PortfolioAI macro research · September 23, 2026

US Recession Risk: Labor Buffer Meets a Growth Test

The labor-market trigger remains quiet and real output is still rising, but a benign snapshot is not a guarantee against a late-year shock.

Executive summary

Base case: continued expansion over the next three and six months; a less certain, but still expansionary, twelve-month outlook. The August Sahm Rule indicator stood at −0.07 percentage points, well below its +0.50 recession-signal threshold. August unemployment was 4.1%, down from 4.4% in December 2025. Real GDP reached $24.27 trillion in chained-dollar annual-rate terms in Q2 2026, versus $24.18 trillion in Q1. These readings argue against treating recession as the central near-term case.

That conclusion is deliberately narrower than a forecast of safety. The last available weekly BCIG reading, 8.3 on May 8, rebounded from 4.0 in early April but is not a September pulse. The most recent recession-probability series observation, 0.76% for July, is a model reading, not a September or twelve-month probability. A separate Polymarket contract priced a roughly 8.5% “Yes” chance on the morning of September 23 UTC; its definition and year-end horizon differ from the model's. Neither measure should be combined or treated as a precise probability for the next twelve months.

Next 3 months

Recession not the base case. Watch whether layoffs spread beyond isolated sectors and push the Sahm gauge toward +0.50.

Next 6 months

Expansion favored, with material downside risk. A sequential GDP contraction paired with rising unemployment would change the assessment.

Next 12 months

More uncertainty than the year-end market price suggests. That contract ends in 2026; a full-year horizon runs into September 2027.

Assessment is qualitative, not a calibrated three-, six- or twelve-month probability. Series have different release schedules; observation periods appear below.

Three views of the cycle

Unemployment and the Sahm Rule

The labor deterioration signal has retreated; the +0.50 line is a historical rule of thumb, not a forward forecast.

Monthly observations, latest available: unemployment and Sahm August 2026. Sources: UNRATE, SAHMCURRENT.

Model-estimated recession probability

The series rose in autumn 2025 and subsequently declined, though its last observation predates this report.

Monthly underlying observations shown as percentages; last available July 2026. Source: RECPROUSM156N. Not the same contract or horizon as Polymarket.

Real GDP level: expansion, but a smaller latest step

Q2's 0.37% quarter-on-quarter rise is about 1.5% at an annualized pace, calculated from the rounded series levels below.

Quarterly chained-dollar level, trillions at annual rates; last available Q2 2026. Source: GDPC1. Level changes are not official headline GDP growth estimates.

Recession risk scorecard

Latest available observations, not synchronized September readings
IndicatorLatest readingInterpretation and limit
Sahm Rule (SAHMCURRENT)−0.07 pp · Aug 2026Below +0.50 historical recession trigger; unemployment would need to deteriorate materially to trip it.
BCIG, weekly8.3 · May 8, 2026Recovered from 4.0 on April 3; too old to establish current September conditions. Series reference.
Recession-probability model0.76% · Jul 2026Low historical model reading; not a live event-market quote or a 12-month forecast.
Real GDP$24.27tn · Q2 2026Up from $24.18tn in Q1; real output growing, but the quarterly increment was modest.
Unemployment rate4.1% · Aug 2026Below late-2025 levels; watch the direction over multiple releases rather than one print.
Money-market fund assets$8.44tn · Q2 2026Up from $8.29tn in Q1; a liquidity/asset-allocation gauge, not a standalone recession signal.
“Recession” search interest2,078 daily searches · Sep 22, 2026Attention gauge only; page also reports 128,005 average monthly searches, a different aggregation.
2026 recession contract~8.5% Yes · Sep 23, 11:49 UTCTradable market price, not a scientific recession estimate; check contract resolution rules and liquidity.

Commentary and outlook

The strongest argument against an imminent downturn is the labor sequence, not the low event-contract price. An August jobless rate of 4.1% and a Sahm reading below zero indicate that the three-month average unemployment rate has not risen enough relative to its prior-year low to flash the rule's warning. That is a meaningful buffer, although the indicator was designed to identify recessions once labor-market damage becomes visible, rather than to eliminate advance risk.

Real output supports the same near-term conclusion but with less room for complacency. The Q2 level exceeds Q1, so the latest two quarters do not show consecutive real-output declines. Yet an increase of roughly $89 billion in annual-rate chained dollars is a thinner margin than some earlier quarters. Growth that is positive but modest can absorb fewer shocks to household spending, credit supply or business investment.

What would change the call? A sustained rise in unemployment alongside a Sahm reading approaching +0.50; two successive negative quarterly GDP prints; and renewed weakness in timely activity measures would make a defensive stance more compelling. In the other direction, stable hiring and continued positive output would reinforce the expansion case. Cash parked in money-market funds is optionality for investors, not proof that the economy is either protected or headed for recession.

For portfolios, favor exposure that can survive both a slower-growth environment and a still-expanding one: scrutinize refinancing needs, balance-sheet quality and earnings sensitivity to demand before making a binary recession trade. A low probability on a 2026-only event contract should not be confused with the risk budget for investments held into late 2027.

As-of date: September 23, 2026. The charts display observation periods rather than publication dates. Polymarket probability is a historical market price observed at approximately 11:49 UTC on September 23 and can move rapidly. This report is analysis, not investment advice.