Measuring Proximity between Firms: A Literature Review

It is well known that to define the firms’ businesses, the codes of the standard industrial classifications are used, as well as to circumscribe the firms’ Research & Development (r&d) activities, reference is made to the technology classes in which patent applications are filed. Once the business or r&d activity of a firm is circumscribed, researchers tend to process the information collected to measure the proximity between firms (or groups of firms) on a multidimensional space or through relational graphs. The objective of this review is to circumscribe the contributions that focus on the definition of business and r&d activity as an operating step to measure proximity, and to systematise the techniques and methodologies adopted in the literature. The selected works are mainly, though not exclusively, related to the strands of related variety and technological proximity. Although they are widely used for statistical purposes, economic activity codes and technology classes show numerous limitations as highlighted by the most recent literature. For this reason, innovative techniques and methodologies based on alternative data sources are increasingly used. Today, huge amounts of data are available, often in unstructured and textual form. Thanks to text mining techniques and text analytics methodologies, it is possible to process information from texts on company websites describing products and/or services, from companies’ social objects or from descriptions in patent applications. Such applications on text data are spreading in large numbers, represent an emerging strand (not necessarily related to the strands of related variety and technological proximity), and are opening an appreciable avenue in regional economics, innovation economics and operations research.