Simulate a bivariate vector follow VAR(2) structure

[FeiGu]

Hi,

Can someone help me verify that the following code generate a bivariate vector "z" that follows a VAR(2) structure? The first AR matrix is

0.6  0.0

0.0  0.6

and the second AR matrix is

0.18  0.00

0.00  0.18

This vector series also has a linear time trend with coefficient vector, g =

0.3

0.5

Thanks!

Fei

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

PROC IML;

mean_eta  = J(1,2,0);

sigma_eta = {1.0 0.5,

                    0.5 1.0};

eta       = randnormal(300,Mean_eta,Sigma_eta);

f    ={

0.6  0.0,

0.0  0.6};

g    = {0.3, 0.5};

t    = T(do(1,300,1));

z = J(300,2,0);

do i = 3 to 300;

  z[i,] = T( f*T(z[i-1,]) + 0.3#f*T(z[i-2,]) + g*t ) + eta[i,];

end;

Quit;

[udo_sas]

Hello -

This might give you some ideas:

Taken from http://support.sas.com/documentation/cdl/en/etsug/60372/HTML/default/viewer.htm#etsug_varmax_sect003...:

The following IML procedure statements simulate a bivariate vector time series from this model to provide test data for the VARMAX procedure:

   proc iml;
      sig = {1.0  0.5, 0.5 1.25};
      phi = {1.2 -0.5, 0.6 0.3};
      /* simulate the vector time series */
      call varmasim(y,phi) sigma = sig n = 100 seed = 34657;
      cn = {'y1' 'y2'};
      create simul1 from y[colname=cn];
      append from y;
   quit;

Thanks,