Jianxian Wua, b Hui Jinc Yaping Luo

---

Grounded in signaling theory and market microstructure scholarship, this study conceptualizes artificial intelligence (AI) not as an algorithmic trading tool but as a policy signal accelerator—an information- processing layer that compresses the time and informational loss between policy signal generation and price-level internalization. We develop a theoretical framework identifying two causal pathways through which AI improves carbon market price discovery— accelerating overnight signal instantiation into opening prices and filtering non-informational noise from intraday price movement. Exploiting the staggered rollout of China’s carbon trading pilots across 288 cities, we employ a quasi-experimental design to estimate the causal effect of AI adoption on policy-driven emission reductions. Results show that AI amplifies the carbon trading emission-reduction effect by 2.1 percentage points. Mechanism analysis confirms that AI significantly raises the price efficiency ratio, indicating that carbon permit prices move more decisively in the direction of net informational content. These findings provide evidence that AI structurally enhances the signaling architecture of
market-based climate policy.

Keywords: artificial intelligence; carbon emission trading; signaling theory; policy signal accelerator; price efficiency