The Structural Spread, the DTI Ceiling, and the Two-Instrument Regime

The standard market narrative treats interest rate movements as the complete story: the RBA adjusts, retail rates follow, borrowing capacity shifts. That narrative is not wrong so much as incomplete. The certified record of Interest Rates (21210) exposes three structural features that the announcement-to-announcement framing systematically misses — and each of them matters more for portfolio and credit strategy than the next quarter-point move.

The structural spread

Across the full fifteen-year certified baseline, the retail lending rate (μ = 6.2400%) sits approximately 388 basis points above the cash rate (μ = 2.3558%). At the certified June 2026 observation the gap is 442 basis points. The spread has survived two complete easing-then-tightening cycles without compressing — it did not narrow when the cash rate fell to 3.60% in 2025, and it did not widen as the rate returned to 4.35% in 2026. This stability tells us the spread’s driver is structural — bank capital requirements under APRA’s Basel III implementation, wholesale funding cost pass-through, and competitive margin — not cyclical. The practical consequence: every model that mean-reverts the borrower’s rate toward the cash rate plus a cyclical premium will systematically overstate the benefit of future easing and understate the ongoing cost of holding.

The quantity ceiling

APRA’s February 2026 activation of the 20% high-DTI origination cap introduced a second, independent constraint instrument operating on a fundamentally different axis. The price instrument (the cash rate, transmitted through the buffer) adjusts the cost of each dollar of credit. The quantity instrument caps the volume of high-leverage credit the regulated system may originate, regardless of price. The certified high-DTI vector shows that high-leverage origination was running at −1.07σ at the June 2026 observation — well below its 12.5% historical average — consistent with the cap continuing to bind. The cap appears to have crystallised existing behaviour rather than forcing new behaviour, which means it will bind most visibly if conditions ease: a rate cut will lower the price of credit but leave the quantity ceiling exactly where it is. Any borrower segment that requires DTI ≥ 6 will find the ceiling unresponsive to monetary policy — a structural break from the pre-2026 regime in which price was the sole instrument.

The 2025–26 round trip as a natural experiment

The rate cycle running from early 2025 to mid-2026 constitutes a near-perfect natural experiment for the spread and the cap, because the cash rate ended the June quarter 2026 where it began, at 4.35%, before a further rise to 4.60% on 30 September. Three findings emerge. First, the retail spread was stable in both directions — it is not an artefact of lags or stickiness but a structural feature. Second, the cap arrived in the tightening leg, when high-DTI origination was already well below its long-run average: it ran at 5.1% to 6.6% of new lending through 2025, against a baseline average of 12.5%. Third, the forward yield vector did not return to its pre-easing level when the cash rate approached its pre-easing level — at the certified June observation it stood at +1.85σ, against +1.40σ in December 2024, pricing a more persistent restrictive regime than the one markets expected in late 2024. The forward curve’s information content is that this cycle is settling into a structurally higher floor, not a return to the prior equilibrium.

The wholesale funding pass-through and duration sensitivity

The structural spread above is stable at the level of the whole retail book. This section addresses a different question: what determines how quickly and by how much a given bond-yield move reaches the fixed-rate retail book specifically, as distinct from the variable book.

A bank writing new fixed-rate loans funds that book, at the margin, against swap and government bond yields of matching duration — not against the cash rate. The pass-through is close to immediate for new originations: a bank cannot profitably write a 3-year fixed loan at yesterday’s swap rate if today’s swap rate has moved against it, so new fixed-rate quotes reprice within days of a material bond-market move. The pass-through to the back book — loans already written — is by construction zero; a fixed-rate loan does not reprice mid-term regardless of what happens to yields afterward. This asymmetry means the practical impact of a bond-market repricing event is concentrated entirely on new originations and refinancing decisions, not on the existing fixed-rate loan stock, which is the opposite of the variable-book transmission profile addressed above.

The certified March 2026 record is consistent with this mechanism: the forward-yield vector’s Z-score (+2.17σ) exceeded the cash-rate vector’s Z-score (+1.20σ) by the widest margin since September 2022, which is exactly what a genuine funding-cost-driven repricing of the fixed book — as distinct from a cash-rate-driven repricing of the variable book — would produce. Portfolio and pricing models that treat “the interest rate” as a single input will misprice this dynamic; a duration-aware model that separates the fixed book’s funding cost from the variable book’s policy-rate exposure is required to capture it correctly.

Advanced practitioner implications

Credit analysis: sensitivity models should stress not only rates but cap headroom — a borrower who passes serviceability but falls into the DTI ≥ 6 band faces a constraint that is invisible in standard affordability calculators. Portfolio strategy: the two-instrument regime means the next easing cycle, when it arrives, will not replicate pre-2026 transmission — the cap clips the high-leverage tail, which historically amplified the price signal on the way down. Modelling the next cycle on pre-2026 relationships will overestimate the recovery. Valuation: capitalisation rate assumptions anchored to a long-run average cost of capital should adjust for the spread permanence — the borrower’s cost of capital has a structural floor approximately 440bp above whatever the RBA sets, and it is on the borrower’s cost, not the policy rate, that asset valuations clear. Fixed-rate book management: lenders and analysts modelling fixed-rate revenue should track the forward-yield vector directly rather than assuming it tracks the cash rate with a fixed lag — the March 2026 divergence shows the two can decouple materially within a single certified quarter.