ABSTRACT
This conceptual paper addresses Value-at-Risk (VaR),
an alternative approach to measuring market risk. Though VaR can be seen as an
‘ideal’ risk estimating technique, its application and reliability lies heavily
on specific distribution assumptions and parameter settings. Highlighted issues
within its framework include the effect of heavy-tails, volatility clustering,
multivariate condition, roles of confidence level, holding period and
performance evaluation. Thus, more in depth assessment which integrates VaR
methods with volatility models from the perspective when the return’s
distribution is either normal or non-normal are needed to provide a better
maximum loss forecast for an investment. In addition, the practice of VaR must
also be complemented with stress testing and backtesting which helps to validate
the selected models.
Keywords: Value-at-Risk,
volatility modelling, holding period, confidence level,
stress test, backtesting.
No comments:
Post a Comment