Re: st: Averaging coefficinets across reg3 output

[Maarten Buis <maartenlbuis@gmail.com>]

Sorry, forget about my previous answer. You already have everything in
one model. So you can just use -lincom- to compute the average
coefficient and its standard error:

*---------------- begin example -------------------
// get data
webuse supDem, clear

// estimate the model
global demand "(qDemand: quantity price pcompete income)"
global supply "(qSupply: quantity price praw)"
reg3 $demand $supply, endog(price)

// see how coefficients are called
reg3, coefl

// compute the average effect of price
lincom (_b[qDemand:price] + _b[qSupply:price] )/2
*-------------------- end example -----------------
(For more on examples I sent to the Statalist see:
http://www.maartenbuis.nl/example_faq )

-- Maarten

On Tue, Jul 12, 2011 at 8:53 AM, Maarten Buis wrote:
> As a general strategy I would stack the data and estimate the entire
> system in one model with interaction terms to allow T and X to have
> different effects for different equations. I would use effect coding
> for those dummies (see for example: -findit xi3-), such that the main
> effects of T and X are the effects for an "average" equation. I don't
> know how well that works with models like -reg3- though.
>
> -- Maarten
>
> On Mon, Jul 11, 2011 at 6:29 PM, Guy Grossman wrote:
>> Hi -
>>
>> I am reposting a question that was not responded. Assistance would be
>> highly appreciated. In short, I am using Stata 11.2 on mac to estimate
>> the following system of equations:
>>
>> Y1 = alpha_1*T+beta_1*X +e_1
>> Y2 = alpha_2*T+beta_2*X +e_2
>> Y3 = alpha_3*T+beta_3*X +e_3
>> .
>> .
>> Yk = alpha_k*T+beta_k*X +e_k
>>
>> where Y1-Yk are standardized dependent variables, X is an exogenous
>> covariate and T is a binary endogenous variable, instrumented by Z. I
>> am using reg3 (three-stage estimation). My main interest is the
>> coefficients on T (the alphas). Y1--Yk are thought to be from the same
>> family of outcomes. My goal is to conduct the following analysis:
>>
>> 1. calculate a mean effect value that captures the average
>> relationship between T and the K different outcomes that belong to the
>> same family
>>
>> 2. calculate the standard error of the mean effect value which depends
>> on the variance of of each of the individual alpha and the covariance
>> between
>> alpha_1,k and alpha_1, --k
>>
>> 3. test the cross equation hypothesis that the average effect of the K
>> coefficients equals zero
>>
>> My Stata code starts with:
>> global y1 "(q1: y1 t x)"
>> global y2 "(q2: y2 t x)"
>> global y3 "(q3: y3 t x)"
>> global y4 "(q4: y4 t x)"
>>
>> reg3 $y1 $y2 $y3 $y4, endog(t) exog(z)
>>
>> I am not sure however, how to continue to stages (2) and (3).
>> suggestions would be highly appreciated!
>> Regards,
>> Guy
>>
>>
>> _____________________________
>> Guy Grossman
>> Dept. of Political Science
>> Columbia University, NY
>> www.columbia.edu/~gsg2102
>>
>> *
>> *   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/
>>
>
>
>
> --
> --------------------------
> Maarten L. Buis
> Institut fuer Soziologie
> Universitaet Tuebingen
> Wilhelmstrasse 36
> 72074 Tuebingen
> Germany
>
>
> http://www.maartenbuis.nl
> --------------------------
>



-- 
--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany


http://www.maartenbuis.nl
--------------------------

*
*   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/