Using comprehensive data on natural disasters, government spending, and election returns, I show that voters reward disaster relief spending but not disaster prevention spending. This aspect of voter behavior creates a large distortion in the incentives that governments face, since the data show that prevention spending substantially reduces future damage.
The paper has some nifty graphs showing the incumbent party's vote share (and change in vote share) as a function of relief and prevention spending. The slope for relief is sharply positive; the slope for prevention is flat. Given these incentives, it's hardly surprising that politicians spend about fifteen times as much on relief (which attracts votes) as they do on prevention (which doesn't).
Of course, if prevention were useless, this would be a pretty good outcome. But Healy presents additional evidence that disaster prevention spending has high returns:
Perhaps voters appreciate the ineptness of government prevention efforts. To consider this possibility, I estimate the effectiveness of government prevention spending...
Given mean annual prevention spending of $195 million and mean disaster damage of $16.5 billion, the regression estimates that a $1 increase in prevention spending resulted in a $8.30 decrease in disaster damage, and this estimate captures only benefits that occur in the five years from 2000-2004.
My main quarrel with this paper is that Healy ends by discussing an exception to the rule that voters don't reward prudent prevention. This might be a valuable topic for a follow-up paper, but it's a shame to dilute the message of the main results: Even when government spending could greatly improve efficiency, it fails to do so. In the hands of rational voters, government is strong medicine for public goods problems. But in the hands of real voters, government is usually a big bottle of snake oil.