FK83 / bvarsv

Analysis of the Primiceri (REStud, 2005) model
28 stars 14 forks source link

Steady state solution for 3 variable TVP-VAR #7

Open AEdlerfi opened 5 years ago

AEdlerfi commented 5 years ago

Hi,

It is my understanding that we can compute the steady state values of the TVP-VAR by doing the following in each time period:

Yss = (I-B_1-B_2)^-1*C

Where, Yss is the steady state vector of n variables, I is the the nxn identity matrix, B_1 and B_2 are the time varying coefficient matrices and C is the time varying vector of intercepts.

If we were to do this using the example in the vignette, would the following code be correct:

`bvar.fit <- bvar.sv.tvp(usmacro, p =2, nburn = 5000, nrep = 50000)

ssvals <- list(ISTAR = NA, USTAR = NA, RSTAR = NA)

for(i in 1:dim(bvar.fit$Beta.postmean)[3]){

ss <- solve((diag(3)-bvar.fit$Beta.postmean[,2:4,i]-bvar.fit$Beta.postmean[,5:7,i]), bvar.fit$Beta.postmean[,1,i])

ssvals[["ISTAR"]][[i]] <- ss[1] ssvals[["USTAR"]][[i]] <- ss[2] ssvals[["RSTAR"]][[i]] <- ss[3]

} `

Thanks

Adam