From the American manufacturing belt to spatial clustering in transnational networks: the evolution of industrial geography as reflected in Geografiska Annaler

ABSTRACT

This article features the Swedish geographer Sten De Geer’s contribution to the concept and delimitation of the American manufacturing belt, as published in Geografiska Annaler in 1927, and the reception of his article among US geographers. The marked attention paid to this article contributed to the positioning of Geografiska Annaler as one of the leading European journals of geography in the North American academic debate. Later articles published in Geografiska Annaler during the ensuing nine decades illustrate how the field of industrial geography developed in Sweden and internationally. It is demonstrated that today’s economic geographers have made deliberate attempts to more closely cooperate across disciplinary boundaries, particularly towards the field of international business, in order to better understand and explain the spatial concentration and distribution of economic activities in the same spirit as Sten De Geer’s seminal work.

KEYWORDS:

• American manufacturing belt
• industrial geography
• spatial clustering
• Sten De Geer

Economic geography in general and the geography of manufacturing, in particular, did not occupy a prominent position in the early years of Geografiska Annaler. Despite high ambitions to be a ‘general’ science, the strong natural-science bias of geography as a discipline was naturally reflected in the collected volumes of geographical research, in the research profiles of geography scholars and, subsequently, in geography-related journals in both Sweden and on the international front. One notable early exception to the vast majority of leading academic geographers who focused mainly on physical phenomena on the surface of the earth was the Swedish professor of geography, Sten De Geer.

In this article, I discuss De Geer’s contribution to the concept and delimitation of the American manufacturing belt, as published in Geografiska Annaler in 1927, and the reception of his article among US geographers. The marked attention paid to this article contributed to the positioning of Geografiska Annaler as one of the leading European journals of geography in the North American academic debate. Thereafter, I highlight a number of articles published in Geografiska Annaler during the ensuing nine decades that illustrate how the field of industrial geography developed in Sweden and internationally. Special attention is paid to a paper published in 1996 by Anders Malmberg, Örjan Sölvell, and Ivo Zander on ‘Spatial Clustering, Local Accumulation of Knowledge and Firm Competitiveness.’ In retrospect, their article can be seen as a forerunner to deliberate attempts by economic geographers to more closely cooperate across disciplinary boundaries in order to better understand and explain the spatial concentration and distribution of economic activities in the same spirit as Sten De Geer’s seminal work.

Sten De Geer and the American manufacturing belt

Sten De Geer (born 1882) had a solid background in the natural sciences with a specialization in geomorphology. He defended his doctoral dissertation, which was focused on one of Sweden’s main rivers, Klarälven, at Uppsala University in 1911, and he initially walked in the footsteps of his father, the internationally recognized geologist and geomorphologist Gerard De Geer. After losing the competition for Lund University’s chair in geography to Helge Nelson in 1916, the younger De Geer held teaching positions at Stockholm University College and the Stockholm School of Economics until 1928, when he was appointed to the August Röhss’ Chair in Geography with commercial geography and ethnography at the University College of Gothenburg. According to the donation’s prerequisites, the chairholder was expected to also teach at the Gothenburg School of Economics and Business Administration. This side assignment was not a burden for De Geer, as early in his career he had demonstrated an interest in the broader field of spatial distribution phenomena (Martin and James 1993, 277). Moreover, the environment of the traditional ‘Handelshochschule’ in Stockholm, where he taught future business leaders in practical and commercially applicable aspects of geography, may have further piqued his interest in examining physical phenomena from human and social perspectives.

