The Map

A map can show the mountains that separate a farm from a port; it cannot, without another layer, show who controls the road between them. The distance and the authority both enter the farmer's calculation. So do the diseases that interrupt work, the means of preventing them, the title under which the land is held, and the expectation that a buyer will pay. The earlier Roots asked what lies beneath the capital, knowledge, and skill through which growth occurs. This plate follows its geographical and institutional answers toward the point where they meet. Nature gives production conditions; societies organize ways of living within them and changing their consequences. Explaining a nation's poverty requires identifying those mechanisms, rather than assigning its position on the income table to the place where it happens to stand.

Geography matters through particular costs and opportunities, each with its own possible remedy. A mountain may make a road expensive; a distant market may make a perishable crop unprofitable; conditions favorable to disease may make sustained work harder. These are propositions about channels of production, transport, and health. Their strength does not require the claim that every tropical or landlocked society shares a fate. A bridge changes the journey without moving the mountain. Reliable treatment can change the burden of disease without changing the latitude. But building and maintaining those responses requires resources and collective decisions, and the cost can remain higher than elsewhere even when the decisions are well made. The geographical explanation becomes useful when it specifies the obstacle and its consequences; a colored band across an atlas is only the beginning of that account.¹

Institutions enter the same decision by determining which returns a producer can expect to keep. In an invented project, an investment of sixty promises sales of one hundred. Transport costing thirty leaves ten after investment; transport costing ten leaves thirty. Now suppose an arbitrary seizure would take the entire revenue remaining after transport, with a one-half probability. Expected net returns become minus twenty-five and minus fifteen. A better road improves both cases but, under these assumptions, induces investment in neither. Reliable protection against seizure makes the high-transport-cost project worthwhile even before the road improves. The example places physical and institutional obstacles inside one calculation. It also makes their interaction visible: when the producer can retain the proceeds, twenty saved in transport adds twenty to the incentive to invest; under the seizure risk, it adds only ten.²

The difficult empirical task is to separate institutional effects from the prosperity that can itself improve institutions. Acemoglu, Johnson, and Robinson approached this problem through colonial history. Their 2001 study proposed that mortality faced by Europeans affected settlement, settlement affected the institutions colonial authorities established, and those institutions persisted. They used historical mortality as an instrument for contemporary institutions and estimated a large effect on income. The method depends on a demanding condition: after the included controls, historical settler mortality must affect later income through institutions rather than through another remaining channel. The authors recognized the possibility that a persistent disease environment could violate that condition and offered historical arguments and statistical checks in response. Their result is an argument with an identification strategy, whose assumptions can be examined; it is stronger than a mere correlation and more conditional than a verdict that the landscape never matters.³

The dispute becomes sharper when the geographical variable measures a proposed mechanism rather than location alone. Sachs's 2003 response used malaria risk and an ecological instrument, reporting an association with income that remained after institutional quality entered the model. This challenged the inference that geography operated wholly through institutions. It did not remove the need for causal assumptions: the ecological instrument must not influence income through omitted pathways that the estimate incorrectly assigns to malaria. Nor does controlling for an imperfect institutional index hold every institution constant. These are problems to investigate on both sides. The decisive evidence must discriminate between direct effects, indirect effects, and shared causes, with measures suited to each. An argument that rainfall or disease helped shape an institution is compatible with an argument that they also continue to affect production through other routes.⁴

A national ranking of institutional quality still leaves the reformer with choices that the ranking cannot make. A promise of secure property may protect a smallholder against confiscation, or secure an exclusive privilege against competitors. A formal right to sue has different value when the court can decide promptly, when judgment can be enforced, and when a powerful defendant can prevent either. The relevant institution includes the way the promise operates. The project above likewise does not tell us whether a statute, a change in enforcement, or a different distribution of political power would make retention credible. It helps locate an incentive; it does not supply the means of altering it. To move from explanation to reform, we need evidence about those means, the people able to obstruct them, and the distribution of the gains they would create.⁵

The map of wealth consequently records several histories laid over the same terrain: the inherited physical setting, the technologies that alter its constraints, and the settlements of power that decide which remedies are attempted and who receives their benefits. Their entanglement makes the explanation harder, but also makes fatalism premature. A costly road may be worth building; a working road may still serve only the favored; a change of law may remain ineffective without the capacity to enforce it. Each possibility asks for a more exact account of what binds. Mountains and rules differ in how they can be changed, but neither permits us to read a nation's future directly from its present poverty. The task is to find the constraint that can be relieved, and the arrangement that will let the relief reach the people living under it.