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

# Re: st: SVAR estimation with Stata

 From Nick Cox To statalist@hsphsun2.harvard.edu Subject Re: st: SVAR estimation with Stata Date Wed, 4 May 2011 08:48:19 +0100

```It would have been helpful to all to have

1. Posted your second message as a reply to people who replied to you,
not as a new post with the same title.

2. Thanked them for their contributions.

Nick

On Wed, May 4, 2011 at 4:17 AM,  <ting.wang@yale.edu> wrote:

> But I posted new questions about Impulse Responses. And the instructions
> indicated previously in the response was the ones I tried before but failed.
>
> 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?
>
>
>
> Quoting Nick Cox <njcoxstata@gmail.com>:
>
>> This seems to overlap with a question asked and answered earlier.
>>
>> Nick
>>
>> On Tue, May 3, 2011 at 10:25 PM,  <ting.wang@yale.edu> wrote:
>>>
>>> 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?

*
*   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/
```