In July 2022, the 10Y-2Y Treasury spread went negative. Every financial journalist in the world wrote about it. "Yield curve inverts, recession coming," the headlines said, more or less. Clients asked about it. LinkedIn filled up with hot takes. The inversion became the most watched signal in markets.
Then markets bottomed in October 2022 and ripped higher. Soft landing narratives took hold. By mid-2023, the inversion was at its deepest point in four decades, touching -108 basis points in early March 2023. And yet the recession never arrived on schedule. Some analysts declared the signal broken. The inversion had cried wolf.
This is where I think people are reading the curve wrong. The inversion is not the signal that the recession has started. It is the signal that the conditions for a recession are being created. The actual damage, the credit contraction, the unemployment spike, the GDP decline, tends to materialise after the curve uninverts. And by the time it uninverts, most investors have already lowered their guard.
The pattern across every major cycle
Go back and look at the three major US recessions of the past four decades. In each case, the 10Y-2Y inverted months or years before the recession started. But in each case, the NBER-dated recession start came after the curve had already steepened back toward positive territory.
1990 recession: Curve inverted late 1988. Uninverted mid-1989. Recession began July 1990, roughly 12 months after the spread crossed back to zero.
2001 dot-com recession: Curve inverted early 2000. Uninverted late 2000. Recession began March 2001, roughly 4 months after the uninversion.
2008 financial crisis: Curve inverted mid-2006. Uninverted mid-2007. Recession began December 2007, approximately 5 months after the spread turned positive again.
Current cycle: Inverted July 2022. Deepest point: -108bps, March 2023. Curve has been gradually flattening since mid-2023, re-approaching zero as of early 2026.
The lag between uninversion and recession start has ranged from four months to over a year. It is not a precise timer. But the direction of causality is clear: the recession does not precede the uninversion. The uninversion precedes the recession.
Why the mechanism works this way
To understand why uninversion is the more dangerous moment, you need to understand what actually causes the curve to invert, and what causes it to uninvert.
When the Fed raises short-term rates aggressively (as it did from March 2022 onwards), the front end of the curve rises faster than the long end. Short-term yields jump because they track the fed funds rate directly. Long-term yields are anchored by growth and inflation expectations, which do not move as quickly. The result: the 2-year yield climbs above the 10-year yield. Curve inverts.
The inversion then starts doing damage through a very specific channel: bank net interest margins collapse. Banks borrow short and lend long. That is their entire business model. When short rates exceed long rates, the margin on every new loan shrinks. Banks respond rationally, they tighten lending standards, reduce credit availability, and pull back from riskier borrowers. Credit conditions tighten across the economy. This does not happen overnight. It takes twelve to eighteen months to fully transmit into economic activity.
The inversion creates the conditions. The uninversion is when the Fed acknowledges those conditions have done their work, and by then, it is usually too late to avoid the damage.
Now here is the critical part. The curve uninverts because the Fed cuts rates. And the Fed cuts rates because it is seeing the leading indicators of distress, rising credit card delinquencies, weakening payrolls, slowing retail sales, widening high-yield spreads. The rate cuts cause the front end to fall faster than the long end, which steepens the curve back toward positive territory.
But the credit tightening that was set in motion during the inversion does not reverse immediately. Banks do not instantly reopen their lending books the moment the Fed cuts 25bps. Corporate borrowers that had been refinancing expensive short-term debt suddenly cannot. Consumers who relied on revolving credit find terms worsening. The economic damage that was being slowly loaded into the system finally discharges, right as the curve is uninverting and investors are thinking "all clear."
The velocity of the uninversion matters
Not all uninversions are equal. The speed of the steepening contains information about how worried the Fed actually is.
A slow, grinding uninversion, where the spread drifts from -80bps to -40bps to 0 over twelve to eighteen months, gives the economy more time to absorb the tightening. The Fed is moving carefully. It sees problems but nothing acute. The recession, if it comes, may be shallow.
A fast uninversion, where the Fed is cutting rates at consecutive meetings, the front end is collapsing, and the spread swings from -100bps to +50bps in a matter of months, is a much more dangerous signal. That velocity indicates the Fed has seen something serious enough to act aggressively. In 2008, the Fed cut rates by 325bps between September 2007 and April 2008. The curve steepened sharply. The financial system collapsed six months later.
The speed of Fed cuts is inversely correlated with the economy's health at the time of cutting. The faster they move, the worse the outlook. A curve that steepens because the Fed is panicking is not a recovery signal, it is a distress signal wearing recovery's clothing.
Where we are now
As of early 2026, the 10Y-2Y spread has been on a gradual path back toward zero after its historic inversion. The Fed began cutting rates in September 2024, and the front end has been falling. The long end has been somewhat sticky, there is a credible argument that term premium is rising as fiscal deficits remain elevated and bond supply increases.
The uninversion, in this cycle, has been happening slowly. That is the relatively reassuring version of events, it suggests the Fed is not in emergency mode, credit conditions are tightening but not collapsing, and any recession risk is back-loaded rather than imminent.
But the historical playbook says: do not relax simply because the curve is moving back to normal. Normal is where recessions tend to actually start. The credit damage from two-plus years of inversion is still working through the system, in commercial real estate, in small business lending, in consumer credit quality at the lower end of the income distribution.
The practical implication
Most macro commentary treats the yield curve as a binary: inverted equals bearish, uninverted equals neutral or bullish. That framing is wrong, and it is wrong in a way that costs investors real money.
If you got defensive only when the curve inverted in 2022, you were early, markets rallied sharply in 2023 and 2024 despite the inversion. If you then relaxed as the curve started uninverting, you may have missed the lagged damage that was still accruing.
The more useful framing: inversion tells you a recession is being loaded. Uninversion tells you it may be about to fire. The window to watch most closely is not the moment of maximum inversion, but the twelve months after the spread crosses back through zero.
And one more thing. The yield curve does not predict the severity of a recession, only its timing. A curve that was inverted for two years, reaching -109bps at its trough, has loaded considerably more credit stress into the system than one that briefly dipped to -20bps. Depth and duration of inversion matter for the scale of subsequent damage. The current cycle's inversion was both deep and long. That context does not disappear simply because the curve is normalising.
Watch the curve. But watch the right part of it.