How we made The £500,000 Question

Published 21 September 2026.

What the investigation can answer

This is an original comparison of published aggregate statistics, not a property valuation service or an individual-level causal study. It compares housing budgets across England and Wales, domestic migration between regions, and evidence about working patterns. It does not match people to purchases or to their working arrangements.

Editorial decision: proceed with angle B, London has a large domestic net outflow but most departures do not go to northern England. There is also a measurable rise in northbound gross moves, a temporary 2021 peak and a marked age contrast. We reject a claim that homeworking caused these moves or that migration failed to respond to prices: neither causal hypothesis is identified here.

Housing data and the £500,000 test

Source: ONS Median house prices for administrative geographies, year ending March 2026 edition, released 17 September 2026. Matching sales counts use the same edition and period. We use the published median sale prices, not asking prices or the mix-adjusted UK House Price Index.

The four types are detached, semi-detached, terraced, and flat/maisonette. The all-property median is a separate published statistic, never an average of the four. The data include the mix of new and existing dwellings represented by this ONS series. Prices are nominal pounds. A rolling year includes 12 months of recorded sales; registration lags can produce revisions. We did not substitute our earlier Price Paid Data extraction or any EPC records.

We extracted all 318 England/Wales local authorities before selecting editorial examples. Tables 3a–e supply district-level values, 2a–e county values; other sheets are checked by GSS geography code. We select the column headed “Year ending Mar 2026”, never assume it is the last spreadsheet column. Both prices and sales carry their original source values in the derived files.

For a valid median p, the budget ratio is 500000/p. A ratio of 1.45 means the budget is about 45% above the median; it is not 45% more floor space. We neither infer bedrooms nor claim the illustrated dwellings can be bought for this sum. Taxes, moving costs, mortgage access, maintenance and renovations are not included in the £500,000 benchmark.

Map categories and small samples

We count the number of type-specific medians at or below £500,000, from zero to four. Equality is included. This avoids imposing a monotonic flat–terrace–semi–detached ordering that the data do not always follow. Eight authorities have at least one reversal of that conventional ordering, including some thinly traded categories. A count of four means all four medians meet the threshold, not that all homes do.

We require at least 20 sales for each type before giving an area a map category. This is a transparent TRS editorial caution threshold, not an ONS confidence guarantee. Fifteen areas therefore appear as insufficient volume. The other 303 divide into 2 areas with zero medians below/equal to budget; 17 with one; 20 with two; 65 with three; and 199 with four. These are counts of authorities, not population-, stock- or transaction-weighted shares. The table retains published low-volume medians with their counts and an explicit warning. Missing/suppressed values remain unavailable, never zero.

Boundaries and places

Housing data and the map use current 2025 local-authority identities. ONS December 2025 BGC boundaries are simplified by the official service to a 400-metre display tolerance and projected to British National Grid for plotting. All 318 map codes must join the housing table exactly. Sheffield and Barnsley changed boundaries and codes in 2025 (E08000039 and E08000038). The migration source intentionally uses 2023 boundaries (E08000019 and E08000016). We do not join their housing and migration observations as if these boundaries were identical.

Newcastle upon Tyne is a local authority, not the North East. Manchester is not Greater Manchester. London is the nine-digit ONS region comprising its local authorities. Cornwall is a unitary authority. Devon's county statistic includes Exeter, excludes Plymouth and Torbay, and is taken from the official county table; we never average district medians to create it. Devon appears as an additional comparison, not a 319th district in the national table.

Burnley and Hyndburn were chosen after national extraction as two low-price inland contrasts with substantial all-property sale counts. They are not a ranking of liveability. Featured cities and counties differ in size, housing mix and administrative extent. We do not construct a city attractiveness league table.

Affordability: a different date and denominator

Source: ONS ratio of house price to residence-based earnings, 2025/current workbook, released 26 March 2026. District tables 5a–c and county tables 3a–c provide price, earnings and published ratio. Prices refer to the year ending September 2025; earnings are 2025 median gross annual full-time employee earnings. The affordability chart uses those matching source prices, not the newer March 2026 medians. The national table labels both dates.

Residence-based means employees living in the area, wherever they work. Workplace-based describes employees working there, wherever they live. We use residence-based consistently for the local affordability argument and downloaded the workplace workbook to check the distinction. The regional values in the retrieved workbooks are London 11.35 and North East 4.94 on a residence basis; workplace equivalents are 10.61 and 5.00. These replace the approximate preliminary figures in the brief. They must not be mixed.

The ratio divides two separate medians. It does not represent a typical actual buyer, household disposable income, a borrowing limit or years someone must save. Earnings exclude the self-employed. Local estimates are subject to sampling uncertainty; small differences should not become ranks. City of London and Isles of Scilly have unavailable residence-based ratios in the source and remain unavailable in our table.

Migration: directions, years and reconciliation

Source: ONS Internal migration in England and Wales. The extraction uses the rebased 2012–2021 workbook, the revised 2022 and 2023 workbooks published 30 July 2025, and the current 2024 and 2025 summary tables. The 2025 release was published 29 July 2026. Each year ends 30 June; these are moves, not a count of unique people or houses bought.

