Saturday, September 17, 2005

Why is disaster relief so badly done?

In the wake of Katrina, everyone has their favorite politician to blame. Bush, Blanco, Nagin, Brown, Arthur Lawson, etc. Correctly affixing blame is important, but if you want that, there are dozens of other blogs you can read. Instead, I want to spend a bit of time thinking about structural reasons why disaster relief is so bad. Politicians come and go, but the system remains.

Leon Fuerth, Al Gore's foreign policy advisor, is blogging guest-blogging on this subject at Kevin Drum's place. He's hit the rather obvious point that government focuses on the short-term rather than the long term, and says we need the public to start telling politicians to pay attention to the longer term. This seems to be ignoring the hard part of the question, which is of course *how* you change the public.

Furthermore, it's not at all clear that you even want public pressure to be the dominant force providing incentives for long-term planning. People don't know much about disasters. This is partly due to the abysmal state of science reporting in this country, as I've been harping on recently. But more importantly, this is rational ignorance--the cost to the individual of, say, researching levee failure mechanisms, or reading about engineering buildings for eathquake safety, far outweighs the benefit. So even if the public did pressure politicians to do long-term planning, we would expect to see symbolic fixes for problems that people happen to be worried about, rather than fixes for problems that are most likely to do damage. Airport security costs increased by several billion per year after 9/11, while levee funding in NOLA fell. I would guess that a large fraction of the money spent on airport security upgrades was unnecessary, and levees... well, no more need be said about them.

So if the public can't bail us out, let's look at the government. What is it about its current design that makes it work badly in this area?

We can start with the fact that politicians have suboptimal incentives, as pointed out by Tyler Cowen.

Are democratic governments simply not very risk averse when it comes to very bad, low probability events? The model behind this conclusion is simple. Politicians would have to spend the money on protection no matter what, and lose the benefits of spending that cash elsewhere with p = 1. The chance of reelection goes up only with a small probability, namely if the bad event happens and voters can tell their representatives were suitably cautious. Why not instead spend the money with a higher chance of boosting reelection prospects? The key stylized fact is that if a politician messes up very badly, there is no penalty worse than removal from office, which is a penalty (roughly) fixed in value. And since the value of holding office may not fall in proportion to the suffering caused by the disaster, politicians' utility maximization will not bring optimal spending either.

Giving proper incentives to politicians is obviously a big part of the problem. Another way of looking at this is that there's a time lag between the disaster-causing political action and the actual disaster. If (say) levees are allowed to decay over decades, most of the politicians who allowed that decay will no longer be in office when they finally fail. The mayor of New Orleans didn't have a good plan to evacuate NOLA, but he didn't inherit one from his equally unprepared predecessors either. Thus most politicians avoid even the modest penalty of losing political office.

Also, the presence of term limits means that lame ducks don't run for reelection. People who (say) blame Bush for running FEMA as a political patronage operation can't punish him at all.

How can we redesign government to get around this? Unfortunately, I can't think of any solution that doesn't cause even greater problems. Can you? Everything I can think of leads back to the solution I bashed earlier, namely, depending on "the people" to hold politicians accountable.

Another issue is federalism. Disaster relief is handled by state, local, and federal agencies, all with overlapping areas of expertise and responsibility. As anyone who's worked in a group knows, overlapping responsibilities are a good way to ensure that nothing gets done. Everyone does as little work as possible, in the hope that someone else will pick up the slack. No one has a complete picture of the situation, so they can't tell what's really needed. Coordination is difficult. And blame can easily be shifted around, as anyone paying attention to politics post-Katrina can see. It seems to me that bringing all disaster relief responsibility to the federal level might be helpful. (Not the state or local level--that would lead to wasteful duplication.) However, I think this is a subject where actual knowledge of the structure of the various government agencies involved is necessary, so I should shut up.

I have a few more thoughts on the subject but I've got to go grade papers. Yech.