Stress Testing Interest-Rate Exposure

Identify the curve that prices the position

A portfolio can depend on government yields, overnight discount curves, swap rates, and contractual floating-rate indexes in several currencies. A policy-rate increase does not specify how every maturity or reference rate moves. Identify the curve used for each valuation and define the shocked tenors. Otherwise, a seemingly simple rate scenario can apply inconsistent assumptions to economically related instruments.

Keep units explicit. A 100-basis-point increase is a one-percentage-point rise, entered as 0.01 in a decimal-yield duration formula. It is not a 100% relative increase in the starting yield. Specify compounding and day-count conventions when repricing cash flows. A unit error can overwhelm the economics while still producing a clean-looking numerical table and an apparently plausible percentage loss.

Use duration as a local valuation approximation

For a plain fixed-rate bond, modified duration gives an approximate percentage price change equal to negative duration times the yield change. The SEC's interest-rate-risk bulletin explains the inverse relationship between fixed-rate bond prices and market rates. Duration quantifies a local version of that relationship; it does not capture every feature of a complex bond or guarantee accuracy for large moves.

The sign matters. A positive duration exposure usually loses value when the relevant yield rises and gains when it falls, holding other inputs fixed. A short bond or a derivative can reverse that sensitivity. Calculate signed money sensitivity from the actual position, not from a generic asset-class label, and distinguish market value from face amount when converting a percentage change into currency units.

Compare parallel and curve-shape shocks

A parallel shock moves all modeled tenors by the same amount. It is useful for a first check but cannot reveal a portfolio's response to steepening, flattening, or a localized maturity move. Key-rate durations describe sensitivity to specified curve nodes under a chosen interpolation method. Their definitions matter because different curve construction methods can allocate sensitivity differently among neighboring maturities.

For a key-rate approximation, multiply each node's duration by its own yield change, sum the products, and apply a negative sign. The sum of node durations should be checked against the parallel duration under compatible definitions. Large discrepancies can indicate inconsistent curves, units, or coverage. A sensible total duration does not prove that the maturity distribution of risk has been mapped correctly.

Calculate three hypothetical valuation checks

Take a synthetic USD 400,000 bond sleeve with modified duration six. A parallel 100-basis-point yield rise gives an approximate return of -6%, or -USD 24,000. If convexity is 50 using decimal-yield units, the second-order term is one-half times 50 times 0.01 squared, equal to +0.25%. The duration-plus-convexity estimate is therefore -5.75%, or -USD 23,000.

Now use a separate first-order curve-shape test with key-rate durations 0.8, 2.2, and 3.0. Apply rises of 50, 100, and 150 basis points respectively. The estimated return is negative (0.8 times 0.005 plus 2.2 times 0.01 plus 3.0 times 0.015), or -7.1%, equal to -USD 28,400. These are alternative calculations, not losses to be added together.

Separate price risk from horizon cash flows

CFA Institute's Interest Rate Risk and Return distinguishes price effects from reinvestment effects and relates their importance to the investment horizon. A mark-to-market loss today is not the same object as the return from holding an instrument through a future date. That future return can include coupons, principal, reinvestment, and a sale value, all requiring explicit assumptions about timing and rates.

Holding a bond to maturity does not make its intermediate market-value sensitivity disappear. Nor does a higher future reinvestment rate immediately reimburse a current price decline. Keep an instantaneous valuation test separate from a horizon return projection. If the portfolio has financing liabilities, model their repricing on their own contractual schedule rather than automatically netting them against asset duration.

Escalate optional and floating-rate instruments

Callable bonds, mortgage-related instruments, and options can change their effective cash flows or sensitivities when rates move. A fixed duration borrowed from a calm period can miss that change. Use a model that represents the relevant exercise or prepayment behavior and challenge those assumptions. A numerical repricing is still conditional on the model; it is not an observed future transaction price.

Floating-rate securities may have limited benchmark-rate duration between reset dates, but they can retain spread, basis, liquidity, and credit exposure. Caps, floors, reset lags, and discounting conventions can matter. Avoid assigning them a zero total stress loss merely because their coupon resets. Identify what the rate calculation covers and route remaining exposures to separately specified scenario components.

Make the report useful for model challenge

Present losses by currency and maturity bucket, along with the principal curve assumptions. Reconcile sensitivity-based totals to full repricing for representative positions where feasible. Explain any difference caused by convexity, accrued interest, optionality, or curve interpolation. Use stable sign conventions so a positive cash gain cannot be confused with a positive loss number when results are combined across instruments.

List unsupported contracts and identify whether credit spreads, foreign exchange, income, and funding were held constant or excluded. The final result should show which rate movements drive the portfolio and where approximations become unreliable. It is a conditional valuation exercise, not a forecast of central-bank decisions, a recommendation about bond ownership, or evidence that a regulatory interest-rate framework has been satisfied.

Review checkpoints

• Define curve, currency, tenors, and units before applying a rate shock.
• Parallel duration cannot describe every curve-shape movement.
• Separate immediate valuation from reinvestment and contractual financing effects.

Source notes

1. SEC Investor.gov: Fixed Income Investments - Interest Rate Risk
2. CFA Institute: Interest Rate Risk and Return

Continue the analysis