Empirical Evaluation of Methods Used to Improve Historical Simulation Value-at-Risk Using Sudanese Stock Market Data

No Thumbnail Available
Date
2015-07-05
Authors
Dawod, Abdaljbbar Babiker Abdaljbbar
Journal Title
Journal ISSN
Volume Title
Publisher
uofk
Abstract
The research has three major purposes: rstly, reviewing historical simulation VaR. Secondly, reviewing the exponential-weighing model, ltered historical simulation model, and volatility-adjusting model with GARCH, EWMA, EGARCH and GJR-GARCH models in order to improve the accuracy of historical simulation VaR. Thirdly, application of these weighing strategies for the Khartoum Stock Exchange using the Sudatel Telecom Group as case study, where R package has been used to estimate the models. To determine the tness of these models, Unconditional Coverage test (LRUC), Independence test (LRI ), and Correct Conditional Coverage test (LRCC) as backtesting methods have been adopted. The results showed variations in accuracy for the di erent con dence levels: under the current circumstances of the local market, there is no signi cant di erence in the accuracy of historical simulation VaR, if the equal-weighing model, exponential-weighing model, EGARCH model or, GJR-GARCH model are used at 95% or 90% con dence levels, whereas they are inaccurate at 99% and 99.9% con dence levels. It also showed that the ltered historical simulation is an accurate model to estimate historical simulation VaR in Sudanese markets at 95% and 90% con dence levels. Finally, GARCH model and EWMA model with = 0.94, are accurate models for the local markets at 90% con dence level, but their performance is poor at 99.9%, 99% and 95% con dence levels. Depending on the results, the research recommends to estimate historical simulation VaR in the local nancial market at high con dence levels ( i.e. 95% and 90%)
Description
Keywords
Empirical Evaluation of Methods Used to Improve Historical Simulation Value-at-Risk Using Sudanese Stock Market Data
Citation
Collections