De Geer had also shown a keen interest in geography as an academic discipline from an early point in his career. In an important article in Geografiska Annaler, he addressed the delicate and sensitive issue of defining the field and noted that he refused to categorize geography as within either the natural sciences or the humanities. Instead, he labelled it ‘a general science along with Statistics, Mathematics, Philosophy and even History in its widest signification’ (De Geer 1923, 6). His own pragmatic definition indicated that ‘geography is the science of the present-day distribution phenomena on the surface of the earth’ (De Geer 1923, 2; Hägerstrand 1982, 122). In order to distinguish geography from history, he suggested that the past only needed to be considered in order to understand the present (De Geer 1923, 2f; Martin and James 1993, 279). Although he initially approached human and social geography through studies of population distribution mapping, as well as urban areas and ports around the Baltic Sea, he soon broadened his research interests to include global political geography and provided contributions on the ‘New Europe’ after the First World War. He also published an article on ‘The Subtropical Belt of Old Empires’ in Geografiska Annaler (De Geer 1928). In several of his research papers and books, he showed an impressive ability to create new cartographic techniques, such as the use of various-sized dots and circles to indicate population sizes and production volumes. However, his major contribution to economic geography at the time was his extensive study entitled ‘The American Manufacturing Belt,’ which was published in Geografiska Annaler (De Geer 1927).

By today’s journal standards, it is difficult to see De Geer’s paper as a conventional, compressed article. It comprises 127 pages as well as 2 plate maps, or roughly 60,000 words. In this respect, it should be classified as a semi-monograph rather than a paper. The concept of an American manufacturing belt consisting of the densely populated, rapidly growing industrial region in the eastern part of the United States – from southern Wisconsin and western Illinois to the Atlantic coast (Pennsylvania, New York, and Massachusetts) – and Canada’s southern Ontario was not new. Several American geographers had previously described the dynamic development of industrialization and urbanization within this belt based on three large raw-material bases: food, cotton, and coal/iron ore. One leading text in this regard was J. Russell Smith’s ‘Industrial and Commercial Geography’ (1913). A number of textbooks used in high school geography had also highlighted this topic. Despite the availability of a number of descriptions of the geography of American industrialization, De Geer remarked that ‘surprisingly little has been written about this manufacturing belt’ (1927, 236). De Geer’s own unique contribution to the literature was the fact that his article offered the first quantitative delimitation of the American manufacturing belt. It also included a meticulous systematization and categorization of towns, cities, and separate manufacturing districts within the ‘Belt.’ De Geer made use of a number of physical, human, and social criteria to put together a coherent geographical delimitation of the entire industrial region.

It is striking that this thorough and detailed study was built on fairly cursory fieldwork. After lecturing at the University of Chicago during the summer of 1922, De Geer spent late August and early September making ‘direct observations during systematically planned, but rather hurried journeys through the belt’ (1927, 234). The keyword here is ‘systematically planned.’ De Geer could utilize his solid background in the natural sciences to understand the physical conditions surrounding industrialization and urbanization, and he benefitted from his previous studies in population and urban geography. Furthermore, he took advantage of his ability to synthesize a vast range of statistical and geographical data, which he transformed into an elegant pedagogic and cartographic description. He also put the American manufacturing belt into a wider global context by relating it to an extensive description of the development of the world’s manufacturing belts in general, including a detailed comparison with Swedish manufacturing districts.

The starting point of his study was his observation ‘that the large national manufacturing districts of Europe are situated so close together that in fact, they form one single European manufacturing belt’ (De Geer 1927, 233). Accordingly, a comparison with the situation in North America could be used as an analytical tool in order to ‘theorize’ (although De Geer did not use this term) the growth of manufacturing regions. According to De Geer,

the belt has developed during a comparatively short and rather homogenous historical period and may therefore be expected to have been influenced by geographical laws more manysidedly than has been the case in Europe with its complicated system of states and its very old traditions. For these reasons, the American manufacturing belt is certainly a more suitable subject to regional geographical study than the European belt. (1927, 234)

