Guest Blogger: Phil Polgreen on the Avian Flu Prediction Market
On March 1, 2007, the Iowa Health Prediction Market launched a market for the Avian Flu. This builds on an earlier grant Pioneer made to the Iowa team to run prediction markets on where, when and what strain of domestic seasonal flu would strike in Iowa. The state flu market has since caught on in North Carolina as well, and the grantees published their first peer-reviewed journal article on their findings.
The avian flu market has gotten a great deal of press coverage and we asked Phil Polgreen, M.D., one of its directors, to tell us more about it. He answered:
"In 1988, a couple of economists had lunch at a pizza place called the Airliner in Iowa City. The presidential primaries were in full swing. Thus, the conversation that day surrounded the polls, and how accurate, or sometimes inaccurate they tend to be. Economists have long known that prices contain information, and so one of them suggested, half-jokingly, that they should open a futures market to predict election results. That off-hand comment lead to several drawings on napkins and was the birth of the first prediction market, the Iowa Electronic Markets.
Almost twenty years later, the IEM has a track record of predicting elections substantially more accurately than national polls or surveys, yet at the time of its inception, a political scientist the University of Iowa, when asked about the project, predicted that these markets would never amount to much of anything.
In 2003, I was at a party in downtown Iowa City (not far from the Airliner). I had heard about the IEM, and I started talking to its co-founders, Forrest Nelson and George Neumann after my wife introduced me to them. They asked what I did; I told them that I was an infectious disease physician; and we started talking about SARS. At the time there was a lot of information about SARS, but none of it was in one place, and different people around the world had access to different kinds of information. There was a great deal of uncertainty, and no one knew whether the illness would fade away or emerge as a persistent world-wide health problem. We started talking about opening a SARS prediction market, but SARS disappeared.
At the time, I remember thinking that prediction markets could have all kinds of healthcare applications, especially in the field of infectious diseases where all sorts of information exists, but is often disparate, and held by different groups of people who rarely talk to each other.
Surveillance is routinely done for infectious diseases and other adverse health outcomes, but no one ever asks those gathering or using the surveillance information what disease activity will be like in the future. Traditional forecasting methods have a difficult time capturing subjective information, but this is just what prediction markets do very effectively.
Forrest, George and I decided to open a flu market about a year after we first met, against the advice of almost everyone. We got a lot of weird looks and a lot of rejected grants, but this is exactly the same thing that happened when the IEM began. Our goal is to try to change the way people think about information aggregation. The support of the Pioneer Portfolio of the RWJF is helping us to test our hypothesis that prediction markets can be used to forecast infectious disease activity.
In the few years since we started our flu project, the popularity of prediction markets has grown. Dozens of corporations are now using prediction markets to help make business decisions. For example, Best Buy just hired a project leader whose sole responsibility is to develop prediction markets to forecast trends in consumer electronics. We can envision all sorts of applications in the fields of public health and clinical medicine, but what we would really like to hear are some of yours."
Edited 3/21/07 to add: The market launch on March 1 generated a number of conversations about the concept on other blogs, including Effect Measure.
Also, the Avian Flu Market is actively recruiting traders to take part in the market. Interested participants need to have some degree of expert knowledge related to if/how the disease will spread. To find out more, visit here and see the last paragraph about how to sign up.