Interpreting Yield-Curve Shapes and Changes

Identify the curve before describing its shape

Treasury coupon yields, fitted zero-coupon rates, overnight-indexed swap rates, and corporate yields are different curves. Their credit, liquidity, collateral, and cash-flow properties differ. A slope calculated from two Treasury yields cannot simply be compared with a slope using a swap rate at one end and a policy target at the other. Start with a consistent instrument family and observation convention.

Specify whether points are observed securities or fitted constant-maturity estimates. A benchmark security can change over time, while a fitted curve depends on its estimation method. State currency, maturity, timestamp, compounding, and any interpolation. This avoids attributing a construction change to economics and helps distinguish a genuinely unusual curve shape from a mismatch between the series used to draw it.

Measure level, slope, and curvature separately

A curve's level describes the general scale of yields, its slope describes differences across maturities, and curvature describes the middle relative to the ends. Two curves can share the same ten-year-minus-two-year spread while differing substantially at five years or in their overall level. A two-point slope is a summary statistic, not a complete representation of the financing environment.

Define the subtraction order. A ten-year yield of 4% minus a two-year yield of 4.5% gives negative 0.5 percentage points, or negative 50 basis points. Those are synthetic values. The curve is inverted between those points, but another maturity pair could remain upward sloping. Avoid announcing that the entire curve is inverted when the evidence refers only to one selected segment.

Work through two different steepenings

Starting from the illustrative 4.5% two-year yield and 4% ten-year yield, suppose they decline to 3.5% and 3.75%. The slope rises from negative 50 to positive 25 basis points. This is a falling-yield, or bull, steepening: both yields fall, but the shorter rate falls more. Existing bond prices would have different sensitivities according to their cash-flow profiles, not just the slope change.

In a separate synthetic case, two-year and ten-year yields rise to 4.7% and 4.8%. The slope rises to positive 10 basis points, a rising-yield, or bear, steepening. Both cases become more upward sloping while producing opposite yield directions. Calling either simply a steepening discards information essential to interpreting price changes, policy expectations, and the role of compensation for holding longer maturities.

Separate expected short rates from term premiums

The New York Fed's Treasury Term Premia resource describes yields as reflecting expected future short rates and a term premium. The premium is not directly observable and must be estimated. Consequently, an increase in a long yield need not imply that the expected policy path rose by the same amount. Changed compensation for duration risk can also contribute to the move.

For an illustrative decomposition, an expected-short-rate component of 3.2% and a term-premium component of 0.8% sum to a 4% long yield. If only the premium rises to 1.3%, the yield reaches 4.5% with unchanged assumed short-rate expectations. This is a synthetic approximation for explanation, not an estimate from the ACM model. Preserve the model name and estimation date when using actual decompositions, which can differ across methods.

Interpret forward rates as prices under conventions

A forward rate can be derived from discount factors and summarizes the rate implied for a later interval. Under illustrative annual compounding, a one-year zero rate of 3% and a two-year zero rate of 4% imply a second-year forward of 1.04 squared divided by 1.03, minus one, approximately 5.01%. Coupon yields cannot be substituted into this formula without constructing compatible discount factors.

That implied rate is not automatically an unbiased forecast of the future realized short rate. Risk premiums, supply and demand, and measurement conventions affect its interpretation. Similarly, a positive forward spread is not free extra return from rolling shorter bonds. Any comparison must consider the future rate actually realized, reinvestment uncertainty, and the relevant horizon rather than assuming the implied path will necessarily occur.

Use recession evidence without changing its meaning

The New York Fed's Yield Curve as a Leading Indicator resource uses the ten-year-minus-three-month Treasury spread in a model with a stated twelve-month horizon. It explicitly distinguishes the estimates from official Federal Reserve forecasts. A different maturity pair, frequency, or country is not the same model. An inversion is an input to a historical statistical relationship, not a date-stamped guarantee of recession.

The bank's research on uncertainty in recession probabilities illustrates that estimated probabilities themselves have uncertainty. Parameters, sample choice, and model specification matter. Report the published definition and avoid converting a point estimate into certainty or a precise market turning point. Historical economic associations and profitable asset timing are separate claims, particularly because prices may already incorporate concern about the outlook.

Build a curve comparison that answers a question

Present starting and ending curves, changes at major maturities, selected slopes, and the source conventions. If the question concerns a bond position, map those changes through its maturity-specific sensitivities. If it concerns economic interpretation, show the evidence for expectations and premium explanations separately. Do not let one visually dramatic slope substitute for the analysis required by the underlying research question.

Retain the timestamp, downloaded observations, and any fitted estimates so another reader can reproduce the picture. Identify whether a conclusion rests on a single day, a monthly average, or a sustained change. End with the range of plausible mechanisms and the observations that would distinguish them. This produces a defensible interpretation without claiming that the curve reveals a uniquely determined future for either growth or investment returns.

What to retain

• A slope requires named maturities, consistent instruments, and an explicit sign convention.
• Bull and bear steepening describe different yield movements despite similar slope changes.
• Forwards and recession probabilities require interpretation, not literal certainty.

Source notes

1. Federal Reserve Bank of New York: Treasury Term Premia
2. Federal Reserve Bank of New York: The Yield Curve as a Leading Indicator
3. New York Fed Liberty Street Economics: How Uncertain Is the Estimated Probability of a Future Recession?

Related research