HSBC results: how reinvestment timing and deposit rates change net interest income

Read one rate change through valuation and interest income

In its October 27, 2026 investor presentation, page 16, HSBC reported a $2.6 billion third-quarter pretax OCI loss on hold-to-collect-and-sell (HTC&S) bonds from higher rates. Page 18 models a $2 billion twelve-month banking NII increase for a parallel 100 bp rise, as of June 30, 2026, assuming static balances, no management actions and 50% deposit pass-through.

Subtracting these numbers cannot establish a payback period. One is a valuation movement during a completed quarter; the other changes net interest income under assumptions applied to assets and liabilities at a different reference date. A common dollar denomination does not align their periods, tax treatment, balances or future actions. The useful question is how the same rate change reaches the bank through different channels and at different times.

An existing fixed-rate bond normally does not pay a larger coupon simply because market rates rise. Its sale price can change immediately. Principal returned at maturity, however, can be invested at the then-prevailing rate. If deposit costs rise first, net interest income may deteriorate before enough assets have rolled over. Saying that higher rates benefit banks leaves out this sequence.

The calculations below apply stated fictional assumptions to a fixed-coupon bond, quarterly asset rollover and deposit repricing. Readers can record maturity dates and funding terms from their own documents, identify the first quarter of additional interest and find the pass-through rate that removes the gain. This does not reconstruct a bank’s internal model. It identifies the evidence needed to interpret a disclosure.

Put three clocks beside the comparison

The first clock asks what future payments are worth today. The second records when contractual interest receipts or payments can change. The third identifies the reporting period in which those changes reach earnings or capital. A single chronology can reveal whether interest receipts fell on the valuation date or whether only funding costs reset. It follows a contract through time rather than merely listing different indicators.

Place a rate change at the start of a year, a coupon at year-end and principal repayment further out. The new discount rate changes the opening valuation, while the contractual year-end coupon remains fixed. If deposit terms reset in between, interest paid can increase during the year. Treating unmatured assets as if they already earn the new rate removes precisely that intervening burden.

Accounting presentation also differs. OCI is other comprehensive income, presented separately from profit or loss. For FVOCI bonds, AASB 9, paragraph 5.7.10, separates interest and valuation recognition. Adding valuation movements to net interest income mixes interest receipts with asset measurement. Retaining line-item names prevents that addition error.

A chronology does not turn historical observations into a fresh forecast. Balances and market rates can change between a sensitivity reference date and quarter-end, producing different results for an equal shock. In a later disclosure, check the balance-sheet reference date and rates as well as the publication date. The guide to results, guidance and market expectations explains how to align those comparisons.

Fixed payments acquire a different price when the discount rate changes

Start with a bond of face value 100, an annual coupon of 2 and four years remaining. Assume payment in full, no tax or fees, no early-redemption option and unchanged credit quality. It pays 2 at the end of years one through three and 102 at the end of year four. Discounting at 2% gives a price of 100. Discounting the same payments at 3% gives about 96.28.

The equation is price = 2/1.03 + 2/(1.03²) + 2/(1.03³) + 102/(1.03⁴). Displayed results use two decimal places; intermediate calculations remain unrounded. A higher yield applies a larger discount when bringing future payments back to today. That is why an unchanged contract can fall in price. It does not mean the principal promise has immediately been broken or the coupon has fallen by about 3.72.

With the same coupon, a one-year bond is worth about 99.03 and a seven-year bond about 93.77. Under these assumptions, principal received later is more exposed to the higher discount rate. Maturity alone does not determine the loss on every bond. Coupons, payment frequency, redemption options and the rates at individual maturities affect sensitivity. Average remaining life cannot simply be relabelled as duration, a measure of price sensitivity.

The essential sequence is to fix the payment schedule first and then change the rate. For an actual bond, identify payment dates, coupons, redemption and accrued interest as well as face value. Use yields with matching currency, credit conditions and maturity. Substituting a policy rate for every discount rate is not a sufficient repricing calculation.

Only principal already returned can earn the new rate

Now leave the pricing example aside and divide an asset base of 100 into four fictional maturity tranches. Each 25 matures at the start of one of the four quarters. This is a maturity ladder. All assets initially earn 2% annually. Market rates rise by one percentage point at the start of the year, and returned principal can immediately be reinvested at 3%. There are no new inflows, credit losses, fees or hedges, and total assets do not grow.

