PREDIKSI VALUE-AT-RISK MODEL RISIKO RERATA DENGAN TEORI KREDIBILITAS

• Main Author: Puspitasari, Rizka
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/83283
• Institution: Institut Teknologi Bandung

Risk is a potential loss that requires quantitative analysis. This analysis can be conducted through a risk model, which is a statistical model designed to predict the magnitude of risk and provide an overview of potential losses. In this thesis, a mean risk model constructed from the sample mean of risks is used. Two types of models will be developed: Type I with independently and identically distributed random samples, and Type II with a stochastic process whose parameters depend on time. The mean risk model will be used to determine the measure of risk. One commonly used risk measure is Value-at-Risk (VaR). VaR is defined as the maximum potential risk value that may occur within a certain time period at a certain confidence level. However, VaR prediction using the mean risk model is less stable if the data has high variation. Therefore, the classical credibility theory approach is used to improve the stability of risk predictions. Credible Value-at-Risk (CreVaR) is a VaR prediction in the mean risk model that involves credibility theory to achieve more accurate results. The objective of this research is to construct Type I and II mean risk models, determine VaR and CreVaR predictions for both types of mean risk models, and test the accuracy of these predictions. Based on the simulation results, it is found that the constructed mean risk models can be used to predict VaR more accurately. Additionally, the use of credibility theory in making predictions also enhances the accuracy of the results, leading to more stable predictions.