The Wealth of Nations, Mapped

Angus Maddison spent much of his life trying to measure the size of every economy in the world, as far back as the evidence would go. The project that carries his name now holds 21,366 estimates of GDP per head for 169 countries, in dollars of the same buying power, from Roman Italy to 2022. Put those numbers on a map with a slider and you can watch the shape of the world economy change: the long centuries when China and India made almost half of everything, the rise of Britain and then the United States, the sudden take-offs of Japan and Korea, and China's return.

The map also shows something the usual charts hide: how little we know about most of the world before 1820. Where Maddison has no estimate for a year, the country is hatched, and before 1820 most of the map is hatched. I have left it that way on purpose.

01 · The mapPlay twelve thousand years

Press play, drag the slider or jump to a moment. Switch between the size of each economy, its output per person, its share of the world and how fast it was growing. The time scale is uneven: the ten thousand years before AD 1 pass quickly, the years since 1950 slowly. Before AD 1 the map is empty and the panel beside it shows the world as a whole; from AD 1 to 1820 striped regions show Angus Maddison's older regional estimates, with the newer national estimates on top where they exist. The strip under the slider shows how many countries have an estimate in each year. Click a country for its whole history, and add another to compare.

02 · Before nationsTen thousand years in one line

Before AD 1 nobody has estimated the output of any country or region, and the map stays empty. What does exist are estimates for the world as a whole, and they are mostly reasoned guesses. Population comes from reconstructions such as HYDE, whose central estimate for 10,000 BC is 4.4 million people and whose range for 3000 BC runs from 7.8 to 81 million; an older reconstruction by McEvedy and Jones puts 2000 BC at a third of HYDE's figure. Output per head comes from Brad DeLong's estimates, which start near subsistence by assumption: in his figures it barely doubles, 2.1 times, between 10,000 BC and 1800. Since 1820 it has grown 15.5 times, and the world's population is about 1,795 times what HYDE estimates for 10,000 BC.

From AD 1 to 1700, Angus Maddison's own regional estimates fill the gap between the world and the nations. They are drawn on the map as striped regions, and they are as much assumption as measurement: at AD 1, 7 of his 11 regions sit at round figures of 400, 425 or 450 dollars a head (in 1990 prices), his assumed level for economies near subsistence. On those figures the Indian subcontinent and China made 57% of world output in AD 1, Western Europe, then mostly Roman, 13.7%, falling to 9.0% in 1000. Maddison's 2010 figures and the 2023 database also disagree where they overlap: his China makes 33.0% of the world in 1820, the newer research 28.6%.

03 · The great divergenceHalf the world, then a tenth, then a third

In 1820, the first year with a world total, China made 28.6% of the world's output and India 16.6%. They were the two largest economies on Earth not because people there were rich, but because so many people lived there. Britain, the richest large country with an estimate, produced about $3,306 a head; China $882 and India $937.

What followed is what economic historians call the great divergence. China's share fell to 10.0% by 1913 and to its lowest, 4.2%, in 1961, at the end of the famine of the Great Leap Forward. India's fell to 3.0% in 1979. Together the two went from 45% of the world to 8% in 1973. Over the same years Western Europe and the "Western offshoots" (the United States, Canada, Australia and New Zealand) rose from 27% to 59% in 1913.

Then it turned. By 2024 China's share is back to 21.4% and India's is 8.7%: together 30%, almost exactly the West's 30%.

04 · Changing placesWhen the leaders swapped

The United States and Britain. In Maddison's figures the American economy became larger than Britain's (within today's borders) in 1862, in the middle of the Civil War, and never fell behind again. Output per head took longer: the United States first passed Britain in 1880, swapped places with it several times up to 1898 and again in 1933 and 1934, in the Depression, and has been ahead every year since 1935. Britain's share of world output peaked at 9.3% in 1870; in 2024 it is 1.9%. The American share peaked at 29.7% in 1944, with the rest of the industrial world at war.

The United States and China. China was the largest economy in every year Maddison can compare until the 1880s. There is no estimate for China between 1870 and 1887, and by 1887 the United States was larger. It stayed the largest until 2014, when China passed it again in these purchasing power figures. (At market exchange rates, which measure something different, the United States is still larger.)

05 · Take-offsJapan, Korea, Taiwan and China

Most of the time economies grow slowly: two per cent a year doubles output per head in a generation. A few have grown far faster for far longer. In 1950 Japan's GDP per head was $3,062, 28% of Britain's. It then grew by 8.1% a year for 23 years, multiplying 6.0 times, and by 1973 had reached 95% of the British level. It passed Britain in 1980.

South Korea started lower and went further. In 1960 its GDP per head, $1,548, was a quarter of Japan's and lower than Ghana's ($2,197); 88 of the 122 countries of a million people or more with an estimate were richer. It grew by 7.0% a year to 2000, passed Japan in 2019 and Britain in 2020, and by 2024 was 28 times as rich as in 1960. China's GDP per head has grown 12.2 times since the reforms of 1978.

