Wednesday, July 30, 2003
Wham, Bam, Thank You PAM
posted by James
[update: Glen Whitman makes essentially the same point at Agoraphilia]
I was going to write an angry, incredulous post about just how unfair most criticism of the Policy Analysis Market (PAM) has been, but it looks as though James Surowiecki has already done a pretty good job over at Slate. Slate is perhaps redeeming itself; yesterday it ran a frustratingly inconclusive piece by Daniel Gross, many of whose questions are answered by Surowiecki. The idea behind PAM is fairly simple; it's a market in which you invest in “futures” that pay off depending on certain events in the Middle East (assassinations and political intrigue, as well as mundane things like economic growth). The government would monitor the market and use the information to prevent certain things (presumably the assassinations) and ameliorate others (economic slowdown). In fact, the program sparked a political firestorm, and has been canceled. I expect Democrats to use the program as a rhetorical punch line whenever they criticize Bush's foreign policy. It's too bad, because PAM might well be among the better ideas Bush's team has come up with.
There are some good reasons to think PAM is a bad idea. One is that it depends on people profiting from tragedy; exactly how big a problem this presents has been discussed over on Muxalicious. Additionally, like any market PAM would be subject to manipulation. Since the market would automatically feed back into US policy, it could become a self-fulfilling prophecy, and it might even destabilize political regimes by amplifying any bad news. I'll examine some more shortcomings, but first I should explain why it's such a beautiful idea in the first place.
One principle objection raised against PAM is that terrorists would benefit by gambling on a bombing, and then carrying it out. As Surowiecki points out, though, any action on the market conveys information to everyone else (including the CIA). This is why the market would work; market profits would constitute a powerful incentive to share all accurate information, and the information is then available to the United States. People with no sympathy for us and no inclination to help our cause would nonetheless help us merely by pursuing their own self-interest. Terrorists who attempted to profit off their attacks would essentially be begging the US to anticipate and foil their plans. In reality, terrorists would probably go to great lengths to assure that no one with knowledge of their plans participated in PAM.
The chief objection I've seen to PAM is that it would be self-defeating once the CIA started to take action based on the information. Daniel Gross concentrated on this in his piece in Slate. For instance, if the prices showed that a lot of people were betting on an assassination, and the CIA prevented the assassination, the market would obviously no longer be rewarding accurate information.
This would only stop the markets from functioning properly if every piece of accurate information were acted upon perfectly by the CIA. It's easy to see that no one will bet on a bombing if the very fact of betting precludes the bombing. However, imagine that the CIA is only 90% effective when it gets the information. Now, most terrorist activities are hugely unlikely ex ante. I would expect that short-term futures on assassination would be quite cheap, because assassinations are rare. Thus, you could get very good “odds” for most drastic terrorist activities. Now, you should make a bet if its expected value is positive. In a market like PAM, this means that you multiply the probability that an event will happen (p) by the amount you stand to make if you win. Then you multiply (1 – p) by the cost of the bet, and if the expected gains outweigh the expected losses you take the bet.
Thus, if the CIA is only 90% effective, p will be quite low. Nevertheless, because of the great odds that you can get on unlikely events, this high probability can be overwhelmed by the high ratio of returns to initial money down. Of course, eventually the odds will be driven to the point that the bet is no longer profitable. This will often happen too late, from the terrorists' point of view, because the CIA will be tipped off by the very fact that the odds are shifting in favor of whatever the event is. Note that it's not so much the collected wisdom of thousands of minds that will provide the real information bonanza; it's the ability of people within terrorist organizations to exploit their "insider information" that brings the really sensitive information to market.
Of course, the CIA wouldn't be the only group watching the market. One disadvantage of a market like PAM is that it makes all the information public. Terrorists would know everything the CIA would know about the market, and thus they would be able to see how the CIA responds to different knowledge. The terrorists, if they were well-organized, could then manipulate the markets to paint the picture they wanted the CIA to see.
One final objection might prove to be the best. Fighting terrorism isn't just a matter of killing bad people and stopping attacks. It also depends on the political will and social cohesion of the United States. A program like PAM, while probably effective, might unleash cynicism and disillusion within America, ultimately sapping morale and undermining the war on terrorism. Regardless of the actual morality of PAM, its usefulness could be destroyed by its perceived immorality. It is probably this consideration that has scuttled a program that showed unexpected cleverness from an administration not noteworthy for its economic acumen.
