It's natural and efficient to look just one step ahead and then accept that result as "the answer". Why waste energy thinking about higher-order effects, when most of the time they can be safely ignored? In simple circumstances such effects are negligible. But when things are even slightly complicated and when feedback loops add nonlinearity, higher-order phenomena become essential in a correct analysis.
Consider, for example, a widespread electrical power outage, a paralyzing blizzard, a general strike, an Internet worm, or any other semi-extraordinary large-scale phenomenon that interrupts commerce. Factories and shops are closed for several days. Most media coverage simplemindedly takes the average dollar amount of daily business done in the affected region, multiplies it by the duration of the disruption, and calls that "The Loss".
How ridiculous! If something isn't done right now and as usual then it doesn't ever happen?
No account is taken of shifts in making and spending, of delays (or advances) in trade, or of the value of new things substituted for old, expected activities. Time spent at home with family can be precious --- and maybe a lot more important, in the long run, than shopping. So are reading, thinking, learning, resting, and a host of other nonmonetary recreations. Just because they can't (easily) be measured doesn't mean that they aren't worthwhile.
So I have to laugh when I read the staggeringly huge "cost to the economy" of whatever the latest front-page headline is in a frenzy about ...