The RATS Software Forum

Hi all,
I am wondering if anyone has encountered an issue with dominated autoreggressive term. eg
Y(t) = intercept + a * Y(t-1) + b1*X1(t) + b2*X2(t)
where a >> b1 and b2, when values of Y(t-1) are comparable to those in X1(t) and X2(t), or
where a * Y(t-1) >> b1 * X1(t) and b2 * X2(t)

Is there a way in RATs we can reduce the effect of the Autoreggressive term?

Thx!

Do you have a reference on that? First, it's not clear why that would be "reducing" the effect of the autoregressive term, since the condition puts a lower bound on the AR term, not an upper bound. Also, how do you define >> operationally?

Hi, thank you for getting back to me. I found a way to get around the problem. .. though have not solved the problem

Here is my question.... say we have the following AR(1) model:
home sales change (t) = 0.992 home sales change (t-1) + 0.002 GDP change (t), and assume both home sales and GDP changes are within -6% to +6%
then, since the term 0.992 * home sales change is a lot bigger in value (vs 0.002 * GDP change), it dominates and the 0.002* GDP change term has little effect on the model.
is there a way to constrain the coefficient of the home sales change (t-1) autoregressive term so that other exogenous variables can show up "stronger"?

For an ARDL model, the coefficient on the explanatory variable itself doesn't really determine the magnitude of the effect. Instead, in

y(t)=a*y(t-1)+bx(t)

the long-run effect of x is b/(1-a), so if a is close to 1, this could be quite large even if b is apparently small.