I was going to write an angry, incredulous post about just how unfair most criticism of the Policy Analysis Market (PAM) has been, but it looks as though James Surowiecki has already done a pretty good job over at Slate. Slate is perhaps redeeming itself; yesterday it ran a frustratingly inconclusive piece by Daniel Gross, many of whose questions are answered by Surowiecki. The idea behind PAM is fairly simple; it's a market in which you invest in “futures” that pay off depending on certain events in the Middle East (assassinations and political intrigue, as well as mundane things like economic growth). The government would monitor the market and use the information to prevent certain things (presumably the assassinations) and ameliorate others (economic slowdown). In fact, the program sparked a political firestorm, and has been canceled. I expect Democrats to use the program as a rhetorical punch line whenever they criticize Bush's foreign policy. It's too bad, because PAM might well be among the better ideas Bush's team has come up with.
There are some good reasons to think PAM is a bad idea. One is that it depends on people profiting from tragedy; exactly how big a problem this presents has been discussed over on Muxalicious. Additionally, like any market PAM would be subject to manipulation. Since the market would automatically feed back into US policy, it could become a self-fulfilling prophecy, and it might even destabilize political regimes by amplifying any bad news. I'll examine some more shortcomings, but first I should explain why it's such a beautiful idea in the first place.
One principle objection raised against PAM is that terrorists would benefit by gambling on a bombing, and then carrying it out. As Surowiecki points out, though, any action on the market conveys information to everyone else (including the CIA). This is why the market would work; market profits would constitute a powerful incentive to share all accurate information, and the information is then available to the United States. People with no sympathy for us and no inclination to help our cause would nonetheless help us merely by pursuing their own self-interest. Terrorists who attempted to profit off their attacks would essentially be begging the US to anticipate and foil their plans. In reality, terrorists would probably go to great lengths to assure that no one with knowledge of their plans participated in PAM.
The chief objection I've seen to PAM is that it would be self-defeating once the CIA started to take action based on the information. Daniel Gross concentrated on this in his piece in Slate. For instance, if the prices showed that a lot of people were betting on an assassination, and the CIA prevented the assassination, the market would obviously no longer be rewarding accurate information.
This would only stop the markets from functioning properly if every piece of accurate information were acted upon perfectly by the CIA. It's easy to see that no one will bet on a bombing if the very fact of betting precludes the bombing. However, imagine that the CIA is only 90% effective when it gets the information. Now, most terrorist activities are hugely unlikely ex ante. I would expect that short-term futures on assassination would be quite cheap, because assassinations are rare. Thus, you could get very good “odds” for most drastic terrorist activities. Now, you should make a bet if its expected value is positive. In a market like PAM, this means that you multiply the probability that an event will happen (p) by the amount you stand to make if you win. Then you multiply (1 – p) by the cost of the bet, and if the expected gains outweigh the expected losses you take the bet.
Thus, if the CIA is only 90% effective, p will be quite low. Nevertheless, because of the great odds that you can get on unlikely events, this high probability can be overwhelmed by the high ratio of returns to initial money down. Of course, eventually the odds will be driven to the point that the bet is no longer profitable. This will often happen too late, from the terrorists' point of view, because the CIA will be tipped off by the very fact that the odds are shifting in favor of whatever the event is. Note that it's not so much the collected wisdom of thousands of minds that will provide the real information bonanza; it's the ability of people within terrorist organizations to exploit their "insider information" that brings the really sensitive information to market.
Of course, the CIA wouldn't be the only group watching the market. One disadvantage of a market like PAM is that it makes all the information public. Terrorists would know everything the CIA would know about the market, and thus they would be able to see how the CIA responds to different knowledge. The terrorists, if they were well-organized, could then manipulate the markets to paint the picture they wanted the CIA to see.
One final objection might prove to be the best. Fighting terrorism isn't just a matter of killing bad people and stopping attacks. It also depends on the political will and social cohesion of the United States. A program like PAM, while probably effective, might unleash cynicism and disillusion within America, ultimately sapping morale and undermining the war on terrorism. Regardless of the actual morality of PAM, its usefulness could be destroyed by its perceived immorality. It is probably this consideration that has scuttled a program that showed unexpected cleverness from an administration not noteworthy for its economic acumen.
Comments:
Post a Comment
