2s10s Yield Curve Inversion History (1976–2026)

The U.S. Treasury yield curve — specifically the spread between the 10-year and 2-year Treasury yields (the “2s10s”) — is one of the most closely watched recession indicators in macroeconomics. It reflects broader dynamics in monetary tightening, rate expectations, and credit transmission. For a comprehensive framework on rate regimes and monetary transmission, see our monetary policy and interest rate analysis framework.

Every sustained monthly inversion of the 2s10s spread since 1976 has preceded a U.S. recession — a 5-for-5 record across confirmed episodes. The notable current exception is the 2022–2024 inversion, the longest on record at 26 months, which has not been followed by a downturn. This page compiles every major sustained 2s10s inversion documented in the Federal Reserve’s FRED T10Y2Y series, the observed lead times before each NBER-dated recession, and the key statistical takeaways for researchers, analysts, and investors.

Complete History of 2s10s Yield Curve Inversions (1976–2026)

The table below lists every major sustained inversion of the 10-Year minus 2-Year Treasury spread since the FRED T10Y2Y series began in June 1976. The same record exists in animated form: watch the yield curve bend through five decades. “Inversion start” refers to the first month in which the monthly average spread turned negative on a sustained basis (minimum two consecutive negative months). Trough spread is the daily minimum within the inversion window. Recession dates follow the NBER Business Cycle Dating Committee.

¹ Lead time is calculated from the start of inversion (first month of sustained negative monthly average) to the start of the NBER-dated recession (business cycle peak).
² Daily trough: −241 bps on March 20, 1980. Monthly average trough: −214 bps (March 1980).
³ The inversion was briefly interrupted in July 1989 (monthly average +0.20%). Sustained negative runs: January–June 1989 (6 months) and August–September 1989 (2 months). Duration of ~9 months counts the full Jan–Sep 1989 window inclusive of the interruption.
⁴ The inversion was briefly interrupted in April–May 2006 (monthly averages +0.10% and +0.14%). Sustained negative runs: February–March 2006 (2 months) and June 2006–March 2007 (10 months). Duration of ~14 months counts the full Feb 2006–Mar 2007 window inclusive of the interruption. A subsequent isolated negative daily reading occurred in May 2007 but did not qualify as a new sustained episode. Related reading: our study on inflation regimes and their structural drivers.

Sources: Federal Reserve Bank of St. Louis (FRED series T10Y2Y, daily data), National Bureau of Economic Research (recession dates). Trough spreads are daily minima within each inversion window. Last updated: February 2026.

Note on the August 2019 Brief Inversion

The T10Y2Y spread registered negative daily readings on three consecutive trading days in late August 2019 (August 27–29), reaching a minimum of −4 bps. This event attracted significant media coverage at the time. However, the monthly average for August 2019 remained positive (+0.055%), and no month in 2019 or early 2020 recorded a negative monthly average spread under the FRED T10Y2Y series. By the methodology applied in this study — which requires a minimum of two consecutive months of negative monthly average — this episode does not qualify as a major sustained inversion and is therefore not included in the table above.

The February–April 2020 recession (triggered by the COVID-19 pandemic) was exogenous in origin and not driven by the credit-channel transmission mechanism that typically links yield curve inversions to recessions. The absence of a qualifying inversion in the 12–24 months preceding the 2020 downturn is consistent with this assessment.

What the Data Show: Four Key Takeaways

1. Five for five: every confirmed sustained inversion preceded a recession

Across the five confirmed historical episodes, each major sustained monthly inversion of the 2s10s spread was followed by an NBER-dated recession. Lead times range from 10 to 22 months, with a median of 16 months. This consistency across radically different macroeconomic environments — from the Volcker disinflation to the dot-com bust and the subprime crisis — supports the view that the inversion captures a structural credit mechanism rather than a cycle-specific signal.

2. The un-inversion is often the more actionable signal

A counterintuitive but well-documented pattern: recessions tend to begin not during the inversion itself, but after the spread has returned to positive territory. This normalization — typically driven by a rapid decline in short-term rates as markets price in Fed easing — preceded the onset of recession in the 1989–1990, 2006–2007, and other episodes. The un-inversion can therefore be viewed as a more immediate leading indicator than the initial inversion. For researchers tracking recession risk in real time, the re-steepening of the curve deserves at least as much attention as the initial flattening. Further reading: Our note on the four curve shapes.

3. Inversion depth does not predict recession severity

The deepest inversion on record (−241 bps in March 1980) was followed by a relatively brief six-month recession. The shallowest confirmed inversion (−19 bps in 2006) preceded the worst financial crisis since the Great Depression. The 2022–2024 inversion — the second deepest at −108 bps — has not been followed by any recession at all. The magnitude of the negative spread is not a reliable predictor of the subsequent downturn’s depth or duration.

