Extreme weather events inflict substantial economic and human costs globally. The frequency and intensity of these events are increasing, with a sevenfold rise in reported disaster losses since the 1970s, as documented by the World Meteorological Organization. While factors like improved reporting and increased population density in vulnerable areas contribute to this rise, a significant portion is undeniably linked to anthropogenic climate change, as confirmed by the Intergovernmental Panel on Climate Change. This study utilizes Extreme Event Attribution (EEA) methodology to quantify the climate change component of these costs. EEA examines how human-induced greenhouse gas emissions have altered the likelihood and/or intensity of specific extreme weather events. This approach, initially conceptualized by Allen (2003) and applied to the 2003 European heatwave, compares event probabilities in our current climate with those in a counterfactual world without anthropogenic emissions. Two approaches are commonly used: a risk-based approach using the Fraction of Attributable Risk (FAR) metric, and an intensity-based approach focusing on the climate change contribution to event intensity. The economic costs of extreme weather are assessed through direct economic damages (easily quantifiable stock) and indirect losses (more challenging to quantify flows). This study primarily focuses on direct damages, acknowledging the limitations of excluding indirect economic losses. The main objective is to demonstrate the use-value of the attribution-based methodology rather than presenting definitive global estimates due to current data limitations. The study aims to provide evidence suggesting that many Integrated Assessment Models (IAMs) are significantly underestimating the current economic costs of climate change.

The paper reviews existing methodologies for estimating the economic impact of climate change, primarily focusing on Integrated Assessment Models (IAMs). These models, such as DICE and FUND, typically relate economic impacts to changes in average annual temperature, often neglecting or underestimating the costs of extreme weather events. The study highlights the limitations of IAMs, such as their focus on average temperature deviations rather than changes in the distribution of extreme weather events, leading to potential underestimation of the costs. Existing studies on EEA and economic cost attribution of individual extreme weather events are reviewed, laying the groundwork for the current study's global aggregation and extrapolation approach. The study also discusses the limitations of existing economic datasets like EM-DAT, highlighting the challenges of incorporating indirect economic losses and the uneven geographical coverage of economic data.

This study combines data from Extreme Event Attribution (EEA) studies with economic impact data from the Emergency Events Database (EM-DAT). The EEA studies provide Fraction of Attributable Risk (FAR) estimates, indicating the portion of an extreme weather event's risk attributable to anthropogenic climate change. The FAR is calculated as 1 - P₀/P₁, where P₀ is the probability of the event without anthropogenic greenhouse gases, and P₁ is the probability with current levels. The study uses a frequentist approach for FAR aggregation, excluding Bayesian attribution studies due to their limited availability. The economic costs are calculated by multiplying the FAR for each event by its total economic cost (including direct damages and mortality costs using a Value of Statistical Life (VSL)). The dataset includes 185 events from 2000 to 2019 covering heatwaves, floods, droughts, wildfires, and storms. Two extrapolation methods were employed: a global average extrapolation and a regional average extrapolation. The global average method calculates an average FAR for each event type globally and applies it to all similar events in EM-DAT. The regional method calculates average FARs per event type and continent, providing a more refined geographical perspective. However, data limitations necessitate using global averages for many region-event type combinations. The Value of a Statistical Life (VSL) is used to monetize mortality costs, utilizing a consistent value across all countries for simplicity. Data selection prioritized higher-ranked studies based on the Scimago Journal Rank (SJR) when multiple studies were available for the same event. The final dataset was carefully curated to ensure the best available estimates for each event, considering the spatial and temporal match between EEA studies and EM-DAT's economic data.

The analysis reveals that a net 60,951 deaths are attributable to climate change across the 185 events in the dataset, amounting to US$431.8 billion in economic costs. This represents 53% of the total observed damages. Storms constitute the largest portion (over 64%) of the attributable damages, followed by heatwaves (16%), floods (10%), droughts (10%), and wildfires (2%). Extrapolating these findings globally, using the global average FAR method, reveals a total climate change-attributed cost of US$2.86 trillion over 2000-2019, averaging US$143 billion per year. This includes nearly US$90 billion for loss of life and US$53 billion in economic damages annually. The median FAR method produced a larger estimate (US$167 billion annually), suggesting a possible bias in the available data. Annual costs exhibit substantial variability, with peaks driven primarily by high-mortality events like the 2003 European heatwave, 2008 Cyclone Nargis, and the 2010 Russian heatwave/Somali drought. The highest annual cost reached US$620 billion in 2008. The distribution of costs across income groups shows high-income countries bearing the highest climate change-induced economic costs (47%), mainly due to high asset exposure in countries like the United States. However, the observed distribution may also reflect data availability and measurement biases, with lower-income countries likely underreporting economic damages. The annual climate change-attributed costs are estimated to range from 0.05% to 0.82% of global GDP. Notably, low-income countries experience comparatively higher relative costs, at almost 1% of GDP annually, primarily due to higher mortality rates. The study compares these findings with estimates from Integrated Assessment Models (IAMs) like DICE and FUND, revealing a considerable underestimation of extreme weather costs by IAMs. For example, DICE's estimate (US$4.04 trillion) is 40% larger than the attribution-based estimate. This difference is partially explained by IAMs focusing on average temperature changes rather than changes in extreme weather distributions.

The findings demonstrate the potential of an attribution-based approach to quantify climate change-related costs of extreme weather. The study highlights the significant underestimation of these costs by traditional macroeconomic models like IAMs. The significant discrepancy arises due to IAMs failing to capture the changes in extreme weather events, which represent a substantial part of the climate change impacts. The attribution-based approach offers a complementary perspective by focusing on the observed impacts of individual events. The substantial yearly variability in climate change-attributed costs underscores the importance of adapting to extreme weather events. The higher relative costs experienced by low-income countries underline the need for targeted adaptation strategies and international support. Further research is essential to improve the accuracy of these estimates, especially by addressing data gaps in low-income countries and improving the representation of various extreme weather event types in attribution studies. These improvements will enhance the reliability of the attribution-based methodology and refine the understanding of the global economic costs of climate change from extreme weather.

This research presents a novel global estimate of the economic costs of extreme weather events attributable to climate change, highlighting the significant underestimation by existing models. The findings underscore the need for increased mitigation efforts to reduce the fraction of attributable risk and for enhanced adaptation strategies to minimize the economic and human costs of extreme weather events. Future research should prioritize expanding the availability of high-quality EEA studies, particularly in low-income countries and for under-represented event types, and refining economic data collection to capture indirect losses more accurately. This improved data will enable more precise and comprehensive estimations of the global economic cost of climate change from extreme weather.

The study acknowledges several limitations. The main limitations stem from data scarcity and uneven geographical coverage of EEA studies and economic data. Many region-event type combinations lack sufficient attribution studies, necessitating the use of global averages in the extrapolation process. The focus on direct economic damages and mortality excludes indirect economic losses, potentially underestimating the true costs. The use of a uniform VSL across countries might not fully capture variations in the value of life across different income levels. Finally, potential biases in event selection for attribution studies might affect the generalizability of findings. Addressing these limitations requires more comprehensive EEA studies and better economic data collection, especially in low-income regions.