Zastosowanie zmienności zrealizowanej i modeli typu ARCH w wyznaczaniu wartości zagrożonej
Modeling value-at-risk when realized volatility and ARCH-type models are used

Summary/Abstract: When forecasting value-at-risk the important issue is the method of estimating unobservable volatility. In the article we compare the performance of two types of VaR forecasts: parametric, where volatility is based on two GARCH-type models (GARCH and APARCH) and nonparametric, where volatility is calculated as realized volatility with and without night return. Our sample consists of returns of the most liquid stocks and WIG20 index in the period 2007.09 - 2011.04. Surprisingly, for the financial series considered in the paper the value-at-risk based on GARCH-type models performed better than value-at-risk based on ARFIMA models for realized volatility. As the use of nonparametric measures of volatility is more time-consuming and data requirements are serious we recommend the parametric methods for value-at-risk estimation.

• Details
• Contents