A1072
Title: Modeling innovation ecosystem dynamics through interacting reinforced Bernoulli processes.
Authors: Federico Nutarelli - IMT Lucca (Italy) [presenting]
Irene Crimaldi - IMT Lucca (Italy)
Andrea Ghiglietti - Universita degli Studi di Milano-Bicocca (Italy)
Giacomo Aletti - University of Milan (Italy)
Abstract: Understanding how capabilities grow into core strengths, and how those strengths can harden into rigidities, is central to innovation strategy. Yet formal modeling is difficult because innovations in one area often reshape others, making specialization endogenous. This challenge is acute in ecosystems where firm performance depends on managing interdependencies and complementarities. The aim is to introduce a formal model based on interacting reinforced Bernoulli processes that tracks how patent wins propagate across technology categories and how those categories co-evolve. The model reproduces key empirical regularities: sublinear growth in cumulative success, convergence of success shares across fields, and declining cross-category correlations over time. Using GLOBAL PATSTAT data (1980-2018), the model is validated, the structural interaction matrix is estimated, and a statistical test is developed for the strength of cross-category effects under a mean-field approximation. By endogenizing specialization, the framework offers a practical tool for policymakers and managers who must steer complex, co-evolving innovation systems.
