Department of Accountancy, Faculty of Business Studies & Finance,
Wayamba University of Sri Lanka, Kuliyapitiya, Sri Lanka
The stock markets of a country play a vital role in its economy. Stock market indices are vital fragments of information for investors. It is very important to develop models that reflect the pattern of the stock price movements for different sectors since it becomes very significant to investors and policymakers. Therefore, the aim of this research study was to develop models to forecast different sector indices in Colombo Stock Exchange and to compare sector-wise models. The investigation was performed using secondary data for a sample of ten listed sectors in the Colombo Stock Exchange (CSE) for the thirty-four years from 2nd January 1985 to 31st March 2019. Secondary data were collected by using the data library maintained by Colombo Stock Exchange. Financial time series data analysis techniques were used to analyze the collected data. It was applied the ARCH family models in this research study, which included the Autoregressive conditional heteroscedasticity model (ARCH), Generalized Autoregressive conditional heteroscedasticity model (GARCH), Threshold Autoregressive conditional heteroscedasticity model (TARCH), Exponential generalized autoregressive conditional heteroscedastic model (EGARCH), Integrated Generalized Autoregressive conditional heteroscedasticity model (IGARCH) and Power Autoregressive conditional heteroscedasticity model (PARCH) since the sector indices are financial time series data. Findings revealed that the appropriate model to forecast the sector indices of Oil Palms sector, Services sector and Stores & Supplies sector as PARCH (2,1) model, Beverage, Food & Tobacco sector as PARCH (1,1) model, Chemicals & Pharmaceuticals sector as PARCH (2,2) model, Banking Finance & Insurance sector and Investment Trusts sector as IGARCH (2,2) model, Footwear & Textiles sector as EGARCH (1,1) model, the Manufacturing sector as EGARCH (1,3) model and Hotels & Travels sector as TARCH (1,1) model. The findings of this research study are useful to policymakers and investors for their decision-making.
Keywords: Colombo Stock Exchange, Stock price indices, Time series analysis