A yield curve represents the relationship, at a particular time, between the yields on debt instruments and their remaining maturities. Its horizontal axis measures time to repayment, while its vertical axis measures an annualized interest rate. A meaningful curve generally compares instruments denominated in the same currency and having similar credit quality and contractual characteristics. Government bond curves are especially prominent, but separate curves exist for corporate borrowing, interbank transactions, and derivatives.
Yield curves connect the prices of securities traded today with the valuation of payments occurring in the future. They provide benchmarks for issuing bonds, pricing loans, valuing derivatives, and managing financial institutions’ exposures. Their shape also contains information about expected monetary policy, inflation, economic activity, and the compensation investors require for bearing risk.
There is no single curve that fully describes an economy’s borrowing conditions. A coupon-bond yield curve, a zero-coupon curve, and a forward-rate curve express different aspects of the same underlying market, while sovereign, swap, and corporate curves incorporate different risks. Understanding these distinctions is essential when interpreting a steepening curve, an inversion, or a change in long-term yields.
Yield concepts and the instruments behind a curve
The word “yield” can refer to several different quantities. A bond’s current yield is its annual coupon payment divided by its market price, but this ignores capital gains or losses at repayment and the timing of intermediate payments. The more widely used yield to maturity is the single discount rate that equates a bond’s promised payments with its price. It is an internal rate of return on those promised cash flows, not a guarantee of the return an investor will ultimately realize.
For a bond priced at , with payments at times , its annually compounded yield satisfies
The last payment includes repayment of principal. Actual quotation conventions vary: many markets use semiannual compounding, and short-term instruments may use simple-interest or discount-rate conventions. Consequently, rates cannot always be compared directly without conversion. Realized holding-period returns also depend on the price at which a bond is sold, whether promised payments occur, and the rates available for reinvesting coupons.
A zero-coupon bond makes only one payment at maturity. Its yield therefore isolates the price of transferring money between the present and a particular future date. A zero-coupon, or spot-rate, curve is more suitable for valuing separate future cash flows than a curve of coupon-bond yields. Two coupon bonds with identical final maturities can have different yields because their coupons distribute payments differently across the underlying spot-rate curve.
A par yield is different again: it is the coupon rate that would make a hypothetical bond trade at its face value. Published government curves often report fitted par yields at standardized maturities rather than yields on particular outstanding securities. A forward curve describes rates implied today for periods beginning in the future. Each representation is useful, but they should not be treated as numerically interchangeable.
The choice of instruments is equally important. Treasury bills, government notes, inflation-linked bonds, corporate debt, and swap contracts do not promise identical economic payments. Differences in credit risk, tax treatment, liquidity, collateral arrangements, or embedded options can alter yields independently of maturity. A curve built from heterogeneous securities may therefore reflect changing instrument characteristics rather than a pure term structure of interest rates.
Discount factors, forward rates, and curve construction
The fundamental object underlying a yield curve is the discount factor, : the present value of one unit of currency payable at time , within a specified valuation framework. For deterministic promised cash flows valued using a common curve,
This formulation discounts each payment at its own maturity rather than using one yield for the entire bond. It is the foundation of discounted cash-flow valuation. With continuous compounding, the spot rate is defined by .
Market participants generally observe prices for a finite set of instruments, not a complete sequence of discount factors. Bootstrapping constructs the curve progressively. A short-dated instrument may identify one discount factor directly. A longer-dated coupon bond can then identify another after the values of its earlier coupons have been established. In actual markets, payment dates rarely line up perfectly, so interpolation is usually required alongside the sequential calculation.
For example, suppose an annual-payment bond with face value 100, a coupon of 4, and two years to maturity trades at 100. If the one-year discount factor is 0.97, then
giving . Its two-year annually compounded spot rate is therefore approximately . This example illustrates why a bond’s coupon rate alone cannot determine a spot rate: earlier payments must first be valued.
