Revealed Relatedness: Mapping Industry Space

Abstract

In this paper we measure technological relatedness between industries using a dataset on product portfolios of plants. For this purpose we first develop a general methodology to extract data on co-occurrences of classes (e.g. industries) in a single entity (e.g. a plant) to construct estimates of the relatedness between the classes. The core assumption, in line with the concept of economies of scope, is that if two products are produced in the same plant, this is an indication of relatedness between the industries the two products are a part of. Unlike earlier methods, we arrive at a Revealed Relatedness (RR) index that can be interpreted on a ratio scale, allows for the use of indirect (i.e. not directly observed) information on industry relatedness, and conceptualizes relatedness as being asymmetric or directed. Direction of relatedness provides information on, for example, the most likely direction of spillovers between two classes. We also graph the RR matrices using methods borrowed from social network analysis. The result is a visualization of the