4. The 2022–2024 episode challenges the indicator’s track record

The July 2022 to August 2024 inversion lasted 26 months — the longest in the FRED T10Y2Y series — and reached a trough of −108 bps. Yet the U.S. economy posted real GDP growth of 2.9% in 2023 and above 3% on an annualized basis in Q2 and Q3 2024, according to the Bureau of Economic Analysis. Several structural factors may explain this anomaly: households and corporations had locked in low borrowing rates before the Fed’s tightening cycle, the post-Covid fiscal impulse remained unusually large, and AI-driven productivity gains may have extended the expansion. This episode is the first in the FRED T10Y2Y record to produce no recession within any reasonable forecasting horizon — and supports the view that yield curve inversion is a necessary but not sufficient condition for recession.

The Transmission Mechanism: Why Inversions Precede Recessions

The most widely accepted explanation for the inversion-recession link rests on the bank lending channel. Commercial banks borrow short (deposits, money markets) and lend long (mortgages, corporate credit). Their net interest margin — the spread between funding costs and lending yields — is their primary source of profit.

When the yield curve inverts, this margin compresses or turns negative. Banks respond by tightening credit standards and reducing loan supply — not as a strategic choice but as a profitability constraint. This credit contraction transmits gradually to the real economy: business investment slows, consumer credit tightens, and employment eventually deteriorates. The 10-to-22-month lag range observed in the data corresponds to the time required for this mechanism to propagate through bank balance sheets and spending chains.

The expectations hypothesis offers a complementary framework: inversion reflects bond market participants pricing in future rate cuts and economic slowdown. In this reading, the inversion does not cause the recession but rather announces it. Both channels likely operate simultaneously, with the banking channel providing the causal mechanism and the expectations channel providing the informational content.

The 3-Month/10-Year Spread: A Complementary Signal

The Federal Reserve Bank of New York uses the spread between the 10-year Treasury yield and the 3-month Treasury bill rate (FRED series T10Y3M) as the basis for its recession probability model. The empirical case for preferring the T10Y3M over the 2s10s as a forecasting variable — including the Engstrom and Sharpe forward-spread evidence — is laid out in our comparison between T10Y3M and T10Y2Y inversions. This measure behaves somewhat differently from the 2s10s.

The 3M–10Y spread responds more directly to Fed policy decisions, as the 3-month yield tracks the federal funds rate closely. It tends to invert later than the 2s10s but produces what some researchers consider a “cleaner” signal. During the 2022–2024 episode, the 3M–10Y spread remained inverted from October 2022 through December 2024 — roughly two months longer than the 2s10s — and reached a trough of −190 bps (March 2023), significantly deeper.

Research published by Fed economists Engstrom and Sharpe (2018, 2022) suggests that the “near-term forward spread” — which isolates monetary policy expectations over a six-quarter horizon — outperforms both traditional measures. Once this forward spread is included in forecasting models, the 2s10s becomes statistically redundant. However, the 2s10s remains the most widely cited measure in financial media and market commentary due to its simplicity and long track record.

Common Misinterpretations

Treating brief daily inversions as sustained signals. As demonstrated by the August 2019 episode, a brief intraday or multi-day touch of negative territory does not constitute a major inversion. The methodology applied here — and used by the New York Fed — requires a minimum of two consecutive months of negative monthly average. An isolated daily dip of a few basis points carries no demonstrated predictive value under the historical framework. Related analysis: our 2-year-versus-10-year comparison.

Selling risk assets immediately upon inversion. The S&P 500 has historically continued to rally for months — sometimes years — after the onset of inversion. During the 2022–2024 episode, the index hit multiple all-time highs while the curve was deeply inverted. The inversion is a regime signal, not a market-timing tool.

Ignoring the un-inversion. The normalization of the curve is often a more immediate warning than the initial inversion. The return to positive territory indicates that short-term rates are falling — typically because the Fed is easing or the market expects it to — which often coincides with the early stages of economic deterioration.

Confusing correlation with causation. The inversion does not “cause” recessions. It reflects a constellation of conditions — restrictive monetary policy, growth slowdown expectations, compressed bank margins — that, taken together, tend to produce economic contractions. The 2022–2024 episode demonstrates that when key transmission channels are impaired (e.g., if credit does not contract sufficiently), the recession may not materialize.

Pre-1976 Inversions

The FRED T10Y2Y series begins in June 1976. For earlier episodes (1950s–1970s), academic studies — notably by Campbell Harvey (1986) and Arturo Estrella and Frederic Mishkin (1996) — rely on reconstructed data or adjacent spreads (1Y–10Y, 3M–10Y). These studies document inversions preceding the recessions of 1957–1958, 1960–1961, 1969–1970, and 1973–1975, as well as a brief episode in 1966 that is generally regarded as the only false positive prior to the current episode.

