Well, though still on the road, I’m back in Canada after a conference trip to South Africa where I was part of a session on the analysis of late nineteenth/early twentieth century wealth in Britain and its Dominions using probate records. For a summary of my trip, click here. There was a multitude of interesting work and sessions at this conference. Two sessions at the 2012 World Economic History Conference in Stellenbosch South Africa (July 9-13) were exceptionally thought provoking given the concerted efforts to acquire better data on long run economic growth and development.
First there was a session dealing with the Maddison Project, which provided an update on a collaborative international project to continue the work of Angus Maddison on measuring economic performance over time and space. In particular, there is emphasis on re-estimating growth before 1820 to shed light on what is known as the Great Divergence, which was the topic of its own session.
The Great Divergence is the phenomenon in which economic growth between the West (particularly Europe) diverged from the rest of the world during the period between the early Middle Ages to the 1850s. Of course, assessing the extent of the divergence in per capita income requires estimates and considerable effort is being expended on estimating historical per capita incomes in Japan, China and Europe. On the European side, a key project funded by the Leverhulme Trust focuses on reconstructing the national income of Britain and Holland c. 1270/1500 to 1850. One of the most interesting results is that when working backwards to estimate income, it turns out that living standards in the late medieval period were well above bare bones subsistence.
Per capita income estimates assembled by Steve Broadberry, Bruce Campbell, Alexander Klein, Mark Overton and Bas van Leeuwen use agricultural output data, industrial production in metals and mining, textiles and leather and other industries, trade flows and government revenue data. Jan Luiten van Zanden and Bas van Leeuwen have also been constructing estimates for the Netherlands using similar approaches. In addition, data from probate inventories has been used particularly for some aspects of the agricultural output side.
This work has found that per capita incomes in the late middle ages for England and Holland are more than double than those done by Maddison for the year 1000. GDP per capita in 1990 international dollars was close to 1,000 dollars compared to estimates of about 400 dollars by Maddison. Estimates are also being construted for Japan for the 730-1870 period by Jean-Pascal Bassino, Kyoji Fukao and Masanori Takashima and for India before 1850 by Bishnupriya Gupta.
While the term Great Divergence applies to the divergence in incomes between the West and the rest of the world over the period 1270 to 1850, it can also be applied to the range of estimates of per capita income that have been made. After all, when revisions find that income was more than double what was previously estimated, one can be tempted to conclude that we can never really estimate output that far back given the fragmentary and diverse nature of the records. That would be the wrong conclusion. The challenges of estimating historical GDP using a sectoral output approach cannot be under-estimated given the diversity of the data sources as well as regional and urban variations within Europe at this time. Progress in understanding past development and the roots of long run growth requires better data and getting better estimates is also a long-term process.
Another approach to estimating long -term output is apparently underway by researchers such as Greg Clark, Jeffrey Williamson and Peter Lindert in which probate records are being used to derive an estimate of national wealth with national income then estimated as a return to that wealth. Here, the challenge is not only to estimate per capita wealth but to then decide on the range of estimates for a rate of return to apply to the wealth estimates. Having done a lot of work collecting and analyzing Ontario probate records, this is naturally an approach I am interested in seeing the outcome of. An advantage of this approach is the consistency (relatively speaking) of the data source as it involves one source rather than a quilt of records across sectors. If we can benchmark a few probate GDP estimates with years for which we have a good GDP estimate already, we could then use the benchmark to work backwards. Another advantage of the probate record approach is that one could sample urban and rural areas as well as different regions to take those variations into account when constructing the estimate.
In Canada, we have GDP estimates going back to 1870 (The Urquhart/Green numbers) as well as an earlier set going back to 1850 done by O.J. Firestone. Can we go before 1850 using probate records? These inventories certainly go back to the 1600s for Quebec and can be used to construct an estimate of per capita wealth and ultimately per capita income for Quebec. Probate inventories also exist for Atlantic Canada and they date back to the 1790s for Upper Canada (Ontario). It would be possible to mount a more extensive effort to estimate a per capita income series for Canada for the 1600 to 1870 period as a complement to all the work being done for Europe and Asia. It would be a big project.