Introduction

Previous macroeconomic assessments of climate change typically project mild to moderate impacts on global GDP. These assessments often inform integrated assessment models (IAMs) that recommend slow emissions reductions, inconsistent with the Paris Agreement targets. A key limitation of many existing models is the assumption that only local weather affects a country’s economy. This paper examines the role of global weather conditions and finds that they significantly worsen projected economic impacts, indicating that interconnectedness via trade and supply chains makes countries vulnerable to weather events in other regions.

Methods and data

Data
Historical weather data (1962–2022) was sourced from the CRU TS v4.07 dataset and population-weighted at the country level. Climate projections (to 2100) were drawn from 22 CMIP6 models. Annual GDP per capita data was obtained from the World Bank. A total of 169 countries were included, with analysis averaging 50 years of observations per country.

Methods
Three established econometric models—Burke15, Kahn21, and Kotz24—were adapted to include global weather variables in addition to local ones. The models were then used to estimate future economic growth and project global GDP under high emissions (SSP5-8.5) compared to a low-emissions scenario (SSP1-2.6). 2,500 simulations were run across different climate models and resampled historical datasets. A revised damage function based on the updated models was incorporated into the DICE 2023 IAM to assess policy implications.

Results

Adding global weather increased projected losses substantially across models. By 2100, median GDP losses under SSP5-8.5 relative to SSP1-2.6 were:

• Burke15: from −28% (local only) to −86% (with global weather)
• Kahn21: from −4% to −19%
• Kotz24: from −11% to −40%

Uncertainty also rose with the inclusion of global weather. The increased projections reflect greater economic vulnerability to widespread and simultaneous climate impacts.

Country-level results varied widely. Without global weather, losses ranged from −82% (Mauritania) to +139% (Greenland) under Burke15. With global weather, most countries—particularly in the mid-latitudes including Europe, the US, China, and India—faced higher losses. These results suggest global weather exacerbates regional vulnerabilities, even in traditionally resilient economies.

Discussion

Why does global weather matter so much in the empirical models?
In a globalised economy, climate shocks in one country affect others via trade and supply chains. Global weather also captures climatic phenomena not reflected in annual country averages, such as widespread drought or heat. This may reduce the effectiveness of trade as a buffer, leading to simultaneous global supply disruptions. The findings remained robust when controlling for El Niño effects and CO₂ concentrations.

What are the implications of our results?
Revising the DICE 2023 model with the new damage function reduced the welfare-optimal level of warming from 2.7°C to 1.7°C, aligning with the Paris Agreement. It also implied steeper emissions cuts and higher optimal carbon prices post-2030. The results suggest that excluding global weather from IAMs leads to underestimation of risk and misaligned policy recommendations.

Limitations
The analysis assumes extreme warming (SSP5-8.5), and while robust across different baselines, outcomes under milder scenarios may differ. Projections rely on extrapolating historical data, which may not fully capture the effects of future extreme climate conditions. The inclusion of global weather was done in a simplified manner, and future research is needed to refine this approach using more granular and trade-weighted variables.

Conclusion

Including global weather in economic models leads to significantly higher projected damages from climate change. The results highlight the need to reconsider current modelling approaches and align policy frameworks with the greater risks posed by global interdependence under severe warming scenarios.