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: Redundant control variables in ivreg first stage
From
[email protected]
To
[email protected]
Subject
st: Redundant control variables in ivreg first stage
Date
Sat, 3 Dec 2011 23:02:59 +0100
Dear All
I want to regress the cost of capital (COC, dependent variable) on voluntary disclosure quality (VDQ, endogenous instrumented variable). If I am using the STATA command ivreg, I have to specify the dependent variable, the endogenous instrumented variable(s), the control variables and the excluded instrument(s). Specifically, the control variables specified will appear in both the first and the second stage. In principle, the reduced form of a 2SLS regression should look exactly the same as an OLS regression of COC on VDQ and the control variables when treating VDQ as exogenous. Therefore, the control variables in an IV estimation are specified with the second stage in mind, meaning that they are correlated with COC. However, those same control variables may be useless in explaining VDQ. This implies an overidentification of the model such that the adjusted R-squared of the first stage may be suboptimally low, which could be shown by conducting an F-test. Can you tell me whet!
her there is an option in ivreg that allows for excluding redundant control variables in the first stage?
Thanks in advance for your help,
Andreas
*
* 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/