Measured by the best 25 years of growth since 1950 in any country of ten million people or more, South Korea (8.2% a year from 1965), Japan (7.4% from 1950) and Taiwan (7.4% from 1962) are the fastest sustained climbs in the record. Only Iraq and Angola rank higher, and for other reasons: Iraq's run starts in the collapse after the Gulf War of 1991, Angola's during its civil war, and both are oil economies.

06 · The gapHow far apart the richest and poorest are

Among countries of a million people or more with an estimate, the richest had 5.2 times the output per head of the poorest in 1820 (United Kingdom and Nepal), 11.6 times in 1913 and 29.5 times in 1950. In 2024 it is 251 times: Qatar ($146,965) against the Central African Republic ($585).

The very top is often a small oil state, so a steadier measure compares a country nine-tenths of the way up the ranking with one a tenth of the way up. That ratio rose from 3.0 in 1820 to 10.6 in 1950 and 28.1 in 2000, and has eased a little since, to 26.3 in 2024. Part of the early rise is coverage: Maddison has estimates for 41 such countries in 1820 and 150 in 2024, and the poorest are the least likely to have early figures. Even so, the spread between nations is one of the defining facts of the last two centuries.

07 · Before 1820What the record can and cannot say

Play the map from AD 1 and it is nearly all hatched. Maddison has estimates of GDP per head for AD 1 for 14 present-day countries, all in or beside the Roman world, and several share a single regional figure. In 1000 there are 6, in 1500 11, in 1700 17. The long annual series that do exist come from a handful of European archives (England from 1252, France and Spain from the 1270s, Sweden from 1300, the Netherlands from 1348, then Poland, Germany and Portugal), two colonial series for Mexico and Peru, and estimates at intervals for China, Japan and India. There is no world total before 1820 in the current release, so shares of world GDP start there.

Among the countries that do have estimates, the Netherlands was the richest in 1600 and 1700, the trading republic of the Golden Age, at $3,377 a head in 1700. Britain took the lead by 1820 ($3,306) and held it until the 1870s. The second-richest "country" in 1700 is South Africa, which here means the small Dutch settler colony at the Cape: a reminder that some early figures describe a sliver of today's territory.

08 · How I did thisMethods and caveats

The data. All values come from the Maddison Project Database 2023 (Bolt and van Zanden), which compiles estimates of GDP per head and population from hundreds of studies into one series per country, in 2011 international dollars at purchasing power parity. It has 21,366 country-year estimates of GDP per head for 169 units: 55 countries in 1820, 75 in 1913, 146 in 1950 and all of them from 1950 or soon after to 2022. I carried the series forward to 2024 with the IMF's World Economic Outlook (October 2025): each country's real GDP per head at PPP and population grow as the IMF says they did, from Maddison's 2022 level. That covers 166 units; Cuba, North Korea and Syria stay at 2022. For some countries the IMF's latest figures are still estimates, flagged in the popups.

Total GDP and shares. Total GDP is GDP per head times population. Where Maddison has GDP per head but no population for a year, I interpolate population between its estimates if they are at most 30 years apart (409 country-years). Shares divide by Maddison's own world totals, published for 1820, 1850, 1870, 1900, 1920, 1940 and every decade from 1950; between those years before 1950 the world total is interpolated, and from 1950 it is the sum of countries, scaled to Maddison's totals (the scaling is never more than half a per cent). In 1820 the countries with estimates hold 66% of the world's people; by 1870, 93%.

Before 1820. For the world as a whole before AD 1 I use Brad DeLong's (1998) estimates of world output and output per head (his variant without the Nordhaus adjustment for new goods, which is closest to Maddison's concept), and for population the HYDE 3.2.1 reconstruction (Klein Goldewijk and others, 2017) with its low and high range, alongside McEvedy and Jones (1978). From AD 1 to 1700 the regional figures are Angus Maddison's own (2010), for 11 regions; his "Western Europe" also includes Greenland, which I leave undrawn. All of these I took from a compilation of the published series by David Roodman (2020). They are in 1990 dollars, so I rescale them by one factor (1.69) that makes Maddison's 2010 world output per head in 1820 equal the 2023 database's. The world numbers in the bar above the map are interpolated between the authors' estimates, marked ≈. None of this is measurement in the modern sense: the farther back, the more it rests on assumptions about subsistence and on population guesses that differ by a factor of two or more.

The map. Colours are painted on present-day borders (Natural Earth), because that is what the numbers describe. Former states are drawn whole until the year they broke up. A country without an estimate in the chosen year shows its nearest estimate up to 10 years away, dotted, or hatched if there is none. The borders of the time come from historical-basemaps, an open project drawing approximate state borders at 52 dates; they are context, not data, and are drawn faintly.

What this cannot tell you. Estimates before 1820, and for much of Africa and Asia before 1950, rest on thin evidence: wages, prices, harvests, guesses at population. Purchasing power comparisons over two centuries are approximations, and a 2011 dollar of rice in 1820 China is not easy to compare with a 2011 dollar of bread in London. GDP measures market output, not well-being or how it is shared. Present-day borders mean "India" in 1820 excludes Pakistan and Bangladesh, and "Germany" excludes Prussia's eastern lands. Many details depend on revisions: earlier Maddison releases gave different numbers, and so will later ones.