U.S. recession analysis · September 28, 2026
Slower Output, Steady Jobs: The Recession Risk Divide
Growth has lost some speed, but the labor-market trigger remains quiet. The next recession test is whether a modest output slowdown becomes a sustained rise in unemployment.
Executive summary
Next 3 months
Recession not the base case
August unemployment was 4.1% and the Sahm indicator was −0.07 percentage point, well below its +0.50 trigger. Watch weekly claims and hiring breadth for a change in direction.
Next 6 months
Slowdown risk exceeds imminent-contraction risk
Real GDP rose in the latest available quarter, though its pace eased. A persistent employment deterioration alongside weaker output would change this assessment.
Next 12 months
A wider uncertainty band
The year-end event-market price cannot be stretched into a 12-month forecast. Energy, credit and household income will determine whether the current expansion survives into late 2027.
As of September 28, the Polymarket U.S. recession by end-2026 contract priced “Yes” near 8.5%. That is a tradable price for a specific resolution and deadline, not the probability of a recession at any point over the next 12 months. The New York Fed's model series stood at 0.76% for its July observation; its horizon and methodology differ from the event contract. Both are low, but neither cancels the need to monitor incoming labor and spending data.
Three views of the cycle
Labor has not crossed the recession threshold
Unemployment rate (%) and Sahm rule (percentage points), monthly observations through August 2026.
Output expands, but at a slower pace
Real GDP, chained 2017 dollars, billions; quarterly observations. Q2 2026: +0.37% quarter on quarter, about +1.5% annualized.
Model risk has stayed subdued
New York Fed recession probability series (%), monthly observations; last distinct observation July 2026. Repeated daily quotes are not new forecasts.
Recession risk table
| Indicator | Latest observation | Reading for investors |
|---|---|---|
| Sahm rule | −0.07 pp · August 2026 | Below the +0.50 pp recession-onset threshold; it is a confirmation rule, not an all-clear forecast. |
| Unemployment | 4.1% · August 2026 | Down from 4.5% in November 2025. A sustained reversal would be more consequential than one noisy release. |
| Real GDP | $24,269.6bn · Q2 2026 | Above Q1's $24,180.4bn; roughly +1.5% annualized sequential growth. Positive output, slower than Q1's roughly +2.1% annualized rise. |
| NY Fed model | 0.76% · July 2026 observation | Low model-implied risk under its own definition and horizon; not directly comparable to event-market odds. |
| BCIG cyclical gauge | 8.3 · May 8, 2026 | Rose from 4.0 in early April; an older weekly reading, so it should not drive a September call on its own. |
| Money-market fund assets | $8.44tn · April 2026 observation | Substantial liquidity, but not proof of impending recession: cash can reflect yields and portfolio allocation. |
| Recession searches | 1,287 U.S. searches · September 27 | Down 57% versus about seven days earlier on this daily series. A Sunday count is seasonally noisy; attention is not economic output. |
| Prediction market | About 8.5% Yes · September 28 | Contract expires at year-end; prices move and embed trading conditions as well as beliefs. |
Observation dates denote the latest distinct readings in the underlying series, not the date they were read. Quarterly GDP is carried forward on daily displays between releases; money-market assets are shown in trillions after converting the series' millions-of-dollars units.
Professional commentary and outlook
The tension in this dashboard is not between a booming economy and a collapsing one. It is between decelerating production and labor-market stability. Second-quarter real GDP advanced only about 0.37% from the first quarter, while unemployment remained at 4.1% in August. That combination supports a slow-growth base case rather than a near-term recession call. The Sahm measure moving below zero strengthens the near-term case, but its lagging construction means it cannot rule out a future downturn.
Market pricing reinforces the near-term distinction: the year-end recession contract is in the single digits. Yet an 8.5% event price and a 0.76% model reading are not competing estimates of the same event. Investors should avoid averaging them or presenting either as a universal 12-month recession probability. Search interest is also falling on the latest comparison, but a single Sunday search count says more about attention than payrolls or real income.
What would change the call?
- Labor: several months of rising unemployment, weaker payrolls and a Sahm reading moving toward +0.50 pp would challenge the baseline.
- Demand: weaker real consumption and another sharp step down in output would suggest that the slowdown is spreading beyond one quarter.
- Financial transmission: tighter credit, a persistent energy-price shock or widening corporate spreads would raise the chance that weak growth turns into job losses.
For portfolio construction, the current data favor stress-testing cyclical exposure rather than making an all-or-nothing recession bet: compare company balance-sheet resilience, financing needs and earnings sensitivity to slower consumer demand. Revisit the thesis when new labor, GDP and credit releases arrive, rather than treating stale weekly or quarterly series as a fresh daily signal.