DETERMINATION OF PURE PREMIUM FOR TOTAL LOSS ONLY AND COMPREHENSIVE COVERAGE IN MOTOR VEHICLE INSURANCE USING GENERALIZED LINEAR MODEL (GLM)

• Main Author: YANUARSA RAHARDIAN (NIM : 10109030); pembimbing : Dumaria R. Tampubolon, M.Sc, Ph.D, FUAD
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/15740
• Institution: Institut Teknologi Bandung

A car insurance business is one class of businesses in general insurance. Based on the type of coverage, a car insurance coverage may be classified into Total Loss Only (TLO) and Comprehensive (which covers total loss and partial loss). Partial loss is when a loss is, approximately, less than 70% of the coverage; and total loss is when the loss is greater than or equal to 70% of the coverage or when the car is stolen. An insurance company needs to be able to correctly predict the amount of premium in order to have enough funds to pay future claims. This final project used Generalized Linear Model (GLM) to model the amount of total loss and to model the probability of a total loss to occur. The data used as a case study is car insurance data for underwriting year 2010 from an insurance company. At 5% significance level, it is found that the amount of claims follows an Inverse Gaussian distribution with parameters ^(theta) = 125470981 and ^(alpha) = 34:1278, whereas the number of claims follows a Bernoulli distribution with parameter ^(lambda) = 1:68 (pi) 103. To predict the amount of claims for total loss coverage, it is found that, at 5% significance level, the predictors are: harga pertanggungan (coverage); tahun kendaraan (the year when the car was produced) 2000, 2003, 2006, 2008, 2009, 2011; and biaya klaim (the cost to process the claim). To predict the probability of a total loss to occur, it is found that, at 5% significance level, the predictors are: several levels of harga pertanggungan and diskon premi (premium discount). Using the models, the pure premium for total loss coverage is determined. To determine the pure premium for comprehensive coverage, the results from a research carried out by Mochammad Ramadhyan T (2013), which analyzed partial loss coverage, are used.