U.S. Recession Risk Monitor

The U.S. macro picture in mid-May 2026 is a study in divergence. Real GDP continues to expand—real output reached roughly $24.2 trillion in the latest reading—and the unemployment rate holds near 4.3%, a level consistent with late-cycle cooling rather than collapse. The Sahm rule recession indicator sits at 0.13, well below the 0.50 trigger that has historically flagged the start of downturns within months.

Yet other gauges are flashing louder. The New York Fed's recession-probability model has surged to 1.82% on a daily basis—up sharply from the sub-0.5% range that prevailed through early 2026—while the Bloomberg U.S. Conditions Index (BCIG) recovered from an April trough near 4.0 to 8.3 by early May, signaling improving but still fragile business conditions. Prediction markets on Polymarket assign roughly a 25% probability that the National Bureau of Economic Research will declare a U.S. recession by the end of 2026.

Public anxiety, meanwhile, has faded. Daily Google search volume for "recession" registered 3,012 queries on May 18—down 63% from a year earlier and nearly 10% below the prior month—suggesting households are not yet pricing acute contraction risk even as quantitative models and betting markets have repriced the tail.

The soft-landing consensus is being stress-tested

For most of 2025 and early 2026, the dominant macro narrative was a controlled slowdown: inflation moderating, the Federal Reserve holding or easing gradually, and employers trimming headcount without mass layoffs. That story still fits the hard labor data. Unemployment near 4.3% and a Sahm reading of 0.13 are not the profile of an economy already in recession.

What changed in April is the simultaneous jump in model-based recession probability and the BCIG's plunge to multi-year weekly lows before rebounding. The NY Fed's daily gauge moving from roughly 0.48% to 1.82% in a matter of weeks is unusual outside of financial-stress episodes. It likely reflects tighter financial conditions, softer forward-looking survey data, and the lagged effect of prior rate hikes working through credit-sensitive sectors.

Prediction markets vs. public search behavior

Polymarket's ~25% odds for a 2026 recession sit meaningfully above the sub-10% readings common during mid-cycle expansions, yet well below the 50%+ peaks seen during tariff-shock scares in early 2025. Traders are effectively saying: recession is a real tail event, not the base case. Households, judging by search interest, disagree—daily recession queries are down sharply year over year. That disconnect matters for markets: if labor data deteriorates, retail concern could catch up quickly, amplifying risk-off moves.

Sector and positioning implications

• Defensive quality—consumer staples, healthcare, and utilities with pricing power—remains the natural hedge if model odds stay elevated while growth persists.
• Financials face a bifurcated setup: net-interest income durability helps money-center banks in a slow-growth world, but credit losses rise sharply if unemployment breaks above 4.5%.
• Long-duration Treasuries benefit if recession odds convert into actual easing; they underperform if growth re-accelerates and term premia widen.
• Cyclicals and small caps are most exposed to a BCIG-style air pocket—conditions can weaken faster than headline GDP suggests.

What would change our view

A sustained Sahm reading above 0.50, unemployment rising above 4.6% on a three-month average, or BCIG falling back below 5.0 would shift our 6-month outlook from low-moderate to elevated. Conversely, BCIG holding above 9.0 with Sahm stable would reinforce the soft-landing thesis and argue for adding selective cyclical exposure.

Sources: Federal Reserve Economic Data (FRED) series via PortfolioAI indicators; DailySearchVolume.com recession keyword data (EN-US); Polymarket "US recession by end of 2026" market pricing.