The last two notes described a curve rising in steps of seven to thirteen basis points a fortnight. This week it stopped stepping. GBP SONIA swaps rose sharply in the week to 10 September: the five-year is now around 4.75%, the three-year a shade over 4.7%, the ten-year a little under 4.9%. That is roughly thirty basis points on the five-year in a single week, about three times the size of the moves we have been recording, and it puts the five-year some thirty basis points above the late-July peak we spent August measuring everything against. The July high is no longer the reference point. It is simply a level the curve passed on the way up.
The cause this time is not hard to find, and it is not a mortgage-market story. Gilts sold off hard. The ten-year gilt yield touched roughly 5.3% on 10 September, its highest since 2007, and the thirty-year reached levels last seen in the late 1990s. The drivers are mostly global: a broad bond sell-off, oil, and renewed inflation worry. The domestic ingredient is the October Budget, now six weeks away, with fiscal headroom being eroded in real time by the very yields the market is setting. Swap rates are not gilt yields, but they are priced off the same expectations and they rarely part company for long.
What makes the week notable is what did not happen. The direct lender we follow most closely reissued its product guide this week with the five-year limited-company range untouched since the start of the month. The broker-panel lists we sample have not been refreshed since 1 September. Every product rate we can currently see was therefore set against a swap curve around thirty basis points lower than today’s. The spread between the lender’s no-fee five-year and the five-year swap has narrowed by roughly that amount in a week, and at the cheaper end of the panel lists it is now thin enough that we would be surprised to see those products survive the next refresh unchanged.
We have made this argument before, and it has so far resolved the same way each time: a lender holding pricing set against a funding cost that has since risen underneath it is carrying a margin it did not intend to carry, and the correction has only one plausible sign. The size of this week’s move makes it likelier to arrive quickly and less politely than the whole-range reprice we recorded at the start of the month. Short-notice withdrawals are the usual form.
It also changes how we read last note’s oddity, when broker-panel pricing drifted down into a rising curve. We suggested then that it was more likely a lag artefact than a signal. With swaps now thirty basis points higher still, that reading looks considerably safer. Whatever those lists were catching up on, it was not this.
A methodological note. Our primary swap source has put its two-year point behind a paywall, and the secondary source has still not refreshed its data since mid-August. This week’s reading therefore rests on one source and three tenors, and the two-year, the point that matters most for two-year fixes, is unobserved. We would normally discount a single-source reading of this size. We do not here, because the gilt market independently moved hard in the same direction over the same days. That is better corroboration than we had a week ago. We will look for a replacement two-year source before the next note.
The practical reading extends the one we have given since late July. Pricing secured before the summer continues to look better than anything on the shelf, and this week widened that gap again, before the shelf has even reacted. The more useful observation now concerns anyone whose short fix ends in the next year or so. Fixes set in mid-2025 were priced against a very different curve, and a refinance in 2027 will be priced against whatever this one becomes. With the Budget the next obvious catalyst in either direction, and new pricing typically only securable a few months ahead, the sensible work at this distance is budgeting for a higher rate rather than shopping for one.