:42:43 +0000
> > Von: Christopher Baum <kit.baum@bc.edu>
> > An: "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>
> > Betreff: Re: Re: st: Regression with different firms
> > &#xA
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
```