Conference Agenda

Accelerating low-carbon technological innovation is central to achieving global climate targets, yet robust evidence on whether national climate policies effectively trigger real-world innovation remains limited. Existing research largely focuses on single technologies or individual policy instruments, limiting its ability to generalize across heterogeneous technological systems. Here, we develop a machine-learning-assisted analytical framework to systematically assess policy-induced innovation dynamics across 34 economies, nine major low-carbon technologies, and more than 4,100 policy interventions over the period 1990~2023. By combining structural break detection with a multidimensional characterization of policy intensity, stability, and policy mix, we identify statistically significant innovation take-offs and trace the policy conditions under which they emerge. We find that only 3~32% of climate policies lead to measurable innovation take-offs, with pronounced heterogeneity across technologies. Mature renewable technologies exhibit low marginal policy sensitivity but generate large innovation surges once triggered, whereas high-cost and long-cycle technologies respond weakly and often with substantial delays of 1.5~2.5 years. Nearly 60% of observed innovation take-offs occur under relatively stable policy environments, highlighting the importance of long-term institutional credibility. Nevertheless, well-timed policy strengthening can still induce breakthrough effects during specific technological opportunity windows. Overall, our findings demonstrate that climate policy does not automatically translate into effective technological innovation; policy effectiveness instead depends on alignment with technological maturity, cost structures, development horizons, and national institutional contexts to enable sustained progress toward net-zero targets.