Broken windows fallacy?

I’ve seen various bloggers attack the assertion that Japan’s earthquake will be good for the economy (see, e.g., David Bernstein’s post below) as an example of the broken window fallacy.

Now I like the story behind the broken window fallacy as much as the next guy. Yes, GDP might go up, but this only shows how meaningless GDP is. GDP is just, roughly speaking, the sum of prices in transactions, so anything non-market goes uncounted. If you bomb a building and then hire people to rebuild it, the bombing counts as 0 for GDP (not a negative) and the rebuilding counts as a positive; so you can bomb the building ten times and have it rebuilt each time for a super boost to GDP! But clearly, from a real point of view, you’re just using up real resources without creating anything on net. Or, in Bastiat’s terms, consider the “jobs created” by the bombing; without the bombing, that money would have been spent in other ways, perhaps creating jobs elsewhere, but that’s unseen.

Still, I’m not sure the broken windows fallacy is necessarily a “fallacy”. Yes, there are people who don’t understand the seen/unseen distinction as described above. But there are more sophisticated claims one could make, which might be true:

  • The rebuilding money goes to particular people, e.g. construction workers. The money otherwise might have gone elsewhere, e.g. symphony orchestras. The spending induced by the disaster clearly has a redistributive effect, which one might consider desirable if one has particular redistributive goals.

    Yes, maybe one could accomplish those redistributive goals directly. But (1) the existing tax system may not be able to identify the same set of beneficiaries — what if your target group is not “the poor” but “construction workers”?; and (2) even if one could imagine some tax break or subsidy that would benefit exactly that group, it might not be politically palatable in the absence of the disaster. So the disaster might eliminate a political barrier that otherwise would have existed.

  • The rebuilding money goes to consumption, whereas without the disaster it might have gone to savings or investment. Now savings and investment may well be good because they increase future productivity; but there’s also the paradox of thrift and, generally, the view, associated with Keynes, that consumption spending is inefficiently low during recessions. Now maybe Keynes is wrong, but I don’t have a super-strong view on that.

    And, as before, it would be better not to have a disaster but to just vote to spend the money, but there might be political resistance to this by taxpayers, libertarians, Tea Partiers, etc. So one can argue — as, indeed, (perhaps) Keynes and (definitely) some others do — that, if the only way to get that spending is to have disasters, wars, etc., then at least those events will have good economic consequences by removing the political barrier that would otherwise exist.

  • Independent of the recessionary point, there might just be some productive government spending out there, like perhaps some infrastructure spending, that’s efficient regardless of whether we’re in a recession. So if the rebuilding replaces old infrastructure with new and better infrastructure, then that will be an improvement. And unlike the previous points (redistributive policy or counter-cyclical policy), this one will affect long-term growth.

    Yes, it would have been better to have just upgraded the infrastructure without the disaster, but, I think you see this coming by now, political barriers.

  • David himself has suggested, in the comments to his post: “To the extent that the government is about to waste money on nonsense, say, high-speed rail from Tampa to Orlando, and a natural disaster forces the government to redirect that spending to rebuilding basic infrastructure that was in need of replacement anyway, the economic consequences of the natural disaster could be offset by the appropriate redirection of government priorities.” David thinks this is unlikely — “But more likely the government will just add new spending on top of whatever nonsense it would doing before.” — but of course that’s an empirical question.

    Here, the political problem was in a pro-government-spending direction, but it was in the direction of bad government spending, and now we’d have more productive government spending.

None of this is to deny that there is a fallacy somewhere in there: if someone seriously doesn’t understand that you have to take the resource destruction into account, that’s fallacious, and once you take it into account, it’s unlikely you’ll have anything to show for it in the end.

But if any of the above are true — if there’s a valid distributional argument in favor of the people who would get the post-destruction money; if counter-cyclical expenditure is a good idea; if infrastructure spending is a good idea and the disaster causes an upgrade; or if the disaster diverts funds from worse government projects to better ones — and if those problems couldn’t be better addressed in other ways because of political barriers, then, indeed, disasters, wars, etc., may have beneficial economic effects on net.

UPDATE: Commenter Le Poisson points out correctly that destruction of physical does get taken into account. It shows up in the national income accounts as “consumption of fixed capital”. So just ignore the GDP-related prologue to the argument.

UPDATE 2: Commenter awp points out that I was too quick in my correction, so my original version was right after all! (So I’ve uncorrected my previous correction.) GDP – consumption of fixed capital = net domestic product. So NDP falls when a building is destroyed, but GDP doesn’t. So the destruction has a zero effect on GDP, and the reconstruction is all positive.

UPDATE 3: Welcome, visitors from Instapundit, and happi Pi Day!