Inverted Yield Curve: What It Means and Whether It Predicts Recession

A yield curve lines up interest rates by the time until a debt security comes due. This guide focuses on U.S. Treasury yields, which are often used as reference rates because they are marketable obligations of the federal government. A curve can slope upward, remain nearly flat, or slope downward. The word “inverted” describes the last pattern over the maturities being compared: the shorter-maturity yield is higher than the longer-maturity yield.

The curve is not a direct chart of future GDP. It is a set of yields associated with different maturities observed at a particular time. Those yields reflect bond prices and the terms of the securities, along with investors’ expectations and the compensation they require for risks. The curve’s shape can therefore change even when the economy itself has not yet turned into a recession.

The U.S. Treasury’s published curve is a par-yield curve. Treasury derives it from indicative bid-side prices for recently auctioned securities and fits rates between the available maturity points. These are not a list of completed trades, nor does every point represent a specific outstanding bond with exactly that time remaining. That matters when a chart is treated as a precise forecast rather than a constructed market reference. Treasury describes its inputs and fitting method in its [yield-curve methodology]({source:treasuryYieldCurveMethodology}); the [daily rate table]({source:treasuryDailyParYieldCurve}) publishes the resulting maturity estimates.

Use the order of subtraction shown by the series name. A common convention is:

Term spread = longer-maturity Treasury yield − shorter-maturity Treasury yield

Under this convention, a positive result means the longer yield is higher, zero means the two yields are equal, and a negative result means the curve is inverted between those points. Some tables define a spread in the opposite order, so check the labels and formula rather than inferring the sign from the word “spread.”

Consider two made-up sets of yields observed on the same hypothetical date:

The calculations are 4.00% − 4.40% = −0.40 percentage points and 4.00% − 4.80% = −0.80 percentage points. One percentage point equals 100 basis points, so −0.40 percentage points is −40 basis points. These inputs are illustrative only; they are not current Treasury rates, a forecast, or a recommendation.

The two spreads are not interchangeable. They share a 10-year yield but use different short maturities. In this example the 3-month yield is higher than the 2-year yield, so the 10-year/3-month spread is more negative. A different set of yields could produce a different gap or even a different sign for one comparison. “The yield curve is inverted” is incomplete unless the maturities, source, date, and calculation are clear.

The short end of the Treasury curve is sensitive to current monetary policy and expectations for near-term policy rates. A 10-year yield, by contrast, relates to a much longer horizon. In a simplified expectations framework, longer yields reflect the expected average path of future short-term rates plus a term premium. The term premium is compensation investors may require for holding a longer-maturity bond while rates, inflation, and bond prices can change.

If short rates are high today but investors expect them to be lower over much of the coming years, the longer yield can sit below the short yield. Expectations of future rate cuts may be connected to lower inflation, slower growth, or an anticipated policy response, but the yield curve alone does not reveal which explanation dominates. The same observed spread can be consistent with more than one outlook.

The expected path is not the whole story. Treasury supply, central-bank holdings, demand from domestic and foreign investors, safe-haven flows, liquidity, inflation uncertainty, and risk appetite can affect yields and the term premium. Term-premium estimates are not directly observed; models infer them from assumptions and data. The Federal Reserve Board’s [term-structure model documentation]({source:federalReserveTermStructureModel}) explains that longer yields combine expectations and term premiums, and cautions that model-based expectations and premiums are approximate. It would therefore be too strong to translate a negative spread directly into “the market expects a recession.”

The 10-year/2-year spread compares a long yield with a short-to-intermediate yield. The 2-year rate can reflect expectations about the policy path over more than one meeting or quarter. The 10-year/3-month spread compares the long rate with a rate closer to the current short-rate environment. It is the specific term spread used in the New York Fed’s recession-probability model.

That model applies the 10-year minus 3-month spread to estimate a probability of recession twelve months ahead. The New York Fed updates the series monthly and states that the estimate is not an official forecast of the Federal Reserve Bank of New York, the Federal Reserve System, or the FOMC. A model probability summarizes an estimated historical relationship under a chosen specification; it is not a date on a calendar or a guarantee about the next year. Read the [New York Fed indicator and its explanation]({source:newYorkFedYieldCurveLeadingIndicator}) before comparing its output with a 2-year spread chart.

