Is timber insurable? A study of wildfire risks in the U.S. forest sector using spatio-temporal models.

摘要

In the U.S. forest products industry, wildfire is one of the leading causes of damage and economic losses. While individual wildfire behavior is well studied, new literature is emerging on broad-scale (e.g., county-level) wildfire risks. Our paper studies wildfire risks using crucial informational variables across both spatial units and time periods. Several statistical models are used to quantify the risks. We develop several maximum likelihood estimation methods to account for spatio-temporal auto-correlation in conditional risks. A group index insurance scheme is proposed, and its associated actuarially fair premium rates are estimated and presented. Implications for wildfire management policies are also discussed.