Estimation portfolio var with three different methods: financial institution risk management approach

Abstract

Value at Risk (VaR) technique is very important for the measurement and control of market,
credit and also operational risks. VaR has become an essential tool for financial institution
risk management. VaR forecasts of financial instruments are also of great importance. The fact
that volatility seems to cluster over time in a predictable way has very important implications for
investors, risk managers and financial institutions. In order to generate daily VaR forecasts of
equity portfolios for S&P 500, FTSE ALL SHARE and NIKKEI 500 and of over ten-years
Goverment bond portfolios for US and the UK we use the exponentially weighted moving average
(EWMA). EWMA model emphasizes in most resent observations by assigning heavier
weights to them than to those from the distant past. In the latter EWMA estimates are used as
inputs in three VaR estimation methods in order to produce forecasts based on each one of
them. The three methods are: the Variance Covariance approach used by JP Morgan Risk Metrics,
the Historical Simulation approach and the Hybrid method. Finally we use six backtesting
techniques for the validation of the estimation and the evaluation of the forecasting performance.
The results indicate that Historical Simulation out performs the other two competing approaches
with close second the Risk Metrics approach.