Strong regional clusters are increasingly seen as a response to economic globalization by policy makers and regional development agencies. The reasoning of competitive advantages of countries and regions with enterprises organized in clusters has mainly been popularized by Porter (1990, 1998, 2000). As well-working clusters are associated with high productivity growth and innovation potential, the cluster approach has become appealing in different fields of economic policy. In particular cluster-based instruments are in integral part of EU regional policy (Christensen et al., 2011). In most EU countries cluster-oriented policy plays an important role at the national and regional level (Oxford Research, 2008). This also holds for Germany where diverse national and regional programmes were set up to promote cluster development (Török, 2012). Although the cluster approach is based on the agglomeration theory, a variety of definitions of a cluster exists (Martin/Sunley, 2003). The present paper aims at improving the strategy of regional cluster identification. First, at the national level, the dominant related sectors of R&D intensive industries are basically discovered by qualitative input-output analysis (QIOA). Yet it has to be allowed for the fact that usually not all enterprises of these sectors belong to the respective value added chains. Thus, QIOA has to be supplemented by quantitative input-output analysis in order to avoid distortion effects that arise from defining too heterogeneous clusters. Here downstream and upstream sectors are considered according to their involvement in the production activities of the key industry. Secondly, at the local level, it has to be settled whether and how spatial externalities and spillovers should be allowed for in locating regional clusters. Most applied cluster studies ignore the presence of spatial interaction between interrelated geographical units. If geographical units are considered to be spatially independent in the presence of spillover, however, spatial clustering tends to be underestimated (Guillain/Le Gallo, 2007). Feser/Koo/Renski/Sweeney (2001) and Feser/Sweeney (2002) were the first to explicitly accounting for spatial interaction between regions in an applied cluster study for the US state of Kentucky. In a follow-up study, Feser/Sweeney/Renski (2005) extended spatial analysis to the United States as a whole. Both studies make use of the Getis-Ord statistic to measure and test for local spatial clustering (Ord/Getis, 1995). Recently, Pires et al. (2013) utilize the local Moran test for localizing industrial clusters in Brazil. A major drawbacks of both local methods is the necessity of fixing the environments of the regions in advance. reach of the geographical extent of potential spillover effects in advance. To allow for varying reaches of the geographical extent of regional interaction, here the flexible approach of spatial scanning is adopted (Kulldorff, 1997). On the basis of Kulldorff's scan test, the variable extent of potential regional clusters is accurately captured.
