Fixed climate feedback assumptions systematically underestimate policy-relevant economic risks: Implications for climate resilience

Integrated Assessment Models (IAMs), central to informing climate policy, typically treat the climate feedback parameter as a fixed constant, contradicting robust evidence that it weakens as the planet warms. Here, we show that this structural simplification systematically underestimates policy-relevant economic risks. By integrating a state-dependent feedback mechanism calibrated to CMIP6 Earth System Models into a coupled DICE-FaIR framework, we reveal three key findings. First, dynamic feedback produces a consistent warming amplification of ∼0.2 °C across policy scenarios, demonstrating that conventional IAMs physically underestimate future warming trajectories. Second, this physical bias cascades into economic metrics, elevating the Social Cost of Carbon by up to 50% across diverse structural specifications, indicating a systematic limitation rather than a parameter-specific outcome. Third, we identify a ‘policy space compression’ effect: dynamic feedback simultaneously amplifies the tail risks of exceeding critical temperature thresholds while narrowing the feasible window for net-zero emissions. This physically driven compression is systematically obscured by fixed-feedback assumptions, which overstate the available policy space and understate the urgency of accelerating mitigation timelines. These findings demonstrate that neglecting the temperature-dependent weakening of climate feedback creates a biased foundation for risk assessment, with current resilience strategies potentially calibrated to overly optimistic baselines that inadequately account for the evolving dynamics of the Earth system.