Enclaves

Why does a village belonging to one country sit wholly inside another, and why does nobody simply trade it away?

Somewhere on a quiet border there is a village where the international boundary runs through a café, and a house whose front door decides its nationality. Nothing about the ground suggests it. There is no river here, no ridge, no defensible line — the fields on both sides are the same fields, farmed the same way, by people who are often related.

Ask how a boundary comes to follow terrain and you get a story about cost and defensibility. Ask why one runs dead straight and you get a story about map-makers dividing land they had never walked. An enclave follows neither. And it is expensive: two administrations, duplicated services, absurd policing. So there are two questions here, with different answers. How did such a thing get drawn? And why has nobody simply swapped it away?

Take the origin first, and notice an assumption smuggled into the question. We asked how the enclave was drawn, as though somebody drew it. Suppose nobody did.

What would have to be true for that to work? Only that before there were territorial states, sovereignty did not attach to a continuous patch of ground. And it did not. Under dynastic and feudal arrangements, authority attached to holdings, rights and persons: a lord held these manors and those mills, a bishopric held parcels scattered across three valleys, and each parcel could move independently by marriage, inheritance, purchase, or the settlement of a debt. Nothing required a lord's possessions to touch each other, so over centuries they did not.

Then the modern state arrives, wanting something the older system never produced: a map on which every point is unambiguously one colour. What can it do with the scatter it inherited? It can renegotiate everything, which is enormously expensive and dangerous, or it can trace a line around whatever each party already held. It traces the line. The enclave is therefore not a mistake in a boundary; it is an older kind of boundary preserved inside a newer kind of map.

The two most-discussed cases fit that shape. The parcels at Baarle on the Dutch-Belgian border descend from medieval divisions of manorial holdings between two lords, agreed plot by plot when contiguity was simply not a requirement. The enclaves along the India-Bangladesh border descend from a comparable patchwork of revenue holdings around the princely state of Cooch Behar, which partition then converted into international anomalies. (The charming tale that they were staked in a chess game is a legend.)

Now the harder half. If they are so costly, why do they last?

Ask who pays and who decides. The daily cost of an enclave falls on the people living in it and on two local administrations — real, but small in national terms and attached to almost no votes. The cost of removing it falls on the national government, and it is not mainly financial. Ceding territory typically clears a higher constitutional bar than ordinary business: a supermajority, an amendment, sometimes a referendum. The arithmetic is inverted from what common sense expects. Doing nothing is cheap for everyone who matters politically; doing the sensible thing requires one of the hardest majorities a parliament can be asked for.

Two further frictions push the same way. A state that trades away territory in a trivial case may weaken its own doctrine of territorial inviolability in a case that is not trivial at all, so the precedent is priced nationally while the benefit is purely local. And a swap requires parcels of equal value, which nobody can verify, so each side reasonably suspects the other of extracting.

The claim is therefore institutional, not geographic, and it is worth being blunt about that. Terrain contributes nothing here. Where a boundary follows a river or a watershed, the land really did lower the cost of one line relative to another; an enclave shows how thin that influence is on its own. The same landscape carries a scatter of sovereignties or a single clean line depending on what property regime was in force when the map hardened.

That also sharpens what "path dependence" means here. It is not inertia in the vague sense of nobody having got round to it. It is a specific barrier — the constitutional cost of cession — sitting in front of a change everyone concerned agrees would be an improvement.

Which gives a clean test, and a case that runs it. If the binding constraint is ratification cost rather than the value of the land, enclaves should be resolved not when they become annoying but when a ratification window opens, typically inside a larger settlement worth spending political capital on. The India-Bangladesh enclaves are exactly that experiment. An agreement to exchange them was signed in 1974 and sat unimplemented for four decades, not because anyone thought it wrong but because the Indian side required a constitutional amendment. When that amendment passed, the exchange took effect in 2015, and the transfers happened essentially at once. The land had not changed value. The procedure had cleared.

There is a mirror-image prediction, which Baarle illustrates: an enclave inside a free-movement and customs area stops being expensive, so the pressure to resolve it disappears and it becomes a tourist attraction instead. What would refute the account: finding that enclaves are routinely traded away by simple executive agreement in states with high constitutional bars, or that their survival tracks the value of the parcels rather than the difficulty of ratifying a cession.

• How states administer an exclave they can reach only across foreign ground, and what corridors cost the country that grants them.

Every term the collection defines is gathered in the glossary.