Increase in insurance losses caused by North Atlantic hurricanes in a warmer climate

North Atlantic hurricanes are a significant cause of property damage in the United States and a major peril for the global reinsurance industry, accounting for nearly 30% of yearly global insured property losses since 1970. This percentage has risen recently due to increased coastal development and the effects of climate variability and global warming. Hurricanes are complex, requiring moist air, warm ocean water, a pre-existing disturbance, and low wind shear. Climate variability influences these factors, and a warmer climate may increase their availability. Increased evaporation from warmer ocean surfaces could lead to more efficient water vapor supply, and a warmer atmosphere may allow hurricanes to hold more water vapor. Although significant trends in hurricane frequency and intensity aren't yet evident, there's medium-to-high confidence that climate change will lead to more intense hurricane winds and rainfall, likely resulting in higher storm surges and exacerbating the impact of sea level rise on coastal flooding. Previous studies have explored the economic loss dependency on hurricane wind, precipitation, and storm surge, highlighting the significant impact of hurricane intensification due to global warming on economic losses. However, the effects of climate change on hurricane-induced insurance losses in the United States, particularly considering the combined effects of wind, precipitation, and storm surge, remain unclear. This study uses over 70 years of historical data to quantify the potential impacts of a warmer climate on property insurance losses from North Atlantic hurricanes, combining NOAA hurricane data with insurance loss information from the Property Claim Services (PCS) and the National Flood Insurance Program (NFIP) to analyze wind, precipitation, and storm surge losses for nearly 150 historical hurricanes.

Existing research indicates a strong link between climate change and increased hurricane intensity and precipitation. Studies like Knutson et al. (2020) provide projections of changes in hurricane physical properties, suggesting increases in major hurricane frequency and rainfall. Other studies have investigated the relationship between economic losses and hurricane characteristics, such as wind speed, precipitation, and storm surge (e.g., Murnane & Elsner, 2012; Czajkowski et al., 2017; Tonn & Czajkowski, 2022). These studies highlight the potential for significant increases in average annual losses due to climate change. However, a comprehensive understanding of the combined impacts on insurance losses, considering the interplay between wind, precipitation, and storm surge, has been lacking. This gap in the literature motivates the present study, which aims to provide a more holistic assessment of the effects of climate change on insurance losses.

This study utilizes three primary datasets: 1) the HURDAT2 dataset providing detailed information on hurricanes making landfall in the US since 1851; 2) the PCS insurance industry dataset of property losses since 1950; and 3) the NFIP claims dataset listing paid property flood claims since 1979. These datasets were combined to compile information on nearly 150 North Atlantic storms causing insurance losses in the US from 1950 to 2022, detailing wind, precipitation, and storm surge losses. The study uses a bootstrapping technique to generate stochastic event sets, incorporating the probability distribution of projected changes in North Atlantic hurricane intensity and precipitation for +2°C and +4°C warming scenarios. The maximum lifetime storm intensity (rather than landfall intensity) was used as it aligns with hurricane climatology projections. Historical losses were indexed to 2022 to account for economic inflation, insurance penetration growth, population, and wealth per capita, using a 7% annual indexation factor based on GDP deflator and property insurance gross premium data. This indexation does not, however, consider spatial variations in urban development and changes in vulnerability. For hurricane climatology projections in the +2°C scenario, frequency distributions from several independent studies were used, extrapolated to the +4°C scenario using exponential scaling of global mean surface temperature. The analysis considers median changes in hurricane climatology including impacts on hurricane intensity (increase in major hurricanes), frequency (assumed unchanged in the main analysis, but explored in supplementary notes), and precipitation (projected increase). The bootstrapping methodology involves sampling the yearly number of events, dividing them into hurricanes and major hurricanes based on the proportion of major hurricanes in the scenario, and then sampling wind, precipitation, and storm surge losses considering their correlation structure using a Gaussian copula. This generates stochastic event sets for both present and future climate scenarios allowing for the quantification of return periods of insurance losses. The uncertainty from projected changes in hurricane climatology was assessed by generating multiple event sets from the probability distributions of future changes. Verisk’s North Atlantic hurricane model was used to augment empirical CDFs for return periods longer than the historical dataset.

The analysis reveals that median projected changes in hurricane climatology may increase event losses by 5–15% in a +2°C scenario and 10–30% in a +4°C scenario, with greater impact on high-frequency, low-severity events. The 100-year insured loss (approximately USD 220 billion) could have return periods of 80 years (+2°C) and 70 years (+4°C). A +2°C warming could lead to a 10% median increase in average annual loss (AAL), and +4°C warming could cause a 15% median increase. Precipitation-driven losses show a larger percentage increase than wind and storm surge losses. The contribution of major hurricanes (category 4 and 5) to the 2-year return period annual loss increases significantly in warming scenarios (from approximately 10% to 50–60%). The AAL breakdown by hazard shows that while wind-related losses experience the largest absolute increase, freshwater losses show the largest relative increase. Uncertainty analysis using inter-quantile ranges shows a wide range of potential outcomes for future hurricane losses. Similar loss projections were obtained using Verisk's stochastic event set, corroborating the findings. Critically, precipitation-induced annual losses may show the largest relative increase due to increased precipitation and a higher proportion of major hurricanes. The average annual loss of major hurricanes is expected to increase, while that of lower-intensity hurricanes may decrease. If projected changes in storm frequency were included, the median US property insurance losses are projected to decrease; however, scientific confidence in these projections is low.

These results highlight the significant potential for increased insurance losses associated with North Atlantic hurricanes due to global warming. The study isolates the impact of climate change from other factors like exposure growth, economic inflation, and social inflation. The findings demonstrate that precipitation-induced losses are particularly vulnerable, experiencing the largest relative increase. The increase in average annual loss due to climate change alone is estimated at around 0.5% per annum, relatively small compared to other loss drivers. However, this projection needs to be considered in context with other risks such as exposure growth, economic inflation and social inflation which combined exceeded 10% in 2022. The uncertainties associated with the study include those inherent in hurricane climatology projections, the exclusion of some potential impacts of global warming (changes in hurricane size, forward speed, geographic variability in storm tracks), and uncertainties in the effect of future changes in hurricane precipitation on insurance claims. The projections of freshwater losses are considered preliminary and require further research, particularly regarding the interplay of rainfall, runoff, and hydrological processes. The study also did not account for the separate impact of sea level rise, which is considered to increase flood hazards more than changes in hurricane-induced precipitation and storm surge.

Global warming is likely to increase property losses from North Atlantic hurricanes in the United States. The findings provide valuable guidance for climate adaptation and mitigation policies, particularly in areas prone to wind or flood damage. While insurers should assess the impacts of more intense and frequent hurricanes on their portfolios, other societal issues, like limiting exposure concentrations along the East and Gulf coasts, deserve equal attention.

Several limitations affect the study's interpretation and generalizability. First, the uncertainty in hurricane climatology projections leads to a wide range of potential loss outcomes. Second, the analysis does not include all potential impacts of global warming on hurricane properties, such as changes in hurricane size, forward speed, and geographic variability in storm tracks. Third, uncertainties exist regarding how future changes in hurricane precipitation will affect insurance claims due to complexities in hydrological and hydraulic processes. Fourth, the study did not incorporate the independent impacts of sea-level rise, which is acknowledged as a significant factor impacting flood risk. Further research is necessary to refine these projections and address these limitations.