Abstract:
This paper investigates the nature of volatility characteristics of stock returns in the Bangladesh stock markets employing daily all share price index return data of Dhaka Stock Exchange (DSE) and Chittagong Stock Exchange (CSE) from 02 January 1993 to 27 January 2013 and 01 January 2004 to 20 August 2015 respectively. Furthermore, the study explores the adequate volatility model for the stock markets in Bangladesh. Results of the estimated MA(1)-GARCH(1,1) model for DSE and GARCH(1,1) model for CSE reveal that the stock markets of Bangladesh capture volatility clustering, while volatility is moderately persistent in DSE and highly persistent in CSE. Estimated MA(1)-EGARCH(1,1) model shows that effect of bad news on stock market volatility is greater than effect induced by good news in DSE, while EGARCH(1,1) model displays that volatility spill over mechanism is not asymmetric in CSE. Therefore, it is concluded that return series of DSE show evidence of three common events, namely volatility clustering, leptokurtosis and the leverage effect, while return series of CSE contains leptokurtosis, volatility clustering and long memory. Finally, this study explores that MA(1)-GARCH(1,1) is the best model for modeling volatility of Dhaka stock market returns, while GARCH models are inadequate for volatility modeling of CSE returns.
Machine summary:
"Abdul Wadud2 Abstract This paper investigates the nature of volatility characteristics of stockreturns in the Bangladesh stock markets employing daily all share price index return data of Dhaka Stock Exchange (DSE) and Chittagong Stock Exchange (CSE) from 02 January 1993 to 27 January 2013 and 01 January 2004 to 20 August 2015 respectively.
Table 5 reports the results of the mean and variance equations of the estimated models for all share price indices returns of DSE and CSE.
The significant GARCH term β proves that MA(1)-GARCH(1,1) and GARCH(1,1) are the appropriate model to account volatility on the DSE and CSE respectively, and that volatility in the present period influences volatility in the next period, while the highly significant ARCH term α indicates a positive relationship between shocks and volatility in the Bangladesh stock market.
In terms of diagnostic fit presented in Table 6, the estimated model for DSE satisfies all conditions of the GARCH theory based on Ljung -Box Q statistics and ARCH-LM tests, while EGARCH(1,1) model for CSE fails to satisfy all conditions of the GARCH theory.
Considering the existence of the asymmetric effects of shocks on the return volatility in the Bangladesh stock markets, we also fit the data with the MA(1)-EGARCH(1,1) and EGARCH(1,1) models.
In contrast, the GARCH models are inadequate in modeling the volatility of Chittagong stock market return as Ljung-Box Q and Q2 tests suggest that the estimated models are not free from serial autocorrelation up to 36 lags."