PortfolioAI · U.S. recession analysis · September 29, 2026

Higher Yields Test the Consumer, Not Yet the Labor Market

A rise in borrowing costs and weaker household confidence complicate the soft landing. The decisive question is whether this pressure turns into sustained job losses.

Executive summary

Three months: contraction unlikely

August unemployment was 4.1% and the Sahm rule was −0.07 percentage point, far below the +0.50 trigger. A weak confidence reading alone is not a recession signal.

Six months: demand is the test

Second-quarter real output grew, but at a slower pace. Persistently higher financing costs and caution among households could translate into weaker sales and hiring.

Twelve months: conditional risk

The base case is continued expansion, not immunity. The full-year outlook depends on whether real spending, credit and employment weaken together; year-end event odds are not a 12-month forecast.

Tuesday's market close made the transmission mechanism clearer: the 10-year Treasury yield was near 5.255%, while oil fell to about $89.38. Lower energy costs can help household purchasing power, but a higher discount rate and borrowing rate make housing, refinancing and investment harder. Equities slipping modestly is evidence of a repricing, not proof that aggregate demand has already contracted. The Polymarket year-end U.S. recession contract showed roughly 9% “Yes” in the September 28 reading. Its narrow deadline and resolution rules differ from the New York Fed model's probability series; neither number should be relabeled a universal recession forecast.

Three views of economic momentum

Jobs remain the buffer

Monthly unemployment rate (%) and Sahm rule (percentage points), November 2025–August 2026. The dashed line is the Sahm onset threshold.

Real output still grows

Real GDP, billions of chained 2017 dollars, quarterly; the last observation is Q2 2026, not a September reading.

Model risk remains low on its own scale

New York Fed recession probability series (%), distinct monthly observations through July 2026; repeated values on daily displays do not constitute new forecasts.

Recession risk table

SignalLatest distinct observationWhat it means
Sahm rule−0.07 pp · August 2026Below +0.50 pp onset threshold; a lagging confirmation signal, not a promise of safety.
Unemployment4.1% · August 2026Stable versus July; the direction and breadth of future layoffs matter more than a single level.
Real GDP$24,269.6bn · Q2 2026Up 0.37% from Q1, roughly 1.5% annualized. Positive growth leaves less margin for a spending shock than faster expansion would.
NY Fed recession model0.76% · July 2026A model-specific estimate with a different horizon and definition from the year-end contract.
BCIG cyclical gauge8.3 · May 8, 2026Weekly series' last available observation predates this report by months; too old to drive the near-term call.
Money-market fund assetsAbout $8.44tn · Q1 2026Large cash stock may reflect yields and allocation, not a mechanical recession forecast.
Recession searches1,287 U.S. searches · September 27Down 57% versus roughly a week earlier; Sunday seasonality and headline sensitivity limit interpretation.
Year-end market oddsApproximately 9% Yes · September 28A changing contract price tied to 2026 resolution, not a 3-, 6- or 12-month probability.

Observation dates are the underlying data dates, not the date of this analysis. Money-market values are converted from millions to trillions of dollars. Economic data can be revised.

Professional commentary and outlook

Today’s pressure is financial before it is employment-based. A 10-year yield near 5.25% raises the hurdle for investment and home purchases. Tuesday's roughly 3.5% fall in oil offers a counterweight, but the effect on household budgets takes time and could reverse. Confidence weakness can foreshadow restraint; it becomes a recession case only if actual consumption, corporate revenues and labor demand follow it down.

That sequence also matters for portfolio risk. The September 29 Reddit stock discussion features AI infrastructure, defense and leveraged turnarounds. These are not one macro trade: a company dependent on refinancing may suffer from high yields even without a recession, while a supplier with contracted cash flows can face a different demand cycle. Examine debt maturity, interest coverage, cash conversion and customer concentration rather than treating online attention as a macro indicator.

What would change the assessment?

  • Labor deterioration: repeated rises in unemployment and claims, accompanied by a Sahm reading moving toward +0.50 pp, would turn soft demand into a more credible contraction threat.
  • Spending and credit: falling real household consumption together with tightening lending standards and rising delinquencies would show that financing pressure is spreading.
  • Relief: stable long-term yields, durable energy disinflation and resilient payrolls would make a slow-growth expansion more plausible despite poor confidence.

Do not average the 0.76% model observation and roughly 9% year-end event price. Different event definitions, timestamps and horizons produce different numbers. Stress-test cyclicals and highly indebted issuers against a weaker-sales scenario while waiting for employment and real spending to confirm—or reject—the warning.