You can use *Variance decomposition *for getting the information content. It
is available in proc varmax.
On Fri, May 1, 2009 at 3:40 AM, firstname.lastname@example.org <email@example.com> wrote:
> I am using VAR model from Hasbrouck 1991 which estimates trade and
> midquote change as
> r(t) = a1r(t-1) + a2r(t-2) + a3r(t-3) ... + b1x(t-1) + b2x(t-2) + b3x
> (t-3) .... + e1
> x(t) = c1r(t-1) + c2r(t-2) + c3r(t-3) ... + d1x(t-1)+ d2x(t-2) + d3x
> (t-3) .... + e2
> then from VAR, it needs to be inverted to VMA representation in the
> form of
> [r(t) x(t)]' = [a*(L) b*(L) c*(L) d*(L)]' [ e1 e2]'
> in order to estimate information content of trade. The I can't really
> figure out how to tranform it. Can anyone help me out please? Thank