Utilizing quantile regression for vehicle insurance premium calculation through the quantile premium principle

This research explores the use of Quantile Regression (QR) for determining vehicle insurance premiums based on the Quantile Premium Principle (QPP). Accurate and equitable premium calculation is essential in the insurance sector for effectively managing financial risks. This study evaluates the efficacy of QPP compared to the Expected Value Premium Principle (EVPP) in the realm of premium determination. The analysis incorporates the Zero-Inflated Poisson (ZIP) model for claim frequency, the Inverse Gaussian Generalized Linear Model (GLM) for claim severity, and QR for assessing claim severity quantiles. Key predictor variables such as DriverAge and CarAge are categorized into distinct groups, facilitating the computation of varied insurance premiums across different risk classes derived from these variables. Findings indicate that QR effectively estimates claim severity quantiles, which are then utilized in QPP-based premium calculations. Moreover, the outcomes of premium calculations using EVPP and QPP do not consistently align, due to the differing risk measures applied by these principles. The study concludes that QPP provides a more nuanced identification and accommodation of diverse risks than EVPP, thereby offering a more thorough risk assessment in underwriting. Additionally, the research highlights the significant impact of driver age and car age on the variability of insurance premiums, underscoring the influence of these factors on premium calculations.