The yield curve: the chart that predicts recessions
The answer
The yield curve is a chart of what governments pay to borrow for different lengths of time, from a few months to 30 years. Normally longer borrowing costs more. When short-term borrowing costs more than long-term borrowing, the curve is inverted, and in the United States that has come before almost every recession for 50 years.
Why this costs you money
An investor reads a headline saying the yield curve has inverted and a recession is coming. They sell most of their shares. Two things then go wrong.
The first is timing. The lead time between a US inversion and the recession that followed has ranged from roughly 10 months to roughly 22 months. That is not a signal you can trade on. An investor who sold in February 2006, when the curve first inverted, sat out a further 18 months of a rising market before the recession began in December 2007.
The second is worse. The signal has already failed once. The US curve was inverted from July 2022 until August 2024, 26 months, the longest inversion in the Federal Reserve's 10-year-minus-2-year series. It reached a low of about −108 basis points. No US recession followed. Anyone who treated the inversion as a decision rule rather than as information spent 2 years out of a market that rose.
The expensive mistake is not ignoring the yield curve. It is treating a probability as a schedule.
There is a smaller, more immediate cost that applies to almost every reader. When the curve is inverted, short-dated instruments pay more than long ones. Savers who lock money into long deposits or long bond funds at that moment accept less yield in exchange for more price risk. That is the wrong trade in both directions, and it is made every day by people who have never looked at the curve.
How it works
Plot the yield on government bonds against the time to maturity. The horizontal axis runs from 3 months to 30 years. The vertical axis is the yield. Join the points and you have the yield curve.
The curve has 3 recognised shapes.
Normal, or upward sloping. Long bonds yield more than short bonds. This is the usual state. A lender parting with money for 20 years faces more inflation uncertainty than one parting with it for 3 months, and demands more. That extra payment is the term premium.
Flat. Short and long yields are close together. The market is unsure. This often appears on the way from one shape to another.
Inverted, or downward sloping. Short bonds yield more than long bonds.
An inverted curve is 2 statements at once. The short end is high because the central bank has raised its policy rate to slow the economy, usually to control inflation. Short yields track the policy rate closely. The long end is low because investors expect the policy rate to be lower in future than it is today. A 10-year yield is, roughly, the market's average expectation of short rates over the next 10 years plus a term premium. If that average is below today's rate, the market is saying rates will have to come down.
Central banks cut rates for one main reason. So an inverted curve is the bond market saying: policy is tight enough that it will have to be loosened, and the usual reason for loosening is a weakening economy.
It is also a mechanism, not only a forecast. Banks borrow short and lend long, so when short rates exceed long rates they lend less. The curve is partly prediction, partly cause.
Two spreads are quoted, and they are not interchangeable.
- 10-year minus 2-year. The most widely quoted. Watch this one for the headlines.
- 10-year minus 3-month. The Federal Reserve Bank of New York's recession probability model uses this one, and several researchers consider it the better predictor.
They can disagree for months. When somebody says "the curve inverted", the first question is which curve.
What it tells you, and what it does not
The curve tells you what the bond market currently expects about future interest rates. That is valuable, because the bond market is large, professional and puts money behind its expectations.
It does not tell you the bond market is correct. The 2022 to 2024 inversion is the standing proof. It does not tell you when: lead times have run between roughly 10 and 22 months, and a signal with a 12-month range is not a timing tool. And it does not tell you what shares will do — stock markets have frequently risen substantially between an inversion and the recession that followed.
The sample is also small. Since 1976 there have been about 6 clear inversion episodes in the US 10-year-minus-2-year series, and 1 of those 6 has already failed.
The decision rule
Read the curve as a description of what the market expects, and use it for 2 concrete decisions rather than for a market call.
Decision 1: where to hold your cash. When the curve is inverted, the short end pays more. Take the higher yield at the short end and take less price risk while doing it. When the curve is steep, going longer is genuinely paid for.
Decision 2: how much cycle risk you want. A sustained inversion is a reason to check how much of your portfolio depends on the economy growing, not a reason to sell it.
Unless the inversion is driven by something other than expected rate cuts — heavy central bank bond buying, a foreign demand surge for long bonds, or a supply shock at the short end. Then the signal is contaminated and should be weighted down.
The rule to avoid is the obvious one. "The curve inverted, therefore sell" has failed once in a costly way and has never given a usable date.
Try this now
Three minutes. You will build the curve for your own market from numbers you look up yourself.
- Search for "India government bond yields" or "US Treasury yield curve".
- Write down 4 numbers: the 3-month, the 2-year, the 10-year and the 30-year yields.
- Subtract:
10-year − 2-year. Then10-year − 3-month. - Read the signs. Both positive is a normal curve. Both near zero is flat. Either one negative means that part of the curve is inverted.
- Compare the 2-year yield with the central bank's policy rate — the repo rate in India, the federal funds target in the US. If the 2-year is well below the policy rate, the market expects cuts. If above, it expects rises.
- Look up the yield to maturity on the debt fund you hold, and see which part of the curve you are being paid from.
What you should see. In most periods all your subtractions come out positive and the numbers rise from 3 months to 30 years. That is a normal curve and it is the unremarkable state of the world.
As a reference point: on 1 September 2026 the US 10-year minus 2-year spread was about +0.40%, a normal, modestly upward-sloping curve. The India 10-year yield was about 6.99% on 2 September 2026 against an RBI repo rate of 5.25% .