A forward interest rate follows from the relationship between discount factors. Under an annual-compounding convention, the one-year forward rate beginning one year from now satisfies
The example implies a forward rate of approximately . This is a rate embedded in current relative prices. It need not equal the market’s average forecast of the future one-year spot rate, because risk compensation and other pricing effects may separate the two.
Interpolation methods can materially affect the resulting curve. Straight-line interpolation of yields, interpolation of log discount factors, and spline methods can fit the same quoted instruments yet imply different intermediate forward rates. Curve builders therefore consider pricing accuracy, smoothness, stability, and economically sensible behavior. Arbitrage restrictions matter, but negative interest rates are not automatically violations: discount factors can rise over some intervals when the corresponding forward rates are negative.
Curve shapes and movements
An upward-sloping curve has higher yields at longer maturities. This shape is often called normal, although it is neither universal nor required by financial theory. It may reflect expectations that short-term rates will rise, positive compensation for holding longer-duration securities, or a combination of both. For instance, a three-month yield of 3% and a ten-year yield of 4% describe a positive slope, but those two observations alone do not reveal how expectations and risk premiums contributed to it.
An inverted curve has lower long-term yields than short-term yields. Investors may anticipate future reductions in policy rates, perhaps because inflation is expected to fall or economic activity is expected to weaken. Inversion can also be reinforced by unusually strong demand for long-term safe assets. It does not imply that longer bonds are inherently less risky: their prices can remain substantially more sensitive to changes in market rates.
A flat curve has relatively little variation across the maturities being examined. A humped curve rises and then falls, while a trough-shaped curve falls and then rises. Such descriptions are necessarily approximate because a curve may have several local features. The short end can be inverted even when the very long end slopes upward. A statement that “the curve is inverted” should therefore identify the maturity pair or segment involved.
Changes are commonly expressed in basis points, with one basis point equal to 0.01 percentage point. A parallel shift raises or lowers yields across maturities by roughly the same amount. A steepening increases the difference between long- and short-term yields; a flattening reduces it. Market terminology distinguishes a bull steepening, usually driven by falling short-term yields, from a bear steepening, usually driven by rising long-term yields. “Bull” and “bear” refer here to the general inverse relationship between bond prices and yields.
The slope is often summarized by a spread, such as the ten-year government yield minus the two-year yield, or the ten-year yield minus the three-month rate. These measures can move differently because they sample different parts of the curve. A single spread also discards information about overall yield levels and curvature.
Statistical analysis commonly finds that a small number of factors explain much of the movement in government bond yields. These are often interpreted as level, slope, and curvature. They are useful descriptive categories rather than distinct economic causes: one policy announcement or inflation surprise can affect several factors simultaneously.
Expectations, term premiums, and maturity preferences
A central explanation of the yield curve begins with the alternative ways an investor can lend over a given horizon. An investor could buy a long-term bond or repeatedly reinvest in shorter-term instruments. Under a simplified expectations hypothesis, long-term rates reflect the expected sequence of future short-term rates. A common schematic decomposition is
where is an -period zero-coupon yield, denotes future short-term rates, and is the term premium. The expression is approximate because precise relationships depend on compounding, uncertainty, and the model used.
The term premium compensates investors for the risks associated with holding a longer-term bond rather than rolling over short-term investments. Unexpected inflation can reduce the purchasing power of nominal payments, while unexpected interest-rate increases can cause capital losses. The required compensation depends not simply on how volatile a bond is, but on whether its returns are unfavorable in states of the world when investors particularly value resources.
Term premiums need not always be positive. Long-term government bonds can provide protection against some economic downturns if their prices rise when risky assets fall. Investors may then accept lower expected returns for that protection. Conversely, when inflation shocks make bond prices fall alongside other assets, longer bonds can become less useful as hedges. Changes in these relationships help explain why the same curve slope need not carry the same economic meaning across periods.
