The Yield Curve as a Recession Signal: What It Says and What It Doesn't

An inverted yield curve, in which short-term Treasury yields rise above long-term ones, preceded every US recession from 1969 to 2020, which is why the 10-year minus 3-month and 10-year minus 2-year spreads are among the most watched recession indicators. But an inversion is a statement about expected interest rates, not a countdown: the lag to recession has ranged from about six months to two years, and the long inversion of 2022–2024 was not followed by a recession within that window. The curve is best used as one input to how much risk a portfolio carries, not as a signal to trade on.

What is an inverted yield curve?

The yield curve plots Treasury yields against maturity, from bills maturing in weeks to 30-year bonds. It usually slopes upward, because investors want extra compensation for lending longer, and it inverts when short-term yields rise above long-term ones. Two spreads summarize the slope, both published daily on FRED:

• 10-year minus 3-month (T10Y3M). The research benchmark, popularized by Estrella and Mishkin (1996) at the New York Fed and the input to that bank's recession-probability model. The 3-month bill tracks the Fed's policy rate, so this spread compares long yields with today's policy stance.
• 10-year minus 2-year (T10Y2Y). The market's favorite. The 2-year yield embeds the expected path of policy over two years, so this spread inverts when markets expect lower rates ahead.

Estrella and Trubin (2006) found that the 2-year version tends to invert earlier and more often. Whichever you choose, use it consistently.

How to compute the spread: a worked example

Suppose these illustrative yields, all on a bond-equivalent basis, prevail on a given day; they are not market data.

Each spread is the long yield minus the short one, in percentage points:

$$S_{\text{10y-3m}} = 4.20 - 5.40 = -1.20 \qquad S_{\text{10y-2y}} = 4.20 - 4.90 = -0.70$$

The curve is inverted on both measures: by 120 basis points (bp) against the bill and 70 bp against the 2-year note.

A spread becomes a probability only through a model. Estrella and Trubin's probit model, of the type the New York Fed publishes, maps the monthly average 10-year minus 3-month spread $S$ to the probability of being in recession twelve months later:

$$P = \Phi(-0.6045 - 0.7374\,S)$$

where $\Phi$ is the standard normal distribution function and the coefficients were estimated on 1959–2005 data. At the illustrative $S = -1.20$, the model gives about 61%; a flat curve gives about 27%, and a spread of $+1.00$ about 9%. The signal is continuous, so zero is a convention rather than a cliff, and the output is a probability, not a date.

Why does an inverted curve tend to precede recessions?

A long-term yield is roughly the average short rate investors expect over the bond's life plus a term premium for bearing interest-rate risk:

$$y_t^{(n)} \approx \frac{1}{n}\sum_{i=0}^{n-1} \mathbb{E}_t\left[r_{t+i}\right] + TP_t^{(n)}$$

Expected policy. When investors expect the Fed to cut, expected short rates sit below today's and pull long yields under short ones, and the Fed has usually cut hardest when the economy weakens. Tightening also works directly, raising short rates more than long ones while slowing the economy: before each of the six recessions from 1969 to 2001, Estrella and Trubin found, the 3-month rate had risen in the run-up to the signal. As Bauer and Mertens (2018b) note, causation can also run the other way: fear of a downturn can push investors into long Treasuries, lowering long yields.

The term premium. A common argument holds that quantitative easing depressed long yields through the term premium, so an inverted curve would overstate recession risk. Bauer and Mertens tested this: subtracting a model estimate of the term premium made the 10-year minus 3-month spread a worse predictor, not a better one.

The near term. Engstrom and Sharpe (2018) argued that the signal comes from expected policy over the next year and a half. Their near-term forward spread, the 3-month rate implied six quarters ahead minus today's 3-month yield, outperformed every long-term spread they tested. Bauer and Mertens confirmed that it dominates the 10-year minus 2-year spread but found the 10-year minus 3-month spread at least as reliable.

How long after an inversion does a recession start?

The record is strong but thin. By the 10-year minus 3-month measure, an inversion preceded each of the eight US recessions from 1969 to 2020. Bauer and Mertens (2018a), using the 10-year minus 1-year spread, found a single false positive since 1955, in the mid-1960s, and lead times from 6 to 24 months. Even the hits need care: few would credit the 2019 inversion with forecasting the pandemic that caused the 2020 recession.

Even shallow inversions have counted:

The curve also tends to re-steepen before the downturn arrives, as short yields fall in anticipation of rate cuts: in 1990, 2001 and 2007, the 10-year minus 3-month spread was positive again by the time the recession began. And the NBER dates recessions months after they start, so the signal cannot be confirmed in real time.

