Estimating Dhaka Stock Market Volatility: A Comparison between Standard and Asymmetric GARCH Models

Abstract

This paper compares and estimates standard and asymmetric GARCH models with daily returns data of the DSI index (All Share Price Index) of the Dhaka Stock Exchange from 28 March 2005 to 30 November 201 0. The maximum likelihood estimation (MLE) technique is used to estimate the parameters of the chosen models. Results indicate that GARCH has lower log-likelihood than the asymmetric GJR-GARCH model, which implies that GJRGARCH model is a better performing model to estimate and to forecast volatility. Results from other hypothesis tests indicate that volatility process has unit root i.e. volatility process is nonstationary, expected returns do not always depend on volatility, and the conditional variance (volatility) of future asset price is a symmetric function of changes in price at Dhaka stock market.