MEASURING SPATIAL RISK USING CONDITIONAL AUTOREGRESSIVE (CAR) MODEL

• Main Author: Ibrahim Nyamweya, Didymus
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/54885
• Institution: Institut Teknologi Bandung

Premium charges happen to be at the centre of every insurance business. The amount of premium charged will determine if the company will be in or out of business. Claim frequency and claim severity are the two major components in determining the premium rate. Generalized Linear Models (GLM) have been used in traditional insurance to estimate the claim frequency. GLM have shortcomings that make the insurance estimates inaccurate. By applying the Generalized Linear Mixed Model (GLMM) to model claim frequency, we have shown that it possible to model frequency using GLMM. The GLMM model has a spatial random effect which is modelled using the Conditional Autoregressive (CAR) model which is multivariate normal therefore gives room for having both negative and positive values for the spatial effects. The spatial effects for claim frequency were determined by using CAR model. The unknown parameters were estimated using R packages. Based on the estimates, the claim frequency and severity are simulated and applied to calculate pure premium for COVID-19. The simulation for claim frequency with spatial random effects had better pure premium estimate compared to the claim frequency without spatial effects.