Maddison‐style estimates of the evolution of the world economy: A new 2023 update

This paper surveys the literature on historical national accounting, discusses the importance of relative income benchmarks for, in particular, historical income estimates, and presents an update of long run global economic development with a new version of the Maddison Project Database (MPD). As benchmarks are central to methodologies for global income comparisons over time, and therefore vital to MPD, we analyze the consequences and biases of three benchmarks, the 1990 benchmark, the 2011 benchmark and the multiple benchmark method following the recent Penn World Tables (PWT) methodology, for pre‐1940 income estimates. We develop a methodology to determine which benchmark in combination with time series produces the best anchor for the historical income estimates in the MPD. We conclude that the best way forward for the Maddison Project is to stick to the original 1990 benchmark, yet with two important changes. First, we integrate the 2011 benchmark for the post‐1990 period, and second, we fine tune the dataset for the pre‐1940 period by integrating a new historical benchmark for the US/UK comparison in 1909. By integrating more benchmarks, the MPD moves closer to a multiple benchmark approach as developed by the PWT.