Recession Risk — 2026 and 2027

The probability of an NBER-declared US recession by year-end 2026 was the defining macro contract of the spring. It is now close to settled, and the interesting book has moved out a year.

Kalshi's 2026 binary trades at 1.5%, down from 3.5% at the last review, on $9,605 of 24-hour volume against $880,949 of open interest, a large standing position on a question with little time left. With under two months left in the year there is very little runway for an NBER-dated 2026 recession to begin, be observed and be declared.

The 2027 book prices 20.5%, down from 23.5% but more than thirteen times the 2026 number, on a contract that settles in January 2028. It turned $9,422 in a day on $425,262 of open interest. The market is not saying the expansion is safe; it is saying the risk sits past the horizon of the contract everyone was watching.

The policy backdrop points toward tightening. The Fed raised rates 25bp to 3.75–4.00% on September 16 and held there on October 28. Kalshi's December 9 board prices a 25bp hike at 89.5%, a larger hike at 8.5% and a hold at 4.5%.

The step change, priced directly. Kalshi's NBER-specific binaries at both horizons: 1.5% for a recession starting in 2026, 20.5% for one starting in 2027. Same resolution source, same methodology, twelve months apart. Both fell since the last review, from 3.5% and 23.5%.

Polymarket's broader US-recession binary, read against Kalshi's NBER-specific 2026 contract above. The gap is 1.5% against 2.15%, under a point, from one point at the last review: Kalshi fell two points and Polymarket fell from 4.5%. Polymarket's looser resolution language should be worth more when the definition matters, but the spread it buys is small; its leg turned $6,929 in a day. Beside it, Kalshi's IMF global-recession contract sits at 5%, up from 4%, on $138.

The quarter-of-start Categorical is the timing view in principle; in practice its leading legs name quarters that have already passed without a declaration. The top leg is now Q1 2026 at 2.95%, ahead of Q4 2025 at 2.55%, and the book did not trade in twenty-four hours. Beside it, Kalshi's multi-country book on which economies enter recession before 2027, the cleanest read on whether US risk is idiosyncratic or global. China now leads at 3.25%, ahead of the United Kingdom at 2.55%, down from 5.7%, and India at 1.95%, on $310, all of it on the UK leg.

Recession odds do not move in isolation. The initial-claims and unemployment books reprice weekly and are the highest-frequency inputs into this question; the Fed's next decision, on December 9, is the policy response function. Kalshi prices a 25bp hike at 89.5%, up from 79.5%, a larger hike at 8.5% and a hold at 4.5%, down from 20.5%, on $1,115,245 in a day, $720,445 of it on the hold leg. Both are covered in depth in the jobs and FOMC workspaces; here they serve as the sanity check on a 1.5% print: a market pricing another hike is not pricing a downturn this year.

• 1.5% now, 20.5% next year. The most informative number pair on this page. Both fell since the last review, from 3.5% and 23.5%; the market still puts the risk in 2027, not 2026.
• The cross-exchange gap is under a point. 1.5% on Kalshi against 2.15% on Polymarket, from one point. That spread is the price of resolution language, and it moves when the macro path does.
• The 2026 book traded as much as 2027. It turned $9,605 in a day against $880,949 of open interest; 2027 turned $9,422.
• The Fed decides next on December 9. A 25bp hike is 89.5% and a hold 4.5%, yet $720,445 of the board's $1,115,245 day traded on the hold leg, the one that would move first if the data weakened.
• The quarter Categorical is dead, not bearish. Its leading legs name past quarters, and it did not trade in a day. Read it as an unwound book, not a forecast.
• The declaration lag is the mechanical constraint. NBER dates recessions retrospectively, often a year or more after the trough. A 2026 recession would essentially have to have already started for the 2026 contract to pay, which is most of why 1.5% is not lower.