Why Separate Rate and Spread Effects?
A bond’s yield can change because the benchmark interest-rate curve moves, because investors demand a different credit spread, or both. Separating these drivers helps an analyst explain an estimated market-value change: a rise in benchmark rates can reduce a bond’s value even when its credit outlook is unchanged, while spread widening can reduce value even if benchmark rates do not move.
The decomposition in this tutorial is a first-order sensitivity estimate. It assigns the benchmark-rate shock to interest-rate duration and the credit-spread shock to spread duration. It is useful only when the yield changes and duration measures refer to compatible valuation assumptions. It is not a full repricing of the bond, and the estimate can be inaccurate for large shocks or when cash flows change.
For a small change in rates and spreads, the estimated dollar value change is
Here \(D_r\) is the applicable interest-rate duration and \(D_s\) is spread duration. A positive \(\Delta r\) or \(\Delta s\) produces a negative estimated value change; a negative shock produces a gain, all else equal. Shocks must be expressed as decimal yield changes: 25 basis points is \(0.0025\), not \(25\) or \(0.25\).
Why the Components Add
The first-order formula follows from a two-factor approximation to the bond price. For small changes in the benchmark rate and spread, the total differential of price is the sum of the price’s separate sensitivities to those factors:
Duration is defined as the negative price sensitivity divided by price. In this decomposition, \(D_r=-(1/P_0)(\partial P/\partial r)\) and \(D_s=-(1/P_0)(\partial P/\partial s)\), with each partial derivative taken while holding the other factor fixed. Substituting these definitions into the differential gives
This is an additive approximation because it retains the two first-order terms. It omits curvature and interaction effects, including terms involving the product \(\Delta r\Delta s\). The approximation is most appropriate for modest shocks when cash flows and the bond’s relevant sensitivities are treated as unchanged. For larger changes, scenario repricing or additional nonlinear adjustments may be needed.
Worked Examples
Worked Example: Separating Two Shocks for One Bond
A hypothetical insurer holds a fixed-cash-flow bond with market value \(P_0=\$2{,}000{,}000\). Its benchmark rate is 4.20% and its credit spread is 1.30%, giving an initial all-in yield of 5.50%. Its interest-rate duration is 4.8 and its spread duration is 4.2. In a deterministic scenario, the benchmark rate rises 35 basis points and the spread widens 60 basis points. Estimate the dollar value change. Assume the durations are measured at the initial valuation, cash flows do not change, and convexity and rate-spread interaction are ignored. No mortality model or probability distribution is assumed; the stated shocks are scenario inputs.
Method. Apply each duration only to its corresponding shock, convert both shocks to decimal yields, and then add the dollar changes.
The benchmark-rate shock is \(+35\) basis points, or \(+0.0035\). The rate component is
The spread shock is \(+60\) basis points, or \(+0.0060\). The spread component is
Combining the components gives
The estimated ending value is \(\$2{,}000{,}000-\$84{,}000=\$1{,}916{,}000\). The estimated percentage change is \(-\$84{,}000/\$2{,}000{,}000=-0.042\), or \(-4.20\%\). The components sum to the total dollar change, which checks the calculation. The common mistake this method avoids is using the combined 95-basis-point yield change with one duration, which would ignore that rate duration and spread duration can differ.
Worked Example: A Portfolio With Different Benchmark Shocks
A hypothetical portfolio contains two fixed-cash-flow bonds. Bond A has market value \(\$2{,}000{,}000\), rate duration 4.0, and spread duration 4.0. Bond B has market value \(\$3{,}000{,}000\), rate duration 2.5, and spread duration 2.5. Their initial benchmark rates are 3.00% and 3.40%, respectively; their initial spreads are 1.80% and 0.90%. In the scenario, the relevant benchmark point for A rises 40 basis points and the relevant benchmark point for B rises 8 basis points. A’s spread widens 30 basis points and B’s spread widens 10 basis points. Estimate the total rate effect, spread effect, and value change. Treat the shocks as deterministic; assume durations remain fixed, cash flows do not change, and convexity and interaction effects are ignored. No probability distribution or mortality assumption is needed.
Method. Calculate each bond’s rate and spread component separately, using its own market value, duration, and corresponding shock. The benchmark shocks differ because the two bonds are exposed to different points on the benchmark curve.
For Bond A, the benchmark-rate shock is \(0.0040\), so its rate-related loss is
For Bond B, the benchmark-rate shock is \(0.0008\), so its rate-related loss is
The total rate-related loss is \(\$32{,}000+\$6{,}000=\$38{,}000\).
For Bond A, its 30-basis-point spread widening is \(0.0030\), giving a spread-related loss of
For Bond B, its 10-basis-point spread widening is \(0.0010\), giving a spread-related loss of
The total spread-related loss is \(\$24{,}000+\$7{,}500=\$31{,}500\). Therefore, the estimated total loss is \(\$38{,}000+\$31{,}500=\$69{,}500\), and the estimated portfolio value after the change is \(\$5{,}000{,}000-\$69{,}500=\$4{,}930{,}500\). The arithmetic checks: each of the four component losses is based on its own bond’s market value, duration, and decimal shock, and the rate and spread totals add to the portfolio loss.
Common mistake and correction. An incorrect spread total of \(\$39{,}000\) results from adding \(\$24{,}000\) for A to an incorrectly calculated \(\$15{,}000\) for B. B’s correct calculation is \(\$3{,}000{,}000(2.5)(0.0010)=\$7{,}500\), not \(\$15{,}000\). Thus the correct spread loss is \(\$31{,}500\), and the correct total loss is \(\$69{,}500\).
