Methodology: The New Business Map: UK

Methodology v1.0, 28 October 2026. Applies to report v1.0. This edition uses business births to 2024; it will be updated when newer official data are published. Changes are versioned and corrections are logged at the foot of this page.

1. Research question

Across UK local authorities, how did the rate of new-business formation change between 2019 and 2024, and how does that change relate to local household income, unemployment-related benefit claims and self-employment? The study is descriptive and associational. No causal claim is made or implied.

2. Data sources, vintages and terms

Contains public sector information licensed under the Open Government Licence v3.0. Contains OS data © Crown copyright and database right. Contains Royal Mail data © Royal Mail copyright and database right. Source: Office for National Statistics licensed under the Open Government Licence v3.0.

3. Definitions

• Business birth. ONS counts an enterprise as born in a year when it registers for VAT and/or PAYE for the first time and was not active in the two previous years. Businesses below the VAT threshold with no employees are not included. This is a stricter measure than Companies House incorporations or US business applications, and closer to "a trading business".
• Births per 1,000 adults = 1,000 × births / residents aged 18+ in the same year.
• Change in births per adult = (rate 2024 / rate 2019 − 1) × 100.
• Real GDHI per head change = GDHI per head 2024 / (GDHI per head 2019 × CPIH 2024 / CPIH 2019) − 1.
• Claimant rate change = average monthly claimant rate in 2025 minus 2019, in percentage points.

4. Geography and boundary changes

All results are on the 361 local authorities as of April 2023. Births for earlier years are published on the boundaries in force at the time and were re-aggregated using the official reorganisation orders: - 2019: Bournemouth, Christchurch and Poole; Dorset; East Suffolk; West Suffolk; Somerset West and Taunton. - 2020: Buckinghamshire. - 2021: North Northamptonshire; West Northamptonshire. - 2023: Cumberland; Westmorland and Furness; North Yorkshire; Somerset.

The 2018 Scottish code changes (Fife, Perth and Kinross, Glasgow City, North Lanarkshire) were also applied. Check: for every year from 2017 to 2024, the re-aggregated council births sum exactly to the published UK total. The run stops if any published code is left unmapped.

Median pay cannot be re-aggregated. For merged councils whose boundary equals a 2019 county (Buckinghamshire, North Yorkshire, Somerset), the 2019 county median is used. The four councils created by splitting a county (North and West Northamptonshire, Cumberland, Westmorland and Furness) have no 2019 pay comparison.

5. Universe and exclusion rules

1. Population threshold. Councils with fewer than 50,000 adults in either year are excluded (12 councils).
2. Registered-office hotspots. A council is excluded if its births per adult are at least 3 times the UK rate in either year (Bromsgrove, City of London, Camden, Hackney, Islington, Westminster). In central London, head offices and registered offices rather than residents drive the count; Bromsgrove's rate is similarly out of line with its population. A resident-based rate is not meaningful in these places.
3. Unstable base year. A council is excluded if its 2019 births exceed 1.4 times the larger of its 2017 and 2018 births (8 councils: Telford and Wrekin, Southampton, Hinckley and Bosworth, Redditch, Wychavon, Wyre Forest, St Albans, Denbighshire). These councils show 2019 counts far above both earlier years, more consistent with bulk registrations than with a genuine jump in local start-ups, and they would make any change since 2019 look misleadingly negative. The 99th percentile of this ratio among large councils is 1.704.
4. Mass-registration addresses (caution, not exclusion). Using the Companies House snapshot, 165 postcodes each hold at least 1,000 live companies, 652,645 companies in all. A council is flagged as a caution if at least 25% of its companies sit at such postcodes and its birth rate is at least 1.5 times the UK rate (Tower Hamlets). The share does not exclude on its own: councils with very high shares but normal birth rates (for example West Suffolk) show that ONS births are not driven by company-formation mailboxes.

The universe holds 336 councils with 96.1% of UK adults.

6. Statistical testing and robustness

• GDHI is a modelled estimate published without sampling error. Its relationship with business births is assessed by correlation with bootstrap 95% confidence intervals (2,000 resamples, fixed seed 20261028) and by medians within bands of income change.
• Self-employment (Annual Population Survey) is tested for each council using the published 95% confidence intervals: Z = change / √(SE₁² + SE₂²), SE = half-width / 1.96. Benjamini–Hochberg false-discovery-rate control is applied across councils.
• Pay (ASHE): standard errors come from the published coefficients of variation. Council-level full-time median pay changes are too imprecise to test: no council's real change is statistically clear even before correction. Pay is therefore reported only as a check, and GDHI is the income measure.
• Robustness. Every births result is repeated using three-year averages (2017–19 against 2022–24), which smooths one-off spikes and does not need the base-year rule.
• Business births are published counts (rounded to the nearest 5) with no sampling error.

7. Computation and verification

• Code: a Python analysis script (pandas, numpy, openpyxl, xlrd). Every threshold is a named parameter.
• A separate validation script re-reads the raw files with different parsing logic and no pandas, and recomputes UK totals, the re-aggregated council sums, the universe, the number of councils where births per adult fell, the median change and the income and claimant correlations. All checks match.
• Companies House snapshot: 5,704,711 companies were read; 97.14% were placed in a council via the registered-office postcode.

8. Known limitations

• Cross-sectional association across two periods; no causal inference. The income association is modest (r ≈ 0.2).
• ONS births cover VAT- and/or PAYE-registered businesses only. The VAT registration threshold rose from £85,000 to £90,000 in April 2024, which may lower 2024 births mechanically.
• Births are located at the enterprise's registered address, which can differ from where it trades.
• GDHI is modelled and has no published error; small-area estimates can be revised.
• Councils near the top or bottom of the tables can reflect one-year spikes; read individual ranks with the three-year comparison alongside.
• US and UK figures use different measures (EIN applications against VAT/PAYE business births). Compare directions and patterns, not levels.

9. Reproducibility, licensing and corrections

• All inputs are official UK statistics published under the Open Government Licence; this study redistributes only derived tables.
• Published derived table: one row per council (361 rows), with births, rates, changes, income, claimant and self-employment measures, exclusion reasons and flags. Licence: CC BY 4.0, attribution "Censal Research, The New Business Map: UK (2026)". The analysis code and fuller tables are available to journalists and researchers on request.
• Corrections: errors reported to the contact address are assessed within five working days and corrected in place with a dated note.

Correction log: none yet. v1.0 (28 October 2026) is the first edition.