Zooming In: A Practical Manual for Identifying Geographic Clusters
Executive Summary — The concept of clusters-high concentrations of economic activity in a specific geographic unit-is the foundation for a vast amount of research in economics, management, urban planning, sociology, and public policy. Despite notable exceptions, little research has looked carefully at the key issue of cluster identification. In this paper the authors detail the reasons, procedures, data, and results of their effort to identify geographic clusters. They want to increase awareness of the complexities behind cluster identification, and to provide a concrete method that can help researchers define clusters more accurately. In particular, the authors address three related questions in cluster identification: (1) What economic activity should be measured to determine clustering? (2) What is the appropriate geographic unit over which economic activity should be measured? (3) What levels of economic concentration are high enough for the geographic unit to be labeled a cluster? They answer these questions with a combination of literature review, theoretical discussion, and illustrations with various algorithms. While they use a specific empirical context-the global semiconductor industry-for illustrative purposes, the insights and methodologies are general enough for other contexts. The organic cluster identification methodology they propose is especially useful when researchers work in global settings, where data available at different geographic units complicates cross-country comparison. Key concepts include:
- Given the importance of understanding firms' location strategies and economic geography more generally, it is necessary to look at the factors behind the choice of an appropriate geographic unit for empirical analysis, as well as the implications of such choices.
- This study provides a new method to identify a cluster organically based on the economic activities in the data. The method offers unique advantages in precision, flexibility, and applicability to cross-country studies.
This paper takes a close look at the reasons, procedures, and results of cluster identification methods. Despite being a popular research topic in strategy, economics, and sociology, geographic clusters are often studied with little consideration given to the underlying economic activities, the unique cluster boundaries, or the appropriate benchmark of economic concentration. Our goal is to increase awareness of the complexities behind cluster identification and to provide concrete insights and methodologies applicable to various empirical settings. The organic cluster identification methodology we propose is especially useful when researchers work in global settings where data available at different geographic units complicates comparisons across countries.