constructioncapital/cmb-city-charge-comparisons

City commercial charge comparisons: H1 2025 and H1 2026

Seven of the 14 published city reporting areas switched direction between their Q1 and Q2 year-on-year comparisons. A positive first-half result therefore need not describe both quarters. This small, reproducible dataset lets analysts inspect that distinction using counts, rather than a single headline percentage.

Prepared by Matt Lenzie, Commercial Mortgages Broker, a commercial mortgage brokerage. Release CMB-CH-2026-H1-D02, 11 October 2026. These are derived tables from the same CMB report programme as the regional and county research package, not an independent corroborating study. AI assistance was used in preparation and checking.

What the comparisons reveal

In Nottingham, Q1 rose from 19 to 39 recorded commercial charge instruments, while Q2 fell from 26 to 16. The quarter changes, +20 and -10, combine to an H1 increase of 10, from 45 to 55 (+22.2%). Describing this as sustained growth across both quarters would lose half the story.

Birmingham shows the opposite sequence: Q1 fell from 55 to 51, then Q2 rose from 43 to 85. Its +38 H1 change is -4 plus 42. Sheffield was unchanged across H1 (50 to 50), although Q1 fell by six and Q2 rose by six. An unchanged half-year total can conceal a reversal too.

The seven strict reversals are Birmingham, Cardiff, Leicester, Northampton, Nottingham, Sheffield and Wolverhampton. Northampton, Nottingham and Wolverhampton combine positive H1 growth with a negative Q2 year-on-year change. These are comparisons with the same quarter of 2025, not changes from Q1 2026 to Q2 2026. Brighton's Q2 change is zero, so it is not counted as a strict reversal.

Inspect all 14 areas

Counts are commercial charge instruments in the source-defined cohort. Quarter columns show 2026 minus 2025. Each city name links to its exact CMB report.

City report	H1 2025	H1 2026	Q1 change	Q2 change	Strict reversal?
Birmingham	98	136	-4	+42	Yes
Brighton	44	59	+15	+0	No
Bristol	73	91	+3	+15	No
Cardiff	51	47	+4	-8	Yes
Leeds	73	91	+9	+9	No
Leicester	60	48	-17	+5	Yes
Liverpool	139	77	-42	-20	No
London	1250	1277	+17	+10	No
Manchester	58	66	+6	+2	No
Newcastle upon Tyne	30	46	+3	+13	No
Northampton	41	46	+9	-4	Yes
Nottingham	45	55	+20	-10	Yes
Sheffield	50	50	-6	+6	Yes
Wolverhampton	31	45	+16	-2	Yes

The comparison identifies variation within the first half; it does not establish why a city's recorded counts changed. Registration lag, classification and a small starting count can affect the reading.

Files and loading

• city-comparisons.csv: 14 rows. H1 counts and growth, quarter changes, directions, reversal flag, area definition and exact report/CSV links.
• quarter-counts.csv: 56 rows, one per city, year and quarter. Q1 and Q2 of 2025 and 2026 only.
• source-evidence.json: 98 selected public charge fields, 14 full-source CSV SHA-256 hashes and checked URLs. No names, addresses, lender identities, sales or planning observations.
• codebook.json: field definitions and units for both tables.
• reproduce.py: standard-library Python script that regenerates both CSVs from the archived fields and compares their bytes with the release.
• RIGHTS.md: rights and attribution notes. No new open reuse licence is granted.

from datasets import load_dataset

cities = load_dataset(
    "constructioncapital/cmb-city-charge-comparisons",
    "city_comparisons",
    split="train",
)
reversals = cities.filter(lambda row: row["strict_reversal"])

For plain Python, download the package files and run python3 reproduce.py. Expected result: 14 cities, 56 quarter rows, 28 H1 reconciliations, 14 published-growth checks and seven strict reversals. The Hub's train split is a storage label only; no train/test benchmark or forecasting suitability is asserted.

Definitions and method

The source is the 2026-q3 edition, labelled Jan to Sep 2026. This package compares 1 January to 30 June 2025 with 1 January to 30 June 2026, using the source's Companies House charge snapshot dated 1 October 2026. It does not use provisional Q3 counts. Read the CMB programme methodology alongside the city reports.

For each city, H1 = Q1 + Q2. H1 change = H1 2026 - H1 2025; percentage change = 100 times that difference divided by H1 2025, rounded to one decimal. A strict reversal requires one quarter's year-on-year difference to be positive and the other's negative. A zero difference is classified as flat.

The source builder assigns charges by creation date, not delivery date, although source-page prose uses registration terminology. Geography uses the first recorded charge-property postcode, mapped through a postcode-outcode/local-authority bridge. This is a reporting proxy, not a verified boundary lookup for every secured property. London covers the source's 33 borough areas; the other labels refer to the council areas stated in the area column. Manchester is not Greater Manchester.

The commercial cohort uses the source's borrower-sector classification or a matched commercial-sale property reference. It excludes the separate property-company-only series. One instrument is not necessarily one property, borrower, loan or drawdown; counts reveal neither loan values nor credit availability.

Coverage and limitations

These are the 14 city reports published in this edition, not a census of UK cities or a random sample. The source publication filter requires at least 40 observed sales and 50 commercial charges over its edition window; those thresholds do not apply to each H1 count. The selection depends partly on sales coverage even though this package contains no sales analysis. Missing cities do not imply zero charges. Do not sum these areas into a national figure or compare levels as if population and geography were equivalent.

All 14 source pages and CSVs were retrieved on 11 October 2026. Each CSV hash matched the frozen source release. Arithmetic checks validate the published aggregates; they do not independently audit raw charge records, postcode allocation, deduplication or classification. Later source revisions may change the results. The small, selected panel is unsuitable on its own for lending decisions, causal claims or model-performance claims.

Attribution, citation and version

Underlying register: Companies House. Source classification and reporting: Commercial Mortgages Broker. New work: the tidy city/quarter tables, quarter-direction comparison and reproduction code. See RIGHTS.md; upstream rights are retained and no third-party rights are relicensed.

Suggested citation: Lenzie, Matt / Commercial Mortgages Broker (2026). City commercial charge comparisons: H1 2025 and H1 2026. CMB-CH-2026-H1-D02, release 1, 11 October 2026. Cite the specific source city report for an individual finding and the repository revision for this derived table.

Release 1 contains the 14 published city areas. Later corrections should update this repository with a documented change note. Questions about source definitions can be directed through the contact details on the linked CMB reports.