Liquidity-preference theories emphasize compensation for committing funds for longer periods. Market-segmentation theories stress that different borrowers and investors operate in different maturity ranges. A pension fund matching payments decades ahead has different objectives from a business managing next month’s payroll. Under preferred-habitat theories, investors favor particular maturities but can be induced to move away from them if the price incentive is sufficient.
These approaches highlight the influence of supply and demand within particular maturity segments. Increased long-bond issuance, pension regulation, reserve accumulation, or central-bank purchases can alter yields without an equivalent change in expected short rates. The effects depend on how readily other investors substitute between maturities.
Consequently, an observed long-term yield is not a transparent forecast. Separating expectations from term premiums requires a model, survey information, or additional identifying assumptions. Different decompositions can agree closely on market prices while assigning substantially different shares of a yield movement to expectations and risk compensation.
Statistical and economic models
Yield-curve modeling serves several distinct purposes: fitting today’s prices, describing historical movements, forecasting future yields, and valuing contingent claims. A method well suited to one task may be less suitable for another. A flexible statistical fit can reproduce current bond prices accurately without supplying an economically coherent model of future rates. Conversely, a parsimonious economic model may support interpretation while leaving small pricing errors.
David Durand’s 1942 study of corporate bond yields was an important early contribution to systematic term-structure measurement. Later research connected bond prices with stochastic interest-rate processes and no-arbitrage valuation. Oldřich Vašíček’s 1977 model represented the short rate as a mean-reverting process. Its tractability made it influential in both research and practice, although its Gaussian specification allows negative rates and its simplest form provides only one source of uncertainty.
The Cox–Ingersoll–Ross model, published in 1985 by John Cox, Jonathan Ingersoll, and Stephen Ross, used a square-root process for the short rate. Under suitable parameter conditions, its short rate remains strictly positive. Models of this kind illustrate a central trade-off: restricting the dynamics produces useful mathematical structure but can limit the ability to fit actual yield movements. One-factor models are especially limited when short-, medium-, and long-term rates move independently.
For fitting observed curves, Charles Nelson and Andrew Siegel introduced a widely used specification in 1987. In one common notation,
Here is maturity, controls the decay pattern, and the coefficients govern long-run level, slope, and a curvature component. Lars Svensson’s 1994 extension added another curvature term, increasing flexibility. Such specifications have been used by central banks and debt-management institutions to summarize bond-market information.
Other approaches focus on forward-rate dynamics or on multiple state variables. The Heath–Jarrow–Morton framework, developed in work published in 1992, specifies the evolution of the forward curve subject to no-arbitrage restrictions. Affine models express bond yields as linear functions of underlying factors and can be extended to include macroeconomic variables.
No model directly observes the “true” expected path of future policy rates or the true term premium. Estimates depend on the sample, assumptions about long-run rates, treatment of measurement error, and restrictions on risk pricing. The low-rate and negative-rate periods following the global financial crisis also exposed difficulties for models designed around persistently positive rates, encouraging greater use of lower-bound and shadow-rate specifications.
Central banks and monetary transmission
A central bank normally exercises its most direct influence over very short-term interest rates. In many operating frameworks, administered rates and market operations keep overnight rates near a policy target. Longer-term yields respond through expectations about future policy decisions and through changes in risk premiums. Thus a policy-rate increase need not raise every point on the curve: long yields can fall if investors interpret the decision as reducing future inflation or increasing the likelihood of later rate cuts.
Forward guidance influences the expected policy path by communicating how decisions may depend on economic conditions or, in some cases, by indicating an intended period of low rates. Its effect depends on credibility and interpretation. A commitment that lowers expected future short rates can reduce longer yields, but a statement revealing unexpectedly weak economic prospects may produce a similar market response through a different informational channel.
