Restrict/Constrain coeffcients in a VAR/VECM

[ac_1]

Hi Tom,

If I want to simplify/refine a fitted VAR model e.g. removing insignificant parameters, or in an ad-hoc manner, or ideally optimally; i.e. what would be appropriate reasoning and options? For example, using

Code: Select all

open data e1.dat
calendar(q) 1960
data(format=prn,org=columns,skips=6) 1960:01 1982:04 invest income cons
*
set dinc  = log(income/income{1})
set dcons = log(cons/cons{1})
set dinv  = log(invest/invest{1})
*
system(model=varmodel)
variables dinv dinc dcons
lags 1 2
det constant
end(system)
estimate(sigma,residuals=resids,noftests) * 1978:4

if I wanted to restrict: %betasys(3), %betasys(4), %betasys(7), %betasys(11), %betasys(15) to zero, how would I constrain those coeffcients, and re-estimate?

thanks,
Amarjit

[TomDoan]

You would have to create separate equations for each dependent variable. To get maximum likelihood estimates, you would need to use SUR rather than ESTIMATE.

[ac_1]

You would have to create separate equations for each dependent variable. To get maximum likelihood estimates, you would need to use SUR rather than ESTIMATE.

I can follow SUR.RPF to define the equations and estimate via SUR, but how do I set individual coeffcients to zero?

Code: Select all

equation dinveq dinv
# constant dinv{1 to 2} dinc{1 to 2} dcons{1 to 2}
equation dinceq dinc
# constant dinv{1 to 2} dinc{1 to 2} dcons{1 to 2}
equation dconseq dcons
# constant dinv{1 to 2} dinc{1 to 2} dcons{1 to 2}

group varmodel1 dinceq dinveq dconseq
sur(model=varmodel1) * 1978:4

[TomDoan]

Leave them out of the equations.