When Duration Misleads

Duration is the best one-number summary of interest rate risk—but it is still just a summary. Real bond markets are more complex.

Key takeaways

• Duration assumes a parallel shift in the yield curve (all rates rise or fall by the same amount). Non-parallel shifts are common and create surprises.
• Credit spreads (the extra yield corporate bonds pay over Treasuries) can widen or tighten independently of overall rates, creating returns not explained by duration.
• Bonds with embedded options (callables, MBS) have effective duration that changes as rates move, making static duration forecasts unreliable.
• Basis risk—the gap between a bond's price behavior and a duration forecast—can be significant for corporate, municipal, and international bonds.
• Investors need duration plus context: what is the bond's credit quality, where on the yield curve does it sit, and what options does it contain?

The parallel shift assumption

Duration is calculated assuming a parallel shift in the yield curve: all yields move by the same amount in the same direction. If 1-year yields rise 1%, 10-year yields rise 1%, and 30-year yields rise 1%, it is parallel. Duration works perfectly in a parallel shift environment.

But yield curves rarely shift in parallel. Consider a realistic scenario:

Date 1 (starting point):

• 1-year yield: 2%
• 10-year yield: 4%
• 30-year yield: 4.5%

Date 2 (one month later):

• 1-year yield: 3.5% (up 150 bp)
• 10-year yield: 4.2% (up 20 bp)
• 30-year yield: 4.3% (down 20 bp)

This is a steepening curve: short rates rose sharply, long rates barely moved or fell. This is not parallel. Duration assumes all rates move together, but they did not.

Consequences for a portfolio holding a 10-year bond:

• Duration predicts: A 20 bp rise in the 10-year yield should produce a loss of roughly 20 bp × (−duration) ≈ −0.5% loss (for a 2.5-year duration bond).
• Actual result: The 10-year yield rose only 20 bp, so the loss is about −0.5%. Duration was accurate.

But now consider a 30-year bond:

• Duration predicts: A 45 bp rise in yields (averaging the 10-year and 30-year) should produce a loss of roughly −0.45% × 8 (for an 8-year duration 30-year bond) ≈ −3.6%.
• Actual result: The 30-year yield fell 20 bp, so the 30-year bond actually rose. Duration was very wrong.

The lesson: non-parallel shifts create basis risk, where actual returns diverge significantly from duration predictions.

Yield curve steepening and flattening

Two common curve dynamics are worth naming:

Steepening: Long-dated yields rise less than (or fall more than) short-dated yields. This typically occurs when the Federal Reserve cuts rates (pushing short rates down faster than long rates fall) or when recession fears emerge (safe long bonds rise in price while risky short-term credits fall).

In a steepening environment:

• 2-year bonds underperform 10-year bonds (duration predicts equal moves, but the 10-year rallies more).
• Portfolio managers favoring long-term bonds outperform those holding short-term.

Flattening: Long-dated yields rise more than short-dated yields, or fall less. This typically occurs when the Fed tightens aggressively (short rates rise faster than long rates), or when near-term economic risks subside (short-term bonds rally relative to long-term).

In a flattening environment:

• 2-year bonds outperform 10-year bonds.
• Portfolio managers favoring short-term bonds outperform those holding long-term.

A portfolio manager positioned for a parallel shift but holding a concentrated duration bet (say, all 30-year bonds) could be devastated if the curve steepens or flattens unexpectedly. Duration told a false story because the curve did not move in parallel.

Credit spread widening and tightening

Corporate bonds, municipal bonds, and other credit-risky bonds are priced at a spread (or "option-adjusted spread," OAS) above Treasury yields.

For example:

• A 10-year Treasury yields 4.0%.
• A 10-year A-rated corporate bond yields 4.8%.
• The credit spread is 80 basis points.

This spread compensates the bondholder for credit risk (the chance the corporation defaults). Spreads fluctuate based on:

• Economic outlook: In strong economies, corporate defaults are rare, spreads narrow. In recessions, defaults rise, spreads widen.
• Risk appetite: When investors are risk-hungry, spreads compress (they accept lower credit compensation). When risk-averse, spreads widen.
• Federal Reserve policy: When the Fed is cutting rates, spreads often tighten. When hiking, spreads often widen.

Now consider a portfolio manager holding a 10-year A-rated corporate bond. Duration says a 1% rise in overall rates should produce a −10% loss (for 10-year duration). But what if:

• Treasury yields rise 1% (bad for the bond).
• Credit spreads widen 50 bp (very bad for the bond, because the corporate now yields 5.3% instead of 4.8%).

The corporate bond's actual loss is:

• Duration loss from Treasury rate rise: −10%
• Loss from spread widening: −0.5% × 10 (roughly)
• Total loss: approximately −15%.

Duration predicted −10%, but the actual loss was −15% because spreads widened. This is spread risk, and duration does not capture it. Similarly, spreads can tighten when the manager did not expect it, creating unpleasant surprises.

In the 2008 financial crisis, spreads widened dramatically (corporate bonds fell 40%+) even though Treasury yields did not rise much. Corporate bond investors who relied solely on duration were blindsided.

