The Cloudy Future of Political Futures

S ince 2014, the Commodity Futures Trading Commission (CFTC), a U.S. federal regulatory agency, has allowed Victoria University of Wellington, New Zealand, to operate a prediction market called PredictIt. Prediction markets provide forecasts of winners of future elections, working a bit like the more familiar stock market except participants buy shares of candidates rather than companies.

But earlier this month, the CFTC withdrew its no-action letter that had allowed PredictIt to operate, and it said all positions must be liquidated by February. This decision is a major blow to open and fact-based discourse on matters of public interest.

At a time when many media institutions are rightfully viewed with a skeptical eye, prediction markets are a needed antidote to misinformation and fake news. But the CFTC’s decision eliminates this important counterweight to partisan media coverage and analysis, leaving Americans to once again rely on the same legacy outlets whose loss of public trust necessitates innovative solutions, such as PredictIt, in the first place. These markets reward dispassionate, non-partisan analysis, and so may promote civic discourse at a time of growing polarization.

Even worse is that we are losing this resource at a time we can least afford it, in the midst of a fiercely contested midterm election and quite possibly, if nothing changes, for the duration of the pivotal 2024 presidential election. And the opaque way in which the CFTC has acted may have chilling effects more broadly on financial markets.

Since its inception in 2014, PredictIt has been a remarkable resource for those wanting up-to-the-minute U.S. political information. Even people who did not trade on the site could quickly identify competitive races, who were the frontrunners, and who was gaining or losing support. Relating that information through prices based on people’s willingness to put their own money on the line allowed for concise communication of information that would take ordinary news media much longer to state.

Our preference for PredictIt, and the information produced by prediction markets more generally, over traditional media is based on research showing that PredictIt consistently outperforms polls and other forms of political analysis at forecasting election outcomes. One author’s research on the 2020 election shows that PredictIt prices are more accurate than polling analysis from FiveThirtyEight and other sources. The same conclusion holds for the 2018 midterms. The other author looked at almost a hundred years of historical political prediction markets, and found these ancestors of PredictIt were remarkably prescient before polls even existed.

Above all else, prediction markets make civic engagement educational and fun. Hardcore PredictIt traders have incentive to be more knowledgeable about the ins and outs of elections than most political journalists because if they’re wrong, they lose money (journalists, on the other hand, can get promoted after being wrong). Market participants dig into polling methodologies, read between the lines of political announcements, research antiquated election laws, and mind their own personal biases. Traders are experts on topics big and small, with PredicIt markets providing up-to-the minute forecasts on topics as specific as the Trump emolument case.

While PredictIt and other political futures markets add to the social good, the same cannot be said for the manner in which the CFTC has handled this case. The tersely written withdrawal did not specify the reason for shuttering PredictIt aside from stating that the site had violated one of the nine conditions listed in the original No-Action letter. Later reporting claimed the violations stemmed from straying from the research goals, with some markets being too large and others not focused on politics, and sidelining of academics for research or pedagogy.

These criticisms are incorrect and unevenly applied. Both the authors have used PredictitIt data in their classes, and the company has provided data and trading accounts to students for research projects. And IEM, the one other site which has a CFTC No-Action letter to exclusively operate political prediction markets, has offered dozens of non-political markets on movie box offices, corporate earnings, and economic indicators, and never faced sanctions.

This is not the only time the CFTC has relied on reasoning with little to support it. The regulator denied permission to a previous applicant to run political prediction markets because of the possibility that it could diminish electoral integrity. But more traditional financial markets provide far more opportunity to make money from election tampering. Someone who could alter an election outcome could make millions of dollars trading options (equities and bonds) due to their deep liquidity, while the upside for a political prediction market would be only in the thousands. More to the point, one of the authors found no evidence of this happening in his study of long-running and unregulated historical political prediction markets.

Prediction markets not only provide a treasure trove of free information, they are an alternative to the degrading public discourse. They elevate the level of our public dialogue by promoting informed debate and respectful disagreement. At the same time, the opaque and capricious nature of the CFTC actions lead to confusion and doubt in the agency. The CFTC is a primary regulator of the vast financial-derivative markets and may play a role in shaping emerging markets like cryptocurrency. Reversing course or at least providing more clarity on its reasoning would be a first step in repairing the damage to its reputation as an impartial arbiter of financial market safety.

Harry Crane is professor of statistics at Rutgers University and co-founder of Researchers.One. Koleman Strumpf is the Burchfield Presidential Chair of Political Economy at the Wake Forest University Department of Economics.