Identifying and Mapping Industrial Districts Through a Spatially Constrained Cluster-Wise Regression Approach

Abstract

The aim of this article is to exploit an innovative spatial econometric approach to map and study the evolving patterns of industrial districts (IDs). The procedure can be classified as a $k$-means cluster-wise regression procedure and is designed to detect homogeneous areas of subcontracting activity. These spatially contiguous aggregations of subcontractors are identified in terms of production function homogeneity and are defined as spatial regimes. Using this procedure, it is possible to detect two important sources of agglomeration economies that are commonly associated with the presence of an industrial district. The methodology is tested on a sample of Italian micro and small-sized subcontracting firms operating in the footwear industry, showing its effectiveness in identifying the most commonly known IDs in this sector. Most ID regimes are persistent over time, despite the high turnover rates in the local subcontracting population after the 2008 financial crisis. These results can be explained by the presence of locally rooted competencies and context-specific knowledge bases that persist despite the changing actors operating in the locality. Our evidence also shows that location in an ID does not necessarily entail benefits in terms of performance for subcontracting firms.