Reading the Yield Curve: What the 2026 Un-Inversion Tells Us

The Un-Inversion Nobody Expected

As of March 31, 2026, the 10-year minus 3-month Treasury spread (T10Y3M) has turned positive at 0.60 — ending the longest sustained yield curve inversion in modern history. The spread had been negative since mid-2022, roughly 42 months of continuous inversion that eclipsed the previous record of 21 months set in 1978-1980.

Financial media is treating the un-inversion as good news. "Yield Curve Normalizes" read one headline. "Recession Signal Clears" announced another. This interpretation is dangerously wrong.

The historical pattern is unambiguous: recessions begin AFTER the yield curve un-inverts, not during the inversion. The inversion is the warning. The un-inversion is the starting gun. In every recession since 1970 — 1973, 1980, 1981, 1990, 2001, 2008, 2020 — the curve un-inverted (returned to positive slope) within 2-18 months before the recession officially began. The median lag from un-inversion to recession onset is approximately 5 months.

The mechanism is straightforward: the curve un-inverts because either the Fed begins cutting short-term rates in response to deteriorating economic data (pushing the short end down) or because long-term rates spike on rising inflation expectations (pushing the long end up). In the current case, both forces are at work — the market is pricing in emergency Fed cuts while the oil shock is driving long-term inflation expectations higher.

Inside the Estrella-Mishkin Probit Model

The NY Fed's recession probability model, developed by Arturo Estrella and Frederic Mishkin in their seminal 1998 paper "Predicting U.S. Recessions: Financial Variables as Leading Indicators," uses a probit regression with a single input: the T10Y3M spread. The model's coefficients (β₀ = -0.5333, β₁ = -0.6330) produce a probability via Φ(β₀ + β₁ × spread), where Φ is the standard normal CDF.

At the current spread of +0.60, the model outputs a recession probability of approximately 15-18%. This is actually LOWER than when the curve was inverted — which seems counterintuitive but reflects the model's design. The probit model captures the credit conditions signal embedded in the yield curve. A positive spread means the bond market currently sees easier credit conditions ahead. What the model cannot capture is the lag effect: the economic damage from the 42-month inversion is already baked in and working through the system.

This is the fundamental limitation of the Estrella-Mishkin model in the current environment. It is a real-time thermometer that reads "normal" the moment the fever breaks — even if the underlying infection is worsening. GeoWire's composite model addresses this by incorporating five additional signals that capture the lagged effects the yield curve alone misses.

The Oil Shock Compounding Factor

James Hamilton's 2003 analysis of oil price shocks and economic activity introduced the Net Oil Price Increase (NOPI) framework — a measure of how much oil prices have risen above their previous 3-year peak. When NOPI is positive and large, recessions follow with high reliability.

The current NOPI reading is extreme. Brent crude at $105/barrel represents a roughly 55% increase above its 3-year trailing high of approximately $68. Hamilton's coefficient (-0.15 oil-to-GDP elasticity) implies a GDP drag of approximately 8.25% on an annualized basis — though the actual impact depends on duration and the speed of supply adjustment.

What makes the current situation historically unusual is the combination of a yield curve un-inversion signal AND a major oil shock occurring simultaneously. In 1973, the oil shock preceded the yield curve signal. In 2008, the yield curve inverted well before oil spiked. In 2026, we have both signals firing together, which amplifies the recession risk beyond what either indicator alone would suggest.

GeoWire's Hamilton NOPI model currently contributes the largest single upward adjustment to our composite recession probability — more than the yield curve, more than credit spreads, more than the Sahm Rule. The oil shock is the dominant risk factor.

The Sahm Rule: Labor Market Early Warning

The Sahm Rule, developed by economist Claudia Sahm in 2019, triggers when the 3-month moving average of the national unemployment rate rises 0.50 percentage points or more above its low from the prior 12 months. It has correctly identified every recession since 1970 in real time — often before official NBER dating.

The current Sahm Rule reading is 0.37 — below the 0.50 trigger threshold but rising. The March jobs report showed unemployment at 4.2%, up from 3.7% at its cycle low. The trajectory matters: the Sahm indicator has risen 0.15 points in the last two months alone, an acceleration rate consistent with the early stages of the 2001 and 2008 recessions.

Initial jobless claims, the weekly leading indicator of labor market stress, have crept above 240,000 — not yet alarming in isolation, but up 15% from the 210,000 average in Q4 2025. Energy-sector layoffs in the Gulf Coast and rising construction sector unemployment (as interest rates suppress housing starts) are the primary drivers.

If the Sahm Rule triggers at 0.50, it will be the first time in history that the trigger coincides with both a yield curve un-inversion and an active oil shock. The triple confluence has no modern precedent.

What to Watch: A Triple Threat Without Modern Precedent

The convergence of three historically reliable recession signals — yield curve un-inversion, Hamilton NOPI oil shock, and an approaching Sahm Rule trigger — creates a risk environment that deserves more attention than it is receiving. No single prior recession featured all three signals activating within the same quarter.

For investors, the key thresholds to monitor are: the Sahm Rule crossing 0.50 (currently 0.37 and rising), high-yield credit spreads crossing 500 basis points (currently approximately 420 and widening), and the Philadelphia Fed Leading Index turning negative for three consecutive months (currently at -0.8 for February, following -0.2 in January).

For business owners, the practical implications are more immediate. The oil shock is already flowing through to transportation costs, with diesel up 38% from pre-crisis levels. The steel and aluminum supply disruptions are adding 15-25% to construction material costs. And the semiconductor supply chain — threatened by the helium shortage — could constrain technology equipment availability by mid-2026.

GeoWire's composite recession probability model currently estimates a 62% probability of US recession within 12 months. This figure incorporates all six of our academic models: the NY Fed Probit (yield curve), Hamilton NOPI (oil shock), Sahm Rule (labor market), Philadelphia Fed Leading Index (broad leading indicators), credit spreads (financial conditions), and our geopolitical risk adjustment. The live reading is available on our dashboard, updated hourly with fresh FRED data.

The yield curve "normalizing" is not the all-clear signal the headlines suggest. It is, historically, the final warning.