The History and Return of Prediction Markets

概览

This episode uses a Throughline excerpt to trace prediction markets from modern academic experiments back to older election betting traditions. The core question is why markets that let people bet on outcomes can sometimes predict elections better than polls.

The story begins with the Iowa Electronic Markets in 1988, created after economists noticed polls had missed Jesse Jackson’s Michigan caucus win. It then reaches backward to earlier election betting in Europe and the United States, including public markets outside the New York Stock Exchange.

A key conclusion is that prediction markets are not new: they have repeatedly appeared when people wanted to forecast political outcomes, make money, hedge risk, or demonstrate expertise. The episode also explains why earlier election markets faded: newspapers grew more comfortable with scientific polls, and horse racing offered bettors many more events.

分段落总结

[00:00] Opening and Listener Callout

[事实] The episode opens with a Masters of Scale promo and a Planet Money request for listeners to share how the economy is affecting their lives. [事实] The hosts ask about pandemic-era decisions, stock market gains, rising costs, and personal strategies for getting by or thriving. [推测] This opening frames the show within broader economic uncertainty, though it is separate from the main prediction-market story.

[01:27] Why Prediction Markets Matter

[事实] The host says a story changed how they think about prediction markets. [事实] The episode introduces the idea that prediction markets have roots going back hundreds of years, while modern versions were developed by economists testing market theories. [事实] The host previews a theory that horse racing may have displaced older election betting because races happen far more often than elections.

[03:17] The Iowa Experiment Begins

[事实] In 1988, Robert Forsythe and two economist colleagues discussed election forecasting over lunch in Iowa City. [事实] They were reacting to polls missing Jesse Jackson’s large Michigan caucus win over Michael Dukakis. [事实] The economists decided to create a market where participants could buy and sell shares tied to political candidates.

[05:00] Markets Versus Polls

[事实] The Iowa Electronic Markets let students and faculty trade candidate shares, similar to stocks or commodities. [事实] The idea was that prices might reveal what people actually believed about election outcomes. [事实] The market predicted the 1988 popular vote within two-tenths of one percent and outperformed major polls.

[05:47] Regulatory Limits

[事实] The Iowa team asked the Commodity Futures Trading Commission for permission to operate nationwide. [事实] The CFTC issued a no-action letter allowing the project under limits: small stakes, no deposits over $500, no paid advertising, and no profit motive. [事实] The markets were restricted to presidential elections, with sports betting excluded.

[06:53] Growing Public Attention

[事实] Between 1988 and 2004, the Iowa Electronic Markets became one of the more reliable predictors of U.S. presidential elections. [事实] During that period, its predictions beat traditional polls 74% of the time. [事实] National outlets including NPR, The Wall Street Journal, and the Financial Times covered the project. [推测] The media treated the Iowa project as novel because the older history of election betting had largely disappeared from public memory.

[09:29] The Racetrack Connection

[事实] The episode shifts to economist Coleman Strumpf, who has taught prediction markets for nearly 20 years. [事实] Strumpf says the racetrack is a useful place to understand prediction markets. [事实] He recalls watching bettors as a child and noticing how people used longshot wins to show they were smart. [推测] The episode suggests prediction markets are partly about information and partly about psychology, status, risk, and reward.

[11:21] Betting, Gambling, and Social Science

[事实] The episode notes ongoing debate over whether prediction markets are simply betting under a more sophisticated name. [事实] It does not try to settle the legal definition of gambling. [事实] Strumpf’s point is that bets in contexts like stock markets, insurance, and racetracks can reveal useful information about human behavior.

[12:00] Rediscovering Older Election Markets

[事实] During research on the Iowa Electronic Market, Strumpf’s colleague Paul Rhode told him it was not the first political prediction market. [事实] Rhode had found newspaper stories from October and November 1924 about election markets. [事实] The two researchers then found election betting traditions going back to the 16th century, including papal elections, city-state elections, and U.S. markets reaching back to George Washington.

[13:45] Election Betting on the Curb Exchange

[事实] Around the turn of the 20th century, U.S. election markets gained momentum. [事实] A major venue was the curb exchange outside the New York Stock Exchange. [事实] Reporters could observe prices and identify prominent traders, including political machine figures, bankers, Wall Street people, and hotel owners.

[15:01] Political Signaling and Hedging

[事实] Political leaders were expected to publicly bet in support of their candidates. [事实] The episode describes Democratic machine figures publicly backing William Jennings Bryan while allegedly placing offsetting bets privately. [事实] The markets operated in a legal gray area: friendly election bets were not necessarily illegal, but bettors were not supposed to vote. [推测] Public betting functioned both as forecasting and as political theater.

[16:20] Accuracy of Early Markets

[事实] Strumpf says he initially assumed historical markets would not contain much useful information. [事实] He later concluded he was wrong and that these markets were often very good at indicating winners and the scale of victories. [推测] This supports the episode’s broader argument that prediction markets can aggregate dispersed knowledge even before modern polling.

[18:33] Why Election Markets Faded

[事实] The episode identifies the late 19th century through World War II era as a golden age of prediction markets. [事实] Strumpf says newspapers covered election markets but were never fully comfortable doing so. [事实] In the 1930s, scientific polling by Gallup and others gave the press a more acceptable forecasting tool. [事实] Thoroughbred racing also became more attractive to bettors because it offered many races instead of only a few election events.

[20:09] Four Decades of Silence

[事实] Presidential election markets appear to have gone dark sometime in the 1940s. [事实] Strumpf says he has not found people or books from that period that discussed those markets. [事实] The Iowa Electronic Markets revived public awareness of prediction markets in 1988.

[20:56] Scaling Pressures and Missed Opportunities

[事实] After going national, the Iowa Electronic Markets received calls from people who wanted to invest far more than the $500 limit allowed. [事实] Some callers were not speculators but people trying to hedge political risks that could affect their businesses. [事实] The Iowa team also received suggestions to move offshore, such as to the Cayman Islands, where they could operate without the same restrictions. [事实] Forsythe says the academics turned those opportunities down because they were focused on teaching and research.

[22:11] Modern Platforms and Legacy

[事实] Forsythe says Kalshi and Polymarket are running essentially the same kind of prediction markets on a much bigger scale. [事实] He says they use trading and contract rules similar to those used by the Iowa Electronic Markets in 1988. [事实] The Iowa Electronic Markets demonstrated to modern economists that prediction markets could work. [推测] The episode frames today’s prediction-market platforms as commercial successors to an academic experiment rather than as entirely new inventions.

[22:51] What Comes Next

[事实] Coleman Strumpf says the Iowa Electronic Markets raised the question of whether prediction markets could forecast things beyond elections. [事实] The host says the full Throughline episode includes more on terrorism futures and military applications. [推测] The excerpt ends by suggesting that the biggest unresolved issue is not whether prediction markets can work, but where they should be allowed and trusted.

播客点评/总结

This episode’s value is historical clarity. It shows that prediction markets did not suddenly emerge from modern tech platforms; they sit inside a much older tradition of election betting, financial hedging, media coverage, and public signaling.

A strong part of the episode is the contrast between polls and markets. Rather than claiming markets are always superior, the episode shows specific cases where markets performed well, why people found them useful, and why legal and cultural discomfort limited them.

Its limitation is that this Planet Money version is only an excerpt from Throughline. The episode mentions broader subjects like terrorism futures and military applications, but does not fully develop them in the transcript.

[推测] This is especially suitable for listeners interested in economics, political forecasting, market design, and the relationship between gambling, information, and regulation.