Unequal economic impacts of climate change via idealized carbon dioxide removal: insights from deep learning

[Geoengineering News]

https://www.cell.com/the-innovation/fulltext/S2666-6758(26)00252-3

Authors: Bin Tang, Jianan Wei, Yimin Liu, Bian He, Wen Bao, Yi Yang, Wenguan Wang, Wenting Hu, Anmin Duan

13 July 2026

Abstract

Rapidly reducing carbon dioxide (CO2) levels is essential to meeting the Paris Agreement’s temperature targets.1,2 Previous assessments of CO2 removal (CDR) have primarily focused on the hysteresis and invertibility of climate change itself,3–7 overlooking quantitative analysis of potential economic impacts of climate change via CDR. In this study, we first develop a powerful neural network model, EconClimNet, trained on decades of economic data from 1,554 sub-national regions worldwide and 111 climate indices derived from the fifth-generation ECMWF atmospheric reanalysis (ERA5). Compared to popular machine learning algorithms used in the Earth Science community, EconClimNet achieves superior performance in capturing the intricate relationship between climate indices and economic outcomes. On this basis, we apply EconClimNet to the outputs from idealized CO2 ramp up and ramp down experiments from phase 6 of the Coupled Model Intercomparison Project (CMIP6). It is found that altering climate trajectories through idealized CDR can yield significant long-term economic benefits globally, with Oceania overall positioned to experience the greatest gains with an increase trend about 1.41 [0.73∼8.76] million USD yr-1. Unequal economic effects also emerge across income levels. Rich countries see more benefits (increase trend about 1.56 [0.31∼6.21] million USD yr-1) via CDR, mainly contributed by services (increase trend about 0.03 [0.00∼0.23] million USD yr-1). Our findings underscore the need for governments to foster the innovation, development, and prudent deployment of CDR technologies, as well as to advance global CDR strategies that prioritize vulnerable nations from a climate justice perspective.

Source: The Innovation