His study was mainly built on secondary data from the US national censuses carried out in 1910 and 1920. He used the number of wage earners as the major measure of industrial activity together with the shares of the value of manufactured products. The focal geographical object was cities with more than 10,000 inhabitants, as classified in the Census statistics. Thereafter, he spent a significant amount of energy on grouping the cities and their surrounding manufacturing districts in order to identify a rough delimitation of the belt. In total, he investigated about 400 manufacturing cities, constituting 66 manufacturing city groups in nine larger manufacturing districts in 15 states plus Ontario (De Geer 1927, 264–283). Already at this stage of the analysis, De Geer demonstrated his ambition to cover the entire geographical environment by classifying these city groups based on five physical location criteria: (1) on the shore of the Atlantic Ocean or one of the Great Lakes; (2) on a river, a part of a river, or a river’s tributaries; (3) on a canal or a canalized river; (4) on a hill or in a valley; and (5) on an open plane (1927, 284). Fifty-one of the 66 city groups and 300 of the 400 cities were situated along some hydrogeographical localization line (De Geer 1927, 284). The observation of the importance of access to waterways was elaborated later in the study with regard to access to the railroad network. De Geer also noted how the future development of the automobile and the road network would be ‘of great interest as a complement to the study of manufactures from a geographical point of view’ (1927, 324).

In addition, De Geer investigated the impact of a number of other factors that could have affected the development of separate cities and manufacturing districts within the belt, such as topographic, geological, and climatological conditions; continuously arable land; and immigration and settlement patterns. When it came to the climate, he used a somewhat controversial and later contested study by Ellsworth Huntington on the relation between climate and the emergence of human civilizations, and the estimation of the optimum temperature for physical and mental labour (Huntington 1915). He also showed a striking ‘coincidence of the distribution of cyclonic intensity and frequency with the distribution of human energy and industrial activity’ (De Geer 1927, 303). Using a map of the location of cyclonic storm factors in North America, Europe, and Japan, De Geer concluded that ‘in all three cases the manufacturing region or district lies round the southern border of the most intensive cyclonic area’ (1927, 303).

His article raised considerable attention among American economic geographers and stimulated studies in which some of his categorizations were challenged (see, e.g. Hartshorne 1936; Jones 1938; Strong 1937; Wright 1938). However, it is still viewed as the pioneering work in this field and as the natural starting point in descriptions of the geography of American industrialization (see, e.g. Alexander 1963; Alexandersson 1956; Dickinson 1964; Meyer 2003). Chauncy D. Harris states in the anthology American Geography: Inventory and Prospect, published by the Association of American Geographers in 1954, that

it is interesting to note that just as the first quantitative delimitation of the American Manufacturing Belt was made by a European geographer, the first similar approach to the delimitation of the European manufacturing areas was made by the American geographers Chauncy D. Harris (sic) and Burton W. Adkinson. (Harris 1954, 296)

Harris continues with a critical remark that

it is one thing to recognize and define the American Manufacturing Belt, and another thing to explain it. Although Sten De Geer made a laudable attempt to explain its location and limits, to this day a thoroughly critical and comprehensive, and balanced evaluation of the factors in the localization and development of manufacturing in this belt is lacking. (Harris 1954, 303)

De Geer paid attention to certain phenomena that have also caught the eye of modern-day economic geographers. Such phenomena include the specialization of cities within the manufacturing belt, the shift from specialized city groups toward a pattern reflecting a larger variety of industries, and the relatedness of various manufacturing activities. In this regard, his work serves as a precursor to the contributions to industrial location theory that followed in the ensuing decades, such as Palander (1935), Lösch (1940), and Hoover (1948). His work also serves as the basis for contributions to theories of competitive advantage and the ‘new economic geography’ of agglomeration economies and the clustering of economic activities, as highlighted by business economists like Porter (1994) and general economists like Krugman (1991a).

