Impulse response functions when the system has equations with lag and contemporeneous relations

[Mike Kraft]

Hi
I have a system as following:

A(t)= A(t-1) + B(t-1) + C(t-1) + error(t)
B(t)= A(t-1) + B(t-1) + C(t-1) + error (t)
C(t)= A(t-1) + B(t-1) + C(t-1) + error (t)
D(t)= A(t) + B(t) + C(t)+ error (t)

In other words, all equations in the system show a lag relation except for the last equation where the relation is contemporeneous.In addition D variable does not appear in any of the first three equations. The theory underlying this is that the D variable is only affected by A, B, and C in time (t). It is stock return variable and theory assumes that the market is efficient. My aim is to estimate the impulse response fuctions (IRFs) and graph the responses of D to shocks in A, B, and C.

I understand that the system can be estimated using sur command, however, to report the IRFs, I have to estimate the system as a VAR using var command and then graph the IRFs.

My problem is in how to estimate the var while maintaining the lag and contemporeneous relationships so I can report the impulse responses after estimation??

[Mike Kraft]

Dear All
To clarify, the problem with the system above is that the last equation has contemporaneous relations among the D variable and other variables and hence I am not sure how this equation can be integrated with others in the system so I can estimate the IRFs.
If the last equation was not there I would use something as following:

var A B C , lags(1) dfk        
irf create order1, step(10) set(myirf1, replace)
irf cgraph (order1 A C oirf)(order1 B C oirf)

Is there any way to estimate IRFs when the system has one equation when all variables are contemporaneous?