Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | aslihan arslan <aslihanarslan@yahoo.com> |
To | statalist <statalist@hsphsun2.harvard.edu> |
Subject | st: ML with fixed effects and IV? |
Date | Wed, 18 May 2011 00:54:45 -0700 (PDT) |
Dear Statalisters, I am trying to estimate the determinants of different sources of household income simultaneously using a short panel data with t=2. My main rhs variables of interest are endogenous, therefore I also need to use instruments. I realize these may be too many econometrically difficult estimations put together, but I'd appreciate if anyone has suggestions as to how to do an ML+IV+Fixed Effects regression for all income sources. I have seen Stephen Jenkins' post last November (http://www.stata.com/statalist/archive/2010-11/msg00327.html), but seems like there is no follow-up on that panel-ml yet. Even if panel-ml was available, I'll still need to figure out how to do IV. If I manually include the first step predicted values of endogenous rhs variables in a panel-ml program, would the "robust" standard errors also take care of the IV adjustment? Lastly, I already tried xtivreg2 (IV+GMM) giving up on the system estimation based on Christopher Baum's advice that reg3 is very limited and sur cannot handle IVs, and IV+GMM is advisable in that case (http://statalist.1588530.n2.nabble.com/Re-st-st-Simultaneous-Equations-Model-amp-GMM-Estimation-td5308431.html). These results did not make much sense and I thought taking into account the cross equation correlations of error terms in different income sources may improve results. Thanks in advance for any ideas, Best, Aslihan * * 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/