For methodological rigor, the table above is limited to data directly verifiable on FRED. Pre-1976 episodes are documented in the academic references cited below.

Methodology and Sources

Frequently Asked Questions

What is a yield curve inversion?

A yield curve inversion occurs when short-term government bond yields exceed long-term yields. In the case of the 2s10s spread, it means the 2-year U.S. Treasury yield is higher than the 10-year yield. This is considered abnormal because investors typically demand higher compensation for lending over longer periods to account for inflation risk and uncertainty. When the curve inverts, it signals that markets expect economic conditions to worsen, driving future interest rates lower.

Does the yield curve always predict a recession?

Over the period covered by FRED data (since 1976), all five confirmed major sustained monthly inversions of the 2s10s have been followed by an NBER-dated recession. The 2022–2024 episode — the longest at 26 months — is potentially the first false signal. Academic literature also identifies a 1966 episode (based on reconstructed data) that was not followed by a recession. The historical hit rate for sustained monthly inversions remains 5 out of 5 confirmed episodes, though the 2022–2024 case remains under evaluation.

How long after the inversion does the recession start?

Across the five confirmed episodes, the lead time ranges from approximately 10 months (1980–1982) to 22 months (2006–2007), with a median of 16 months. This variability makes the inversion unsuitable as a precise market-timing tool, but consistent enough to function as a leading indicator of macroeconomic regime change.

What about the 2019 inversion? Did it not precede the 2020 recession?

The T10Y2Y spread touched negative values on just three trading days in late August 2019 (minimum −4 bps), with monthly averages remaining positive throughout 2019. By the methodology applied here — which requires a minimum of two consecutive months of negative monthly average — the 2019 episode does not qualify as a sustained inversion. The February–April 2020 recession was triggered by the COVID-19 pandemic shock rather than by the credit-channel transmission mechanism that typically links inversions to downturns. These two facts are consistent: no sustained inversion preceded the COVID recession, which was exogenous rather than cycle-driven. Eco3min lays this out in our crisis hub.

Why didn’t a recession follow the 2022 inversion?

Several hypotheses have been advanced: households and businesses had locked in low borrowing rates before the Fed’s tightening cycle, limiting the transmission of higher rates to the real economy; the post-Covid fiscal stimulus supported aggregate demand well beyond its initial impact; and AI-driven productivity gains may have extended the expansion. The U.S. economy appears less interest-rate-sensitive than in prior cycles, which has weakened the traditional bank lending channel.

What is the un-inversion and why does it matter?

The un-inversion refers to the yield spread returning to positive territory after a period of inversion. Historically, recessions have tended to begin after the un-inversion — not during the inversion itself. The un-inversion typically reflects short-term rates falling faster than long-term rates, as the Fed begins easing or the market prices in rate cuts. Many analysts consider the un-inversion a more immediate recession warning than the initial inversion.

Is the 2s10s spread still a reliable recession predictor?

The 2022–2024 episode has prompted a reassessment. As Campbell Harvey — the economist who first documented the yield curve’s predictive power in his 1986 dissertation — has noted, the more widely known and anticipated an indicator becomes, the more likely market participants are to alter their behavior in ways that diminish its effectiveness. The 2s10s remains a valuable signal of macroeconomic stress, but it is increasingly viewed as one input among several rather than a standalone predictor.

Conclusion

The 2s10s yield curve inversion remains one of the most robust leading indicators of U.S. recession risk in the post-1976 data. Six major sustained inversions have been identified; all five of the confirmed historical episodes preceded NBER-dated recessions, with a median lead time of 16 months. The 2022–2024 episode — the longest inversion in the history of the FRED series at 26 months — is the first to go unmatched by a recession within any reasonable forecasting horizon.

Two methodological points are worth underscoring. First, the threshold matters: brief daily touches of negative territory (as in August 2019) carry no demonstrated predictive value and should not be treated as equivalent to sustained monthly inversions. Second, the un-inversion — the return of the spread to positive territory — is often the more operationally relevant signal for real-time recession monitoring, as it typically coincides with early-stage monetary easing and economic deterioration.

As Campbell Harvey has observed: the more an indicator is publicized and anticipated by markets, the more its effectiveness tends to erode over time. The 2022–2024 episode may represent the beginning of that erosion — or simply a structural shift in interest-rate sensitivity that has delayed, not eliminated, the recessionary transmission.

The data and analysis presented on this page are provided for informational and educational purposes only. They do not constitute investment advice or a recommendation to take any specific action.

Related Eco3min analysis

• Version française de cette analyse
• US Dollar at the Onset of Major Global Crises (1973–2023) — Historical Dataset
• How monetary policy transmits to the real economy
• Reading economic & financial cycles