Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.

st: SVAR estimation with Stata

 From ting.wang@yale.edu To statalist@hsphsun2.harvard.edu Subject st: SVAR estimation with Stata Date Tue, 03 May 2011 17:25:20 -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.

Here is the Stata codes:

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', exog(`dum') lags(1/6) aeq(A) beq(B)
{`dum' refers to the 11 dummy variables treated as exogenous}

Now create the Impulse Response Function:

irf create model2, set(myirf, replace) step(8)
irf table fevd, noci {noci is to suppress confidence bands}
irf graph oirf, impulse(r) response(cpi) {orthogonalized IRF}
irf graph sirf, impulse(r) response(cpi) {structural IRF}

Questions:
1. Is it the correct way to involve "lags" to the estimation?
2. The convergence is not achieved, any possible reasons for that?
3. When I create the IRF, Stata was stuck and could not move on. Does anybody
know how this might come across?

Thank you very much. Looking forward to any positive reply.

ting

--ting
*
*   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/
```