Recession Risk Outlook 2026

The Sahm rule indicator (sahmrealtime) sits at 0.27 as of February 2026, more than halfway to the 0.50 threshold that has correctly identified every postwar recession in real time. The Conference Board LEI six-month change is mildly negative; it has been negative for 24 of the past 36 months without a recession actually arriving, the longest such period in the LEI series history. The 10Y-2Y yield curve un-inverted in October 2024 after 26 months of inversion, currently at +52bp. The 10Y-3M curve un-inverted in early 2025. Q1 2026 GDP is tracking near +2.0 percent on the Atlanta Fed GDPNow, in line with potential.

Composite recession probability models give mixed reads. The NY Fed yield-curve model implies recession probability near 25 percent for the next twelve months, down from the 60 percent peak in 2023 but well above the unconditional 15 percent base rate. The Convex CRPI sits in the moderate range, consistent with the soft-landing thesis but not committing to it. The Bloomberg Economics model sits near 35 percent. The market-implied probability through HY spreads at 284bp is closer to 10 percent.

The dispersion is the story. Hard data (GDP, payrolls, retail sales) all point to expansion. Soft data (LEI, ISM Manufacturing, consumer sentiment Michigan 56 in early 2025) point to weakness. The Sahm rule, which uses the unemployment rate's three-month average versus its prior twelve-month low, is the cleanest single indicator with a 100 percent postwar hit rate, and it sits 0.23 from triggering. April 2026 is the late-stage of the longest "false alarm" in modern data: every leading indicator that should have flagged recession has, and recession has not arrived.

The first force is the labor market. Unemployment at 4.3 percent has drifted up from the 3.4 percent April 2023 low, a +90bp move that has historically been recessionary by itself. The Sahm rule machinery picks this up. Nonfarm payrolls have averaged near +175K over the past six months, materially below the 250K+ pace of 2022-2023 but still positive. Initial jobless claims four-week average sits in the 215-230K range, low historically. The labor market is decelerating, not deteriorating. The bull case is that this is the soft-landing path; the bear case is that the lag between deceleration and deterioration is shorter than consensus.

The second force is consumer balance sheets. The personal savings rate has fallen to 3.5 percent from the 16 percent COVID-era peak, near the cycle low. Credit card debt is at a record nominal $1.18 trillion with delinquency rates rising. The "excess savings" buffer that supported 2022-2024 spending is largely depleted at the median household. At the same time, top-quartile household balance sheets are in record shape (equity portfolios at all-time highs, home equity buffer extended). The bifurcated economy means aggregate spending stays positive while a meaningful share of households face stress, exactly the setup that produces sudden labor-market deterioration when one variable (jobs) shifts.

The third force is policy buffer. The Fed at 3.50-3.75 percent has 350-400bp of conventional rate-cut capacity if it needs it, more than at any other late-cycle moment in modern history. Fiscal policy is constrained by the existing $2 trillion deficit and political configuration. The rate-cut cushion historically has been the deciding factor: cycles with low policy rates entering downturn (2008, 2020) have seen severe recessions; cycles with high rates have seen Fed-managed slowdowns (1995). The 2026 setup is closer to 1995 than to 2008.

The bull-case template is 1995. After the 1994 hiking cycle (3.00 percent to 6.00 percent in twelve months), the economy slowed to roughly 1.5 percent annualized growth in early 1995 with rising unemployment. The Sahm rule never triggered. The Fed delivered three insurance cuts (July 1995, December 1995, January 1996), totaling 75bp, and the expansion continued for another five years. The S&P 500 returned +37 percent in 1995, +23 percent in 1996, +33 percent in 1997. The 1994-95 episode is the only postwar example of a major hiking cycle that did NOT produce a recession. April 2026 is closer to this template than any other in the historical record, with one important difference: the 2026 cycle has lasted longer, the cumulative tightening was larger (525bp vs 300bp in 1994), and the labor market has cooled more.

The bear-case template is 2007. Through mid-2007, the Sahm rule was below trigger, GDP was running 2 percent, unemployment was 4.5 percent, and the equity market hit new all-time highs in October. The yield curve had already inverted in 2006 and un-inverted by mid-2007 (the same sequence as 2022-2024). HY spreads were 241bp in May, the tightest of the cycle. By early 2008, every variable had broken: BNP Paribas froze funds August 2007, Bear Stearns failed March 2008, Lehman September 2008. The lesson is the late-cycle period of seemingly benign data lasted 12-15 months between the curve un-inverting and the recession starting. Today is roughly 18 months past the October 2024 un-inversion. The historical median lag from un-inversion to recession is 12-18 months.

The bull case is no recession in 2026. Probability roughly 55 percent. The path: payrolls average +125-175K through 2026, unemployment stabilizes near 4.5 percent without breaching Sahm, the Fed delivers 50-75bp of insurance cuts, financial conditions ease modestly, the maturity wall in credit is absorbed at refinancing rates 100-150bp lower than today. GDP runs 1.5-2.0 percent, modestly below potential. Equities trade in a +5 to +12 percent range driven by earnings rather than multiple expansion. Credit spreads stay tight, the curve stays positive, gold underperforms equities. This is the consensus 1995-style outcome.

The bear case is a 2026-2027 recession. Probability roughly 35 percent. The trigger is most likely labor: payrolls slow to +50K monthly in Q3, unemployment crosses 4.7 percent, Sahm triggers, and the consumer pulls back. Default rates rise to 6-8 percent, HY OAS reprices to 500-700bp, equity multiples compress 15-25 percent. The Fed cuts aggressively (200-300bp over 12 months) but the lag between cuts and real-economy recovery runs 6-9 months. Real GDP contracts -1.0 to -2.5 percent over two-to-four quarters. Recession is not catastrophic by 2008 standards but is meaningful for asset prices: SPY drawdown 18-30 percent, IWM drawdown 25-40 percent, gold and Treasuries outperform.

First, the Sahm rule (sahmrealtime, monthly with employment release). Currently 0.27; cross 0.50 and recession has historically already started. Second, monthly nonfarm payrolls. Drop below +75K average over three months historically signals downturn. Third, weekly initial jobless claims four-week average; sustained breakout above 275K is the precursor. Fourth, Conference Board LEI six-month annualized change; below -4 percent has flagged every recession. Fifth, ISM Manufacturing PMI; below 45 (currently low-50s) is the manufacturing recession threshold. Sixth, the 10Y-3M re-inversion or steepening to +250bp+ (severe steepening from inversion has historically been the recession-onset signal). Seventh, retail sales control group three-month annualized change; negative for two consecutive quarters historically aligns with recession quarters. Eighth, the unemployment rate itself versus its trailing 12-month minimum; +50bp move historically aligns with the recession start.