Each quarter is exactly one fourth of a year. In the first quarter, only 25 earns the new rate, adding 25 × 1% × 0.25 = 0.0625. In the second, the first tranche and the next maturity bring the repriced balance to 50, adding 0.125. The third-quarter balance of 75 adds 0.1875; the fourth-quarter balance of 100 adds 0.25. The first-year increase is their sum, 0.625.

All assets earn the new rate by year-end, but they have not done so throughout the year. The first tranche receives the higher rate for one year, the next for three quarters, the next for half a year and the last for one quarter. Summing by maturity gives 25 × 1% × (1 + 0.75 + 0.5 + 0.25) = 0.625. Agreement between the quarterly and maturity-based sums provides two independent arithmetic checks.

The first-year shortfall is not a failure to reinvest. It follows from the contractual maturity structure. If rates remain unchanged in the following year, reinvested contracts continue and the whole base earns 3%, the annual increase over the original 2% becomes 1. Substituting that following-year amount for the first year’s 0.625 counts interest that has not yet accrued.

Maturities within the same year can produce different income

Knowing that 100 matures during the year is insufficient to determine first-year income. Move the quarterly opening maturities toward the front: 40, 30, 20 and 10. Weighting each by its remaining investment period gives 40 × 1 + 30 × 0.75 + 20 × 0.5 + 10 × 0.25 = 75. Multiplying by the one-percentage-point rate increase gives 0.75. Total principal and the new annual rate are unchanged; earlier repayments earn the higher rate for longer.

A back-loaded ladder of 10, 20, 30 and 40 instead produces a time-weighted balance of 50, which is 25 lower, and additional interest of 0.5. The equal ladder’s 0.625 lies between them. The same statement that every asset resets within the year is consistent with both 0.5 and 0.75. Quarterly remaining periods reveal a difference hidden by annual maturity totals.

If maturities occur at quarter-end, principal starts earning the new rate in the next quarter. Equal tranches then add only 25 × 1% × (0.75 + 0.5 + 0.25 + 0) = 0.375 during the year. The rate outlook has not changed; the timing convention has. Assuming opening rather than closing dates without checking the contract overstates first-year interest by 0.25 in this example.

Actual comparisons also require the agreed day-count convention and payment frequency. Equal quarters isolate timing from rates here. A daily-settled contract requires actual days and the contractual denominator. The article comparing sovereign maturities and refinancing costs explains the basic refinancing lag. For bank net interest income, funding reset dates must be placed beside the asset maturity schedule.

Earlier deposit repricing can remove the first-year gain

Add funding costs to the asset increase of 0.625. Assume deposits of 100 initially pay 1% annually. From the start of the year, the bank passes through half of the one-percentage-point market-rate increase. Deposits now pay 1.5%, so annual interest paid rises from 1 to 1.5. The additional cost of 0.5 leaves a first-year net interest income increase of 0.125 after subtracting it from the equal ladder’s additional receipts of 0.625.

Additional funding cost is 0.125 in every quarter. Subtracting it from additional receipts produces −0.0625 in quarter one, zero in quarter two, +0.0625 in quarter three and +0.125 in quarter four. An annual gain can coexist with an initial quarterly decline. Showing only the positive annual total hides interim pressure on funding and earnings plans.

At 75% deposit pass-through, the annual cost increase becomes 0.75, exceeding the same asset gain of 0.625. First-year net interest income falls by 0.125. At 25% pass-through, the added cost is 0.25, leaving a gain of 0.375. Funding terms alone change the sign without changing any asset investment decision. A higher reinvestment yield and a gain retained by the bank are different questions.

Pass-through is not a score of depositor quality. A bank may raise rates to retain funds; holding them down may instead reduce balances. This calculation fixes deposits and changes only their price. If balances move, the amount and price of replacement funding must be added separately. The article on deposit movements and bank funds allocation supplies that structural background.

Derive the pass-through threshold from the maturity sequence

For the equal ladder, the first-year gain disappears at 62.5% pass-through. Additional asset interest is 100 × 1% × the time-weighted repriced fraction of 0.625. Additional deposit cost is 100 × 1% × pass-through. Equating them gives 0.625. This is neither an industry average nor a company forecast. It follows from four fictional maturity dates and deposits resetting at the start of the year.

The front-loaded threshold is 75%; the back-loaded threshold is 50%; the equal quarter-end ladder’s is 37.5%. With the same 50% deposit pass-through, net changes are respectively +0.25, +0.125 for equal opening maturities, zero for back-loaded maturities and −0.125 for equal closing maturities. Moving maturity dates alone creates this ranking under an unchanged rate shock. The critical question is when asset repricing catches up with funding costs.

