Empirical Research on Value-at-Risk Computing Approaches in Evaluation of Financial Risks

LI, KAI (2012) Empirical Research on Value-at-Risk Computing Approaches in Evaluation of Financial Risks. [Dissertation (University of Nottingham only)] (Unpublished)

[img] PDF (dissertation) - Registered users only - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
Download (749kB)


The aim of this dissertation is is to investigate how VAR computing approaches are implemented in evaluation of financial risk. Furthermore, objectives related to various aspects are investigated specifically. The first objective is to investigate how the three traditional VAR computing approaches and implement comparative analysis. The second objective is to investigate how failure frequency test is implemented in evaluating effectiveness of VAR results. The third objective is to investigate how VAR is computed in portfolio which contains either same financial instruments or different types of financial instruments. The fourth objective is to investigate how GARCH model is implemented in computing VAR of time series of financial returns.

By means of collecting historical data of relative stocks and future and implement statistical instruments in use, two sections of findings are demonstrated. In the first section, VAR of stocks of each corporation is computed by Historical Simulation approach, Variance-Covariance approach and Monte Carlo Simulation approach individually at first. At the mean time, failure frequency test suggested by Kupiec is implemented in order to test the reliability of VAR result worked out by each approach. Thus, comparative analysis of the three classical VAR computing approaches is represented. Thirdly, the empirical study of how to compute VAR of portfolio involves only stocks is implemented. In the second section, VAR of FTSE 100 is firstly computed. Afterwards, in order to test characteristics of GARCH model, various tests are implemented including test of normality, Jarque-Bera test, unit root test, autocorrelation and partial autocorrelation test and ARCH effects. Thirdly, VAR of FTSE 100future is computed by GARCH (1,1) – N model, and VAR of the portfolio involves stocks and futures is worked out.

Key words: Value-at-Risk, comparative analysis, GARCH, portfolio

Item Type: Dissertation (University of Nottingham only)
Depositing User: EP, Services
Date Deposited: 08 Apr 2013 11:19
Last Modified: 19 Oct 2017 13:17
URI: https://eprints.nottingham.ac.uk/id/eprint/26006

Actions (Archive Staff Only)

Edit View Edit View