Why does distance still shape who trades with whom?

Because distance is still a cost, and trade turns out to behave a lot like gravity. The gravity model of trade predicts the flow of trade between two countries from their economic sizes and the distance between them: bigger economies trade more with each other, and trade tends to fall with distance. Research shows "overwhelming evidence" for the second half of that.

The basic version is a one-line formula borrowed from physics. Trade flow equals a constant times the economic size of one country times the economic size of the other, divided by the distance between them. It was introduced by Walter Isard in 1954, building on John Quincy Stewart's earlier idea of demographic gravitation from 1941.

Economists do not use it only to predict trade. They use it to measure what else pushes countries together or apart: common borders, languages, legal systems, currencies and colonial legacies. They use it to test whether trade agreements such as NAFTA, or organizations like the WTO, actually raised trade. It has even been applied to migration, traffic, remittances and foreign direct investment.

The mystery is why it works. The model has been an empirical success, accurately predicting trade flows for many goods and services, yet for a long time some scholars believed there was no theoretical justification for it. The answer, as it turned out, is that a gravity relationship can arise in almost any trade model that includes trade costs that increase with distance. That is a strength, since the model fits so widely, but economists such as Alan Deardorff argue it is also a weakness: because so many models lead to it, its fit cannot tell us which theory of trade is right.