BIMSA >
BIMSA Digital Economy Lab Seminar
Neural network-based carbon options: Multi-period pricing with binary terminal collapse and abatement triggers
Neural network-based carbon options: Multi-period pricing with binary terminal collapse and abatement triggers
演讲者
时间
2025年11月14日 15:00 至 16:00
地点
A3-2-303
线上
Zoom 435 529 7909
(BIMSA)
摘要
Global carbon pricing mechanisms are intensifying as emissions allowance futures markets reach maturation. This structural evolution in carbon finance has elevated derivatives trading, particularly those indexed to futures contracts, to central mechanisms for risk allocation. However, pricing carbon options remains challenging due to the limitations of traditional models in capturing policy-sensitive and technology-driven market dynamics. This study addresses three defining characteristics of allowance trading: (a) the no-arbitrage constraint, (b) the binary terminal price convergence property, and (c) the instantaneous triggering of abatement measures below allowance prices. We develop a risk-neutral pricing framework by constructing a stochastic differential equation (SDE) for allowance prices and deriving the governing partial differential equation (PDE) for carbon options, which is subsequently solved using physics-informed neural networks (PINNs). Our research progresses through three phases: First, we formalize the dynamic pricing process by rigorously embedding these characteristics into mathematical models. Second, we establish single-period and multi-period pricing architectures to quantify how iterative allocation rules and policy cycles propagate through option price formation. Finally, the PINN-based solutions facilitate systematic analysis of two critical dimensions: (i) how multi-period frameworks transmit policy uncertainty shocks to long-dated option premiums, and (ii) how abatement technology breakthroughs interact with allowance price thresholds to reshape option valuations and volatility patterns. Crucially, carbon options exhibit significantly lower time value compared to conventional commodity options, attributable to the risk of price collapse toward zero inherent in characteristic (b). This framework advances carbon derivatives pricing by integrating micro-mechanistic rigor with macro-policy responsiveness.