The Effect on Stock Market Volatility on Stock Prices By Using The Methods of Arch, Garch, And Egarch in Colombo Stock Exchange, Sri Lanka

Abstract

Introduction: The purpose of the study is to examine the effect on stock market volatility on stock prices by using the methods of ARCH, GARCH, and EGARCH in Colombo Stock Exchange, Sri Lanka.

Methodology: The researcher has chosen the S&P SL 20 Index in CSE, Sri Lanka for the study. In this regard, the secondary data was collected from the official website of CSE. Daily historical stock prices were the independent variables and forecasted stock volatility was the dependent variable. The model has been done using EViews 12 SV. 10 years span of data has been analyzed to find the most accurate forecasting model and the deviation between actual volatility and forecasted volatility. To check the accuracy between actual and the forecasted volatility, the error measurements called MAE, RMSE, MAPE, and TU were used.

Findings: According to the study's findings, all three models confirmed that there is a significant relationship between actual and forecasted volatility, evident through the model's ability to capture key market patterns, including volatility clustering and persistence. While each model offers unique strengths, the ARCH model emerges as the most balanced option for general use. EGARCH is particularly useful in markets with asymmetrical responses, and GARCH provides reliable short-term forecasting but is less consistent overall.

Conclusion: ARCH consistently performs well, making it the most balanced and reliable model overall. EGARCH effectively captures percentage errors and handles market asymmetries, proving useful for conditions requiring specific percentage accuracy. Finally, GARCH performs best in minimizing large deviations but falls short on other metrics, placing it behind ARCH and EGARCH in terms of consistent forecast reliability. The final result ranking emphasizes ARCH as the top choice for balanced accuracy, followed by EGARCH for specialized contexts, with GARCH providing strong but slightly less consistent performance.