Thursday, October 8, 2020

A new study from the University of Iowa’s prediction market finds that the COVID-19 pandemic and civil rights protests have had the most significant impact on President Donald Trump’s re-election chances, while other factors have had much less effect.

The study, to be published in the journal PS: Political Science and Politics on Oct. 15, finds traders on the Iowa Electronic Markets (IEM) believe that such events as impeachment, the Mueller Report, and the passage of the spring’s pandemic relief bill are having little impact on the president’s re-election chances. 

“Research using data from the IEM suggests to us that neither individual political events nor individual events leading to civil rights protests have significantly affected the 2020 election,” says Tom Rietz, professor of finance in the Tippie College of Business and a member of the IEM steering committee. “While Trump’s chances fell during the stock market collapse and recession associated with the initial wave of COVID-19, they did not rise with the stock market recovery and decreasing new jobless claims. Instead, his chances continued to fall as COVID-19 cases rebounded and protests continued.” 

Now in its ninth presidential election cycle, the IEM, based in the Tippie College of Business, operates like any futures exchange, only instead of putting their money in gold or wheat or pork bellies, investors buy Trump or Biden. By analyzing price movements, the market can reveal such factors as what is helping or hurting a candidate and demonstrate the IEM’s value as a research tool for prediction markets in general.

The IEM has two kinds of contracts for the presidential election. For one type, investors receive $1 if the associated candidate wins the popular vote. The price of the contract at any given time on this Winner Take All market reveals the probability of winning. In the other, prices reflect the percentage of the two-party popular vote the candidate receives, as investors receive $1 times the share of the candidate’s vote on Election Day. Prices on this Vote Share market reflect the trader expectations at any given moment.

Traders can invest up to $500 in real money during any one election cycle, which provides the skin in the game that sets the IEM apart from other prognostication tools. (It should be noted the IEM does not make a profit.)

“The key difference is the question we ask our investors,” says Tom Gruca, professor of marketing in the Tippie College of Business and IEM director. “We don’t ask ‘Who do you plan to vote for?’ or ‘Who do you want to win?’ We ask ‘Who do you think will win?’ and that makes a big difference.”

“It’s easy to tell a political pollster that you planned to vote for Joe Biden,” says Joyce Berg, professor of accounting and a member of the IEM steering committee who has been using the IEM for research since shortly after it was founded. “But are you going to put money on him if you think Donald Trump is actually going to win the popular vote?”

This year’s five election markets—which also include markets that predict which party will control Congress—opened in February 2019, and about 800 traders have executed more than 38,500 trades since June 1. The traders come from 18 countries and have invested more than $70,000 in the markets.

In the new study, Gruca and Rietz look at what’s moved the presidential markets through August 2020 and find that investors on the Winner Take All market mostly shrugged off many of the recent events of Trump’s presidency. Prices moved little as a result of the Mueller Report release, Trump’s impeachment and Senate trial, passage of the pandemic relief bill in March, or five major Federal Reserve actions during the spring. Gruca says those outcomes were likely priced into the market already.

Trump’s probabilities were affected by the stock market collapse and economic shutdown in March, falling from 45 cents to as low as 32 cents, before rebounding to 35 cents by the end of the month.

A summer of market volatility began with the start of civil rights protests began in May.

“Volatility increased dramatically after July 6 with six significant price movements in 29 days as states reopened, protests continued, Federal law enforcement ‘surged’ in U.S. cities, new COVID-19 cases began decreasing, and companies reported Phase I and II clinical trial results, moving into Phase III,” Rietz says. Trump’s largest gains occurred when his price jumped 45 percent after biotech firm Moderna Inc. published successful Phase I results on July 15, and 30 percent when Russia announced October vaccination plans on August. 2.

“Overall, reelection chances appear more affected by protests and the pandemic than economic outcomes or individual political events,” he says.

The accuracy of the IEM is the result of a variety factors: the “wisdom of crowds” theory, which, oversimplified, says that a large group of people knows more than just a few; the “profit motive,” which says that people like to make money; and the “efficient market” theory, which says that the price of an equity incorporates all information that investors have about that equity.

“The market sorts through news, rumor, and fluff, reacting only to the news,” Berg says.

The IEM dates to 1988, when the Rev. Jesse Jackson shocked political observers by winning Michigan’s Democratic Party presidential primary.

“Nobody saw it coming, and the polls didn’t predict it,” says Forrest Nelson, emeritus professor of economics at the UI and one of the IEM’s founders. “We wondered how so many people could have missed this, and if there was some mechanism we could use that would provide an alternative to traditional measures of voter sentiment.”

The question intrigued Nelson and colleagues Robert Forsythe, then a professor of economics, and the late George Neumann, a professor of economics. Would a market that sells political “futures” be a useful tool in determining who is leading an election, they wondered. Could it go so far as to predict the winner?

To test the theory, they created the IEM—then called the Iowa Political Stock Market—in time for the 1988 presidential election. The 100 or so investors from across campus came closer than most polls in predicting the actual popular vote shares.

Two markets have been opened for every presidential election since—a Winner Take All (WTA) market, where traders buy and sell based on which candidate they think will win the popular vote, and a Vote Share market (VS), which is based on each candidate’s vote percentage.

There also have been markets for Congressional control in off-year elections, state Congressional elections, and international elections. A Congressional control market also opened this year.

The IEM has opened non-political markets in the past, too, to test a market’s ability to predict the outcomes of such events as influenza outbreaks, hurricanes, initial public stock offering prices, and movie box office openings.

“It’s a powerful teaching and research tool that provides a window into the black box of how and why markets work well and why people make the decisions they do,” says Berg. “It provides a powerful incentive for students to investigate the underlying phenomena being traded. Movie box office markets, for instance, give marketing students an incentive to think hard about the factors that lead to new-product success.”

The IEM isn’t a foolproof political crystal ball—for example, investors gave a low probability that the Republicans would sweep Congress in 1994, or the Democrats would sweep it back in 2006. But a 2007 study by Berg, Rietz, and Nelson compared the results of the IEM’s Vote Share market to nearly 1,000 public opinion polls in presidential elections from 1988 to 2004 and found the market was more accurate than 74 percent of them.

The market also accurately predicted a Hillary Clinton popular vote victory in 2016, as she won the largest share while losing the electoral vote that gave Trump the presidency.

Media contact: Tom Snee, tom-snee@uiowa.edu, 319-541-8434