Economist Jeremy Horpedahl estimates U.S. third-quarter 2026 growth near 3.2% in a post for Economist Writing Every Day, ahead of the Bureau of Economic Analysis’ (BEA) initial figure due Thursday.
Kalshi’s betting markets, the Atlanta Fed’s GDPNow model and a Wall Street Journal survey were all near that level, Horpedahl wrote. The St. Louis Fed Nowcast, newly included in his comparison, was also close to 3.2%. He described the New York Fed model as generally a weaker predictor.
Horpedahl’s calculations put growth at 3.24% in the advanced estimate by averaging Atlanta Fed and Kalshi projections. For the eventual final figure, he combined Kalshi’s projection with the Wall Street Journal survey to reach 3.15%. These are his combinations of forecasts, not official BEA figures.
The first GDP reading is provisional. Horpedahl said it can shift as more data arrives and benchmark updates are applied, so he also compared the models with the latest available GDP growth figures dating back to 2023.
In that 14-quarter comparison, Kalshi had the lowest average error, narrowly ahead of the Wall Street Journal survey. Kalshi was the strongest predictor in six quarters. The survey was strongest in three, including two in which it matched Kalshi; it led on its own only in the oldest quarter in the sample.
The St. Louis Fed Nowcast had the highest average error among the five predictors, but it was the strongest in five quarters, each time on its own. It also performed well in the latest three quarters, including an exact match for first-quarter 2026. Horpedahl added it to his tracking list to see whether that recent performance continues; he said this quarter’s figures may not distinguish it from the other models because their projections are similar.
The BEA is due to release its initial estimate of third-quarter growth Thursday, Oct. 29. Later releases will revise the reading as more information becomes available.