If assets and deposits differ in size, the threshold changes. More generally, threshold pass-through = asset principal / deposit principal × change in asset yield / change in the funding reference rate × time-weighted repriced fraction. The rate-change ratio disappears only when those changes are equal. If the funding reference-rate change is zero, this division is unavailable; use the original interest-difference equation instead.

A threshold is not an instruction to cap actual deposit pricing at that value. Outflows caused by low rates, service costs, secured borrowing and competing offers are absent from the equation. Its purpose is to identify the assumption that drives the conclusion for a particular maturity schedule. Close to the threshold, a few basis points or a delayed repayment can remove the apparent gain.

A different deposit mix can raise costs despite low pass-through

Even if many account rates stay unchanged, customers moving into higher-paying term deposits can raise the bank’s total interest bill. A comparison of pass-through on the same account can miss that change. Relevant inputs include average balances by product and the period during which each rate applies. Combining repricing and product migration in one line makes the reason for higher costs harder to trace.

Change just one further assumption in the equal-ladder, 50%-pass-through example. Total deposits remain 100, but 10 moves from the start of the year into a product paying an additional percentage point. The extra cost is 10 × 1% = 0.1. Subtracting it from the retained gain of 0.125 leaves only 0.025. Neither asset maturities nor the remaining 90 of funding have changed. A modest mix shift can determine the outcome when the margin is thin.

At that additional rate differential, a migrating balance of 12.5 removes the entire first-year gain. Migration during the year must be weighted by the remaining period. If 10 moves halfway through, the cost is 0.05. Treating it as an opening-year move would overstate the burden at 0.1. The timing principle used for asset reinvestment must also apply to funding.

This is not a forecast of product rates or migration. Customers compare liquidity, convenience, guarantee coverage and alternative investments. Internal transfers and departures to another bank also have different funding implications. Calculate a constant-total product-mix change first, then add net outflows and replacement funding. This avoids charging twice for the same movement of funds.

Recalculate when each reinvestment receives a different yield

So far, the new 3% rate has been available throughout the year. A first step toward a more realistic comparison is to give each reinvestment its own yield before changing the maturity amounts. Suppose its increase over the original 2% is successively 1, 0.8, 0.6 and 0.4 percentage points. Keep equal principal tranches of 25 and the same investment periods.

Additional first-year receipts are 25 × 1% × 1 = 0.25; 25 × 0.8% × 0.75 = 0.15; 25 × 0.6% × 0.5 = 0.075; and 25 × 0.4% × 0.25 = 0.025. Total additional income is 0.5, or 0.125 below the constant one-point shock. Every reinvestment earns more than before, but applying the initial high yield to every tranche overstates the gain.

This does not reset the yield on the whole portfolio each quarter. The first 25 reinvested at 3% keeps that contractual rate when later market yields fall. Lower subsequent rates apply to newly returned principal. Summing by quarter instead gives 0.0625, 0.1125, 0.15 and 0.175, again totalling 0.5. Reconciling the maturity and period tables catches the mistake of repricing previously reinvested assets again.

Short and long yields also need not move together. Choosing a longer new investment can require a different yield under the same policy rate. The guide to Treasury yields and curve comparisons explains matching maturity and observation date. When changing this example, retain the date and tenor of each new investment rather than immediately compressing four rates into an average. That keeps the source of the income change visible.

Trace assets and deposits when rates reverse halfway through

Now assume market rates are one percentage point higher for the first half-year and then return to their original level. The first two tranches of 25 lock in 3% and remain invested to year-end. Principal maturing in quarters three and four is reinvested at the original 2%. Additional first-year receipts are therefore only 0.25 and 0.1875, totalling 0.4375.

If deposit rates also return to their original level halfway through, the extra 50%-pass-through cost applies for only half a year: 100 × 0.5% × 0.5 = 0.25. The net gain is 0.1875. That exceeds the 0.125 gain when rates stay high all year, not because asset yields rise further, but because previously locked-in receipts persist while extra funding payments decline earlier.

If customers have moved into higher-paying fixed-term deposits, however, falling market rates may not reduce costs immediately. If the additional annual cost stays at 0.5, subtracting it from receipts of 0.4375 produces −0.0625. The same market-rate path gives an opposite result because the deposit reset dates differ. Downward repricing terms matter as well as the size of the initial increase.

