Bjornland-Leitemo JME 2009

[TomDoan]

From the description of @VARIRF:

ACCUMULATE=||vector of positions||
Indicates which variables (by position in the model) need to be accumulated (integrated)

With the model set up with

var gap dpi compi2 dlrsp intr

that means the responses of dpi (which is the first difference of the inflation rate), compi2 and dlrsp are all accumulated ("undifferenced"). The DPI is fairly obvious, since then you get the response of the inflation rate rather than the change in the inflation rate. The DLRSP gives you the response of log stock prices (rather than returns), which should have a zero long run response to the monetary policy stock because of the restriction in the SVAR. COMPI2 doesn't really matter much.

What you would do with a different model will depend upon the variables and the model.

[istiak]

Hi Tom,
It seems to me that when we are creating a monetary policy shock and a stock price shock, we are creating a 1% change of the corresponding variables, am I right? Is there any way of changing the code and creating a 1 standard deviation change of the corresponding variables to create the shock?
Thank you.

[TomDoan]

Other than for the decomposition of variance, this does shocks normalized to unit impacts. Prior to the normalization, they're standard deviations. So you just eliminate (or comment out) lines that look like:

*
* Normalize the shocks
*
compute f=%ddivide(f,%xdiag(f))