Yields & Spreads

The 2s10s Spread

The 2s10s spread is the difference between 10-year and 2-year US Treasury yields. When it goes negative — the curve inverts — a US recession has historically followed within 6 to 18 months. It's one of the most studied single-indicator recession signals in modern macro, and the 2022 inversion is the longest on record.

May 13, 2026T10Y2Yyields · recession-indicators · fed-policy · term-structure

SPREAD_2S10S

—

For most of post-war US macroeconomic history, this single number — the gap between two Treasury yields — has been the bond market's most reliable signal that a recession is coming. Not perfectly timed, not infinitely reliable, but more accurate than just about any other one-line indicator economists track. The 2022 inversion is the longest on record. Whether the 2s10s is still a working indicator is one of the most consequential open questions in modern macro.

SPREAD_2S10S

—

What it measures

The 2s10s spread is the arithmetic difference between two yields:

Both legs are constant-maturity Treasury yields published daily by the Federal Reserve, derived from the Treasury's own yield curve fit. The St. Louis Fed publishes the resulting spread directly as series T10Y2Y — daily frequency, denominated in percentage points.

When the spread is positive, the 10-year yields more than the 2-year — the "normal" shape of the curve. When negative, the curve is inverted — the 2-year yields more than the 10-year. Zero is a flat curve, the threshold between the two regimes.

Why it matters

There are two distinct reasons to care.

The forecasting one. Every US recession since 1969 has been preceded by a 2s10s inversion. The lead time has typically been 6 to 18 months. This isn't a structural model — it's a stylized historical fact — but it's stable enough that the New York Fed, the Cleveland Fed, the Conference Board, and most major investment banks all build it into their recession probability dashboards. (The NY Fed's preferred specification uses the 3M-10Y spread, which behaves similarly but inverts slightly later.)

The plumbing one. Banks fund themselves on the short end of the curve — deposits, repo, money markets — and lend on the long end — mortgages, commercial loans, business credit. Their net interest margin, the gap between what they pay and what they earn, depends on the slope of the yield curve. When the curve inverts, banking profitability comes under pressure, lending standards tighten, and the supply of credit to the real economy contracts. The 2s10s is therefore not just a forecast; it's a partial cause of the slowdown it predicts.

What moves it, and what it moves

Moves it:

Fed policy expectations. The 2-year yield is dominated by the expected path of the Federal Funds Rate over the next two years. Hawkish surprises push the 2-year up; dovish surprises push it down.
Growth and inflation expectations. The 10-year yield embeds expected short rates further out plus a term premium. Weakening growth or anchoring inflation pulls the 10-year down.
Term premium itself. The compensation investors demand for holding duration risk. QE compresses it; QT pushes it back out. After a decade of QE, term premia have been unusually low — by some estimates negative for extended periods.
Foreign demand. When central banks abroad accumulate dollar reserves, they tend to buy the long end of the curve, suppressing the 10-year yield.

It moves:

Bank net interest margins, and from there, bank lending appetite.
The 30-year mortgage rate, which tracks the 10-year yield with a roughly stable spread overlay.
Recession probability models, which then feed into Fed policy, corporate hiring plans, and household sentiment.
Equity sector rotation — financials suffer in inverted regimes; defensive sectors typically outperform.
Insurance and pension fund liability valuations, which discount future cash flows along the entire curve.

A worked example: the 2006–2008 inversion

The 2s10s first kissed zero in late January 2006, then re-inverted persistently from August 2006 onward, with both 2-year and 10-year yields hovering in the high 4% to low 5% range. By early 2007 the inversion had widened to roughly −0.20%.

The spread remained inverted or flat through the summer of 2007. In July 2007, two Bear Stearns hedge funds tied to subprime mortgage securities collapsed. In August 2007, BNP Paribas froze withdrawals from three funds — the textbook start of the credit crisis. By October 2007 the S&P 500 had peaked. The NBER would later date the official start of the recession to December 2007 — roughly seventeen months after the persistent inversion began.

By the time Lehman Brothers failed in September 2008, the 2s10s had already re-steepened sharply (the 2-year had collapsed as the Fed cut rates toward zero). The indicator had done its forecasting work a year earlier; the steepening on the way out was a separate signal about how the credit cycle was unwinding.

The current cycle, and the open question

The 2s10s inverted again on July 5, 2022, after the Fed began raising rates aggressively to fight post-pandemic inflation. By July 2023 the spread had reached −1.08% — the deepest inversion since 1981. It remained negative for over two years before un-inverting in late 2024.

As of mid-2026, no NBER-recognized recession has followed. Several explanations are circulating:

The post-QE term premium is structurally lower than the historical norm, distorting the long end and making the curve look more pessimistic than the economy actually is.
Massive fiscal support during COVID and a historically strong consumer balance sheet have delayed the credit-tightening transmission to a degree the historical data didn't anticipate.
The recession was averted by an unusual combination of immigration-driven labor supply, productivity gains from AI investment, and the Fed's bank rescue operations after the regional bank failures of 2023.
It's still coming — just later than any prior cycle, and possibly as a fresh inversion rather than a delayed effect of the 2022 one.

The indicator hasn't been falsified. But its lead time, which used to be measured in months, may now need to be measured in years. Watching whether the next US recession (whenever it arrives) is preceded by a new inversion or whether the 2022 reading turns out to be the prefiguring signal is one of the more consequential open questions in macro right now.

FAQ

How long after the 2s10s inverts does a recession typically follow?

Between 6 and 18 months, historically. The 1989 inversion led the 1990 recession by 13 months. The 2000 inversion led the 2001 recession by 13 months. The 2006 inversion led the December 2007 recession by roughly 17 months. The 2019 inversion led the COVID recession by 6 months (though the pandemic muddies attribution). The 2022 inversion has now persisted for over two years without an officially declared recession, raising real questions about whether the lead time is now measured in years rather than months.

Are all inversions followed by recessions?

Every US recession since 1969 has been preceded by a 2s10s inversion, but not every inversion has been cleanly followed by a recession. The 1998 inversion around the LTCM crisis was brief and did not precede a recession. The current 2022 inversion is the longest on record and remains contested. The base rate of false signals is low but not zero, and the size and persistence of the inversion matter more than the bare fact of crossing zero.

What's the difference between 2s10s and the 3M-10Y spread?

The 3M-10Y spread (FRED series T10Y3M) compares the 10-year yield to the 3-month T-bill yield. The New York Fed's official recession probability model uses this spread, not 2s10s. The 3M-10Y tends to invert later than 2s10s but with a stronger predictive signal in formal models. Most market participants watch 2s10s because both legs are popular benchmarks for the bond market and the data is updated daily by the Fed without lag.

Why does the yield curve normally slope upward?

Investors demand extra yield — a 'term premium' — for locking up money in longer-maturity bonds, since they take on more interest-rate risk, inflation risk, and opportunity cost over a longer horizon. Under normal conditions, a 10-year Treasury should yield more than a 2-year Treasury. Inversion happens when those normal incentives are overwhelmed by expectations that future short rates will be lower than current ones — which usually means the market expects the Fed to cut.

Is the 2s10s still a reliable recession indicator?

Genuinely debated. The 2022 inversion has lasted longer than any prior inversion on record without a recession being declared by the NBER. Some economists argue that years of quantitative easing distorted long-end yields and broke the historical relationship. Others argue the recession is still coming, just delayed by unprecedented fiscal support and consumer balance sheet strength. As of mid-2026, the question is open and the indicator has not been formally falsified — its lead time may simply have lengthened.