Copula-Based Regression Analysis To Estimate the Total Losses on Health Insurance

Abstract

The insurance company is a company that received delegation of the risks it has insured, so that this company needs to pay attention to losses incurred as a result of a claim. Estimating losses of claim is an important task for insurance companies to predict their obligations. Total losses in the company's portfolio is defined as the amount of loss policy. Losses in the health insurance policy can be calculated based on two variables: the frequency and severity of claims. In the literature of Statistics, joint distribution is a method of statistical analysis that can combine two different data distribution, it is Copula. This thesis aims to provide a study of Copula for the estimation of loss claims in health insurance, case study is taken from an insurance company XYZ. Further, the authors conducted a regression between the Generalized Linear Model (GLM) of claim frequency and claim severity using Copula-based Regression Model is estimated by Maximum Likelihood Estimation (MLE). In the end of analysis, Copula-based Regression Model can be used to estimate the projected claim load in the budgeting of claim load the insurance company. This will help provide better estimation results compared to the methods currently used.

Keywords: Health Insurance; Copula; GLM; MLE; Regression.