Geographical location and urbanisation of the Swedish manufacturing industry, 1900–1960: evidence from a new database

Abstract

This article introduces a new database, based on official statistics, of regional manufacturing industries in Sweden. We employ this database to examine the distribution of manufacturing activity across Swedish regions and cities, 1900–1960. Over this period we observe an increasing concentration of manufacturing activities, reaching a peak around 1940, across the northern, southern and western parts (NUTS-I areas) of Sweden. Over the same period, the North-South divide in terms of manufacturing employment grew larger. Across counties (NUTS-III) and cities we, however, observe two shorter periods of convergence of manufacturing activities, in the early twentieth century and in the post-war period, whereas the inter-war period was characterised by divergence. These developments occurred to the backdrop of the urbanisation of industry in Sweden, as the rural share of manufacturing employment declined from roughly 60 to 25% between 1900 and 1960. We also find that the regional patterns of individual industries over time followed different trajectories, suggesting that that the determinants of industry location differed significantly across industries.

Keywords:

• industry location
• industrialisation
• second industrial revolution
• urbanisation
• convergence

JEL Classification:

• R11
• R12
• 033

Acknowledgements

The authors gratefully acknowledge funding from the Swedish Research Council, project numbers 2008-2023, and from the Jan Wallander and Tom Hedelius foundation, project numbers W2008-0357:1 (Enflo) and W2009-0402:1 (Henning).

Notes

¹Crafts, ‘Regional GDP’ (2005). Rosés et al., ‘Regional Inequality’ (2010). Combes et al., ‘Spatial Inequalities’ (2011). Felice, ‘Regional Value Added’ (2011). Henning et al., ‘Trends’ (2011). See also Boschma and Frenken, ‘Economic Geography’ (2006). Boschma and Martin, Handbook (2010).

²Williamson, ‘Regional Inequality’ (1965).

³Krugman, Geography (1991a). Krugman, ‘Increasing Returns (1991b). Krugman, ‘New Economic Geography’ (2000). Fujita et al., Spatial Economy (1999).

⁴Puga, ‘Rise and Fall’ (1999).

⁵For Spain, see Paluzie et al., ‘Geographical Concentration’ (2004). For France see Combes et al., ‘Spatial Inequalities’ (2011) and for the United States, see Kim, ‘Expansion’ (1995). There is also an emerging literature examining the relative merits of the New Economic Geography and Heckscher-Ohlin theories in explaining the location of industries, see Crafts and Mulatu, ‘Location’ (2005). Crafts and Mulatu, ‘How did Location’ (2006). Wolf, ‘Endowments’ (2007). Tirado et al., ‘Economic Integration’ (2002). Rosés, ‘Whole of Spain’ (2003). Martinez-Galarraga, ‘Determinants’ (2012). Betran, ‘Regional Specialization’ (2011). Klein and Crafts, ‘Manufacturing Belt’ (2012).

⁶Bresnahan/Trajtenberg, ‘General Purpose Technologies’ (1995).

⁷Lundquist et al., ‘Decomposing’ (2008). Schön, Sweden's Road (2010).

⁸Söderberg/Lundgren, Ekonomisk koncentration (1982). Jonsson et al., ‘Towns’ (2009). Henning et al., ‘Trends’ (2011).

⁹Kim, ‘Industrialization’ (2005).

¹⁰Jonsson et al., ‘Towns’ (2009).

¹¹Lundh, ‘Wage Forms’ (2012).

¹²Heckscher, ‘Industrialismen’ (1964), 236.

¹³Kim, ‘Expansion’ (1995). Paluzie et al., ‘Concentration’ (2004). Combes et al., ‘Spatial Inequalities’ (2011).

¹⁴For this article a selection of the data was used for analysis. Details about the complete dataset can be found in the Appendix.

¹⁵NUTS is the definition of regions used on Europe-wide basis by Eurostat and several national statistical offices within the EU. They come in three different sub-national levels, from larger (NUTS-I) to smaller regions (NUTS-III). For a detailed description of the Swedish NUTS-regions we refer to the website of Statistics Sweden: http://www.scb.se/Pages/List____257647.aspx

¹⁶These sectors are (1) metals and machinery, (2) earth and clay, (3) wood, (4) paper, pulp and graphical, (5) food, (6) textiles and clothing, (7) leather, hair and rubber, (8) chemicals and (9) power stations, waterworks and gasworks.

