The aim of the project
The purpose of the project is to establish comparable levels of labour productivity in Sweden relative to the US and the UK for manufacturing, market services and agriculture. The project has established those comparable levels for a few so-called benchmark years. To explore the long-run movement of comparable levels, we have used time series projections of output to working-hours ratios. This approach has allowed the project to explore the movement of productivity across a vast time frame, revealing what role manufacturing, market services and agriculture have played in convergence at the whole economy level.

What the project has accomplished
The project has so far established comparable levels of labour productivity within manufacturing and market services (agriculture remains to be done). For manufacturing, the project has used the so-called “industry-of-origin approach” (van Ark 1993). It aims to convert nominal values of value added, expressed in each country’s currency, into comparable levels of real output. This procedure requires information on products: output in physical terms and corresponding sales value. Dividing sales value by tons or units etc. yields so-called unit values. Dividing Swedish unit prices by US unit prices for comparable products yields unit value ratios, used to establish an average price relation for manufacturing as a whole. The average ratio is reminiscent of a fictitious exchange rate and transforms different currencies into a single unit of pay. In contrast to the trading exchange rate, which mirrors prices of tradables, our fictitious exchange rate covers a range of products produced by each country’s manufacturing industry. The project has produced such unit value ratios for Sweden relative to the US and Sweden relative to the UK in 1963. For the pre-World War II era, the project has revised the previous work of Prado (2008). In sum, the project has established comparable levels of labour productivity for the benchmark years of 1909, 1925, 1935 and 1963.

In order to examine the movement of productivity across a wider time span, the project has devoted considerable attention to establish a time series of volume output and working hours for Sweden. The most recent version of Historical National Account offers a series of volume output for manufacturing. On close scrutiny, however, it has several features making it incomparable to the UK and US series of output. The project has therefore produced a new Swedish series pre-1950. The time series of volume output and working hours underlie our time series projections. In this way, we have covered the period between 1870 and 2010. We have extrapolated the time series from the benchmark in 1924, and the additional benchmarks provide cross-checks of consistency between the time series and the benchmarks. These cross-checks bear testament to a satisfactory amount of agreement between the benchmarks and the time series.

The most important result
(1) In order to render convergence in income per capita, a dynamic manufacturing sector appears to be necessary prerequisite (Kaldor 1967; Rodrik 1913). The project has shown that Sweden conforms to this conventional wisdom. The time series projections show that the pairwise productivity ratios across countries change significantly over time. On the Swedish to US comparison, the US maintained its productivity advantage, on average 2-3 times as high level, until the early 1960s. Then, most of the US advantage disappeared within two decades. Today, the Swedish manufacturing enjoys a slight productivity advantage relative to the American. On the Swedish to UK comparison, the UK had an advantage of 80 percent in the 1870s. At the outbreak of World War II, the UK lead was down to 10 percent. The productivity accelerations of the 1960s and the 1992–2010 period turned the ranking of the two countries on its head. Today, the Swedish lead appears to be about 60 percent. Thus, what began as a British advantage of 80 percent ended, after 140 years, by a Swedish lead of 60 percent.

(2) The historiography contains a few long-term studies of sector-wise productivity across countries, pertaining exclusively to developed countries, such as the US, UK and Germany. The existing studies indicate that the productivity ratios in manufacturing across countries have remained quite stable over time. Convergence in GDP per worker-working hours has instead owed to structural transformation (Broadberry 1997). This stability in the productivity ratio of manufacturing contradicts the conventional wisdom of what drives economy-wide convergence. However, in light of the project’s studies, this result, based on the US, UK and Germany, appears anomalous. Besides our Sweden/US and Sweden/UK comparisons referred to earlier, in a companion project with several overlaps (Jan Wallander and Tom Hedelius stiftelse), we have documented the long-term movements of comparable levels of productivity in manufacturing for Sweden/Brazil; US/Brazil; Sweden/Japan; and Sweden/Germany Except for the Sweden/Germany comparison, stability in the productivity ratio of manufacturing is no way discernable. Comparing countries that begin at radically different points of departures in terms of productivity levels, the relative levels change dramatically over time. Convergence in manufacturing productivity across countries is key to understanding what goes on at the economy-wide level. In the Sweden/Germany case, the productivity ratio of manufacturing was close to unity in the early twentieth century; and although subjected to wide fluctuations during the two world wars, it lacked a long-run trend across the twentieth century.

