Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | ting.wang@yale.edu |
To | statalist@hsphsun2.harvard.edu |
Subject | st: SVAR estimation with Stata |
Date | Mon, 02 May 2011 22:22:40 -0400 |
Dear all, I am running Structural VAR model with Stata. Suppose the contemporaneous coefficient matrix is G0, that is, e(t)=G0*u(t), where e(t) is the structural error and u(t) is the error of the reduced form VAR. My inquiry is about how to set up the A and B matrix in Stata based on G0. Is B matrix supposed to be an identity matrix? How do we involve lags in structural VAR estimation? If G0 is considered to be instantaneous, does it really matter how many lags to include in the model? Here is the codes I used in Stata. It does not converge when we carry out a maximum likelihood estimation. Iteration only goes on forever, which forces me to limit the maximum number of iterations. matrix A (exactly the G0 matrix) =(1,.,0,0,.,0,.\.,1,.,.,0,0,0\0,0,1,.,.,0,0\0,0,0,1,.,0,0\0,0,0,0,1,0,0\0,0,0,0,.,1,0\.,.,.,.,.,.,1) matrix B =(1,0,0,0,0,0,0\0,1,0,0,0,0,0\0,0,1,0,0,0,0\0,0,0,1,0,0,0\0,0,0,0,1,0,0\0,0,0,0,0,1,0\0,0,0,0,0,0,1) svar `varlist', lags(1/6) aeq(A) beq(B) I hope someone can help with this. Thank you very much. I really appreciate it. * * 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/