Worked Example: Spread Tightening Offsets a Rate Increase
A hypothetical fixed-cash-flow bond has market value \(\$5{,}000{,}000\), initial benchmark rate 4.00%, and initial credit spread 2.20%. Its rate duration is 6.1 and its spread duration is 3.7. A deterministic scenario has the benchmark rate rise 20 basis points while the spread tightens 35 basis points. Estimate the net dollar change and ending value. Assume both durations are evaluated initially and remain applicable for this first-order estimate; cash flows are unchanged and convexity and interaction effects are omitted. No probability distribution or mortality assumption is required.
Method. A rate increase causes an estimated loss; a spread tightening is a negative spread shock and causes an estimated gain.
The rate shock is \(+0.0020\). Its estimated value effect is
The spread shock is \(-0.0035\). Its estimated value effect is
The net estimated change is \(-\$61{,}000+\$64{,}750=+\$3{,}750\). The estimated ending value is \(\$5{,}000{,}000+\$3{,}750=\$5{,}003{,}750\), a gain of \(0.075\%\) of initial value. As a check on direction, the rate effect is negative and the spread effect positive; the larger spread gain leaves a small net gain. The common mistake avoided is treating spread tightening as a positive shock, which would incorrectly make both components losses.
Conditions and Practical Interpretation
The decomposition is interpretable only when each shock and duration match. Rate duration should represent sensitivity to the benchmark-rate movement being applied; spread duration should represent sensitivity to a change in the bond’s credit spread. A single parallel rate duration is not necessarily appropriate for a nonparallel curve movement. Where exposures are represented by several benchmark points, calculate the rate effect for each relevant exposure and sum those effects, as in the portfolio example.
The estimate also assumes that the bond’s contractual cash flows remain unchanged. A downgrade, default, restructuring, or embedded option exercise may alter expected cash flows or the sensitivity measures. In those situations, simply applying initial durations can miss important effects. The calculations here are deterministic sensitivity estimates, not statements about the probability of a market scenario.
Keep the sign convention explicit: \(\Delta P\) is the change in value, so a loss is negative. If reporting a positive loss amount instead, define it as the negative of the estimated value change. For a portfolio, sum signed dollar changes; do not add absolute values when one component is a gain and another is a loss.
| Scenario movement | Estimated first-order price effect | Reason |
|---|---|---|
| Benchmark rate rises | Negative | Positive \(\Delta r\) multiplied by negative rate sensitivity |
| Benchmark rate falls | Positive | Negative \(\Delta r\) multiplied by negative rate sensitivity |
| Credit spread widens | Negative | Positive \(\Delta s\) multiplied by negative spread sensitivity |
| Credit spread tightens | Positive | Negative \(\Delta s\) multiplied by negative spread sensitivity |
Common Exam Traps
- Using basis points as whole numbers. Convert 40 basis points to \(0.0040\) before multiplying by duration.
- Applying one duration to both factors. Use rate duration for the benchmark shock and spread duration for the spread shock; they need not be equal.
- Double counting. If the rate and spread effects are calculated separately, do not also apply a duration to the full all-in-yield change and add the separate components again.
- Ignoring the sign of a tightening. A spread narrowing is a negative \(\Delta s\), and therefore contributes a positive estimated value change.
- Presenting a linear estimate as exact repricing. The formula omits convexity, interaction effects, changing durations, and potential changes in cash flows.
Check Your Understanding
For each problem, use the first-order decomposition and retain the sign of each value effect.
- A hypothetical bond is worth \(\$1{,}200{,}000\), with rate duration 5.2 and spread duration 4.1. The benchmark rate rises 15 basis points and the spread widens 20 basis points. Estimate the rate effect, spread effect, and total value change. Assume fixed cash flows and unchanged durations. (Answer: rate loss \(\$9{,}360\); spread loss \(\$9{,}840\); total loss \(\$19{,}200\).)
- A hypothetical bond is worth \(\$800{,}000\), with rate duration 6.0 and spread duration 3.5. The benchmark rate falls 25 basis points and the spread widens 40 basis points. Estimate the net change. (Answer: rate gain \(\$12{,}000\); spread loss \(\$11{,}200\); net gain \(\$800\).)
- A portfolio has Bond A worth \(\$4{,}000{,}000\), with rate duration 3.2 and spread duration 2.8, and Bond B worth \(\$2{,}000{,}000\), with rate duration 5.5 and spread duration 4.0. Their benchmark rates rise 30 and 10 basis points, respectively; their spreads widen 50 and 25 basis points, respectively. Estimate total rate loss, total spread loss, and total loss. (Answer: rate loss \(\$49{,}400\); spread loss \(\$76{,}000\); total loss \(\$125{,}400\).)
- A hypothetical bond worth \(\$5{,}000{,}000\) has rate duration 4.0 and spread duration 3.0. The benchmark rate rises 15 basis points. How much must the spread tighten for the first-order estimate to be unchanged in value? (Answer: spread tightening of 20 basis points.)
- A hypothetical bond has market value \(\$2{,}500{,}000\), rate duration 5.0, and spread duration 4.0. The benchmark rate falls 12 basis points and the spread widens 18 basis points. Find the net estimated dollar change and ending value. (Answer: rate gain \(\$15{,}000\); spread loss \(\$18{,}000\); net loss \(\$3{,}000\); ending value \(\$2{,}497{,}000\).)