(3) Our results challenge the conventional wisdom about Sweden’s rank in terms of productivity levels in the fourth quarter of the nineteenth century. A common assumption is that Sweden belonged to the poorest countries of Western Europe in the nineteenth century, manifested by the mass migration to the New World. Part of the explanation to this assumed backwardness is the rich yet far from perfect official statistics. Failing to circumvent the most serious deficiencies of the early official statistics, earlier versions of historical national accounts have tended to underestimate the level of output per capita, leading Sandberg (1979) to conclude that Sweden was an “impoverished sophisticate” (Gadd 1999). Our estimates of comparable levels of labour productivity urge us to reconsider on the one hand the Swedish position relative to other countries in the beginning of our study period, and on the other, how fast Sweden grew during the twentieth century. There is nothing conspicuous about the gap between Sweden and the leading economies of the US and the UK in the late nineteenth century. All countries lagged significantly behind these two productivity frontiers. Sweden was on a par with Germany, which makes her a member of the “late nineteenth century convergence club”. Only the members of this club managed to benefit fully from the macro innovations of the Second Industrial Revolution: electricity and the combustion engine both of which changed completely the conditions for manufacturing, transports and communications. One may underline the favourable Swedish rank by reference to Japan; the Japanese productivity level of manufacturing was merely 30–40 percent of the Swedish by the beginning of the twentieth century.

Avenues for future research
Much of what constitutes service output originates from the public sector. An important question is how to measure the efficiency with which the public sector performs its multiple functions. The Swedish public sector, our investigation into market services shows, was very successful in steering and facilitating investments within transports and communications from the late nineteenth century. Within the filed of research called “quality of governance”, researchers have attempted to trace the origin and development of the modern, and also efficient, Swedish state bureaucracy (Rothstein 2011). One should seek ways to measure also its productivity at least for a benchmark year.

In the future we should use our sector-wise benchmarks for the 1930s to build up a so-called “direct estimate” of GDP per capita, providing a comparison of the Swedish level with that of the US. This direct estimate might offer a methodological contribution to the debate on direct estimates vs. time series extrapolation (Ward and Devereux 2003; Broadberry 2003).   

International dimensions
The project’s prime objective, establishing comparative levels of labour productivity, has spilled over into collaboration with researchers in Montevideo and Sao Paulo (supported by STINT/CAPES and Bromanska). It aims to publish a book under the imprint of Palgrave Macmillan preliminary titled “Brazil and Sweden – A Tale of Two Capitalisms”. One chapter of the book is devoted to a Brazil/Sweden productivity comparison in manufacturing. It will be co-authored with Cecilia Lara from Universidad de la República, Montevideo. She will be a guest researcher at the Department of Economy and Society, University of Gothenburg in June 2018. The collaboration has also resulted in the session “Contrasting Development Paths in Latin America and Scandinavia: What Can We Learn?” in the forthcoming WEHC in Boston, July 2018.

The participants of the project have presented draft versions of the papers at the following conferences:

* WEHC in Kyoto, August 2015.
* EHES, Pisa, September 2015 and Tubingen, September 2017.
* The SSHA Annual Meeting, Chicago, November 2016.
* The Fifth Asian Historical Economics Conference, Seoul, September 2016.
* The Fifth Jornadas Uruguyas de Historia Económica, Montevideo, December 2016 (Co-organiser of the session ”Productivity and real wages in comparative perspective”).
* Swedish Economic History Meeting, Umeå, October 2015 and Stockholm October 2017.
* Sound for Seniors Workshop, Uppsala, August 2017.