chrstdanne / varsignr Goto Github PK

Hi Christian,
I am currently going through this paper "Understanding the turbulence in financial markets"
mentioned here:
https://docs.google.com/viewer?a=v&pid=sites&srcid=ZGVmYXVsdGRvbWFpbnxqdWFuYW50b2xpbmRpYXp8Z3g6MjA1NDgzYWVhYTc1ODI5MA
and trying to implement the SVAR model explained in Page 7 with 4 structural shocks to these 4 endogenous variables. The imposed sign restriction on IRFS of the endogenous vars is given in Table 1 (page 8). The model is estimated using BVAR estimation using Minnesota prior.

I am trying to estimate this model using your package and have certain doubts:
In your example using uhligdata, when constructing the sign restriction vector, you have stated that
first element of constr indicates the shock of interest in the model.

In your example this shock is due to an endogenous variable (interest rate). In my example how do I give this shock of interest of an exogenous shock (like monetary policy,agg demand,agg supply,risk aversion etc.).

Also how do I construct the historical decomposition of shocks explained in page 12.

If we want to plot the historical decomposition of shocks, where the shocks are due to all endogenous variables, we can use this approach
https://stackoverflow.com/questions/36950491/historical-decomposition-in-r
Here the shocks are exogenous.

I would really appreciate if you could help me in clearing these doubts.

Thanks

Dear Christian,

I have gone through your documentation . Could you please explain how your FEVD measure is different from the variance of the k step ahead forecast revision (as per fig 9 of Uhlig 2005) . You have mentioned the same in footnote 9 but it's not clear to me . Could you kindly explain mathematically .

Regards
Arnab Biswas

I am quite new to R language and VARsignR. I wonder if it's possible to recover the input Y using the returns of rwz.reject. According to my current understanding about "Sign restricted VAR", if we can estimate coefficients and residuals, then Y can be recovered (composed back). However, I cannot see anything about the estimated residuals. Once I thought SHOCKS in returns (or goodshocks in the source code) might be what I want, but its dimension is not matched. I have been trying comparing the source code of rwz.reject and RWZAccept to the papers referred to, but still confused so far. So much thanks if anyone could help me!

Hi Christian,
does the RWZAccept fn accepts constraints with multiple shocks.
Lets say if I have a 4 variable and 2 shocks matrix. I assume your rows are shocks and cols as
variables.
then constr=matrix(c(4,-3,2,1,-1,-2,3,-4),nrow=2,ncol=4)
In this case, the length of constrained matrix will be 8 and the iterations will be till 8, but the ik will be a 4 * 1 vector.
Also how to setup a constraint matrix with no restrictions on certain variables for certain shocks