Fama-MacBeth (1973) using asreg

[Yuxi Chen]

Dear all,

I have some questions about the package asreg, thanks Attaullah Shah for sharing this in advance.

First, I have tried to run a simple CAPM, using monthly excess returns (exret) as independent variable and market risk free return (mktrf) as dependent variable.

asreg exret mktrf, fmb

and I get mktrf ommited from regression.

I have read one post explained by Attaullah Shah that the variables need to vary cross sectionally.

market risk free returns are the same every month, so I understand why my variable has been dropped. However, I though FM was once a standard procedure for running CAPM...so how do others manage to do this?

Thanks.

[Attaullah Shah]

Yuxi Chen Thanks for your kind words. This question is asked frequently here, and to answer this question, I wrote this post https://fintechprofessor.com/2019/06...sreg-in-stata/. See the details there.

[Yuxi Chen]

Dear Attaullah Shah,

Thanks for your reply, it explains the procedure well, just I am a bit confused about step2. I get betas for each company first by running

bys company: asreg returns market_returns

so this will be the first steps of FM procedure by doing a time-series regression for each company, and then use fmb command to complete the cross-sectional regressions. I don't get why we need step 2

xtset company month
asreg returns betas, fmb

[Attaullah Shah]

The logic is well explained in the seminal paper
Fama, E. F., & MacBeth, J. D. (1973). Risk, return, and equilibrium: Empirical tests. Journal of political economy, 81(3), 607-636.

[Yuxi Chen]

Thank you.

One last question, it might just be a typo. your step 3 said xtset company month, should that be xtset portfolio month?

Also, is it still eligible to ues FM procedure if I don't create portfolios at all, just using firm-month panel data, and get betas by running TS regression and then CS regression using fmb?