Application of Regression Models to Insurance Premium Pricing in the Nigerian Non-Life Insurance Market

Premium adequacy in the Nigerian non-life insurance market, regulated by the National Insurance Commission, is a persistent challenge, with underpricing contributing to systemic solvency risks and overpricing driving low insurance penetration rates below one percent of GDP. This study applies generalised linear model regression techniques, specifically the Tweedie compound Poisson-Gamma model standard in actuarial pricing practice, to develop risk-differentiated premium models for motor and fire insurance classes using claims data from three insurance companies operating in Lagos and Abuja. Policy-level data including vehicle age, engine capacity, driver age, location, sum insured, policy tenor, and claims history are extracted from insurer data systems for 24,000 policies over three underwriting years. Rating factor significance is assessed through chi-square and likelihood ratio tests, with the final GLM retaining eight statistically significant rating factors for motor and six for fire. Predicted pure premiums from the GLM are compared against current tariff rates, revealing overpricing of up to 40 percent for low-risk vehicle categories and underpricing of 22 percent for commercial vehicles in Lagos. The study recommends transitioning from flat tariff structures to individual risk-based pricing supported by the GLM framework. Keywords: generalised linear model, insurance premium pricing, non-life insurance, actuarial, Nigerian insurance market.