Through quantitative easing, central banks purchase securities, often including longer-term government bonds. One channel works through reduced duration risk held by private investors: if investors require compensation to absorb that risk, removing some of it can lower the term premium. Purchases can also signal future policy intentions, alter market liquidity, or affect the relative scarcity of particular securities. These channels are difficult to disentangle empirically because purchase announcements often coincide with other policy measures.
Yield-curve control differs from announcing a purchase quantity. It targets a yield or yield range at specified maturities, with purchases or other operations used to support that objective. The Bank of Japan introduced yield-curve control in September 2016, combining a short-term policy rate with a target for the ten-year Japanese government bond yield around zero. It modified the framework subsequently and ended it in March 2024. The historical example illustrates that a central bank can target more than the overnight segment, although maintaining such a target can require substantial intervention.
The curve is also an important part of the monetary transmission mechanism. Longer-term market rates influence mortgage pricing, business investment, asset valuations, and government financing costs. The effect on a particular borrower depends on credit spreads and contractual features as well as the benchmark curve.
Monetary policy does not determine the curve independently of fiscal conditions or global markets. Changes in expected government borrowing can affect the supply of securities, while foreign demand can influence longer yields. Central-bank communication, fiscal policy, and international capital flows may therefore reinforce or offset one another within different maturity segments.
Economic information and recession prediction
The yield curve receives particular attention as an indicator of the business cycle. In the United States, negative spreads between long-term Treasury yields and short-term rates have preceded several recessions. Research by Arturo Estrella, Frederic Mishkin, and others helped establish the empirical usefulness of term spreads during the late twentieth century. The ten-year–three-month spread became especially prominent in statistical models estimating the probability of a recession several quarters ahead.
Several mechanisms can produce this association. When monetary policy becomes restrictive, short-term rates may rise above levels investors expect to prevail over the longer run. If tightening is expected to slow demand and inflation, investors may anticipate eventual rate cuts, limiting long-term yields. An inversion can therefore reflect the combination of current restraint and expectations of future easing.
The curve may also be associated with the availability of credit. Financial institutions often transform short-term funding into longer-term assets. A narrowing spread can affect some lending incentives, although actual bank profitability depends on deposit pricing, asset repricing, hedging, credit losses, and business mix. It is therefore inaccurate to infer a mechanical reduction in all bank lending from an inverted government curve.
Forecasting models commonly relate a term spread to the probability of a future recession, sometimes using a probit specification. The probability is conditional on an estimated historical relationship; it is not a direct market quotation. Results vary with the spread selected, the forecast horizon, the sample period, and whether additional variables are included. United States recession dates are conventionally identified retrospectively by the National Bureau of Economic Research, rather than by a rule based solely on two quarters of declining output.
The United States curve inverted before the recessions beginning in 2001 and December 2007. It also inverted in 2019 before the recession beginning in February 2020. That last sequence does not establish that bond investors foresaw the COVID-19 pandemic: a forecasting relationship can coincide with a downturn triggered by a shock quite different from the risks originally reflected in prices.
Predictive performance also varies across countries and monetary regimes. Capital controls, managed bond markets, changing term premiums, and unconventional monetary policies can alter the relationship between slope and activity. A low or negative term premium may flatten the curve even when expected short rates do not imply a major slowdown. Conversely, rising long-term risk premiums can keep the curve upward-sloping despite deteriorating prospects. Interpretation therefore benefits from examining inflation expectations, credit conditions, labor-market information, and the estimated components of yields rather than treating inversion as a deterministic signal.
Valuation, hedging, and financial institutions
Yield curves provide the discount rates and forward rates used throughout financial valuation. For a fixed-rate bond, a relevant curve values each coupon and the principal repayment. For a floating-rate instrument, forward rates help project future payments, while discount factors convert those projected payments into present values. Credit-risky securities require additional treatment of default and recovery; simply discounting promised payments at a government curve would generally overstate their value.
