1. UKM Learning and Research Repository
2. PHD Thesis / Tesis PHD
3. Faculty of Science and Technology / Fakulti Sains dan Teknologi

Title:	Empirical Analysis Of VAR And CVAR By The Utilization Of GARCH Models And Extreme Value Theory: Evidence From Malaysian Stock Market
Authors:	Mohamed Amraja Mohamed (P47395)
Supervisor:	Ahmad Mahir Razali, Associate Professor Dr.
Keywords:	Empirical Analysis Of VAR And CVAR Empirical Analysis Of VAR And CVAR By The Utilization Of GARCH Models Utilization Of GARCH Models And Extreme Value Theory Evidence From Malaysian Stock Market Financial risk management--Simulation methods
Issue Date:	31-May-2013
Description:	This research empirically investigates the Value at Risk (VaR) and Conditional Value at Risk (CVaR) for long and short trading positions by employing the GARCH family models and extreme value theory (EVT) for the stock return series traded in Malaysian stock market. To avoid the issue of aggregation bias, sectoral indices have been used rather than aggregate indices in the analysis. The normal as well as heavy-tailed distributions are commonly assumed for GARCH estimations in the context of VaR and CVaR. Although numerous earlier studies have assumed Gaussian distribution in GARCH estimations to calculate VaR and CVaR, relatively recent papers do not seem to support this assumption due to an existence fat-tailedness in the stochastic errors. Hence, the existing literature proposes a set of conditional densities to address the phenomenon of heavy-tailedness in the disturbances. Overall, this study has taken into account the non-normality facts in innovations by assuming symmetric (GED, student t) as well as asymmetric (skewed Student t) heavy-tailed distributions. Apart from that in order to account for possible asymmetries in the behavior of stock returns we have also applied the univariate asymmetric EGARCH model to estimate VaR and CVaR. Indeed, reliance on the error diagnostics and several selection criteria has led us to select the most favored GARCH-type model for each sector under study. In addition, extreme value theory is being widely used in financial risk management as an alternative technique to GARCH-based VaR and CVaR. There are two general ways of identifying extremes in the financial data. The first approach considers the maximum loss of successive periods such as weeks, months or years. Therefore, the selected loss maxima constitute the extreme events and the method is called the block maxima method (BMM). The second approach focuses on the losses (returns) that exceed a given threshold. In this approach, all losses that exceed the threshold constitute the extreme events. This method is the peak over the threshold (POT) method. Different sectors provide different empirical estimate for the 95th percentile and therefore, providing different threshold. One may note that the threshold should be sufficiently large to satisfy the condition tends towards infinity, so that the EVT is applicable. Generally, this study compares aforementioned models and selects the best one for the underlying financial data. This study contributes to the existing literature through three main aspects. Firstly, we provide a fairly simple approach to estimate the day-ahead VaR for both long and short position traders when the underlying process is described by the econometric model (GARCH models) and the EVT approach. Secondly, we extend our methodology to the CVaR, and show that our procedure can also be applied in a relatively straightforward manner. Lastly, we provide a greater understanding of the estimation procedures of VaR and CVaR for both long and short position traders and provide the interpretations of VaR and CVaR risk measures with respect to the Malaysian share prices. The results show that, First, VaR and CVaR may underestimate the risk of securities with fat-tailed properties and high potential for large losses. Second, VaR and CVaR may both disregard the tail dependence of asset returns. Third, CVaR has less of a problem in disregarding the fat tails and the tail dependence than VaR does.,PhD
Pages:	283
Call Number:	QA276 .M836 2013
Publisher:	UKM, Bangi
Appears in Collections:	Faculty of Science and Technology / Fakulti Sains dan Teknologi

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.