Skip to content

Maddison Project Database

The successor to Angus Maddison’s pioneering long-run GDP-per-capita reconstructions. Maintained by an international consortium of economic historians coordinated through the University of Groningen. Provides estimates of real GDP per capita (in 2011 international dollars) for individual countries, going back as far as the data permits — to 1 CE for some economies, to the medieval period for several European countries, to ~1500 for many. Frequent updates incorporate new national reconstructions (notably the Broadberry-led work on Britain, China, India, and other major economies).

The single most-cited macro-historical dataset in the field. Most cross-country charts of “the Great Divergence” or “the Hockey Stick of Modern Growth” are drawn from Maddison.

  • Time: ~1 CE to present, with progressively more countries as one moves forward. The 2023 release includes ~2,800 pre-industrial data points (vs. 158 in the original Maddison 2003 dataset).
  • Variables: real GDP per capita; population; some derived series.
  • Geography: 169 countries at present.
  • Updates: 2010, 2013, 2018, 2020, 2023, with each release substantially revising pre-1820 estimates as new national reconstructions become available.

(Real GDP per capita, 2011 international dollars unless otherwise noted)

YearUK / EnglandWestern EuropeChinaIndia
1500$1,691~$1,200$600* (Maddison 2009, 1990$) — or $1,103* (Broadberry-Guan-Li 2018, 1990$)$682* (Broadberry-Gupta 2015, 1990$)
1700$2,879~$1,500~$600* (Maddison) — or declining (Broadberry)n/a
1820$4,332~$2,000$600* (Maddison)$1,078*
1870$8,212~$3,400$530*$526*
1913$8,772~$5,500$552*n/a

*Asterisked numbers use the older 1990-international-dollar denomination from earlier Maddison releases or from Broadberry’s parallel reconstructions, not directly comparable to the 2011 international dollar figures for the UK.

  • Unit inconsistency in casual citation. The most-cited “Maddison” figures float between two international-dollar bases (1990 and 2011). When a popular source compares “UK in 1820 had $4,332 vs. China $600,” it is comparing apples to oranges; you cannot subtract these directly. Older 1990$ figures need to be uplifted by roughly 1.5–1.6× for nominal comparability with 2011$ figures.

  • The “China at $600 across centuries” artifact. The original Maddison 2009 China series shows per-capita GDP holding constant at $600 from 1500 through 1820. This was Maddison’s working assumption when he had no underlying data — a placeholder, not a measurement. The Broadberry-Guan-Li 2018 reconstruction (drawing on actual Chinese sectoral data back to 980 CE) shows Chinese per-capita GDP peaking in the Northern Song around 1080 CE, declining through the Ming, and at substantially below European levels by 1700. The Maddison flat-line is still in casual circulation; treat it as an artifact of pre-2018 data scarcity, not as evidence.

  • Pre-modern numbers are reconstructions, not measurements. The medieval and early-modern figures rest on assumed urbanization rates, sectoral output reconstructions, and price proxies — not on historical national accounts in any modern sense. Confidence intervals are wide, and the series has been substantively revised at each update.

  • Country-level series mask within-country variation. “England” or “China” hides large regional differences (Yangzi delta vs. inland China; London-and-the-Southeast vs. the rest of England). For the California School parity claim, the regional comparison is more meaningful than the national one — and Maddison’s national series cannot resolve it.

  • The post-1990 World Bank merger. Our World in Data and several derived sources splice the Maddison Project at 1820–1989 with World Bank data from 1990 onward; the splice point is documented but not always made visible.