Actuarial study and statistical analysis of extreme wildfire insurance claims
Loading...
Date
2024-08-08
Authors
Wang, Jiali
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The wildfire season of 2023 was one of the most devastating in Canadian history and caused significant losses to the insurance industry. This study focuses on the large wildfire losses exceeding CAD 25 million in the last 16 years (2008-2023).
For severity analysis, mainly due to the largest amount of wildfire losses reaching CAD 4 billion, heavy-tail distributions have been applied to find the best fit for the loss claims history. We also estimate loss amounts in different return periods.
Traditional actuarial ratemaking methods in non-life insurance are based on the assumption of the absence of correlation between the frequency and severity of claims. In this thesis, for frequency analysis, logistic regression is used to investigate the relationship between the maximum temperature and the probability of extreme wildfire losses. Consistent with common sense, the probability of a large wildfire occurring is found to increase with increasing temperature. According to the requirements of Solvency II, we derive the 99.5% confidence interval for the claim frequency.
Overall, there are not many published papers and available data on Canadian wildfire insurance losses. Therefore, this thesis lists some possible research directions in property and casualty insurance, reinsurance, and life insurance in the future research directions section.
Description
Keywords
Wildfires, Canada, Insurance, Heavy-tail distributions, Logistic Regression