Embedded options and changing effective duration

Callable bonds, MBS, and convertibles have embedded options whose value changes as rates move. Effective duration, not modified duration, is the right measure—but effective duration itself changes as rates move.

Example: Callable bond in a steepening curve

A callable 10-year corporate bond has:

• Modified duration: 7 years
• Effective duration: 4 years (accounting for the call)

You think rates will fall and buy the bond expecting it to rally. But the curve steepens: short rates fall sharply, long rates rise. The 10-year sector (where your bond sits) falls in yield (good), but the 5-year sector (where the bond might be called) rises in yield (bad—the issuer is less likely to call).

• The bond's price rises due to falling 10-year yields.
• But effective duration increased from 4 years to 5 years (because the call is less likely).

The price rise was larger than duration predicted. You were right about the direction but benefited more than duration suggested. In this case, duration understated your gain.

In the opposite case—curve flattens, long rates fall sharply—effective duration might shrink further, and your gains might be smaller than duration predicted.

Effective duration is not constant; it is a function of the rate environment and market expectations. Duration-based risk models that treat effective duration as static will be inaccurate for bonds with options.

Basis risk: when the bond does not behave like Treasuries

Basis risk is the difference between a bond's price behavior and the price behavior of a hedge or benchmark.

Example: A-rated corporates tracking Treasuries

An investor holds a 5-year A-rated corporate bond and hedges it by shorting 5-year Treasury futures (betting that if the corporate falls, the Treasury will also fall, offsetting the loss).

In theory, both have similar duration, so the hedge should work. In practice:

• If the economy strengthens, Treasuries fall in yield (investors flee to safety), but corporates rise in yield (investors embrace risk). The hedge does not work; the investor loses on both legs.
• If credit spreads widen (recession fears), corporates fall much faster than Treasuries. The hedge partially works but not perfectly.

The basis—the difference in how the corporate and the Treasury behave—creates risk. Duration alone cannot capture this because duration assumes all bonds move the same way.

Currency and foreign exchange basis

International bonds introduce currency risk. A UK investor holding euro-denominated bonds has two sources of return/loss:

1. Bond price change (predicted by duration).
2. Currency move (euro appreciates or depreciates against sterling).

A bond with 5-year duration priced in euros might have very different risk in sterling terms if the sterling-euro exchange rate is volatile.

If euro bonds fall 5% but the euro appreciates 3%, a UK investor's sterling return is −5% + 3% = −2% (better than duration predicts). If the euro depreciates 3%, the return is −5% − 3% = −8% (worse than duration predicts).

This currency basis risk is often large for international bond portfolios. Duration captures interest rate sensitivity but not currency sensitivity.

Liquidity basis: bid-ask widening

During stress (financial crises, rate spikes), bid-ask spreads on bonds widen dramatically. A corporate bond that normally trades with a 0.05% spread might trade with a 0.5% spread during a crisis.

If you need to sell during a stress event, the illiquidity loss (0.5% spread) is often larger than any interest rate loss predicted by duration. This is liquidity basis risk—the risk that the bond will be hard to sell precisely when you want to sell.

Treasury bonds (highly liquid) have much narrower spreads than corporate bonds. If you hold corporates, you are accepting liquidity risk that duration does not measure.

When to worry about duration limitations

You should look beyond duration when:

1. You hold non-Treasury bonds (corporates, municipals, foreign): Spread risk, liquidity risk, and currency risk are all important.
2. You hold bonds with embedded options: Callable corporate bonds, MBS, convertibles all have effective duration that changes with rates.
3. Your hedge or benchmark is different from your holdings: If you hold corporate bonds and hedge with Treasury futures, basis risk is real.
4. You are investing in a specific market segment (e.g., 30-year bonds): Curve risk means your returns will differ from parallel-shift predictions.
5. Economic conditions are changing rapidly (recession fears, Fed pivot): Credit spreads and volatility can move sharply, creating returns not explained by duration.

A more complete risk model

Professional investors often use a multi-factor model:

Bond Return ≈ (Yield Income) + (Duration Effect) + (Spread Effect) + (Convexity Effect) + (Currency Effect) + (Liquidity Effect)

• Yield Income: The coupon and yield income you accrue.
• Duration Effect: −Modified Duration × ΔYield.
• Spread Effect: Change in credit/option-adjusted spread × Duration.
• Convexity Effect: ½ × Convexity × (ΔYield)².
• Currency Effect: ΔForeignCurrency / HomeCall.
• Liquidity Effect: Bid-ask widening or changes in market depth.

This model is more complex than duration alone, but it explains actual returns much better. For retail investors, understanding these factors is useful for avoiding surprises, even if you do not formally calculate them.

Flowchart: When duration is insufficient

Next

Duration and convexity are powerful tools, but they are only the beginning. The final article steps back and provides a one-page summary of everything in this chapter—a cheat sheet for quick reference and decision-making. It ties together all the concepts: what duration is, how it is measured, what convexity is, and how to use both in practice.

Related concepts

• Duration and Convexity in Tandem — using both metrics together
• Effective Duration — duration adjusted for embedded options
• Modified Duration — the standard linear measure