Predicting a Pandemic

A research brief from Iowa Electronic Health Markets features a case study of the successful H1N1 predication market and explores the potential health prediction markets have to protect public health.

One year ago, traders logged onto the Internet to place trades not on oil futures, but on the spread of a frightening virus. The traders, in this case experts on infectious diseases, very quickly sized up the severity of the problem. And they predicted the virus would spread widely but would not turn into a devastating killer.

This is not the plot of a medical thriller, but the true story of an RWJF-funded prediction market launched by the University of Iowa in 2009 to quickly assess the emerging threat posed by a new strain of influenza now known to all as H1N1.

Just three days after initial media reports about the virus, a team at the University of Iowa recruited traders from an existing pool of microbiologists, epidemiologists, public health officials and others who had been active in previous seasonal and avian flu markets. These experts in infectious disease were given $100 each in “funny money” (no real value) and asked to predict the number of future H1N1 cases and the mortality rate for the new strain.

By Thursday afternoon (long before the CDC had posted publicly-available surveillance data), the market had established a prediction: Traders wagered that the disease would be seen in every state by the end of May 2009 and that the number of infections would be high, but that the mortality rate would stay low.

As it turned out, they were right.