New Insights on the Allocation of Innovation Subsidies: A Machine Learning Approach

Abstract

Gaining more insights on how R&D&i subsidies are allocated is highly relevant for companies and policymakers. This article provides new evidence of the identification of some key drivers for companies participating in R&D&i project selection processes. It extends the existing literature by providing insight based on sophisticated, accurate methodology. A metaheuristic optimization algorithm is employed to select the most useful variables. Their importance is then ranked using a machine learning process, namely a random forest. A sample of 1252 cases of R&D&i subsidies is used for more than 800 companies based in Spain between 2014 and 2018.