An interest-rate swap exchanges payment streams, commonly fixed interest against floating interest. The fixed rate that makes a new swap have zero initial value depends on projected floating payments and their discounting. Following the global financial crisis, derivatives markets increasingly adopted multiple-curve frameworks. A discount curve associated with collateral remuneration can differ from the curves used to project floating reference rates. An overnight indexed swap curve is particularly important in this setting.
The first-order price sensitivity of a bond is commonly summarized by duration. Modified duration approximates the percentage price change resulting from a small change in yield:
A duration of 7 implies an approximate 7% price decline for a one-percentage-point yield increase, before allowing for higher-order effects. Convexity improves this approximation by accounting for curvature in the price–yield relationship. These measures are most straightforward for fixed, option-free cash flows.
A single duration measure is insufficient when the curve changes shape. Key-rate durations measure sensitivity to yield changes around selected maturities, allowing a portfolio to be analyzed against twists and localized movements. Two portfolios can have identical overall duration yet behave differently if one concentrates exposure at intermediate maturities while the other combines short and very long bonds.
Banks use curve analysis in asset–liability management because loans, securities, deposits, and wholesale funding reprice at different times. Insurers and pension funds use it to compare the interest-rate sensitivity of assets with that of long-dated obligations. Falling long-term discount rates can increase the present value of those obligations even while raising the market value of bond holdings. Accounting and regulatory rules determine which effects appear immediately in reported measures.
A mortgage-backed security adds an important complication: borrower prepayments change the timing of cash flows. When rates fall, refinancing can return principal earlier; when rates rise, repayments may slow. This creates option-related exposure that cannot be captured by applying a fixed payment schedule to a government curve. Valuation therefore combines the yield curve with models of volatility and borrower behavior.
Credit, inflation, and international comparisons
Government bond curves are often treated as benchmarks, but government debt is not universally free of default risk. Sovereign yields can reflect repayment concerns, redenomination risk, taxation, market liquidity, and legal features. Even within one currency area, different governments can have different curves. In the euro area, member states share a monetary policy but issue debt separately, so spreads between their bonds cannot be interpreted simply as differences in expected central-bank rates.
Corporate curves similarly combine the currency’s underlying interest-rate structure with issuer-specific risks. A credit spread compensates for more than expected default losses: it can include uncertainty about recovery, systematic risk, trading costs, and liquidity. Different issuers may also populate different maturity ranges, creating composition effects. A curve inferred from a broad corporate index can therefore change when the index’s sector mix or credit quality changes, even without comparable repricing of every constituent.
Nominal and inflation-linked government curves permit analysis of inflation compensation. The difference between comparable nominal and real yields is commonly called breakeven inflation. It is related to expected average inflation over the relevant period, but is not a pure measure of it. Inflation risk premiums, liquidity differences, indexation lags, and contractual provisions also matter. For example, United States inflation-protected securities include a principal protection feature at maturity that can have additional value when deflation risk is significant.
Cross-country yield differences are equally difficult to interpret in isolation. A high nominal yield may reflect higher expected inflation rather than a higher expected real return. Investors exchanging currencies also face changes in the exchange rate. Under covered interest parity, forward exchange rates link interest rates across currencies in the absence of relevant arbitrage frictions. In practice, balance-sheet costs, funding conditions, and demand for hedging can produce a cross-currency basis.
Data conventions introduce further complications. An actively traded, newly issued government bond may yield less than a nearby older issue because it is more liquid or more useful in the repo market. Some securities trade at unusually expensive prices because they are scarce as collateral. Bonds with call options, tax exemptions, or unusual repayment schedules are not directly comparable with ordinary fixed-rate securities, even when final maturity is the same.
Published curves consequently embody methodological choices: which securities to include, how to handle outliers, whether to report par or zero-coupon yields, and how to extrapolate beyond liquid maturities. At the very long end, a smooth-looking curve may be supported by relatively few transactions. Its apparent precision should therefore be distinguished from the depth and comparability of the underlying market observations.