Biology Reference
In-Depth Information
States? (2) What geographic areas would be infected in the United States and
worldwide? (3) Would the morbidity and mortality in the United States be
similar to the early reports from Mexico? and (4) How long into the summer
would the 2009 novel influenza A (H1N1) outbreak last?
As of late April, we were unsure of what type of novel influenza A
(H1N1) surveillance would be conducted, for how long it would continue,
and at what granularity the data would be reported. Thus, we designed
contracts that we thought would be easy to verify over subsequent months.
The full list of contracts appears in Table 8.1. All contracts were designed
to be worth 1.00 H1N1 dollar if the specified event occurred and 0 H1N1
dollars if it did not.
On April 27, 2009, we opened five contracts, and we sent a single e-mail
to current and previous participants in our other health prediction markets
inviting them to join (see Table 8.1). We specified that all U.S.-based contracts
would be resolved using data from the Centers for Disease Control and
Prevention, and World Health Organization (WHO) data would be used for
international-based contracts.
Each participant who signed up received 100 H1N1 dollars with which to
trade in any of the novel influenza A (H1N1) related contracts. In addition to
serving as the medium of exchange for the market, this valueless currency
also provided a measure of the trading and prediction success of individ-
ual traders. All trading occurred on our Web site, and the market was open
24 hours each day. Figures 8.1a,b show the beginner and advanced trading
interface. We sent only one reminder to trade on May 11, 2009.
Signing up for accounts were 160 individuals. However, only 64 partici-
pated. Traders from a variety of backgrounds participated, including phy-
sicians, epidemiologists, nurses, pharmacists, and health administrators.
From Tuesday, April 29, to Wednesday, July 1, 2009, a total of 6,753 shares
were traded. The number of orders per trader ranged from 1 to 400, and on
July 2, trading balances range from H1N1$1.66 to H1N1$137.10.
As traders bought and sold contracts over the Internet, in accordance with
their individual expectations regarding novel influenza A (H1N1) activity, a
price was generated for each transaction. Because all contracts were to have
liquidation values of either H1N1$0.00 or H1N1$1.00, these prices generated
by the market can be interpreted as the consensus probability that the event
specified by the contract would occur. For example, if a contract trades for
H1N1$0.90, this translates into a 90% probability of the event occurring.
Figures 8.2-8.6 show how consensus emerged very quickly among trad-
ers for each of our contracts. For example, almost as soon as the market
opened, traders predicted that the mortality rate in the United States would
be less than 1% (Figure 8.2) and that there would be more than 1,100 con-
firmed cases within a month (Figure 8.3). The market also predicted within
a day of opening that more than 40 states and 50 countries would have
confirmed cases by the end of May (Figures 8.4 and 8.5a,b). All of these
events occurred. Traders also predicted, but with a bit less certainty, that