¹⁷See Appendix. Jörberg, Growth (1961).

¹⁸We exclude industries such as mining and iron and steelworks that were not included in the industrial statistics prior to 1913. Due to the fact that the power stations, waterworks and gasworks were removed from the industrial statistics in the 1960s as well as the fact that this sector does not constitute a proper manufacturing industry we exclude this sector in the further analysis.

¹⁹For example Kim, ‘Expansion’ (1995) for the United States and Jonsson et al., ‘Towns’ (2009) for Sweden.

²⁰We would like to thank an anonymous referee for pointing this out.

²¹For some periods, we do have access to gross output data. However, this can reflect different intensities in the use of intermediate or capital goods and thus cannot be used to properly address the geographical characteristics of industrial activity. It is a task for future research to overcome these data issues in order to present longer time series of both employment and value added.

²²The 50 included cities, sorted by population in 1900, are: Stockholm, Gothenburg, Malmö, Norrköping, Gävle, Helsingborg, Karlskrona, Jönköping, Uppsala, Örebro, Lund, Borås, Halmstad, Sundsvall, Linköping, Landskrona, Eskilstuna, Kalmar, Västerås, Karlstad, Söderhamn, Kristianstad, Ystad, Falun, Luleå, Uddevalla, Visby, Västervik, Södertälje, Härnösand, Nyköping, Växjö, Karlshamn, Oskarshamn, Östersund, Kristinehamn, Sala, Vänersborg, Varberg, Lidköping, Arboga, Hudiksvall, Skövde, Köping, Skara, Enköping, Umeå, Eksjö, Mariestad, Filipstad.

²³See, e.g., Bairoch, Cities (1988).

²⁴Available at: http://www2.historia.su.se/urbanhistory/cybcity/. See also Nilsson, Tätortsstatistik (1992).

²⁵The Gini coefficient ranges between zero and one; the Theil index ranges between zero and the natural logarithm of the N number of regions; the Coefficient of Variation ranges between zero and . A low number of the indices indicate an equal regional distribution of manufacturing employment while the reverse is true for higher numbers.

²⁶Northern Sweden consists of the counties of Värmland, Kopparberg, Gävleborg, Västernorrland, Jämtland, Västerbotten and Norrbotten. Eastern Sweden consists of the counties of Stockholm, Uppsala, Södermanland, Östergötland, Örebro and Västmanland. Southern Sweden consists of the counties of Jönköping, Kronoberg, Kalmar, Gotland, Blekinge, Malmöhus, Kristianstad, Halland, Skaraborg, Älvsborg and Gothenburg and Bohus.

²⁷Söderberg, ‘Long-Term Perspective’ (1984).

²⁸Schön, Sweden's Road (2010).

²⁹Using a definition based on population levels in the early 1900s may be sensitive to differential population growth rates across cities, such that cities would migrate between these categories over time. The fact that there is considerable inertia in the city size distribution over time permits us, for analytical purposes, to treat these categories as fixed over this period.

³⁰These location quotients are calculated as in Section 3.2. Note that aggregate manufacturing employment and population thus equals the sum of each variable across the 50 largest cities, rather than for Sweden as a whole as in the previous sections.

³¹Other measures of concentration, e.g., Theil indices or Coefficients of Variation, produces similar results. In order to economise on space these results are not presented here but are available from the authors upon request.

³²E.g., urban agglomerations lacking full city-rights, such as small market towns (köpingar) and historical urban districts (municipalsamhällen).

³³Söderberg and Lundgren, Ekonomisk koncentration (1982).

³⁴Lundh, ‘Wage Forms’ (2012).

³⁵Söderberg, ‘Long-Term Trends’ (2010).

³⁶Schön, Sweden's Road (2010), 208–10.

³⁷We are grateful to Lukas Ahlström, Thor Berger and Linnéa Johansson-Krueger for excellent research assistance in digitising the material.

³⁸Generally cities that housed at least 100 workers were reported in the official industrial statistics.