I have estimated an ADL(2,2) model by using the xtabond2 module in the same STATA-9 program but installed at two different computers. The set of right-hand side variables and their distributed lag terms are treated as exogenous. The gmmstyle instrument set is invoked only for the lagged dependent variable, while other exogenous set of variables are put into the ivstyle instrument set. In particular the command is:
/*ADL(2,2): Treating rer and wrer2 as exogenous*/
xtabond2 rxct L(1/2).rxct L(0/2).(loggdpm loggdpx logrer logwrer) year contig comlang logdist if expid==1, gmm(L.rxct, lag(2 .) collapse) ///
iv(L(0/2).(logrer logwrer loggdpm loggdpx)) iv(year contig comlang logdist, eq(level)) nod robust small
Please note that xtabond2 module is updated with the latest available routines in both the computers. However the estimation results are very different from one PC to another. However this problem is not observed in the case of following estimation command:
/*ADL(2,2): Treating rer and wrer2 as predetermined*/
xtabond2 rxct L(1/2).rxct L(0/2).(loggdpm loggdpx logrer logwrer) year comlang contig logdist if expid==1, gmm(L.rxct, lag(2 .) col) ///
gmm(logrer logwrer, lag(1 .) col) iv(L(0/2).(loggdpm loggdpx)) iv(year comlang contig logdist, eq(level)) nod robust small
I have difficulty to understand the problem. I will appreciate your explanations.
Thanks.
Mizanur Rahman
National Graduate Institute for Policy Studies (GRIPS)
Tokyo, Japan
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/