DETERMINATION OF INSURANCE PREMIUM USING QUANTILE PREMIUM PRINCIPLE

• Main Author: Letisia Moniaga, Yessica
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/74564
• Institution: Institut Teknologi Bandung

Insurance premium is the total of pure premium and safety loading. Insurance promises benefits in the future if an incident occurs. Therefore, there are many risks faced. Risk assessment is important for assessing the adequacy of premiums, where risks can originate from uncertainty over the number and size of claims against expectations. GLM is often used to model insurance claims, for estimating the expected size of claims and the frequency of a policy. However, GLM cannot provide information about quantile (Value at Risk) to describe how risky the policy is, which is important for assessing the solvency of an insurance company. As a solution, there is a method called Quantile Regression that can estimate distribution quantiles. In this final project, data on motor vehicle insurance policy claims are used, obtained from Package 'insuranceData' in R. The first step is to determine the logistic regression model for claims. Next, to estimate the severity of claims, two methods are used, which are Quantile Regression and gamma regression (generalized linear model). The estimated QPP premium is the sum of the estimated pure premium and safety loading which is proportional to the difference between the quantile of aggregate claims and the pure premium. Estimation of the aggregate claim quantile is obtained by multiplying the expected frequency and severity of claims using the Quantile Regression method. Pure premium is obtained by multiplying the expected frequency and severity of claims using gamma regression. In this Final Project, a model is constructed for policyholder data collected between 2004 and 2005 from Australia on a total of 67856 policies, of which 4624 have at least one claim. The modeling results using the Quantile Premium Principle are compared with the Expected Value Premium Principle, which is the sum of the pure premium and safety loading which is proportional to the pure premium. QPP can represent the level of risk at the premium better than EVPP for suitable probability level. This is considering that safety loading is a form of risk from the uncertainty of the number and size of claims on insurance premiums.