Stage-Aware Public Finance for Deep-Tech Transformation: An Agent-Based Model of Congestion, Signalling, and Funding Efficiency

Building on a model presented at EMS 2025, this study develops a three-stage agent-based model (Idea, Prototype, Scaling) of public deep-tech funding to examine when public support crowds in private co-investment and which innovation stage should receive marginal budget increases. Grounded in market failure, systems of innovation, mission-oriented and signalling perspectives, and calibrated to EIC, EIB and OECD data, the model reproduces persistent Prototype-stage congestion. Results show that stronger signalling increases bankable exits and lowers public cost per outcome, and that Scaling-focused budget increases outperform proportional allocation while upstream-only increases reduce throughput, yielding governance heuristics for public innovation funders.