Step 5 shows something most people never see. The 2-year yield is a forecast: the market publishing, free, what it thinks the central bank will do over 2 years.
Three real cases
1. The United States, February 2006 to March 2007 — the signal working, slowly The 10-year-minus-2-year spread went negative in February 2006 and stayed negative for roughly 14 months. The Federal Reserve had raised rates repeatedly to cool a housing boom. The recession the NBER later dated began in December 2007, about 22 months after the inversion started, and ran to June 2009. Anyone who read the inversion as an instruction to sell in early 2006 was right about the outcome and lost roughly 18 months of returns getting there.
2. The United States, 14 August 2019 — the day the signal made headlines The 2-year Treasury yield rose above the 10-year for the first time since 2007. The Dow Jones Industrial Average fell about 800 points, more than 3%, in its worst day of that year. Commentators declared a recession was coming. One did arrive, in February 2020, but its cause was a pandemic nobody in August 2019 was pricing. This case is genuinely ambiguous: the signal was followed by a recession, and the recession had a cause the signal knew nothing about.
3. The United States, July 2022 to August 2024 — the signal failing The curve inverted in July 2022 and stayed inverted for about 26 months, the longest episode in the Federal Reserve's 10-year-minus-2-year series, reaching a low of about −108 basis points. No US recession followed. Real GDP grew about 2.9% in 2023 and above 3% annualised in parts of 2024. This is the first sustained inversion in the modern series not followed by a recession, and any honest treatment of the yield curve has to put it beside the successes.
The question that resolves it
The curve inverts. A novice asks: is a recession coming?
An expert asks: why is the long end low — because the market expects rate cuts, or because somebody large is buying long bonds for reasons unconnected to the economy?
Those 2 causes produce an identical-looking chart and carry opposite information. If long yields are low because investors expect cuts in 18 months, the signal is meaningful. If they are low because a central bank is buying, or because foreign reserve managers need long-dated assets, the shape is a supply and demand outcome and says very little about growth.
What would make this wrong
If the yield curve were a reliable recession predictor, no sustained inversion would pass without a recession. One already has, from 2022 to 2024: a documented failure in the signal's home market, in its most-watched form, at its longest duration on record.
The other honest limits are 3. The sample is tiny — roughly 6 episodes since 1976, so any rule built on it should be held loosely. It is a US result, and it does not transfer automatically. And the mechanism can be broken by policy: years of central bank bond buying have suppressed the term premium, and a suppressed term premium can make the curve appear inverted without the market forecasting anything.
In India
The Indian G-Sec curve is published daily, but it does not carry the same meaning, and pretending otherwise is a common error in Indian financial commentary.
The Indian curve is almost always upward sloping, often steeply. Sustained inversions of the kind seen in the US are rare, for 3 structural reasons. Government supply at the long end is heavy and continuous, which keeps long yields high. Demand at the long end is regulated rather than tactical, because insurance companies and pension funds are required to hold long government bonds. And India's higher inflation and growth path keeps the term premium wide.
The RBI's Financial Stability Report of 31 December 2025 described exactly this pattern: short yields falling with rate cuts and surplus liquidity, while long yields stayed elevated on heavy issuance, widening the term spread.
The practical Indian version of the question is different. Instead of "has it inverted", ask "is the curve steep or flat?" A steep curve means you are paid meaningfully to lend longer. A flat curve means you are not, and short duration funds become the sensible place to be. That decision is available in every Indian market condition.
In the United States
The US Treasury publishes a daily par yield curve for every maturity from 1 month to 30 years. It is free, official and the reference point for the global bond market.
Two spreads dominate. T10Y2Y, the 10-year minus 2-year, is what most headlines mean. T10Y3M, the 10-year minus 3-month, is used in the Federal Reserve Bank of New York's published recession probability model, which converts the current spread into a percentage probability of recession within 12 months and is available to anybody.
The US curve carries information the Indian curve does not. The US Treasury market is the deepest and most liquid in the world, and every maturity trades continuously in both directions, so the curve is a real expectations survey with money attached. American investors act on it directly: when short-term Treasuries pay more than long ones, US money market funds absorb hundreds of billions of dollars, which shows up in published fund flow data.
Where they differ, and what that tells you
The US curve is a forecast. The Indian curve is largely a supply and demand picture.
That is not a matter of sophistication. It follows from 2 structural facts. The Federal Reserve moves its policy rate frequently and in both directions, so short yields carry a lot of information. And the US Treasury market is liquid at every maturity, so long yields are a genuine consensus.
In India, the RBI moves less often, the government's borrowing programme dominates the long end, and much long-end demand is regulatory rather than opinion-driven. An Indian long yield tells you as much about the fiscal deficit and insurance rules as about growth.
What that tells you is a direct instruction: do not import the American rule. Waiting for an Indian yield curve inversion as a recession signal means waiting for something that may never arrive, and would probably mean something different if it did.
The transferable part is smaller and more useful. In both countries, the shape of the curve tells you whether you are being paid to take duration risk. A steep curve pays you to lend longer. A flat or inverted curve does not, and then the correct response is to shorten, take the higher short yield, and hold less price risk. That works in Mumbai and in New York, and needs no recession forecast to be useful.
Carry this
- The yield curve is the price of time. Normally longer costs more.
- Inverted means the market expects the central bank to cut, which usually means trouble. It does not say when, and in 2022 to 2024 it was wrong.
- Use the shape to decide whether you are paid to lend longer. That decision is available every day.