… is from pages 179-180 of the 2012 revised edition of Steven Landsburg’s book The Armchair Economist  (original emphasis; footnotes deleted):
When we are told that we should subsidize the aerospace industry because we’ll need their facilities in the event of war, economists are likely to be skeptical. In ordinary circumstances, entrepreneurs can foresee the probability of war as accurately as government officials can. If there is a one-third chance of a major war in five years, then there is a one-third chance that a factory capable of producing fighter aircraft will be a very profitable thing to own. Why shouldn’t that prospect provide sufficient incentive to keep the factory in business?
Of course, there will be fewer such factories when the chance of war is one-third than when it’s one-half, but that is presumably the outcome that would also be chosen by a wise government. It makes good sense to invest fewer resources to defend against a less probable event.
The proper incentive is missing only if investors expect the government to follow historical precedent and impose price controls in time of war. If we are concerned about our defense preparedness, the problem arises not from too little interference with the market (in the form of subsidies) but from too much interference (in the form of controls). The best prescription for military preparedness might be a constitutional amendment guaranteeing freedom from price controls.
Notice the central role that the probing, economically informed question  plays here in Steve’s analysis.