Iowa Electronic Markets

The Iowa Electronic Markets is an election-forecasting project that turns participants’ judgments into tradable contracts and market prices. Its significance lies in the book’s account of a striking result: a market with only a few hundred traders, and a membership unlike the electorate it predicts, often forecast election outcomes more accurately than polls asking voters directly.

Origins and Design

Created in 1988 and run by the College of Business at the University of Iowa, the Iowa Electronic Markets was established to test whether a betting-style market could provide a competitive alternative to polling. Polls already offered a statistically rigorous way to estimate how people would vote, but a market could draw on many kinds of information, not only survey responses.

The project is open to anyone who wants to participate. Traders buy and sell futures contracts based on their expectations about an upcoming election. In one common format, a contract pays one dollar if a named candidate wins and nothing if that candidate loses. Its price therefore serves as a probability estimate: a contract selling for 50 cents implies a market assessment of roughly even odds, while an 80-cent price implies an estimated 80 percent chance of victory.

A second common format forecasts a candidate’s share of the popular vote. The payoff reflects the eventual percentage: a contract on George W. Bush in 2000, for example, would have paid 48 cents because he received 48 percent of the vote. In this market, a price near 49 cents before an election would represent a forecast close to a 49 percent vote share.

Forecasting Record

The book reports a study covering 49 elections between 1988 and 2000. On election eve, the Iowa Electronic Markets’ average absolute errors were 1.37 percent for presidential elections, 3.43 percent for other U.S. elections, and 2.12 percent for foreign elections. The study also found that the markets generally outperformed major national polls, and did so even months before elections.

The comparison with individual polls was particularly telling. Across presidential elections from 1988 to 2000, 596 polls were released. On the day each poll appeared, the market price was more accurate three-quarters of the time. Poll results tended to swing sharply, while market forecasts changed less dramatically and generally shifted in response to new information. Their relative steadiness was presented as one reason they could be more dependable forecasts.

These results do not mean the market was infallible. The evidence is comparative: its recorded errors were often smaller than those of polls, and its prices generally incorporated information effectively. The book’s case for the project rests on that measured performance, not on a claim that prices reveal certainty about an election’s outcome.

A Small and Unrepresentative Crowd

The market’s predictive record is notable because it did not mirror the population whose votes it sought to forecast. It had never had more than about 800 traders; most were men, and a disproportionate, though shrinking, share came from Iowa. The participants were therefore not simply predicting their own voting behavior. They were forecasting what voters across the country would do, and in the comparisons described, their collective estimates beat answers obtained by asking voters themselves.

The project exemplifies the book’s broader account of how a crowd can make a sound judgment. Diversity supplies different private information and interpretations; independence helps prevent participants from merely copying one another; decentralization lets individuals draw on their own knowledge; and aggregation turns separate judgments into a collective estimate. In a market, prices provide the mechanism that combines those judgments. The point is not that every trader knows the answer, but that the combined estimate can contain more information than any one participant’s view.

Influence and Wider Significance

The Iowa Electronic Markets helped inspire other prediction markets, including the Hollywood Stock Exchange, which allows participants to wager on box-office returns, opening-weekend performance, and Oscar outcomes. The book places the project among examples suggesting that markets can forecast events beyond sporting contests by drawing on information that ordinary polls may not capture.

Its broader significance is also a qualification: a crowd’s accuracy depends on how it is organized. A large membership or a representative electorate is not the only route to a useful forecast. What matters in the account is whether participants bring varied information, judge independently, and have a functioning way to aggregate their judgments. The Iowa Electronic Markets thus serves as a concrete case for the claim that collective prediction can be powerful without being a simple survey of the people whose behavior is being predicted.