Devolved authority

Why do large states hand decisions to local bodies they could just as easily make centrally?

A state that can legislate for its whole territory nevertheless hands schools, refuse collection, licensing hours and hospital budgets to bodies answering to a fraction of it. Nothing compels this. The capital has the lawyers, the statisticians and the money, and in most unitary states it can take the power back by ordinary statute whenever it likes.

So why give away a decision you are perfectly able to make? The stock answer — that local people should decide local things — restates the practice as a preference. What we actually want is the property of a decision that makes it worth moving down, and the property that makes it worth keeping up.

Begin with what a central decision is: one answer for everybody. Suppose two regions, and suppose the first would set some service at 100 and the second at 200 — figures chosen only to make the shape visible. The centre must pick a single number. Pick the midpoint and each region sits 50 away from what it wanted; pick either region's own figure and the other sits 100 away. Devolve, and each takes its own number, so the gap vanishes.

Now change one input and watch the argument die. Let both regions want 150. The central choice now costs nothing, and devolving buys nothing at all. Notice what that tells us: the value of moving a decision downward is manufactured entirely by the spread of what places want, not by centralisation as such. That direction holds no matter how you price the mismatch, which is why it is worth more than any particular number.

A second source of spread is informational. The condition of a bridge, which street floods, which school is short of teachers — these are cheap to know locally and expensive to send upward, because sending them means compressing them into categories the centre already has. Much of what a local body knows does not survive the journey to the capital in a form anyone can act on.

What, then, argues the other way? Ask what happens when a region's choice lands on someone outside it. A council that permits a smoky plant on its boundary collects the jobs and exports the smoke; a region that skimps on flood defence sends the water downstream. When effects cross the line that defines the decider, the decider is systematically choosing on incomplete accounts. Scale cuts the same way: a currency, a coastguard, a vaccine stockpile each cost far less per head at national size. This is the reasoning the European treaties turned into the test of subsidiarity — act at the higher level only where the objective cannot be sufficiently achieved at the lower one — and it is the same logic economists formalised as the decentralisation theorem: absent spillovers and scale economies, local provision is at least as good as uniform provision, and better whenever tastes differ.

Three folk accounts are worth setting aside, each for its own reason. The centre is too busy fails because a busy centre can hire clerks: sending administration downward while keeping the decision is a real and common arrangement, so busyness explains staffing, not authority. Closer to the people is simply better fails on selection — the same governments that devolve schooling insist on holding defence, and an ideology predicts neither split. Regions extracted it is true in particular cases, yet devolution appears in states with no restive region making demands, so it cannot be the general mechanism.

What would falsify the account? It predicts a pattern, not a level: functions whose effects escape the local boundary should be held centrally almost everywhere, and functions whose effects stay inside it should be devolved wherever tastes differ, across states with unrelated constitutions. If the assignment were instead scattered — defence routinely devolved, refuse collection routinely run from the capital — the mechanism would be dead, and we would be looking at something else entirely, most likely history and accident.

One distinction does most of the work, and losing it makes the whole subject foggy. Moving the administration of a decision downward, while the rule and the budget remain the capital's, is deconcentration; moving the decision is devolution. A council that spends a ring-fenced grant on a nationally specified service has staff but no authority. This is why I am setting aside a figure often quoted in this debate — the share of public spending that passes through subnational government, which I can only recall loosely and will not assert. Even if it were exact, spending share counts money handled, not choices made, so it cannot measure the thing under discussion.

Two neighbouring pieces in this collection reach conclusions that sit oddly beside this one, and the reconciliation is instructive. Mission command argues that an army delegates because the situation moves faster than a report can climb and an order descend, and that the delegation is bounded by shared doctrine — everyone wants the same outcome. Coordination cost argues the opposite-seeming point that pushing work outward multiplies the connections to be kept in step. Both are true here, because they concern different objects. Mission command moves the how while the what stays common; devolution moves the what itself, and is worth most exactly where the ends genuinely differ between places. And coordination cost is the bill devolution pays: it is why the case for devolving collapses as soon as effects start crossing boundaries, since then the connections must be maintained anyway.