Modeling Pure Premiums with Generalized Linear Models
Document Type
Thesis
Degree Name
Master of Science (MS)
Department
Mathematics
Date of Award
9-1-2026
Abstract
Modeling pure premium is central to non-life insurance pricing, yet it becomes challenging when claim experience varies widely across policyholder groups and exposure levels. This thesis applies a regression-based framework using generalized linear models (GLMs) to estimate pure premiums and generate interpretable rate relativities for insurance pricing. We analyze two Swedish Wasa insurance portfolios: the busscase dataset for bus insurance and the mccase dataset for motorcycle partial casco insurance. The study uses the frequency--severity approach, where claim frequency and claim severity are modeled using Poisson and Gamma GLMs respectively. Estimated frequency and severity components are then combined to obtain pure premium estimates and multiplicative rate relativities across risk classes. In addition, the Tweedie GLM is applied as a one-part model for estimating pure premiums directly from aggregate claim costs, making it a useful alternative to the frequency--severity framework. The results show that GLMs provide an effective and interpretable approach for modeling claim experience and estimating pure premiums. Overall, the study demonstrates that GLMs are valuable tools for actuarial pricing, risk classification, and the estimation of pure premiums in non-life insurance.
Advisor
Nahid Hasan
Subject Categories
Mathematics | Physical Sciences and Mathematics
Recommended Citation
Osei-Poku, Kate, "Modeling Pure Premiums with Generalized Linear Models" (2026). Electronic Theses & Dissertations. 1394.
https://lair.etamu.edu/etd/1394
