SUEST with fixed effects

[Antonio Miguel]

Dear All,

I want to explain differences in portfolio returns for two types of investors (institutional and retail)
I'm using Suest, as it allows me to get the coefficients for both types of investors and to compute and test for the significance of these differences


In my regressions:
portfolio takes values from 1 to 5 (5 different portfolios)

institutional is 1 for institutional and zero for retail investors


This is the code:

**SUEST

fvset
eststo: reg return i.portfolio (+ controls) ///
if institutional==1

estimates store institutional

fvset
eststo: reg return i.portfolio (+ controls) ///
if institutional==0



estimates store retail

suest institutional retail, cluster (date)

test [institutional_mean]_cons=[retail_mean]_cons
test ([institutional_mean]_cons+[institutional_mean]2.portfolio)= ([retail_mean]_cons+[retail_mean]2.portfolio)
test ([institutional_mean]_cons+[institutional_mean]3.portfolio)= ([retail_mean]_cons+[retail_mean]3.portfolio)
test ([institutional_mean]_cons+[institutional_mean]4.portfolio)= ([retail_mean]_cons+[retail_mean]4.portfolio)
test ([institutional_mean]_cons+[institutional_mean]5.portfolio)= ([retail_mean]_cons+[retail_mean]5.portfolio)


The code above does not include fixed effects

So, I can test for differences between institutional and retail investors for all 5 portfolios ,as portfolio 1 coefficient is given by the constant

But, what if I include time and firm fixed effects in the regression (i.date i.id)?
How can I identify the coefficient of portfolio 1, as the constant coefficient will now be affected by the fixed effects?

Thanks in advance

Antonio