Abstract
Climate change is a major threat to the future of humanity, and its impactsare being intensified by excess man-made greenhouse gas emissions. One methodgovernments can employ to control these emissions is to provide firms withemission limits and penalize any excess emissions above the limit. Excessemissions may also be offset by firms who choose to invest in carbon reducingand capturing projects. These projects generate offset credits which can besubmitted to a regulating agency to offset a firm's excess emissions, or theycan be traded with other firms. In this work, we characterize the finite-agentNash equilibrium for offset credit markets. As computing Nash equilibria is anNP-hard problem, we utilize the modern reinforcement learning techniqueNash-DQN to efficiently estimate the market's Nash equilibria. We demonstratenot only the validity of employing reinforcement learning methods applied toclimate themed financial markets, but also the significant financial savingsemitting firms may achieve when abiding by the Nash equilibria throughnumerical experiments.