Lukas J. Hahn

Overview

Lukas J. Hahn is an actuarial scientist and co-developer (with Marcus C. Christiansen) of the Bayesian Multi-Population Mortality Projection (BMPMP) model. His research focuses on stochastic mortality modeling for multiple interrelated populations, combining Cairns-Blake-Dowd (CBD) logit mortality models with vector error correction models (VECMs) estimated via Bayesian MCMC, without requiring convergence of mortality rates across groups.

Key Contributions / Features

• Hahn and Christiansen (2016): Co-developer of the BMPMP framework — a two-stage model that pairs CBD logit mortality rates (ages 40–100) with a VECM for the latent CBD parameters, estimated via Gibbs sampler / Metropolis-Hastings MCMC. Demonstrated empirically that European national mortality rates do not converge, and that the Li-Lee (2005) common-factor model fails for heterogeneous country groupings. Showed cointegration rank $r$ is the critical model parameter (higher $r$ produces left-skewed posterior predictive distributions); lag order $k$ is irrelevant.

Related

• Marcus C. Christiansen
• Multi-Population Mortality Modeling
• Cairns-Blake-Dowd Model
• Vector Error Correction Model

Sources

• Hahn and Christiansen 2016 — Mortality Projections for Non-converging Groups of Populations