Pricing life insurance using bivariate temperature-mortality seasonal hidden Markov models

Kouton, Samuel Darwin Dona;Barigou, Karim;Nguyen, Thai
(2026) , 26 pages

Files

ISBA_DP_2026-19.pdf
  • Open Access
  • Adobe PDF
  • 1.2 MB
Download

Details

Authors
  • Kouton, Samuel Darwin Dona
    Author
  • Barigou, KarimUCLouvain
    Author
  • Nguyen, Thai
    Author
Abstract
We study the dependence between weekly mortality and temperature through a non-homogeneous Seasonal Hidden Markov Model (SHMM) with trend. The framework jointly models mortality and temperature while allowing for regime switches, explicit seasonality, smooth longterm trends, and time-varying transition probabilities in latent dynamics. We show that the general identifiability and strong consistency results of Touron (2018) apply to our Poissonand Gaussian-emission specifications. We estimate the model on weekly Italian mortality and temperature data using two- and three-state specifications. Empirically, the SHMM improves fit relative to non-switching benchmarks and yields an interpretable state-dependent characterization of the temperature–mortality relationship. We then develop a change-of-measure framework for pricing mortality-linked insurance contracts that incorporates temperature-related mortality risk and latent regime uncertainty.
Affiliations

Citations

  • APA
  • Chicago
  • FWB

Kouton, S. D. D., Barigou, K., & Nguyen, T. (2026). Pricing life insurance using bivariate temperature-mortality seasonal hidden Markov models (LIDAM Discussion Paper ISBA 2026/19). https://hdl.handle.net/2078.5/276676