MODELING PURE PREMIUM AUTOMOBILE INSURANCE BASED ON FREQUENCY CLAIMS AND RANDOM EFFECT USING NEGATIVE BINOMIAL DISTRIBUTION

• Main Author: VIOLA BURHAN , ARINI
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/25784
• Institution: Institut Teknologi Bandung

The determination of the appropriate premiums is important in insurance, including automobile insurance. In the pure premium modeling on this thesis used the frequency of claims history and the factors that affect the rate of claims such as the age of the participants, the gender of participants, the year of the vehicle, vehicle type, and others. Those factors are all that is observed, which are known as covariate. In addition to the factors above, this model also included the unobserved factors, which we refer to as a random effect. Claim frequency distribution and <br /> <br /> random effect in this thesis is assumed to be the negative Binomial and Beta. Modeling this provides a form of appreciation for participants who rarely make a claim or even does not claim. This model was applied to the insurance company XYZ and obtain some conclusions. First, for participant’s gender retrieved that women more often make a claim than men. For the age of vehicle retrieved that the age ranged from four to six years more often make a claim than the age of the other vehicles. And for type of vehicle is obtained that the vehicle of type minibus more often make a claim than other types of vehicles. Second, the type of vehicles have larger contributions than the gender of the participant and the age of vehicle in participant's claim rate. Third, the premium on participants who rarely claim is lower compared to the premium on participants who often claim.