• Corpus ID: 10896891

FOR IDENTIFYING LOCAL AND DOMESTIC INDUSTRIAL CLUSTERS USING INTERREGIONAL COMMODITY TRADE DATA Rainer vom Hofe

@inproceedings{Bhatta2007FORIL,
  title={FOR IDENTIFYING LOCAL AND DOMESTIC INDUSTRIAL CLUSTERS USING INTERREGIONAL COMMODITY TRADE DATA Rainer vom Hofe},
  author={Saurav Dev Bhatta},
  year={2007}
}

• S. D. Bhatta
• Published 2007
• Economics

Defining geographic as well as industrial boundaries in industrial cluster analysis is a challenging task. The most widely used methods for identifying industrial clusters utilize data from input-output tables. Given that inter-industry linkages are not confined to political boundaries, the questions discussed in the relevant literature on input-output based approaches focus on whether to use a regional or a national input-output framework. But neither framework can capture non-local transactions, which may be of major importance for smaller study regions. In the presented research, we expand the regional Chicago input-output framework to an interregional framework which accounts for both regional and non-local (i.e., commodity imports and exports by industry) inter-industry transactions. Applying factor analytical techniques to the data in this inter-regional framework, we are able to derive two sets of industrial clusters which we refer to as local and domestic clusters. In addition, we also identify the key sectors for each cluster based on indicators for industry backward and forward linkages. A comparison of these local and domestic clusters shows that while there are some similarities between these two types of clusters, there are also some significant differences between them. Thus, we demonstrate that it is important to include non-local inter-industry transactions as well in applied industrial cluster analysis 1 Many thanks to Ke Chen, Assistant Professor of Economics, Finance, Geography and Urban Studies at the East Tennessee State University, for her assistance with the factor analysis. The Industrial Geographer vom Hofe and Bhatta 2 INTRODUCTION In recent years, industrial cluster analysis has received increasing attention from regional economists, economic development practitioners and policy makers alike. After decades of first-wave industrial attraction efforts (i.e., smokestack chasing) followed by years of second-wave industrial retention and expansion strategies in the 1980s, the beginning of the 1990s has seen the emergence of what might be called thirdwave economic development strategies (Herbers 1990). These third-wave approaches are characterized by a focal shift from firm and business-level assistance to strategies that stimulate the formation of networks and clusters of industries, workers, social and political institutions, and demand markets (Fosler, 1992). Furthermore, these strategies are industrial-policy oriented, rather than project-oriented, and recognize the importance of community and business collaborations in strengthening entire and globally competitive industrial sectors (Bradshaw and Blakely, 1999). As earlier work by Isard (1959), Czamanski (1974), and Ó hUallacháin (1984) indicates, the concept of cluster-based economic development is not a new economic development theory. Its roots can actually be traced to Marshall’s (1920) ideas of agglomeration economies. But this concept gained currency only in the 1990s when it emerged as the new economic development panacea (Martin and Sunley, 2003). And much of the popularity boost of industrial clusters is attributable to Porter’s (1990) widely-cited Diamond of Advantage in which he identifies four determinants that foster cluster development and where governments can play a proactive role.2 But despite all the 2 The four determinants included in Porter’s Diamond are: firm strategy and rivalry, factor and demand conditions, and a network of related and supporting industries. theoretical and conceptual understanding of the working mechanisms of industrial clusters, there is still no consensus in the relevant literature on the use of quantitative methods for proper cluster identification. One important strain of cluster identification methodologies has evolved around using input-output (I-O) tables. The basic idea here is to identify inter-industry linkages based on vertical purchase-sales relationships, or group industries according to similarities in buying/selling patterns. Though widely applied, I-O based approaches for identifying regional clusters are open to a number of criticisms. One major shortcoming of these methods is that since I-O tables are aspatial by design, clusters identified using I-O based methods do not contain any information on how close firms and businesses are located to each other (Latham 1976). As suggested by Ó hUallacháin (1984), this criticism can be partially addressed by arguing that clusters identified using I-O tables do represent spatial agglomerations of industries when the region under consideration is relatively small. But even when the aspatial nature of I-O tables is not an issue, the identified regional clusters have yet another limitation, namely their inability to account for nonregional industry