This does not support a new claim that cuts always help banks. Rapidly falling new-loan rates, floors on deposit rates or borrower prepayments could reduce receipts first. Recording which contracts reset first on both the upward and downward legs explains competing outcomes better than a single forecast of the rate peak.

Align interest accrual with the dates cash is received

The quarterly increases above allocate interest income and expense to each period. They do not assume that matching cash arrives at each quarter-end. Annual interest receipts funded with monthly payments can require interim cash even when the annual interest difference is positive. That is why accrual and settlement need separate clocks.

In a practical worksheet, separate the date principal returns, the reinvestment date and coupon payment dates. If money waits at a lower yield after maturity, it cannot immediately earn the higher assumed rate. A pre-arranged transaction still requires checking when cash is needed. Differences created by date conventions should not be presented as evidence of superior investment or forecasting ability.

For example, suppose the first 25 remains in a waiting account for the first quarter without the additional yield, then earns 3% from quarter two. Relative to the original equal ladder, the additional-interest reduction is 25 × 1% × 0.25 = 0.0625. This comparison assumes the waiting account earns the original 2%. A non-interest-bearing balance would forgo more. Without stating the waiting yield, the cost of delay can be made too small or too large.

The distinction also reaches companies borrowing from banks. A loan’s interest reset and its cash debit can occur on different dates, changing working-capital needs. The guide connecting the three financial statements provides the earnings and cash background. Applying this article’s quarterly schedule to a company cash plan requires principal repayments and settlement dates alongside the interest calculation.

Holding an asset does not remove its economic exposure

There is a valid element in the argument that a fixed bond’s price decline need not matter if it can be held to maturity. If contractual payments arrive, no interim cash is required and holding remains possible, the market loss need not be realized that day. But that does not remove the opportunity cost of low locked-in receipts or the consequences of needing cash earlier. The ability to wait belongs in the assessment.

The IFRS Foundation classification overview uses asset management and contractual cash flows as classification foundations. In a research note, retain the company’s actual category and record holding objectives and funding dates separately. Substituting a preferred valuation presentation does not change contractual payments or cash needs.

The Basel Committee’s SRP 98, paragraphs 98.1–98.6 and 98.15, distinguishes prospective earnings, economic value and accounting recognition, and addresses mismatched asset and liability reset dates. A bond-price calculation alone does not measure the economic value of an entire bank, which also requires liabilities, derivatives and customer behavior.

Selling early to reinvest at a higher yield is not a free solution. It involves the sale price, transaction costs, funding and accounting effects. Holding may also be unavailable. Check alternatives to sale for short-term cash needs and the collateral they require. Where public information cannot establish that flexibility, an interest sensitivity alone cannot certify adequate liquidity.

How the interest effect reaches loan pricing and funds allocation

A bank whose funding costs reset before asset income catches up may seek a larger spread on new loans. A bank with stable deposits and quicker asset resets may face less pressure under the same rate change. This does not forecast lending across countries from one bank’s sensitivity. It explains why comparing borrowing conditions requires bank funding terms as well as loan benchmark rates.

For a company, loans linked to the same benchmark can still differ in spreads, fees, collateral, maturity and available commitments. A bank need not pass improved interest income into cheaper lending. Retaining capital, distributions, liquidity and existing cost burdens compete for the benefit. Higher bank interest receipts do not establish that companies will find borrowing easier.

Cross-border comparisons must also align currency. Funding deposits in one currency and investing in another introduces conversion and hedging costs omitted by a simple interest differential. Short funding supporting longer assets requires financing terms at successive dates. The same-currency, constant-balance example here is a starting point without those costs, not a forecast of cross-border margins.

Higher interest income is not the same as a higher shareholder return. Credit costs, expenses, tax, capital and expectations already in the share price matter. The guide to P/E, P/B and ROE explains why short-term earnings improvement does not establish cheapness. The pass-through threshold is evidence for examining a rate path, not an equity valuation or trading signal.

Before adding a hedge, identify what it fixes

Actual banks do not manage rates solely by waiting for maturities. Interest-rate exchanges can alter receipts and payments without buying or selling an asset’s principal. Multiplying bond holdings by a rate change therefore cannot reproduce whole-bank net interest income. Identify the rate paid, the rate received and how long those terms remain in place.

A hedge notional is not additional lending capacity or cash on hand. It is the reference amount for contractual interest calculations, not freely available money of the same amount. Adding bond principal repayments to expiring hedge notionals mixes different functions. The task is to align reset dates, not to compare the apparent size of totals.

