The Dumb Agent Theory, namesake of this website, has been put into practice many times. From such diverse purpose as tracking weather patterns and predicting who will win presidential elections, to determining future market trends by corporations such as Yahoo and Google. One market in particular, however, deserves greater attention: the Influenza Prediction Market by the University of Iowa.

This is not about the bird flu, but Influenza which, although not receiving as much negative press as many other notorious illnesses, is still a dangerous and deadly disease. Every year, in the United States alone, over 200,000 people are hospitalized because of the flu, and about 36,000 people die. Obviously, these people would not have been hospitalized, nor would they have died, had they been given access to a vaccine in advance.

This is where the crux of the problem with influenza lies. Unfortunately, due to a shortage of vaccines, irregular (albeit seasonal) outbreaks and no prior knowledge as to where outbreaks will occur, administering vaccinations to people in need can be a very complicated process. The Center for Disease Control publishes information about where and when outbreaks have occurred, but this is 2-3 weeks after their occurrence, so it does not aid very much in this endeavor. If, however, doctors were to know where outbreaks are most likely to occur one to two weeks in advance, they would be able to mobilize staff and resources to cure potential victims. With this in mind, the University of Iowa’s Iowa Electronics Market (IEM) decided to create the Influenza Prediction Market.

Influenza Prediction Market

This market was composed of 60 doctors and nurses, as well as others in the medical industry in Iowa, who were each given $100 with which to trade. These were called Flu Dollars, but were converted to real dollars once the experiment was over. The participants then became investors in this Influenza Prediction Market, and were able to use it like a Futures Market. Their $100 purse would rise or fall depending on their accuracy. The “shares”, or contracts, were actually the weekly color coding system used by the CDC, which is: RED=Widespread, BLUE=Regional, PURPLE=Local, GREEN=Sporadic, YELLOW=No Activity.

The goal of this market, of course, was not to make extra money for the medical workers, but to see if this market could be utilized to discover where potential influenza outbreaks would occur. The results were very positive: the traders were able to predict the level of outbreaks with 80% accuracy with a lead-time of two to three weeks. With preparation four weeks in advance, they were able to get within one color on either side with 90% accuracy. And this was despite the fact that the predictions were for the entire state of Iowa, while almost all of the traders were located near the university, on the eastern side of the state.

This year the market is being expanded to cover the entire country, with doctors and nurses nationwide allowed to trade. The results will be followed closely and, if the trend continues, it should promise to be very useful. Although it is an unproven theory, if it can save lives it will be worth using.

Problems

The main issue with a market of this sort is the self-fulfilling obsoleteness it might engender. For example, if the market can successfully predict all places that influenza will be occurring, vaccines will be sent to these areas and, if done quickly and successfully enough, there will be no outbreak of influenza. Although this may sound good, it would also mean that traders will have been too successful, and will not be as motivated to trade anymore knowing that the vaccines will impede their payouts (as odd as that may sound when spelled out).

The solution here can take place as the problem occurs. For example, if the vaccines can be proven to have prevented an influenza outbreak, the market was successful and a payout can occur despite the fact that sickness was prevented. This would just require small changes in the conditions for payouts, such as “Payout shall occur, regardless of outbreak, if the outbreak was prevented according to predictions of this market,” as an example.

Other Uses

The parallels with Bird Flu are also not lost. With outbreaks now reoccurring in India, and people worrying about where and when the next outbreak will be, any sort of help in identifying where to cull, vaccinate, cure, or evacuate will be useful. With information sometimes very hard to come by, a market like this may prove invaluable for illnesses of this sort.

Did we help you save money? If so, how about donating some of it to DumbAgent?