Between late February and the end of March 2026, FrictionIndex showed insolvency stress building sharply across several UK sectors simultaneously — the most severe episode in our data since records began in 2007. The tables below show exactly what our weekly data captured, and when the official statistics later caught up.
| Month | FrictionIndex MoM | Official (E&W) | Official MoM | Official data published |
|---|---|---|---|---|
| Sep 2025 | -2.0% | 2,000 | -2.3% | 16 Oct 2025 |
| Oct 2025 | -0.4% | 2,029 | +1.4%† | 18 Nov 2025 |
| Nov 2025 | -4.6% | 1,866 | -8.0% | 19 Dec 2025 |
| Dec 2025 | -9.0% | 1,671 | -10.5% | 20 Jan 2026 |
| Jan 2026 | +5.4% | 1,749 | +4.7% | 17 Feb 2026 |
| Feb 2026 | +38.5% | 1,878 | +7.4% | 17 Mar 2026 |
| Mar 2026 | +24.7% | 2,022 | +7.7% | 17 Apr 2026 |
| Apr 2026 | -2.3% | 2,085 | +3.1%† | 19 May 2026 |
| May 2026 | -5.0% | 1,868 | -10.4% | 19 Jun 2026 |
Every month for the past year, FrictionIndex's direction has matched the official insolvency statistics' direction — both rising, or both falling, in the same month — in 9 of the last 10 months. The difference is timing: FrictionIndex is available every week, in real time. The official government figures for the same period aren't published until three to seven weeks after the month ends.
The months where our monthly average technically diverged from the official direction — October 2025 & April 2026 highlight the importance of weekly data rather than monthly. October 2025 reflects a turning point that happened mid-month. FrictionIndex's weekly data shows the rise beginning in the third week of October. A single monthly average partially obscures a move like this. If anything, this is a demonstration of why weekly data matters: a monthly snapshot can blur exactly the kind of inflection point that FrictionIndex shows happening in real time, week by week. April 2026 shows the same dynamic from the other side. FrictionIndex's Real Estate reading had already peaked in late March and was falling sharply through April — because the underlying filing activity had passed. The Insolvency Service's own April release confirmed that the same cluster of connected Real Estate companies spanned both March and April, meaning some of those administrations were still being formally recorded weeks after FrictionIndex had already shown the event happening. This is the lead-time advantage working exactly as intended: FrictionIndex shows the event once; the official count can take two months to fully catch up to it
In most months, FrictionIndex and the official statistics don't just agree on direction — they agree closely on the size of the move too (e.g. Dec 2025: -9.0% vs -10.5%; Jan 2026: +5.4% vs +4.7%). In a handful of months the direction matches but the size of the move differs noticeably (Nov 2025, Feb–Mar 2026, May 2026). In each case the difference is traceable to a specific, verifiable cause — not to the two sources disagreeing about the underlying trend.
October's official figure was artificially inflated by the Insolvency Service's own system migration — they confirmed that compulsory liquidations for October "have been revised more than usual" as a result of moving to a new case management system on 1 November. November's sharp official fall is largely a correction from that distorted October base. FrictionIndex, drawing directly from Gazette filings and unaffected by the Insolvency Service's internal systems, showed the truer, steadier picture.
The Insolvency Service's own March release confirms both series are telling the same story — insolvencies "between November 2025 and February 2026 were lower than levels typically seen," and March's rise "followed four months of numbers that were lower than those typically seen." The size of the moves differs because the two measures have different denominators. FrictionIndex is a ratio — deviation from a 52-week rolling baseline — so a surge off a quiet base produces a large percentage move. The official count is an absolute number always running in the thousands, so the same surge looks proportionally smaller. Additionally, the large connected-company cluster that drove this episode was captured by FrictionIndex in real time as gazette notices were filed through February and March; the official count processed those same cases across two monthly publications, spreading the impact across March and April rather than showing it concentrated as it actually arrived.
April's official count was inflated by the tail of the connected-company cluster still being recorded. May's sharp official fall partly reflects correcting from that inflated April base. FrictionIndex had already captured and reflected the cluster in March, so by May it had less distance to fall — and showed the smaller, more accurate move.
The February–March episode illustrates something beyond the headline percentage differences. The official February figure was +7.4% — a single number with no information about timing, trajectory, or concentration. FrictionIndex's weekly data showed the full picture: a clear, accelerating trajectory that was already visible and deteriorating before the official February figure even existed. Anyone relying on the monthly official data didn't just receive the information late; they received a smoothed average that concealed the shape and velocity of the stress entirely. The monthly figure gives no indication that by month-end conditions were running at a level that would eventually require the Insolvency Service to acknowledge an unusual cluster. The weekly data showed the acceleration as it happened.
Sector-level detail from the February–April 2026 episode
FrictionIndex showed: Index rose sharply over a two-week window
"Mostly driven by more than 100 connected companies in the Real Estate sector entering administration"
FrictionIndex showed: Sustained at an elevated level
Cluster confirmed continuing — connected-company administrations across March and April combined
FrictionIndex showed: Index climbed steadily to a multi-month peak
Monthly construction insolvencies rose 14% in March vs February (301 → 347)
FrictionIndex showed: Index nearly doubled over the episode
161 pub closures across Britain in Q1 2026 — "almost two a day"
FrictionIndex showed: Index more than doubled
Consistent with ongoing trade body warnings of rising manufacturer distress
FrictionIndex showed: Index climbed to a peak — building after the main episode had already peaked elsewhere
Martyn Barratt Transport Ltd (25-year Nottingham haulage firm) — administrators appointed 13 Apr; Gazette notice 21 Apr
Unlike Construction or Real Estate, there's no fast official sector figure for Transport to compare against — detailed industry breakdowns are only published quarterly. FrictionIndex showed this sector's stress building week by week regardless, including capturing the Gazette filing for a 25-year haulage firm the week it happened.
In plain English — no statistics jargon
Looking back over six years of data, when FrictionIndex has been at today's level, UK small business borrowing costs have gone on to rise within the following 12 months roughly 4 times out of 5. We tested this twice, independently, on two separate halves of our history — both times it held.
When our Construction reading has been elevated, banks' real estate losses have historically run roughly three times higher than when it's been normal — averaging £55m a quarter versus £18m. We're not calling this proven yet — but it's a strong early pattern we're continuing to track.
Every time our index has stayed in "Elevated" territory for a month or more, it has fallen back within two months — in every single case across 20 such episodes since 2007. That's exactly what a genuine stress indicator should do.
We hold ourselves to the same standard we'd want from any data source: we test rigorously, and we only publish what survives independent re-testing. Relationships that may have looked promising at first but didn't hold up to our scrutiny and were excluded.
FrictionIndex captures the same underlying corporate filings as they happen — it does not forecast events before they occur. The evidence above demonstrates speed and visibility relative to official statistics, not predictive forecasting. Source for official figures: Insolvency Service, Company Insolvency Statistics (England and Wales, seasonally adjusted), monthly releases, gov.uk.