Methods to measure value at risk of a portfolio: a case study on Sri Lanka Stock Exchange

Abstract

In this study three different methods are used to measure and compare Value at Risk (VaR) of four different portfolios. Twenty securities which are used to calculate S&P 20 SL index in Colombo Stock Exchange were used to construct four different portfolios each worth Rs.100,000. Data was obtained from Colombo Stock Exchange during the period of 1st of January 2012 to 30th of September 2016. Four portfolios were constructed using the securities with the mean closing prices: less than Rs.100, Rs.100 - Rs.200, Rs.200 - Rs.300 and greater than Rs.300. Equally weighted VaR models were constructed for selected portfolio securities. VaR is calculated using three different methods namely Variance-Covariance method, Historical method and Monte-Carlo Simulation method at three different significant levels. When making decisions about a portfolio it is best to make decisions by considering the maximum loss that can be expected when investing in a particular portfolio. By comparing the values obtained in different portfolios the study concluded that, if the VaR is calculated at 90% or 95% of confidence, the best method is Monte Carlo method while if the VaR is calculated at 99% of confidence, the best method is Historical method. Further it could be seen that the portfolio that constructed using the securities with mean closing prices Rs.100-Rs.200 gives the lowest VaR value and the portfolio that is constructed using the securities with mean closing prices above Rs.300 gives the highest VaR value from all three methods.