Due to a lack of complete and reliable secondary data (De Geer 1927, 261), De Geer was unable to give a cartographic account of the branch and product specializations of the cities within the belt. He mentions an unpublished map in which prominent branches of manufacturing are shown by the city. Instead of a map, he offers verbal descriptions of, for example, shoe cities, cotton cities, and metal and machinery cities, which demonstrate a good understanding of how different sectors of various industries were related to and supported each other (1927, 286ff). In this sense, his study can be viewed as a forerunner to modern economic geographic studies of ‘industry relatedness’ (Neffke, Henning, and Boschma 2011). However, as Hägerstrand (1982, 123) notes, De Geer’s approach was deliberately ‘a-historic’ in the sense that he described the ‘present’ situation. He did not attempt to examine the evolutionary process of industrial development in order to demonstrate how separate decisions made under different preconditions and different extents of limited knowledge constituted a cumulative pattern of economic activity. Such studies, later labelled ‘evolutionary economic geography,’ became popular in the early 2000s (see, e.g. Boschma and Martin 2010). However, they were not entirely original, as similar ways of explaining economic patterns had basically been mainstream in the field of economic history along with the tradition established by the Swedish economist Heckscher ([1936]1949). They were also a natural part of studies in historical economic geography. Many examples were manifested by the ‘Uppsala school’ of industrial geography and published in Geografiska Annaler, especially in the 1950s (Arpi 1953; Eriksson 1953, 1957, 1960; Lindberg 1953). In a paper on manufacturing belts in the United States, Krugman notes that ‘if there is one single area of economics where path-dependence is unmistakable, it is in economic geography – the location of production in space’ (1991b, 80), although he does not explicitly refer to De Geer.

Although some of the ideas De Geer raised while attempting to determine the most favourable conditions for the concentration of manufacturing activities have not survived over time, his systematic, multifaceted approach to understanding ‘present-day distribution phenomena’ of industrial districts has been widely adopted. It is no surprise that his study gained a significant amount of attention from American geographers and city planners, and it was used as a seminal work on industrial geography for many decades. During his relatively short stays in the United States, De Geer was successful in building up an extensive network of colleagues, which helped to forge closer links between Swedish and American geographers. In the following years, these contacts were substantiated in contributions to Geografiska Annaler (Atwood 1929; Cahill 1934; Jefferson 1934; Whittlesey 1930). These early contacts may also have been important when it came to the ‘new wave’ of close relations between American and Swedish geographers in the 1950s and 1960s.

In 1996, Geografiska Annaler, series B, published a paper entitled ‘Spatial Clustering, Local Accumulation of Knowledge and Firm Competition,’ which was authored by Anders Malmberg (an economic geographer at Uppsala University), and Örjan Sölvell and Ivo Zander (both from the Institute of International Business at Stockholm School of Economics) (Malmberg, Sölvell, and Zander 1996). The peer-review process for this paper was not straightforward. It is always a challenge for scholars to bridge different disciplines, as reviewers have a tendency to presuppose that a certain journal should build on certain disciplinary traditions and a familiar strand of literature. If a paper aims to cross these invisible borderlines, referees might be tempted to suggest that ‘it should seek another journal.’ After a few rounds with a number of referees offering different conclusions, as well as approaches to several colleagues who declined to review the paper because it was ‘beyond the scope of their competence,’ I decided to publish it anyway, as I found it to be a relevant case of a research problem involving two academic disciplines that had previously acted in parallel without utilizing each other’s experiences. It was the authors’ ambition to identify a common denominator for economic geography and international business studies in order to investigate the phenomena of spatial clustering, accumulation of knowledge in local milieux, and the competitiveness of the corporate firm.

This endeavour was carried out by integrating economic geography theories on spatial agglomeration with theories on international business and innovation processes. The authors applied a truly multifarious approach by neatly bringing together literature from many fields and categorizing various contributions in order to identify overlapping areas and research gaps. Transaction efficiency and flexibility, knowledge accumulation, agglomeration of economic activity, and spatial clustering of related firms and industries were placed in a four-field diagram in order to pedagogically demonstrate how knowledge emanating from different research areas – manufacturing belts, creative, entrepreneurial and learning regions, regional production systems, industrial districts, innovative milieux, and industry clusters – could interact and support each other (Malmberg, Sölvell, and Zander 1996, 89). This allowed for new and sometimes unexpected insights into forces of agglomeration and spatial clustering. One observation was that ‘it is important to focus upon knowledge accumulation effects of spatial clustering rather than exclusively upon potential benefits in terms of (short-term) transaction efficiency and mere flexibility’ (Malmberg, Sölvell, and Zander 1996, 94). When discussing knowledge accumulation within spatial clusters, the authors argued that transnational corporations ‘are dependent upon strong local milieu – or home bases – in the knowledge accumulation necessary for their long-term competitiveness’ (Malmberg, Sölvell, and Zander 1996, 94f). As such, there is no opposition to the notion of local knowledge accumulation. On the contrary, the increasing importance of a TNC in the global economy increases the relevance of the mutual benefits of innovation processes within the local milieu, the diffusion of knowledge to the external business environment, and the inflow of external knowledge (Malmberg, Sölvell, and Zander 1996, 93ff).

