Updating Maddison


Jason Collins


January 23, 2013

Angus Maddison’s estimates of per capita GDP - from 1 AD through to the 2000s - are one of the most commonly used data sets in the examination of long-term economic growth. While Maddison passed away in 2010, a group of his colleagues created the Maddison Project, with the goal of continuing Maddison’s work.

The project has just produced one of its first major outputs, an update of the original Maddison dataset, including estimates of economic development across the world from 1 AD to 2010. A working paper describes the results, and you can download the new dataset (xlsx).

Maddison’s work has not been without its critics, and the working paper contains judgment on some of these points. In particular, Maddison’s estimates of the gap between Asian and European incomes before industrialisation are supported:

One of the central questions in this literature was whether the level of economic development (in terms of GDP per capita) in China (and India and Japan) before industrialization was comparable to Western Europe (Pomeranz 2000). Maddison’s estimates for that period have been criticized because they show an already substantial gap in real incomes between the different parts of EurAsia; in Western Europe the average GDP per capita was about 1200 dollars, whereas China and India were estimated at between 500 and 600 dollars. …

Summing up, a substantial amount of new work has been published in the past ten years which is generally consistent with the picture Maddison put forward in his 2001/2003 framework. The most severe criticisms at his estimates by Pomeranz (2000) and other specialists on Asian economic history, that he systematically underestimated real incomes in large parts of Asia in the 18th and early 19th century, has generally been proven wrong: detailed research by scholars working on India, Indonesia, Japan and China has shown that the magnitude of the real income gap as estimated by Maddison was about right. Another important result is that Maddison might have overestimated growth in Europe between 1300 and 1800, and that levels of real income were already quite high during the late Middle Ages.

As someone who uses Maddison’s datasets a bit, I’m pleased to see this work going on.