trade (Feser and Luger 2002). As a result, local clusters that cross regional boundaries might not necessarily be identified using these methods, a particularly relevant limitation in these times of increasing globalization. The primary goal of this paper is to present a revised I-O-based methodological framework for identifying meaningful industrial clusters and apply it to the Chicago Metropolitan Statistical Area (MSA). The contribution of this research to the existing body of cluster literature is that our approach uses an extended regional I-O table that incorporates commodity trade between the region and the rest of the nation. This modification allows The Industrial Geographer vom Hofe and Bhatta 3 us to identify two types of clusters, which we refer to as local and domestic clusters. Following Czamanski and Ablas (1979) and Feser and Bergman (2000), we define an industrial cluster as a group of industries that are linked by significant inter-industry trade, similarities in buying/selling patterns, or similarities in markets for resources and outputs. A local industrial cluster is defined as a cluster confined to the region under consideration while a domestic cluster could be part of a supra-regional cluster with a strong presence in the local economy. An important aspect of this research is the recognition that clusters do not represent purely local or purely national phenomena. As the focus of many cluster studies is on sub-national regions, such as metropolitan, state, or multi-state regions, non-regional trade activities are often very important for the study area. Therefore, clusters cannot be properly studied in isolation from the surrounding national economy. Focusing on selected clusters for the Chicago Metropolitan region, we will show that a comparison of local and domestic clusters can help us to understand the existing strengths of the local economy as well as provide insights for devising potential cluster-based economic development strategies. The next section provides a brief overview of industrial cluster concepts and cluster identification methodologies based on inputout tables. The relevance of this research is outlined in section III, followed by a description of the data and study region in section IV. The next two sections present the methodology and findings of this study. And the final section presents some concluding remarks. IMPORTANT CONCEPTS AND METHODOLIGIES IN INDUSTRIAL CLUSTER ANALYSIS In his widely cited book Principles of Economics, Marshall (1920) described three agglomeration forces which explain the existence of spatial conglomerations of firms. These economies of localization are knowledge spillovers among firms, labor market pooling, and cost advantages produced by the sharing of industry-specific non-traded inputs. Subsequently, Hoover (1948) extended Marshall’s concept of external localization economies by identifying two more agglomeration forces, namely, the existence of large, diverse markets (external economies of urbanization) and the availability of large and specialized factors of production increases (internal returns to scale). These theories of agglomeration economies have played a key role in shaping the definitions of industrial clusters. Cluster concepts that focus on regional industrial specialization or the similarity among industries can trace their roots to Marshall’s and Hoover’s theories of agglomeration economies. Hanson (2000), for instance, emphasizes the role of localization economies in the formation of clusters and states that “the existence of localized externalities implies that firms prefer to be near large agglomerations of other firms in their own industry or related industries” (p. 4). Similarly, Hill and Brennan (2000) define a competitive industrial cluster as a “concentration of competitive firms or establishments in the same industry” (p. 67). Based on similar concepts, Munnich et al. (1998), Rex (1999), Botham et al. (2001), and Peters (2004) measure regional specialization and identify industry The Industrial Geographer vom Hofe and Bhatta 4 agglomerations through location quotients computed using industry employment data. However, as pointed out by Doeringer and Terkla (1995) and Rosenfeld (1997), location quotients are by no means sufficient to identify industrial clusters as they fail to account for the role of inter-industry trade linkages in the spatial grouping of industries. Isard et al.’s (1959) pioneering work on industrial complex analysis is the first approach to grouping firms into clusters that do not necessarily belong to the same or similar industries. What Isard et al. loosely referred to as “combinations of industrial activities for a region” (p. 1) became the foundation for cluster analysis based on production value chain linkages derived from input-output tables. Intensive research on the use of input-output tables for identifying industrial clusters was done in the 1970’s and 1980’s by scholars like Czamanski (1974, 1976), Roepke (1974), and Ó hUallacháin (1984). Input-output tables continue to be widely used in industrial cluster analysis to this day. However, scholars like Feser and Bergm