A contract fixing investment returns may expire and be renewed before the underlying asset legally matures, changing rate sensitivity on another date. Conversely, a hedge may persist after a bond matures, so reinvesting principal does not fully reset the combined rate terms. Without the contractual combination, converting all “maturities this year” into one repricing fraction can make the threshold meaningless.

Keep a public-data reconstruction narrow when disclosures are aggregated. If receipt and payment reset dates are unavailable, show results under explicit alternative assumptions rather than attaching the company’s name to a fictional ladder. A large difference from the disclosed model does not immediately demonstrate an error. Check omitted contracts, changing balances, the yield curve and customer behavior first.

Capital adds another condition between valuation and lending

Improved net interest income does not create an equal amount of lending capacity. Regulatory capital has defined adjustments and a risk-based denominator. The interest worksheet contains neither. It also omits expenses, credit losses, tax and dividends between interest income and retained earnings. A positive ladder result cannot simply be copied into regulatory capital or new lending volume.

When tracing fair value into capital, align pretax valuation amounts with capital movements that may include tax, currency and other adjustments. Dividing one valuation number by risk exposure does not necessarily explain a disclosed ratio change. A numerator can rise while a faster-growing denominator lowers the ratio; a falling ratio alone does not establish that the capital amount fell.

The article on large-bank relative scores and intermediation provides related context, but this calculation does not use other banks’ assessment scores as its denominator. It begins with the earnings consequences of asset and funding reset dates, then examines capital allocation. Connecting opening and closing balances for the same bank under the same definition is more useful than immediately ranking ratios across institutions or regimes.

Even ample capital need not produce more lending if demand or returns are weak. Conversely, other revenue, capital raising or portfolio changes can absorb pressure on interest income. Assessing actual lending conditions also requires capital, liquidity and borrower-demand evidence. The calculation here isolates rate timing rather than certifying a bank’s soundness.

Turn a maturity schedule into a comparison that evidence can overturn

For a personal worksheet, fix a comparison horizon such as one year and align currency and units. Record principal, maturity or reset date, old yield, new yield and the investment period remaining within the horizon. Each row’s additional interest is principal × yield difference × time. Sum those rows, then subtract funding balance × rate difference × applicable time over the same horizon. Convert percentages to decimals: 1% becomes 0.01.

Next change only the maturity sequence. A large front-versus-back difference makes individual dates more important than the annual total. When changing the rate path, check that already fixed contracts have not been repriced again. When changing pass-through, either keep balances constant or add outflows and replacement funding separately. A result that changes several assumptions at once cannot be attributed to only one cause.

Finish with the evidence that would change each result. Later-than-expected resets delay additional receipts. Migration into higher-paying deposits raises costs. Earlier funding-rate cuts reduce the burden after reversal. These are observations that identify particular worksheet inputs, not a generic caution list. More announcements alone need not alter the conclusion.

The table is for manual copying and use. It does not submit inputs, save results automatically or determine a bank’s eligibility. Users provide their own maturity and contractual data and interest-rate evidence for the comparison date. Confidential inputs need not be sent to this article. Keep them in notes you control and vary assumptions to identify missing information.

SG Group View: examine the time remaining, not just the direction

SG Group’s view is that rate analysis should move from the yield reached at year-end to the period during which that yield can actually be earned. In the equal ladder, every asset earns the new rate by year-end, yet first-year additional interest is only 0.625 on principal of 100. Funding resets immediately, allowing the first quarter and full year to have different signs. Retaining that sequence takes the analysis beyond a binary gain-or-loss label.

The view is supported where documents establish dispersed maturities, fixed asset receipts and earlier funding resets. Evidence of rapidly repricing assets, delayed deposit changes or offsetting contracts could show that the fictional lag overstates exposure. Conversely, lower reinvestment yields or movement into more expensive deposits would weaken the expectation that additional receipts remain with the bank.

At the next reporting date, look beyond one large earnings number. Compare the repriced asset share, maturity distribution, applicable funding-cost periods and deposit mix under unchanged definitions. Do not fill missing dates or contractual terms from a headline or average remaining life. Without specific reset dates, retaining front-loaded and back-loaded scenarios is more defensible than presenting the threshold as a forecast.

The broader market implication remains conditional. If many banks incur higher funding costs first, pressure on lending spreads or allocation may persist. But adding individual sensitivities does not establish a global credit-contraction amount. The useful takeaway is a procedure: for the same rate change, identify which principal receives new terms, from what date and for how many months, rather than assert a market direction.