As the paper by Malmberg, Sölvell, and Zander addressed economic geography and international business research, it was cited in both ‘silos’ and consequently gained a broader readership than a more conventional article (see, e.g. John Dunning’s overview of ‘The Key Literature on IB Activities: 1960–2000’ in the Oxford Handbook of International Business (Dunning 2001, 45). It was also cited in an official report by the Swedish Government (Glimstedt 1999, 29; in SOU 1999:83) and reprinted in a widely read anthology edited by John Cantwell, a well-known international business scholar (Cantwell 2004, chp. 8).

Although the Malmberg-Sölvell-Zander paper can be seen as a harbinger of a new approach to industrial geography that would more explicitly be inspired by neighbouring disciplines, a few examples of novel ways of describing and explaining patterns of industrial activity had already emerged. In 1968, Gunnar Törnqvist published a revised version of his inaugural lecture, which was given on the occasion of the installation of the chair in economic geography at Lund University, in Geografiska Annaler, B (Törnqvist 1968). In his lecture, he outlined the contours of a research programme that was to become an important part of a new field of research in industrial economic geography – the geography of information flows and contact patterns. In this respect, he preceded the many studies that focused on the spatial behaviour of firms in industrial networks that were later published in Geografiska Annaler, B (see, e.g. Conti 1993; Grotz and Braun 1993; Malecki and Veldhoen 1993).

Another dimension of industrial geography that had basically been absent in the scholarship of De Geer and his successors, that might have been one source of inspiration for Malmberg, Sölvell, and Zander was the focus on the individual firm, rather than the industry, as the key actor in locational decision making. In his work ‘Behavior and Location,’ the American geographer Allan Pred suggested a model for an industrial location that focused on the decision maker within the firm (1967/1969). A more explicit investigation of the firm itself was provided by Krumme (1969), which was then extended by others (see, e.g. Laulajainen 1981; Laulajainen and Stafford 1995; Nilsson 1996). In another broadening of the perspective of locational decision-making, Håkanson (1981) brought the research and development functions within the multinational firm into the limelight and, in so doing, forged a link to management and organization studies, and to international business research. Notably, the ‘corporate geography’ approach was far from uncontroversial. Schoenberger (1989) demonstrated the gap between business dilemmas and social dilemmas in an attempt to broaden the firm’s decision making to a larger societal and public context. Walker (1989) lectured a ‘Requiem for Corporate Geography’, while Hagström (1990) suggested the ‘Unshackling of Corporate Geography’.

Despite these different views regarding corporate geography as a field of study aimed at revealing spatial patterns of industrial activity, the trend continued in Geografiska Annaler, B, especially in studies of international acquisitions through foreign direct investments (Bagchi-Sen 1995; Green and Meyer 1997; Ivarsson and Johnsson 2000). Another important development was the growing focus on the ‘invisible’ service production in relation to manufacturing (Bryson 2007; Daniels 2000). This shift was natural, given not only the increased role of the service sector in terms of employment and the share of GDP in advanced economies but also by the shift toward high-technology industries (Boschma and Van der Knaap 1999). This shift required a better understanding of the location factors that affect how added value is created (not only in mainstream technologies), how global financial flows are distributed and allocated (Clark 2005), and the creativity process itself (Power 2010; Pratt 2008; Scott 2010; Törnqvist 2004). A more recent trend has been the forging of evolutionary economic geography with environmental and sustainability aspects (see, e.g. Patchell and Hayter 2013). In this respect, it can be argued that economic geography is moving back toward its original close relation with the natural sciences but in a new context. I am sure that Sten De Geer would not object!

Disclosure statement

No potential conflict of interest was reported by the author.