Mixed approach to detect share price volatility: an empirical analysis of DSE and CSE

Abstract

This study aimed to determine which models would best predict volatility in Bangladesh's Dhaka Stock Exchange Ltd. (DSE) and Chittagong Stock Exchange Ltd. (CSE). Previously, scholars conducted substantial research on Bangladesh's stock markets. Nevertheless, following a thorough examination, the researcher has identified several areas to contribute to the existing body of literature. Typically, most research conducted in or outside of Bangladesh used the most fundamental combinations of order in the adopted models, specifically (1,1). Furthermore, the majority of previous studies did not take into account a wide range of error distributions. Despite conducting a thorough search, the researcher was unable to locate any studies that combined the two major stock exchanges in Bangladesh—the Dhaka Stock Exchange Ltd. (DSE) and the Chittagong Stock Exchange Ltd. (CSE)—into a single research endeavor to conduct a comprehensive study on volatility forecasting. Meanwhile, experts in stock market volatility persistently strive to improve their forecasts' accuracy by incorporating microeconomic, macroeconomic, and other external factors into their models. However, even after a careful search, no prior study that included so many of those exogenous variables in one study could be located. Furthermore, no previous studies in the research field of volatility estimation were found to have used a "Mixed method" approach that combined the stakeholders' perspectives with quantitative outputs. To address the existing gaps in the research, the researcher has constructed a set of research questions and established specific research objectives, as outlined in Chapter One. The researcher has reviewed a wide range of domestic and international literature, which chapter two includes. The second section of this chapter also consists of an initial examination of the chosen models. The researcher utilized four volatility models, specifically GARCH(p,q), EGARCH(p,q), TGARCH(p,q), and PARCH(p,q) models, to assess and compare their effectiveness in forecasting volatility. The comparison was conducted for DSE and CSE, using three error distributions: Normal, Student's t, and Generalized Error Distributions (GED). Subsequently, the researcher added sixteen microeconomic and macroeconomic variables and two intra-market variables as exogenous variables in the best-suited volatility forecasting models. The purpose was to determine if including these variables improves the accuracy of the predictions made by the selected models. In addition, the researcher incorporated the perspectives of other stakeholders, including officials from the DSE & CSE, officials from the Bangladesh