Can prediction markets improve economic news?
The Congressional Budget Office (CBO) updates its 10-year “Budget and Economic Outlook” report, one of the most widely cited sources of economic health, about twice a year; however, the economy it describes changes every day. That mismatch is the premise behind the Citizen Debt Forecast (CDF)—a decentralized economic model based on prediction-market data.
The model takes the CBO’s fiscal baseline, holds its projected primary deficits fixed, and replaces the agency’s assumptions about growth, inflation, and interest rates with values implied by market prices on its markets. Prediction markets allow users to place wagers on economic forecasts on a day-to-day basis. Those bets then either succeed or fail based on the actualized performance of the economy in the wager, thus creating a new way for economic predictions to be made with real skin in the game.
The result is a 10-year debt-to-gross domestic product (GDP) projection that reprices continuously. At launch, it put federal debt held by the public at 118 percent of GDP in 2036—just below the CBO’s estimate of 120 percent. Within weeks, the live reading surpassed the official figure.
The specific number matters less than the question it raises. A commercial exchange—in this case, a prediction market—has an obvious interest in promoting the usefulness of its own data, and a company-built model deserves the same scrutiny as any other model in the industry. But can this general approach, in which prices from real-money markets supplement official statistics and surveys, improve the information the public receives about the economy? With significant evidence now behind the CDF, researchers are starting to take the method more seriously.
What the Federal Reserve Research Shows
The strongest evidence comes from a source with no commercial stake in the outcome: a working paper published in February through the Federal Reserve Board’s Finance and Economics Discussion Series. The paper, which evaluated the exchange’s macroeconomic contracts against the benchmarks upon which economists and journalists already rely, carries the standard disclaimer that it “represents the views of the authors” and “does not indicate concurrence” by the Board.
Its findings are nonetheless notable. Several months ahead of each Federal Open Market Committee (FOMC) meeting, market-implied forecasts for the federal funds rate were about as accurate as the Federal Reserve Bank of New York’s survey of professional market participants. In fact, the most likely outcome in the market’s distribution has matched the actual decision on the day before every meeting since 2022—a record that neither surveys nor federal funds futures have achieved.
The results for other indicators were similar. In a modest but statistically significant improvement to headline inflation predictions, the market’s average error on release day was roughly 0.06 percentage points compared with 0.08 for the Bloomberg consensus of professional forecasters. The two performed about equally on both core inflation and unemployment, and in no case were the market forecasts significantly worse.
But accuracy is only part of the contribution. Surveys arrive on fixed schedules and usually report a single number. Markets update within minutes of a data release or a policy speech, and because contracts trade across a range of outcomes, they produce a full probability distribution. The paper also notes that no other market-based distribution exists for GDP growth, core inflation, unemployment, or payrolls. For example, in mid-2025, the consensus forecast called for 1.4 percent growth for the year while the market assigned a probability near 40 percent to growth below 1 percent. The two figures were consistent with each other, yet the second conveyed how much uncertainty surrounded the first.
Why Economic Coverage Needs Better Inputs
That kind of context is frequently missing from economic news. The state of the economy is a staple of daily coverage, and there are signs that coverage has drifted from the underlying data. Analysis from the Brookings Institution found that, since 2018, the tone of economic news has been consistently more negative than fundamentals like growth, inflation, and unemployment would predict. The authors caution that the direction of causation between news tone and public sentiment remains unclear. Whatever the cause, a reporting environment built on periodic point estimates, selectively quoted analysts, and narrative framing leaves considerable room for distortion.
Market prices offer a partial correction because they are difficult to spin. The theoretical case is an old one: Prices aggregate dispersed knowledge that no single forecaster or agency possesses, and participants bear a financial cost if they are wrong. A commentator who predicts recession every quarter pays no penalty for being early by a decade, while a trader who does the same loses money. Similar logic, even with reputational rather than financial outcomes, underlies other decentralized information tools. Previous R Street research on X’s Community Notes program has shown that distributed systems with built-in accountability mechanisms can produce reliable information at scale without a central arbiter.
The Limits Are Real
None of this makes any particular product authoritative. Not every market has sufficient volume and liquidity to create a truly decentralized prediction, and the debt forecast illustrates the limits. The Federal Reserve paper validated short-horizon contracts, such as the next inflation release or the next few FOMC meetings, but it did not test 10-year projections, where trading is thinner and feedback on accuracy arrives slowly. The CDF also draws its longer-dated interest rate path from conventional swap markets and depends on several modeling assumptions that its designers acknowledge. They describe the output as a conditional scenario with sensitivities as its most useful feature. Under the current calibration, an additional percentage point of sustained real growth lowers the 2036 debt ratio by roughly 10 points, while a persistent increase of the same size in refinancing rates raises it by about six. For its part, the CBO has a forecasting record comparable to that of private forecasters, and its current-law baseline serves a legislative purpose that no market can replace.
A Complement to Existing Sources
The central finding of the Federal Reserve research is modest, but it matters. Prediction markets on economic outcomes performed as well as or better than the forecasts that currently anchor economic reporting, doing so continuously and with a full accounting of uncertainty. That establishes, at least in principle, that market-implied data can sit alongside official projections, surveys, and expert analysis as a regular input to news coverage of the economy. Emerging research suggests that prediction markets can even forecast weather, sports, and elections with greater accuracy than even experts can at times. Journalists and policymakers who treat prediction markets as one signal among several and who understand its limitations can provide a more accurate picture of economic conditions to the public than any single source.