Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: Re: restricting sum of coefficients in IV


From   Christopher Baum <[email protected]>
To   "[email protected]" <[email protected]>
Subject   st: Re: restricting sum of coefficients in IV
Date   Sat, 7 Aug 2010 04:07:23 -0400

<>
On Aug 7, 2010, at 2:33 AM, Burak wrote:

> First, in the original specification (unrestricted), I get the following
> coefficients
> 
> F.gnd = .5361618
> L.gnd = .4614466
> 
> (they are all significant), so they already sum up to approximately one.
> 
> Then, when I try to restrict it by using the approach you have shown (D.gnd
> on the left hand side and F.gnd-L.gnd on the right hand side), I get
> a coefficient FLgnd = .4480393 (significant)
> 
> Isn't this a bit off? I mean, it drops from .53 to .44 even though the
> unrestricted model's coefficients almost sum up to one.
> 
> Besides that, I couldn't manage to use lincom to get the coeff on L.gnd,
> however, shouldnt 1-(FLgnd's coefficient) should be enough to get it?

Well, with that restriction imposed, you are solving a different optimization problem. For instance, the correlation
of FLgnd with the instruments would certainly be different than that of F.gnd. So it doesn't surprise me.

You should be able to do

lincom 1 - FLgnd

after estimation to get the estimate of the other coefficient.

Kit



Kit Baum   |   Boston College Economics & DIW Berlin   |   http://ideas.repec.org/e/pba1.html
                              An Introduction to Stata Programming  |   http://www.stata-press.com/books/isp.html
   An Introduction to Modern Econometrics Using Stata  |   http://www.stata-press.com/books/imeus.html


*
*   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/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index