AGGREGATION OF GARCH PROCESS: FORECASTING V@R AND IMPROVED V@R

• Main Author: (10113009), ANISA
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/21116
• Institution: Institut Teknologi Bandung

In investment, investors want to minimize the risks that can be done with aggregation. Value at Risk (V@R) is one of the most widely used risk measure. V@R aggregation is one of application that defined as the worst lost to be expected of aggregation at given confidence level. This final project, presents V@R measures based on appropriately specified GARCH(p,q) process that has important properties <br /> <br /> <br /> of model, such as fat-tailed distribution. This fat-tailed distribution will answer to minimized risks for aggregation. To forecast the V@R of aggregation, parameter <br /> <br /> <br /> estimation GARCH(p,q) is required. Then, use Monte Carlo Sampling Errors (MCSE) approach to find errors of parameter estimation. To check the accuracy of the prediction, use Correct V@R. From the results, it can be concluded that, general GARCH(1,1) approach performs better than the other. However the accuracy of risk is key to successful risk measure. So, Improved V@R is needed. This final project <br /> <br /> <br /> presents Improved V@R to forecasting risk for GARCH(p,q) process. Improved V@R is done by using coverage probability. Improved V@R can be expressed as the sum of V@R and moments. It is proved that improved V@R is more accurate than V@R because improved V@R gives a smaller value than V@R.