The effects of a change in policy can't be predicted from historical data on things that aren't invariant with respect to the policy.
Robert Lucas: "Given that the structure of an econometric model consists of optimal decision rules of economic agents, and that optimal decision rules vary systematically with changes in the structure of series relevant to the decision maker, it follows that any change in policy will systematically alter the structure of econometric models."
e.g. You can't predict how raising taxes on something will affect revenue, because when you raises the taxes people will stop doing that thing to an unpredictable extent.