Sunday, May 02, 2004

Betting on Terror

Wired News just published an interview with Admiral Poindexter. I had thought of him as some Captain Birdseye type, but he is in fact a physics ace and ex-student of Richard Fenyman. Last year, under his direction, DARPA floated the idea of creating the Policy Analysis Market, a variation on the Iowa Electronic Markets that would allow the anonymous users to trade on their perceptions of various terror events happening.

I read stories in Wired, (see here and here) last year about this and I was surprised when it was killed.

Market foresight (i.e. the Efficient Market Hypothesis) is a controversial thesis, but hardly one unsupported by real-world evidence. No statistical techniques seem to have any profitable forecasting ability in real world tests. Very few mutual funds consistently beat the market. Most hedge funds have a limited life-span, no matter how smart their managers are (or in the case of Mr. Soros, how many muddled books they write.

Thomas Malone writes in the current Harvard Business Review of using markets internally for allocating resources, including investment capital and production capacity, more efficiently.

Robin Hanson, an economist at George Mason University - if Chicago is a "freshwater" faculty, these guys are distilled water - published a paper on decision markets in 1999.

This paper from the U. Iowa suggests how data from the Iowa Electronic Markets could be used for real-world decisions.

One common objection against this is that by allowing someone to profit from the prediction of a terrorist act, these might provide the incentive to commit it themselves. Of course, Hilary Clinton would think that; "a futures market in death".

First, as fanatical Islamists, are al-Qaeda any more likely to gamble than to carry bacon sandwiches or vodka in their cabin baggage?

Second, would the PAM make terror an attractive proposition? Well, are John Kerry and George W Bush raising the money they need for their campaigns by betting on themselves at the Iowa Electronic Markets? Are they motivated to run by the chance of winning a bet that they themselves have made?

Third, would anyone seriously think that the transaction history would allow any trader to cover his tracks from the CIA, any more than they would from the SEC or IRS?

It's pretty complicated to implement, I think, much more so than I thought when I first heard of it. There seems to be some confusion between political risk and terrorism losses. A host of institutional problems makes it difficult to implement.

Someone told me of a site called TerrorBet, which seems to be part of an Irish sports betting exchange is preparing to do something like this in Europe, which is more logical given the absurd constraints on gambling under U.S. law.

I doubt that it could be a serious money-making proposition though, because the number of trades and the volume would be absolutely tiny from individuals. Larger players would need to deal in huge sizes to lay off their risks. Say if Bechtel or Kellogg Brown & Root need to hedge some part of their Iraq risk?

Also, to build a serious marketplace for corporations and professional traders, you need to operate in and between the very different legal and economic frameworks of the markets for insurance, derivatives, and international public and private law.

Predicting or pricing terror risk may well be impossible using analytical solutions or past history. The losses could be so large and unpredictable from selling protection that it simply couldn't be priced in any way at all, leaving society to pass the responsibility to governments who have both the resources and the legal powers that individuals and companies don't.

U.S. insurers are trying to pass on their risk this way right now by removing the legal requirement for them to offer terrorism insurance. Remember how their airlines got an obliging Congress to give them a retroactive and zero-premium insurance policy in the form of a subsidy package after 9-11? Why can't I get one for my motor insurance?

Another possibility is that if Canary Wharf, BA, BAA and the other owners of terror-prone assets had to self-insure, they could create a private entity to spread the risk, which could be purely commercial or government-sponsored. This could then use the civil law, with its lower burden of proof, to go after the assets and persons of characters like those prosthetic-limbed clergymen or Arab princes who propagandise and finance al-Qaeda.

I think that the potential market probably works best as an information mechanism, and an academic motivation may draw in people that wouldn't be partial to a gambling site. I'd expect these would be useful mechanisms for analysing closed political systems. Why not use them for forecasting things like breakdowns in paramilitary cease-fires in Northern Ireland? Or changes at the top of the Chinese Communist Party hierarchy?

Peter 笔德