In table T6, columns are origins and rows destinations. London→North sums the London column for North East (E12000001), North West (E12000002) and Yorkshire and the Humber (E12000003). The reverse uses the London destination row. Net to North is northbound minus southbound. The South East (E12000008) and East of England (E12000006) are calculated separately in both directions. Annual outputs cover 2012–2025; period means use 2017–19, 2020–21 and 2022–25. We compare means rather than totals across unequal-length windows.

London's outward denominator includes moves to all eight other English regions, Wales, Scotland and Northern Ireland. It excludes moves within London and international moves. We reconcile each year's column sum against the published regional outflow, allowing rounding differences of up to 12 moves. Destination shares use this common London origin, so a resident-population rate would not alter the destination comparison. We do not compare raw regional totals as migration propensity or publish unverified per-capita rates.

The city observations sum bilateral LA flows between each city and London's authorities in T2. We never sum total LA inflows to construct regional gross inflows, which would double-count moves within the region. The supplementary city table is alphabetical and descriptive, not ranked. Small net differences, such as York's approximately balanced exchange, should not carry a strong interpretation.

Age analysis

We stream the 2025 detailed origin–destination workbook by single year of age and sex, selecting only London↔North pairs. Both sexes are summed, then ages grouped into 0–19, 20–24, 25–29, 30–34, 35–39, 40–49, 50–64 and 65+. Ages refer to 30 June. Each region's all-age detailed total reconciles with the summary matrix to less than one move after rounding.

Age analysis is a 2025 cross-section, not a claim about a changing age profile since 2019. The annual trend is all-age. We cannot identify households, homeownership, students, graduates, parents or working patterns. People aged 30–49 being more likely to hybrid work in another survey does not make the movers in this table hybrid workers.

Changes in migration measurement

The ONS quality and methods guide explains the health-registration sources, student adjustments and scaling from transitions to moves. Census rebasing revised 2012–2021 estimates. Earlier register sources differ from the PDS-based approach used from 2017; the rebased series improves consistency. The revised 2022/2023 files incorporate source data unavailable for their first releases. We use those revisions.

GP registration lags, student estimation, rounding, disclosure controls and future revisions mean these are estimates. The shaded 2020–21 period marks disruption, not a causal intervention estimate. The more comparable 2017–19 baseline is the main pre-period; 2012–16 provides historical context. The 2026 population revisions chiefly concern international migration and should not be confused with a measured internal migration effect.

Working patterns and the historical series

The ONS hybrid-work article and supplementary tables use the Opinions and Lifestyle Survey in Great Britain, 8 January–30 March 2025. Hybrid means both homeworking and travel to work in the preceding seven days. The fourth category, employed but neither during that week, is retained. Home-only for a week is not necessarily permanently remote employment. ONS confidence intervals for the overall categories are retained in data/hybrid.json; estimates are rounded survey percentages.

We also extracted the Annual Population Survey monthly regional homeworking series, April 2017–March 2025. It measures usual homeworking using HOME categories 1–3, a different definition and UK geography. Crucially, its notes say interviewers were told to refer to pre-pandemic usual patterns from April 2020 until October 2021. Weighting also changes across the series. We therefore do not draw a spliced pandemic trend combining APS with OPN, or interpret that interval as contemporaneous uptake.

Monthly estimates for all regions are supplied in data/homeworking-monthly.csv, with the guidance regime labelled. March 2025's descriptive snapshots are London 34.0%, North East 27.1%, North West 22.6%, and Yorkshire and the Humber 27.2%. They are not precise region rankings or significance tests; monthly samples can be small, these data are not seasonally adjusted, and they are not hybrid percentages. The stable OPN definition since March 2022 supports ONS's assessment that hybrid work increased over that period; no synthetic continuous rate is manufactured here.

Interpretation and external evidence

Reuschke and MacLeavy (2026), Changing work/places, distinguishes remote intensities and identifies associations with mobility; it does not identify the working arrangements of the people in our ONS migration totals. McCollum and colleagues' ESRC Centre for Population Change report (2025) combines longitudinal survey analysis with case studies and finds limited redistribution of skilled residential mobility. Richard (2026) offers London property evidence and a general-equilibrium model of space/commuting mechanisms. Centre for Cities and Imperial College's Office politics (2023) supports discussion of economic concentration and proximity.

We reviewed distance/commuting research but found no commensurate national median move-distance statistic to join to this ONS series. The region data do not establish that commuters moved farther. No bespoke centroid-distance proxy is passed off as an observed journey. The “expanded commuter belt” remains a plausible interpretation to investigate, not our proven headline. Family, care and belonging appear as editorial questions, not survey findings about these movers.

Reproducibility, rights and review

sources.json records source URLs, retrieval times, file sizes and SHA-256 hashes. derive.py builds the analysis; charts.py exports SVG/PNG charts; build.py builds the feature and supporting package. requirements.txt records the small Python dependencies. Raw public aggregate workbooks and reproduction scripts are retained in the separate research archive. No restricted addresses, EPC linkage or non-commercial UCL dataset is used.

ONS statistical data are reused under the Open Government Licence v3.0 unless otherwise specified. Map attribution: Source: Office for National Statistics licensed under the Open Government Licence v3.0. Contains OS data © Crown copyright and database right 2026. Academic texts are cited and briefly paraphrased; their datasets/figures are not republished under an assumed OGL licence. Architectural images are original generated editorial illustrations of domestic possibilities, not listings, measured plans, occupied homes or representations of research participants. Art-direction summaries and master provenance are retained; original generation prompts remain in the creation conversation.