This paper addresses the question of whether the VaR estimation technique, originally prescribed by Riskmetrics and recommended after adaptation by the NSE, adequate enough to estimate VaR in the changing Indian market scenario, especially during the past one year when the markets exhibited considerable activity. This study describes the empirical investigation carried out centered on computing and backtesting VaR estimates of several indices designed by the NSE and Sensex values using recommended Exponentially Weighted Moving Average (EWMA) method and the dynamic volatility model based on GARCH. The results indicate that the GARCH-based VaR estimation method outperforms the officially recommended EWMA method; hence, the study advocates re-examination of the recommended method by appropriate bodies.

Short-term market players need to analyze the risk involved in trading by using appropriate methods. Value-at-Risk (VaR) is one of the most popular risk quantification frameworks generally followed by banks and other financial institutions for maintaining capital adequacy norms imposed by the regulators as recommended by Basel Committee on Banking Supervision (1988, 1996 and 2006). The NSE recommended a technique for VaR estimation, as developed in Varma (1999) based on EWMA technique. This method is easy to implement using software application, either real time or off-line, for modulating fund management strategies of investors. There are many methods available in the literature, such as Historical Simulation (Non-Parametric), Monte Carlo Simulation (Non-Parametric), Filtered Historical Simulation (Non-Parametric), Variance Covariance Method (Parametric), Ordinary Least Square (OLS) Method (Parametric), Kalman Filter Technique (Parametric), for the purpose of estimating VaR (Shah and Moonis, 2003; and Lehikoinen, 2007). The VaR framework also dictates application of backtesting (Campbell, 2005) for checking the reliability, goodness, and consistency of the VaR model adopted for risk measurement for the institutions or market.

In the past two years, capital market in India experienced a state of boom and some indices in different markets in India doubled their values during this short period. Such indices may be considered as `virtual portfolio' for testing VaR estimation framework. Investors often understand market movement through the variations of different indices designed by bourses for different specific sector of the market as well as for general market. Backtesting the applied VaR estimates can show the reliability of specific estimation techniques. Verifying the prediction capability can be an interesting topic for investigation under a particular model.

The IUP Journal of Applied Finance, VaR Computation Method, National Stock Exchange of India, Indian Market Scenario, Empirical Investigation, NSE and Sensex values, Exponentially Weighted Moving Average, EWMA, Dynamic Volatility Model, Short-term Market Players, Monte Carlo Simulation, Variance Covariance Method, Fund Management Strategies.