Modelling and Forecasting Volatility of Returns on the Ghana Stock Exchange Using Garch Models

Abstract

This paper models and forecasts volatility (conditional variance) on the Ghana Stock Exchange using a random walk (RW), GARCH(1,1), EGARCH(1,1), and TGARCH(1,1) models. The unique 'three days a week' Databank Stock Index (DSI) was used to study the dynamics of the Ghana stock market volatility over a 10-year period. The competing volatility models were estimated and their specification and forecast performance compared with each other, using AIC and LL information criteria and BDS nonlinearity diagnostic checks. The DSI exhibits the stylized characteristics such as volatility clustering, leptokurtosis and asymmetry effects associated with stock market returns on more advanced stock markets. The random walk hypothesis is rejected for the DSI. Overall, the GARCH (1,1) model outperformed the other models under the assumption that the innovations follow a normal distribution.

Modeling the Effects of the Global Financial Crisis on the Malaysian Market

“…Awartani and Corradi [17] find that when allowing for asymmetries, the MODELING THE EFFECTS OF THE GLOBAL FINANCIAL CRISIS ON THE MALAYSIAN MARKET Amir Angabini and Shaista Wasiuzzaman GARCH(1,1) model is beaten by the asymmetric GARCH models, but when not allowing for asymmetries it was the best model compared to other GARCH models. Magnus and Fosu [18] reject the random walk hypothesis for the Ghana Stock Exchange and support the superiority of the GARCH(1,1) model compared to other models "under the assumption that the innovations follow a normal distribution." Shamiri and Abu Hassan [19] model and forecast the volatility of the Malaysian and the Singaporean stock indices using Asymmetric GARCH.…”

“…Table 4(a) and 4(b) below present the estimation for the different GARCH(p,q) models for both periods. We assume that the innovation term follows a normal distribution as was done by Magnus and Fosu [18]. Here Alpha refers to the value of previous square error term and Beta refers to the value of previous variances.…”

IPO Stocks Performance Imperfection: A Review of Models and Empirical Works

“…However, both hypotheses constitute the random walk model which was modelled as t t r µ ε = + (4) So that the mean ( ) µ value of the returns is expectedly to be insignificantly different from zero while the epsilon ( ) t ε represents the random error term. However, because of the naive nature of the RWH to assume that stock returns follow random walk and so future expected returns cannot be determined using previous time series of information, the assumption has been debunked by several studies example Fama [5] and Magnus and Fosu [6] etc. This made the EMH by Fama [5] to expand on the information relevance for returns forecast using the RWH as:…”

“…Furthermore the Threshold GARCH of Glosten, Jagannathan and Runkle [19] modifies the original GARCH specifications using a dummy variable with the assumption that unexpected changes in the market returns have different effects on the conditional variance of the returns. Such that good news goes with an unforeseen increase contributing to the variance through the coefficient β instead of an unexpected decrease which is presented as a bad news and contributes to the variance with the coefficient α + y, so that, if y > 0, the leverage effect exist and news impact is asymmetric if y ≠ 0, Magnus & Fosu, [6]. The TGARCH is, = < = > It is obvious to not that the IPO stocks performance and generally equity stocks returns have consistently undergone series of studies to determine the ex ante and post ante returns yet, the volatility persist and locks in the stock market.…”

ARIMA With GARCH Family Modeling and Projection on Share Volume of DSE

“…The extension of ARCH through GARCH is like the extension of the AR to ARMA model, since the introduction of ARCH and GARCH models has been extensively used in this literature. Magnus and Fosu (2006) modeled and forecasted volatility by taking an individual index and using the models or specifications like GARCH (1, 1), EGARCH (1, 1), and TGARCH (1, 1). Rafique and Kashif-ur-Rehman (2011) studied the volatility clustering, excess kurtosis, and heavy tails of the time series using ARCH, GARCH, and Nelson's EGARCH processes (1991).…”