Falling temperatures are not the only thing carbon dioxide removal might buy. A modeling study published in The Innovation suggests that pulling CO2 back out of the air after a period of overshoot would also lift economic output almost everywhere on Earth — with the largest gains flowing to the richest countries.
Efforts to cut emissions are struggling to meet the Paris Agreement's temperature targets, and many assessments now assume that carbon dioxide removal (CDR) — deliberately drawing CO2 out of the atmosphere — will make up the difference. Whether CDR would pay off economically, however, has been hard to judge. Traditional econometric models, the standard tool for pricing climate damage, were built for a handful of variables and struggle once every temperature and rainfall statistic becomes a predictor.
A team led by Wenting Hu of the Institute of Atmospheric Physics, Chinese Academy of Sciences, and Wenguan Wang of Zhejiang University built EconClimNet, a deep learning model designed for exactly this problem. It adapts the widely used Vision Transformer architecture so that regions are connected in two tracks — loosely across borders and more tightly within them, mirroring how economies actually correlate. Trained on decades of economic output and ERA5 climate data, it predicts economic indicators from climate indices more accurately and more stably than traditional econometric approaches and mainstream deep learning baselines.
In idealized simulations where CO2 concentrations first rise and then are drawn back down, four key climate indices recover more than 80% on average across most of the globe. Recovery is incomplete, though: daily temperature variability rebounds most easily, while mean temperature and the number of wet days lag behind — leaving the world warmer and wetter than preindustrial, especially in northern North America and central Eurasia.
The economic feedback follows the same asymmetry. While CO2 climbs, global output declines; once removal begins, output grows in nearly every region. But the ledger is uneven. Oceania suffers the deepest losses during warming and gains the most during removal, yet over the long run only Asia and Europe recoup what warming cost them. Across income levels, richer countries gain more. Services benefit most, manufacturing comes second, and agriculture contributes least — in rich nations it responds negatively as temperatures fall.
The authors are explicit about scope: the scenarios are idealized, and deploying CDR at the assumed scale remains a formidable technical challenge. The study, they write, is not a near-term roadmap but a quantitative basis for long-term research and policy planning — one they hope will be read through the lens of climate justice, prioritizing support for vulnerable countries that see the smallest gains.
The Innovation
Computational simulation/modeling
Not applicable
Unequal economic impacts of climate change via idealized carbon dioxide removal: Insights from deep learning
1-Oct-2026
The authors declare no competing interests.