RE: st: xtivreg2- dropping instrumnets form the instrument matrix

[Nick Cox <n.j.cox@durham.ac.uk>]

It's quite orthogonal to the original question or Kit's points, but I go queasy when I see transforms of temperature, here the logarithm. What scale is temperature measured on here? A logarithm for anything with an arbitrary zero seems very unphysical. You may be using the Kelvin scale, however. Perhaps it's an empiricism that linearizes the growth-temperature relationship, at least roughly, but most plants have some kind of optimum temperature for yield so I would expect a turning point somewhere. The turning point could be very important because it would be fairly silly to grow any crop in a serious way in conditions some way from the optimum, even if it were physiologically or agronomically possible. 

Nick 
n.j.cox@durham.ac.uk 


-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Christopher Baum
Sent: 31 October 2011 18:52
To: statalist@hsphsun2.harvard.edu
Subject: re: st: xtivreg2- dropping instrumnets form the instrument matrix

<>
I want to exclude some instruments (that is all exogenous explanatory variables in my model) from the instrument matrix in the xtivreg2 estimator.

e.g. 
xtivreg2 Price  (  Crop= LogTemp) year GDP, fe first robust

here the first step regression includes the variable year and GDP
I want to suppress this.
Is this possible?

I want to isolate the effect of temperature on the country's crop yield and hence on the price.


No, it is not possible nor desirable, from either econometric or economic grounds. The IV estimator requires that you include all exogenous variables in the Z matrix, including those already in the equation's X matrix. See the Baum-Schaffer-Stillman papers in Stata Journal 2003, 2007.