Characterising innovation policy mixes in innovation systems

Abstract

Along with the rising economic relevance, scope, and complexity of innovation policy, attention to the concept of the innovation policy mix has surged. Yet, knowledge about how innovation policy mixes relate to innovation systems is limited. Drawing on expert survey data on 4710 policies, this paper uses natural language processing to map innovation policy mixes and represent them as mixtures of 25 distinct topics whose proportions vary across countries. It identifies research, innovation in firms and supporting coordination in the innovation system as key focal areas in these mixes and analyses how these focal areas vary in relation to different performance aspects of innovation systems as well as structural and institutional country characteristics. The results indicate that policymakers design policy instruments supporting innovation in firms to mitigate comparative weaknesses of the innovation system, while they design policy instruments supporting research to complement comparative strengths of the innovation system. The results also indicate that the design of policy instruments strengthening coordination in the innovation system is partly shaped by factors unrelated to innovation. These findings illustrate the utility of a novel dataset and natural language processing methods for innovation policy studies and advance the debate about the determinants of innovation policy design.