Functional versus administrative regions

Why do the regions people actually live and work across so rarely match the ones on the administrative map?

Somebody sleeps in one municipality, works in a second, takes a child to hospital in a third and shops in a fourth. The transport authority, the health board and the tax office each draw a different line around them, and none matches the boundary printed on the map in the town hall.

It is tempting to call this administrative failure — boundaries left too long unrevised. Let us test that, because if it were only lag, a recent reform would fix it, and reforms do not.

Start by asking what a functional region is, because it is not something you can go and look at. You observe flows: how many people commute from A to B, how many patients travel from A to hospital B. That gives a matrix of interactions, not a map. To get a map you group places until most of the interaction is captured inside the groups — and "most" is a number somebody picks. Britain's travel-to-work areas are built to a self-containment target somewhere around three-quarters, traded against a minimum workforce size; I am recalling those parameters rather than quoting them, and nothing here depends on the figures. What matters is that the threshold exists and is chosen.

Move the threshold and the regions move. That is exactly the problem modifiable-areal-units.md sets out, and this piece is its mirror: there, any zoning of fixed data yields a different statistic; here, even the "true" functional map is one of a family generated by a choice of cut-off. So "just find the real region" is not available as a fix.

Now the timing argument, the one most people reach for. Functional geography moves with transport and with where firms locate: a new rail line, a widened road, a shift to two days a week in the office, and last decade's labour market area is wrong. Administrative geography moves by statute. Any redraw redistributes revenue, staff, offices and identity, producing concrete, identifiable losers who can lobby, litigate or vote, against beneficiaries whose gain is diffuse. That asymmetry is a strong status-quo bias, so boundaries change once every few decades while flows change continuously, and between reorganisations the mismatch can only grow.

But push on that, because it predicts something checkable: if lag were the whole story, a jurisdiction that redrew its boundaries last year should show almost no mismatch. It does not, and the reason is structural rather than temporal. Administration needs a partition — exhaustive, non-overlapping, one authority per address, stable enough to legislate and tax over. Function produces nothing of the sort. Commuting fields overlap; hospital catchments have fuzzy edges; retail catchments are graded rather than bounded. Worse, they do not coincide with each other — the area within which people find jobs is not the area within which they find a hospital. So even a current administrative map must fail, being asked to be one partition matching several non-coincident, overlapping fields. No line can do that.

That gives a sharper falsification test than "boundaries are old". The structural account predicts, first, that the several functional maps of one territory will disagree with each other, not merely with the official map — so find a region where labour markets, hospital catchments and school catchments delineate the same boundaries and the account is in trouble. Second, it predicts direction under change: improving transport into a metropolitan core should lower the self-containment of outer municipalities, pulling their functional region outward. If a new fast rail link raised outer self-containment — if people commuted less far once travel got cheaper — the mechanism I have described is not the one at work.

Two neighbours diverge at identifiable points. borders.md asks why lines run where they do — rivers, ridges, surveyors' coordinates — and answers from the history of who was drawing. This piece takes the lines as given, whatever their origin, and asks why they mismatch present-day flows; a border drawn perfectly on functional grounds in 1900 would mismatch today just as badly. district-boundaries.md concerns a partition drawn deliberately to convert votes into seats — intentional design against a known objective, whereas the mismatch here appears with nobody intending anything.

Three folk explanations deserve retiring. "Historical accident and inertia" is true and insufficient: it cannot explain why mismatch reappears immediately after a careful, evidence-led redraw. "Just annex the suburbs" fails twice — the functional edge keeps moving outward, so the annexed city is behind again within a decade, and the functions disagree about where the edge is in the first place. "Better data will reveal the true region" fails at the definition: the region is threshold-defined, so more data sharpen the flow matrix without producing a unique map.

One honest qualification, and one omission. I have treated commuting as the canonical case because it is the best measured; whether a labour market area is a sensible unit for planning health or water is contested and plainly differs by service. And there is a real defence of stable administrative boundaries this piece has not made — legal certainty, accountability, and comparing a place with its own past all require lines that do not move each time the flows do.

• How metropolitan governance bodies try to act on a functional region without redrawing anyone's boundary.
• Whether mobile-phone and transaction data change the picture by exposing rhythms commuting surveys miss.

Every term the collection defines is gathered in the glossary.