CREDIBLE VALUE-AT-RISK RISK MEASURE

• Main Author: PUTRI (NIM:20816008), INSANI
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/28068
• Institution: Institut Teknologi Bandung

Insurance is protection to minimize the risk. To make a solvent insurance company, they should manage the risk of claim filed by policyholders. Risk measures are tool to predict risk and help company to determine capital allocation. Risk measures can be represented in distortion function. One of popular distortion risk measures is Value-at-risk (VaR). VaR can be defined as a maximum risk that can be tolerated at level of confidence. VaR is function of random variable's parameters that can be assume as constant and random value. In practice, parameters should be estimated and we obtain VaR prediction. VaR prediction can be consider as random varible because parameters estimation is function of random variables. In parameters estimation, there is parameter variabilities such as bias and mse. Therefore, VaR prediction should be tested for it's accuracy. In addition, it can be shown that VaR prediction is robust. The credibility model is a model that minimizes the parameter variabilities to obtain an accurate prediction. The concept of a credibility model can be adopted to obtain VaR prediction that is credible named as Credible-VaR (Cre-VaR). As a result, compared to VaR prediction values, Cre-VaR values are close to true value of VaR.