R/fn.impulse.R

In VARsignR: Sign Restrictions, Bayesian, Vector Autoregression Models

fn.impulse <-
function(Bh,swish,nn){
nvar <- nn[1]
lags <- nn[2]
imstep <- nn[3]
#
ll <- lags + 1
n1 <- nvar + 1
nl <- nvar*lags
#
Ah <- t(Bh)
#
imf <- matrix(nrow=imstep, ncol=nvar*nvar)
#
M <- matrix(nrow=nvar*imstep, ncol=nvar)
#
M[1:nvar,] <- t(swish)
Mtem <- M[1:nvar,]
#
imf[1,] <- t(as.vector(Mtem))
#
ims2 <- imstep - 1
ims1 <- min(c(ims2, lags))
t <- 1
while(t <=ims1){
   nt <- nvar*t
   ntt <- nvar*(t+1)
   tt <- t+1
   Mtem <- Ah[,1:nt] %*% M[1:nt,]
   M[n1:ntt,] <- M[1:nt,]
   M[1:nvar,] <- Mtem
   imf[tt,] <- t(as.vector(Mtem))
   t <- t+1
}
#
for(t in ll:ims2){
   nt <- nvar*t
   ntt <- nvar*(t+1)
   tt <- t+1
   Mtem <- Ah[,1:nl] %*% M[1:nl,]
   M[n1:ntt,] <- M[1:nt,]
   M[1:nvar,] <- Mtem
   imf[tt,] <- t(as.vector(Mtem))
}
return(imf)
}