Non-consecutive number of lags in VAR (R package "vars")

Question

Is it possible (in package vars or maybe in some other R package?) to include non-consecutive lags into the var model, i.e., just lags 1 and 3.

So far, it looks like when I set p = 3 under function VAR, it includes all consecutive lags between 1 and p (i.e., 1:3).

Answer 1

You can use restrict from vars package for estimating a restricted VAR. This method requires estimate the model twice: 1) the unrestricted model with all the "consecutive lags" and 2) a restricted model with only the lags you want. This is so, becasue restrict function takes as input an object of class 'varest'. See my alternative:

> library(vars)
> data(Canada) # some data
> model <- VAR(Canada[,1:2], p=3) # The unrestricted VAR
> #Bcoef(model) The restriction matrix have to have the same dimension as dim(Bcoef(model))

# Building the restriction matrix
> Restrict <- matrix(c(1,1,0,0,1,1,1,
                       1,1,0,0,1,1,1), nrow=2, byrow=TRUE)

# Re-estimating the VAR with only lags 1 and 3 
> restrict(model, method = "man", resmat = Restrict)

VAR Estimation Results:
======================= 

Estimated coefficients for equation e: 
====================================== 
Call:
e = e.l1 + prod.l1 + e.l3 + prod.l3 + const 

      e.l1    prod.l1       e.l3    prod.l3      const 
 1.2029610  0.1885456 -0.2300286 -0.1299485  1.8382368 


Estimated coefficients for equation prod: 
========================================= 
Call:
prod = e.l1 + prod.l1 + e.l3 + prod.l3 + const 

       e.l1     prod.l1        e.l3     prod.l3       const 
 0.05511963  1.13333804 -0.03338699 -0.18646375  1.22037293 

See ?restrict for further details on this function.