Tuesday, July 27, 2010

How Bad Investing relates to Terror Policy

I hear a lot of people talk about how "dumb" it is that America will spend a trillion dollars a year on terror-prevention but won't spend the same amount on healthcare, when the number of victims of terror over the last 10 years is one to two orders of magnitude smaller.
For this particular argument, I'm going to set aside the actual terror-prevention we got, and the actual healthcare reform we got, and look at these more generally. There are many arguments stating that our anti-terror actions have made us more vulnerable to terror in the last decade; I'm not enough of a foreign policy expert to assess those claims either way. For our purposes, let us assume that terror-intervention actually reduces terrorism by some amount. Similarly, I have argued a number of times that the healthcare reform that passed will kill more people than it saves by restricting healthcare technological innovation, significantly slowing the economy and reducing future stability domestically and abroad; for the purposes of this argument, I will assume instead that a better healthcare bill had passed and it really did insure everyone, cut down on the 'pre-existing' and dropped coverage debacles, and save a lot more people without these negative effects (anyone interested my general vision of what this bill would have to look like is welcome to click on the "favorite links" section to the right).
Let's also ignore the fact that a) good healthcare and good anti-terror are not mutually exclusive if designed to be fiscally appropriate and b) cost-benefit on a lives basis is certainly not the only way to think about these things.  The argument treats them as alternative uses of the same cash, and uses a cost-benefit on a lives basis, so that is the argument I will assess.
There are two major, major fundamental problems with the line of thinking that universal healthcare is automatically a better investment for the country than anti-terror spending because of expected lives lost to lack of insurance vs terror. Firstly, the assumption ignores tail risk, and secondly, the assumption ignores correlational stability.
The tail risk issue has an important analog - securities investing. One of the hallmarks of bad investing is that it weights returns heavily, routine risks somewhat, but significantly underweights tail risks. When you buy a company, you often think of how much you can earn first, what the probability is that you'll lose money due to routine competition second, and you probably don't think of megaproblems (An Enron-style scandal, an earthquake under their only factory, massive counterparty default). However, it is the megaproblems that define the competitive landscape in the long term. As a potential example, Citigroup may have had all sorts of competitive advantages over Wells Fargo in the years leading up to 2008 in terms of profitability, growth, etc, but Wells is the stronger company by far for the long run because Wells was better situated for disaster.
Similarly, terror spending is not about "routine" terror. If terror attacks were only about the lives lost (more about this in a second), then the occasional death of dozens, hundreds or even thousands of people would in fact be less than the thousands or tens of thousands of people who die annually from lack of health insurance (best number we have is about 20,000 a year, I think, though this number is disputed).
Terror outcomes have a distribution, and the tail of that distribution can stretch a long, long way. A successful EMP attack would require only one nuke, exploded high in the atmosphere, and could potentially disable the entire country's electronics. We're talking 100 million deaths. If that has a 1% chance of happening in the next 20 years, that's still an annualized 50,000 people in expected value. A biological attack could do something similar - perhaps that's another 50,000 people. And onwards it goes into other types of attacks. Casualty counts are hard to estimate, probabilities even more so, but if you're looking at an expected value, looking at the last 10 (or even 200) years for casualty counts does not adequately capture the tail risk of a first-in-history-but-still-possible type event. That expected value far exceeds the casualty count of healthcare. Perhaps there's an argument for spending on cutting tail risk instead of cutting overall terror risk, but that's not the argument opponents are making.
The second issue is less intuitive, but it has to do with correlational stability. Every death is certainly a tragedy, but clusters of deaths, for a society, are far more difficult to deal with. If an additional 1 person in every town or borough in America (25,375 total) passed away tomorrow, it would be really, really sad, but those towns would survive. If 25,375 people from one town passed away tomorrow, it would kill the town. If somebody thinking of moving to that town, or starting a business in that town, or selling their wares in that town, had to worry consistently about large scale numbers of deaths in that town happening at random, their own risk minimization would result in the town itself becoming highly unstable and a terrible place to live as everybody refuses to build out infrastructure for fear of it being rendered useless in a flash... even before any actual deaths happened. So while "routine" terror attacks may kill far fewer people than lack of healthcare, the effects of a lack of healthcare are roughly randomly distributed throughout the population, whereas terror attacks have the prospect of striking anywhere, and wherever they strike will have a large number of people disabled and infrastructure taken out. Terror is thus far more destabilizing than lack of healthcare and may kill more people in the long run by making people unwilling to invest in infrastructure that would save lives in the future.
I don't wish to be callous in a cost-benefit analysis; every death is a tragedy. But to respond to a cost-benefit argument on its own terms requires a cost-benefit argument, and if the spending is exclusive, the cost-benefit argument is not as completely in favor of healthcare as its supporters think. In real life, there's a balance to be struck (they're not mutually exclusive initiatives, as long as spending on other issues can be reduced), but I'd like to hear more talk of tail risk and correlational stability, and less talk of absolute historical numbers and present conditions when people talk of the war on terror - it'd force us to focus better on the issues that really matter to the US anyway.

No comments:

Post a Comment