DOSE: Subnational Economic Output for 1,500 Regions, Because National Averages Lie

There is a version of the world economy told in national averages. In that version, Brazil has a GDP per capita of roughly $9,000. But DOSE reveals a different reality. Sao Paulo state produces output comparable to a mid-income European country, while Maranhao in the northeast operates at levels closer to sub-Saharan Africa. Both are Brazil. Both are averaged into a single number that describes neither. DOSE was built to end this statistical fiction.

The dataset is deceptively simple in concept but monumental in execution. Researchers at the Potsdam Institute for Climate Impact Research collected officially reported gross regional product data from 166 national statistical agencies, harmonized the definitions, adjusted for purchasing power, and assembled a panel stretching back decades. The result is 1,661 subnational observations that let researchers work with economic geography rather than political geography for the first time at global scale.

The implications ripple outward from economics into climate science, development policy, and conflict studies. When climate damage functions use national GDP, they systematically underestimate harm to poor regions and overestimate it for rich ones. When development agencies target countries, they miss the pockets of poverty inside middle-income nations. DOSE makes these invisible economies visible, and the research community has responded — 19,000 downloads in its first year suggest that the appetite for subnational truth was enormous.