Data frequency also matters. FRED’s [Treasury spread series notes]({source:fred10YearTreasurySpreads}) describe the 10-year/2-year series as daily and the 10-year/3-month series shown there as a monthly average. A daily spread can cross zero briefly while a monthly-average spread remains positive, or a monthly average can stay negative after some daily readings have moved back above zero. That is not necessarily a contradiction: the observations summarize different frequencies and maturity pairs.

There is no single spread definition that answers every economic question. If you use an official model, reproduce its exact maturities, source series, averaging convention, and horizon. If you are comparing two charts, label those choices so a 10-year/2-year move is not mistaken for the New York Fed’s 10-year/3-month input.

An inverted curve can summarize expectations that short-term rates will eventually fall from a high level. If investors expect weaker activity or lower inflation, they may expect monetary policy to ease, which can lower longer yields relative to short yields. Tight current policy can also make short borrowing rates high while longer yields incorporate a different view of the years ahead. These channels offer an economic reason why the curve may contain information about later activity.

The relationship is predictive, not a mechanical cause. An inversion does not itself make businesses cut production or households reduce spending. Nor does a yield-curve model identify the shock that will determine output, employment, or credit conditions. It measures a pattern in rates that has been associated with later recessions in historical data; changes in market structure, policy, inflation risk, or investor demand can alter how the pattern should be interpreted.

Timing varies. A spread can become negative well before an officially dated downturn; it can also return above zero before the economy contracts. The “un-inversion” can happen because short rates fall, long rates rise, or both move. Looking only at whether the spread is negative today misses the path, the underlying rates, and the time horizon of the question. A model probability, a chart threshold, and a recession date are three different things.

A negative spread does not say that a recession is certain, that it will begin in exactly twelve months, how deep it would be, or which industries and households would be affected. The New York Fed’s horizon is a model design choice, not a universal lead time for every inversion. A 2-year/10-year spread does not automatically inherit the 10-year/3-month model’s probability because the two use different inputs.

It also does not settle the official recession chronology in real time. The National Bureau of Economic Research’s Business Cycle Dating Committee defines a recession as a significant decline in economic activity spread across the economy and lasting more than a few months. It weighs depth, diffusion, and duration across multiple indicators and dates turning points retrospectively. Two consecutive quarters of falling real GDP are not a fixed NBER rule. See the NBER’s [business-cycle dating explanation]({source:nberBusinessCycleDating}) for its criteria and timing.

An inversion is also not a personal financial instruction. It does not specify whether an investor should buy or sell a bond, whether a business should borrow, or whether a household should delay a major purchase. Those choices depend on the person’s goals, cash needs, risk tolerance, and the terms available. The spread can frame one macroeconomic question; it cannot replace an analysis of an individual decision.

Start by checking the exact maturities and subtraction order. Confirm that the figures are Treasury yields from the same date and curve family, and whether a point is a par yield, constant-maturity estimate, or yield on a particular security. Do not assume that labels such as “10-year rate” always mean the yield of one identical bond; constant-maturity rates are interpolated to fixed tenors.

Next, check the observation frequency. A daily 10-year/2-year series, a monthly average, and a 12-month-ahead model probability answer different questions. Make sure units are percentage points or basis points, and distinguish a negative spread from a negative interest rate. In the example above, both yields are positive even though their difference is negative.

Then inspect both underlying yields, not just the gap. The spread can narrow because the short rate rises, the long rate falls, or both change. It can widen after an inversion because short rates fall, long rates rise, or the two move at different speeds. Those paths carry different information about the policy outlook and bond-market pricing. A single line showing the difference hides that decomposition.

Finally, compare the curve with evidence from the real economy and use a clearly named forecast horizon. Employment, income, spending, output, credit, and inflation measures provide separate information; none alone proves that a recession has started. For context, compare U-3 and U-6 labor-market measures, nominal and real GDP, and the CPI, PCE, and GDP deflator. Those indicators describe different parts of the economy and should not be collapsed into one yield-curve signal.

When you report an inverted curve, state the date, maturities, source series, frequency, and spread formula. For example: “The hypothetical daily 10-year-minus-2-year spread is −40 basis points: the 10-year yield is 4.00% and the 2-year yield is 4.40%.” That sentence makes the sign and the underlying rates auditable. If you use the New York Fed’s model, identify it separately as a monthly 10-year/3-month model estimate for the probability of a recession twelve months ahead.

Then say what the signal can support: a negative spread is a market-based indicator worth comparing with other evidence. State what it cannot support: certainty, a specific start date, a causal explanation, or a personal trading decision. This keeps the curve in its proper role as one clue about expectations and risk compensation rather than a standalone verdict about the economy.