False signals and the 2022–2024 inversion

To fight the highest inflation in four decades, the Fed raised its policy rate from near zero in March 2022 to 5.25–5.50% by July 2023, and the curve inverted on all three measures:

• The 10-year minus 2-year spread, from July 2022 (after a brief dip that April) to early September 2024.
• The 10-year minus 3-month spread, from late October 2022 to mid-December 2024.
• The near-term forward spread, from November 2022; Reuters reported it near −170 bp in April 2023.

Yet the economy kept growing through 2023 and 2024, unemployment stayed low by historical standards, and the NBER has identified no recession in that period. None began within 24 months of either long-term inversion, the outer edge of the historical range.

Three explanations stand out. Rate cuts can follow falling inflation as well as a weakening economy, and the Fed began cutting in September 2024 with the economy still growing. Many households and companies had locked in low fixed rates in 2020 and 2021, which slowed the pass-through of higher short rates. Some argued that the recession was delayed rather than avoided, a defense that weakens with every passing month.

Dismissing the signal has its own record. In March 2006, Fed Chair Ben Bernanke argued that, to the extent the curve's flattening or inversion reflected a smaller term premium, its implications for future activity were positive rather than negative. The recession that began in December 2007 was the deepest since the 1930s. Arguments that this time is different were wrong in 2006 and, within the historical window, right in 2022–2024. The curve deserves neither blind faith nor quick dismissal.

How should an investor use the yield curve?

The steps below treat the curve as one input to a risk process; they are educational, not a recommendation to buy or sell any security.

1. Translate the spread into a probability with a published model, such as the New York Fed's monthly estimate, rather than treating zero as a switch.
2. Diagnose the inversion. Is it driven by rising short rates (tight policy), falling long rates (a growth scare or flight to safety) or a low term premium? Do near-term forwards and fed funds futures price cuts within about 18 months?
3. Look for confirmation in the labor market, credit spreads and financial conditions. Our Signals dashboard shows the 10-year minus 2-year spread beside Treasury yields, the fed funds rate, unemployment and the Chicago Fed's financial conditions index, all from FRED.
4. Adjust at the margin, within limits set in advance. A rising probability is a reason to review duration, credit quality, liquidity, leverage and stress tests, not to abandon a long-term allocation.
5. Keep watching after it un-inverts, since the re-steepening has often come close to the recession.

Do not sell everything on the first inversion: stocks have often kept rising for months, and the S&P 500 set new records after the inversions of 2006, 2019 and 2022. Do not trade daily blips, switch spreads until one agrees with you, or read depth as a timetable. The most robust use of any recession signal is to need it less, with a portfolio built to hold up across environments, as described in Four Economic Seasons.

Limitations

• A small sample. Eight recessions is little evidence, and a rule judged with hindsight on so few events can look better than it is, a trap described in How to Read a Backtest Like a Skeptic.
• A changing bond market. Central-bank balance sheets and shifts in who holds Treasuries can move long yields for reasons unrelated to growth, and with short rates near zero, as in 2009–2015, the curve could hardly invert.
• Blind spots. The curve cannot see pandemics, wars or financial accidents, and it says little about a recession's severity or how markets will respond.

Key takeaways

• On the 10-year minus 3-month measure, an inversion preceded every US recession from 1969 to 2020, with lead times of roughly 6 to 24 months.
• It works largely through expected monetary policy and gives a probability, not a date.
• The 2022–2024 inversion lasted about two years without a recession inside the historical window.
• Use the curve to shape risk at the margin, confirmed by other data, not to time markets.

Sources

• Barbuscia, D. (2023). "'Powell's curve' plunges to new lows, flashing US recession warning." Reuters, April 6.
• Bauer, M. D., and Mertens, T. M. (2018a). "Economic Forecasts with the Yield Curve." FRBSF Economic Letter 2018-07, March 5.
• Bauer, M. D., and Mertens, T. M. (2018b). "Information in the Yield Curve about Future Recessions." FRBSF Economic Letter 2018-20, August 27.
• Bernanke, B. S. (2006). "Reflections on the Yield Curve and Monetary Policy." Remarks before the Economic Club of New York, March 20.
• Engstrom, E. C., and Sharpe, S. A. (2018). "(Don't Fear) The Yield Curve." FEDS Notes, Federal Reserve Board, June 28.
• Estrella, A., and Mishkin, F. S. (1996). "The Yield Curve as a Predictor of U.S. Recessions." Federal Reserve Bank of New York, Current Issues in Economics and Finance 2(7).
• Estrella, A., and Trubin, M. R. (2006). "The Yield Curve as a Leading Indicator: Some Practical Issues." Federal Reserve Bank of New York, Current Issues in Economics and Finance 12(5).
• Federal Reserve Bank of New York. "The Yield Curve as a Leading Indicator." Recession-probability model.
• Federal Reserve Bank of St. Louis, FRED. Series T10Y3M and T10Y2Y.