Proposed Models for Comprehensive Automobile Insurance Ratemaking in Egypt with Parametric and Semi-Parametric Regression: A case study

The research proposes alternative tariff systems to estimate the pure premium for The Misr Insurance Company –the biggest insurance company in Egypt, which has 46.39% of the automobile insurance market share– for comprehensive automobile portfolio. The proposed tariff systems construct insurance rate for each risk class according to the risk factors that might affect the loss instead of a fixed rate that is applied by the company. Three different statistical models are used: Generalized Linear Model (GLM), Generalized Linear Mixed Model (GLMM) and Generalized Additive Model (GAM) using Gamma and Poisson distributions. The data consists of 576,381 cases during the years 2013, 2014, 2015 and 2016. Every case represents an insurance contract.The research found that GLMM is the most convenient model for ratemaking for Misr company because it has the lowest value of Akaike’s Information Criterion (AIC) and takes into consideration the nature of the most insurance data that contains repeated measures. Keywords