Border prosperity discontinuity

Why can crossing a single line on the ground change incomes sharply when the land on both sides is identical?

Walk a road that crosses a national border in open country and the land does not change. Same soil, same rainfall, same hills with the same ore in them, often the same surnames on both sides. Yet the houses, the vehicles and the measured incomes can change sharply within a few hundred metres. Every physical input is continuous across that line, and the outcome is not. What kind of thing can produce a step where the ground shows none?

Begin by doubting the observation, because most borders are not drawn at random. They follow rivers, ridgelines, the edge of a language, the furthest reach of a conquest. If a line was placed where two populations or two landscapes already differed, then the step at the border is a record of a difference the line did not cause. Any account that says "the border did it" has to survive that objection first. How would you get past it?

You would want lines drawn by people who neither knew nor cared what lay beneath them. There are some. The 38th parallel across Korea was settled in Washington over a few hours in August 1945 by officers working from a small-scale map. The line through Nogales was fixed by a boundary purchase in the 1850s and ran straight through one settlement, leaving relatives on either side. In cases like these the divergence opened after the line, so it cannot be the land's doing, and it cannot be the people's either.

So what does actually change at the line? Not the geology — the rules. Which court hears your contract, whether your title survives a change of government, what tariff and currency you trade in, who may take your workshop and on what pretext. Now ask the connecting question: how would a rule reach into a hillside and alter what it yields? Through one channel above all, and it is a slow one. A person who expects to be expropriated does not plant an orchard, sink a well, or buy a machine that pays back over twenty years. They hold wealth in forms that are liquid, portable and easy to hide. Repeat that decision across a population for two generations and you have a different capital stock — and a capital stock, plus the skills that accumulate around it, is most of what "income per head" is measuring.

That is only half the mechanism, though, and the missing half is the part that makes it a border story rather than a rules story. Suppose the same gap in rules appeared between two counties of one country. It would not last. People would move to the better side, employers would follow, and wages would be dragged toward each other until the gap was mostly gone. A border blocks precisely that repair: it stops labour moving, deters capital, and taxes goods. So a national border does two things at once — it makes the rules jump, and it forbids the arbitrage that would otherwise erase the consequence.

That gives us something to test. Remove the barrier while leaving the history in place and the gap should begin closing. Germany after 1990 is the case: with movement, currency and law unified, eastern productivity climbed toward the west's and then largely stalled below it, at somewhere around four-fifths by my recollection. Convergence but not completion is what the two-part account predicts, since a capital stock takes longer to rebuild than a law takes to change.

The claim is about which rules apply, not about which government is more admirable. It also does not identify which strand of the bundle does the work: the border experiments hold soil, latitude, disease and often ancestry fixed, which is what makes them powerful, but they cannot separate courts from schooling from tariffs, because all of them switch at the same line.

The causal reading is also genuinely contested. One tradition treats geography as primary — disease burden, soils, distance to a navigable coast — with institutions largely downstream. Another argues that an educated population produces good institutions rather than the reverse. The blind-drawn borders are the strongest evidence for the institutional reading, and they remain evidence rather than proof.

The model predicts absence too. Along much of the French, Belgian and German frontier there is no visible step, because the rule bundles are similar and movement is free. A theory that predicted a jump at every border would be the wrong theory.

Two neighbours sit close by. borders.md asks why lines lie where they do; here their placement is taken as given and only their effect is at issue. persistent-poor-neighbourhoods.md describes a poor place that stays poor while its residents turn over completely, and differs at exactly the point that matters here: there people can move freely, so the explanation must run through sorting. Here they cannot, which is why blocked arbitrage carries so much of the weight.

• Why the gap narrows but rarely closes once a border is dissolved.
• Whether the same discontinuity logic applies to internal boundaries — state lines, special economic zones, municipal limits — where movement is free.

Every term the collection defines is gathered in the glossary.