Three Signals That Typically Precede Recessions — And What They're Saying Now

Signal

The US Treasury yield curve — the difference in yield between 10-year and 2-year Treasury bonds — uninverted in late 2024 after one of the most extended inversions on record: twenty-six consecutive months from July 2022 to September 2024. The uninversion, in which longer-term rates rise above shorter-term rates, ended the inversion phase. It did not, by historical precedent, end the recession risk.

The Conference Board's Leading Economic Index has declined in thirteen of the past twenty-four months. The LEI — a composite of ten forward-looking economic indicators including building permits, manufacturing orders, and consumer expectations — has a track record as one of the more reliable pre-recession signals when it declines by more than 4 percent from its prior peak. It has declined by 6.2 percent since its 2023 peak.

Consumer credit delinquency rates have risen to their highest level since 2011. The Federal Reserve's Q4 2025 data shows 90-day delinquencies on credit cards at 3.6 percent, up from 1.7 percent in 2021. Auto loan delinquencies are at 3.1 percent, up from 1.4 percent. These are not crisis levels, but they represent a sustained trend in the direction of credit stress that historically precedes reduced consumer spending.

Interpretation

Each of these signals has an imperfect track record. The yield curve inversion has preceded eight of the past ten recessions since 1965, but the lead time is variable — ranging from six months to twenty-two months — and it has produced false positives. The LEI has a similar track record with similar caveats.

The significance of the current moment is not any single indicator but the simultaneous elevation of multiple signals that have historically been associated with recession risk. When the yield curve, the LEI, and consumer credit stress all move in the same direction simultaneously, the historical base rate for recession within twenty-four months is approximately 68 percent.

The current cycle has several factors that make historical comparisons imperfect. The labor market remains unusually strong for an economy showing these signs of strain — unemployment at 4.1 percent is historically inconsistent with recession onset. This has generated significant debate among economists about whether the traditional recession models are applying correctly to the post-COVID, AI-disrupted labor market. Some argue the labor market strength provides a buffer; others argue it is a lagging indicator that will deteriorate as the credit and spending signals propagate.

Scenario

Scenario A — the soft landing — involves the leading indicators stabilizing and reversing without tipping into recession. Consumer spending holds; labor market strength provides income support that prevents the credit stress from becoming self-reinforcing; the Fed has sufficient rate-cutting room to stimulate if needed. The 2023-24 period provided some evidence that soft landings are achievable even from inverted yield curves if the Fed acts proactively.

Scenario B — the moderate recession — involves a two-quarter contraction in GDP, unemployment rising to 5.5 to 6.5 percent, consumer spending declining, and a corporate earnings recession that produces elevated but not catastrophic financial market stress. This is the historical modal outcome when the current combination of signals is present.

Scenario C — the deep recession — involves a financial accelerator: credit stress in consumer and commercial real estate loans triggers bank balance sheet problems; bank stress produces credit tightening that deepens the economic contraction; the self-reinforcing cycle produces unemployment above 7 percent and a multi-year recovery period. This scenario requires a financial system shock that is not currently visible in the data; its possibility cannot be excluded.

Probability

Polymarket's recession probability contract — defined as two consecutive quarters of negative real GDP growth beginning before Q3 2026 — was trading at 38 percent as of March 2026. Metaculus's equivalent forecast was 41 percent. Professional forecaster surveys from the Philadelphia Fed give a mean recession probability over the next twelve months of 34 percent.

These numbers cluster around a third to two-fifths probability — higher than baseline but not indicating that recession is the modal outcome. The range is consistent with genuine uncertainty: the soft landing scenario is plausible, the moderate recession scenario is approximately equally plausible, and the deep recession scenario is a meaningful tail risk.

Kalshi's contract on whether Federal Reserve rate cuts in 2026 will exceed 100 basis points — a signal that the Fed is in recession-response mode — was trading at 47 percent. Whether the Fed cuts aggressively is both a predictor of recession risk and a potential mechanism for preventing it.

Indicators to Watch

— Initial jobless claims (weekly): the most real-time signal of labor market deterioration; sustained readings above 260,000 per week would be a meaningful inflection — JOLTS job openings: declining job openings precede unemployment increases by approximately three months — ISM Manufacturing and Services PMI: PMI readings below 50 for three consecutive months signal broad-based contraction — Senior Loan Officer Survey: bank lending standards tightening sharply is a leading indicator of credit contraction — Corporate high-yield spreads: spread widening above 500 basis points signals that credit markets are pricing material default risk

The signals are elevated. The outcome is not determined. Readers who are making financial, career, or business decisions with two-to-five-year horizons should be aware that the historical evidence places material probability on a significant economic slowdown in the window.

---

Miles Thornton is a contributing writer at The Auguro covering financial markets, monetary policy, and macroeconomic forecasting.