A yield curve inversion occurs when short-term Treasury yields rise above long-term yields, producing a negative spread. The most widely tracked measure is the 2s10s spread (10-year minus 2-year yield), though the 3-month/10-year spread is the Federal Reserve's preferred indicator.
Inversions are significant because they have preceded every U.S. recession since the 1960s, typically with a lead time of 6 to 24 months. The logic is direct: when markets expect the Fed to cut rates in response to economic weakness, long-term bond prices rise and long yields fall, while short-term yields remain anchored by current policy rates.
An inversion reflects several forces operating simultaneously:
The depth and duration of the inversion influence its predictive value. Shallow, brief inversions carry less signal than deep, sustained ones.
Three spreads dominate the professional conversation, and each carries distinct properties:
3m10y (3-month vs. 10-year). This is the New York Fed's preferred recession indicator. Academic research, including the work of Estrella and Mishkin, shows it has the strongest historical track record for predicting recession within a 12-month window. The logic is clean: the 3-month yield closely tracks expected Fed policy over the near term, so a large negative 3m10y reflects market conviction that the Fed will cut sharply.
2s10s (2-year vs. 10-year). The most widely cited spread in financial media and among market participants. It is a reliable indicator, but with a longer and more variable lead time than the 3m10y. The 2-year yield incorporates roughly two years of rate expectations, so the 2s10s responds earlier to shifting rate outlooks than the 3m10y does.
2s5s (2-year vs. 5-year). A shorter-horizon indicator that often leads the 2s10s by several months. Portfolio managers who watch curve dynamics early in a tightening cycle often monitor the 2s5s as a leading signal within an already-inverting curve.
The key rule across all three: every U.S. recession since at least the 1960s, including the 1969-70 recession, was preceded by inversion in at least one of these measures. But the lead time ranges from 6 to 24 months, making inversion a directional signal rather than a precise timing tool.
The record across cycles is consistent. In 1978, the 2s10s inverted and the recession began in January 1980, an 18-month lead. In 1989, inversion preceded the July 1990 recession by approximately 12 months. In 2000, the curve inverted before the March 2001 recession by roughly 10 months. In 2006, the 2s10s went negative and the recession did not begin until December 2007, again an 18-month lag.
The 2019 inversion is the most contested entry. The 2s10s inverted briefly in August 2019. The recession that followed in March 2020 was exogenous, triggered by the COVID-19 pandemic rather than the credit and demand deterioration that normally follows a financial-cycle inversion. Most analysts treat this case as a coincidence rather than a confirmation of the model.
The 2022 cycle is still being evaluated. The 2s10s inverted in March 2022. As of mid-2025, the economy has slowed materially but has not entered a formal NBER-defined recession. Whether this becomes a false positive or a long-lag confirmation remains open.
The 1998 inversion is the clearest false positive in modern data. The 2s10s briefly inverted during the Russian debt crisis and Long-Term Capital Management episode. No recession followed.
Three structural caveats apply to all inversion analysis. First, the 3m10y has a better historical track record than the 2s10s for near-term recession prediction. Second, the lead time is variable enough that acting on inversion as a market-timing signal has historically been unreliable. Inversion tells you the direction of risk, not the date. Third, QE and global demand for long-duration assets can depress term premiums artificially, compressing long yields independently of rate expectations. In that environment, inversion may occur earlier than the economic cycle warrants.
The 2022-2024 inversion cycle was the most extreme since 1981. The 2s10s spread reached -108 bps in July 2023. The 3m10y spread reached -189 bps in May 2023. Both figures represent the deepest readings in over four decades.
The curve began normalizing in late 2024 as the Fed initiated a rate-cutting cycle. By mid-2025, the 2s10s had returned to a modestly positive range. Whether the normalization reflects a soft landing or simply the passage of time before a delayed recession remains the central debate among fixed-income strategists.
Does every inversion lead to a recession?
No. The 1998 inversion is the most cited counterexample in the post-1980 data. However, every U.S. recession since the 1960s was preceded by inversion, which makes the relationship asymmetric: recession without prior inversion is rare, but inversion without recession occurs occasionally.
How long after inversion does recession typically begin?
The historical range is 6 to 24 months across cycles. The median across post-1970 episodes is roughly 12 to 14 months. The variability is large enough that inversion alone cannot time a portfolio hedge.
Which spread should I monitor?
For recession prediction, the 3m10y is the academically preferred measure. For market sentiment and relative value, the 2s10s is the more liquid and widely quoted spread. YieldCurve.pro tracks both in real time at /spreads/2s10s and /spreads/3m10y.