KWAME NKRUMAH UNIVERSITY OF SCIENCE AND TECHNOLOGY

KUMASI
COLLEGE OF SCIENCE
FACULTY OF PHYSICAL AND COMPUTATIONAL SCIENCE
DEPARTMENT OF STATISTICS AND ACTUARIAL SCIENCE

COMPARATIVE GARCH VOLATILITY FORECASTING FOR RISK

ASSESSMENT OF THE MSCI WORLD INDEX AND EQUITIES ON
THE GHANA STOCK MARKET.

KENNETH AYEAKOR HEDZE

DUNCAN FRIMPONG BUABENG
JUDITH AKUA OFORIWAA
NANA AMISSAH AMOAH

A PROJECT SUBMITTED TO THE DEPARTMENT OF STATISTICS AND

ACTUARIAL SCIENCE, KWAME NKRUMAH UNIVERSITY OF SCIENCE AND
TECHNOLOGY IN PARTIAL FULFILLMENT OF THE REQUIREMENT FOR THE
DEGREE OF BACHELOR OF SCIENCE, ACTUARIAL SCIENCE

August 14, 2025

 Declaration
We hereby explicitly state that this dissertation is our own work completed for the purpose
of receiving the Bachelor of Science(BSc) degree in Actuarial Science under the capable
supervision of Prof. Akoto-Yaw Omari Sasu and that, to the best of our knowledge, it
does not contain any material that has been previously published by another person or
material that has been accepted for the award of any other degree from the university,
with the exception of situations where appropriate acknowledgement has been made.

i
 DEDICATION
We respectfully dedicate our work to our loved ones, family, friends, and the entire
KNUST community in our own unique way. We acknowledge that we are not self-made
and that the collective efforts of some family members and friends have allowed us to
advance in our academic careers. As a result, we would like to dedicate this work to each
and every person who has made immeasurable contributions in one way or another to its
completion.

KENNETH HEDZE AYEAKOR ..................... ..................

Index Number(8693221) Signature Date

DUNCAN FRIMPONG BUABENG ..................... ..................

Index Number(8691321) Signature Date

JUDITH AKUA OFORIWAA ..................... ..................

Index Number(8696121) Signature Date

NANA AMISSAH AMOAH ..................... ..................

Index Number(8688721) Signature Date

Certified by:
[Link]-YAW OMARI SASU ..................... ..................
Supervisor Signature Date

Certified by:
PROF. GABRIEL ASARE OKYERE ..................... ..................
Head of Department Signature Date

ii
 Acknowledgements
We owe a special obligation of thanks to God Almighty for providing us with the life,
vigor and information we needed to complete this difficult academic task. We also owe
a debt of gratitude.... May the Good Lord bless all of the instructors in the Department
of Statistics and Actuarial Science.

iii
 Abstract
This study analyzes and compares the volatility patterns of the MSCI Index and some
equities on the Ghana Stock Exchange using GARCH models.. By comparing Ghana’s
market to the MSCI World Index using a GARCH volatility forecasting approach, it
sheds light on how stable or unstable these markets really are, and what that means for
both investors and policymakers. The primary objective is to identify the most effective
volatility forecasting models for each equity, to enhance risk assessment and investor
confidence in Ghana’s stock market. Understanding market volatility isn’t just a technical
exercise, it’s key to keeping investor confidence high and making smart investment choices.
Using historical daily prices from June 2015 to May 2025, the research looked at the MSCI
World Index as a global benchmark and compared it to Ghana Stock Exchange equities.
Data came from Alpha Vantage and the GSE’s official site, and the best forecasting model
was chosen based on statistical accuracy tests like the AIC, BIC, and log-likelihood values.
The results showed that GARCH(1,1) is the best model for forecasting volatility with
GARCH(0,1) being the exception for Tullow oil. Also, the MSCI World Index showed a
steady decline in volatility, pointing to relative stability. Ghana’s market, on the other
hand, saw a spike in volatility before settling down, a sign of short-term uncertainty. One
exception was Unilever Ghana, which showed a persistent rise in volatility over time.
These insights can help investors decide how to diversify their portfolios and manage
risk, while also giving policymakers a clearer picture of where market stability needs
strengthening.

iv
Contents

Declaration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i

Dedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i

Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i

List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi

List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1 Background of Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Objectives of the Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4 Justification of the Study . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.5 Limitations of the Study . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.6 Organization of the Study . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Conceptual Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3 Theoretical Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.1 Overview of Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

v
 3.2 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.2.1 Source and Type of Data . . . . . . . . . . . . . . . . . . . . . . . 16
3.2.2 Method of Collection . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.3 Model Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.3.1 Test for Stationarity . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.3.2 Testing for ARCH Effect . . . . . . . . . . . . . . . . . . . . . . . 18
3.3.3 Test for Autocorrelation . . . . . . . . . . . . . . . . . . . . . . . 19
3.3.4 The ARCH model . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.3.5 The Generalized ARCH model . . . . . . . . . . . . . . . . . . . . 20
3.3.6 Information Criterion . . . . . . . . . . . . . . . . . . . . . . . . . 20

4 Results and Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.2 Preliminary Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.2.1 Descriptive Analysis of Data . . . . . . . . . . . . . . . . . . . . . 23
4.3 Main Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.3.1 Stationarity Test . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.3.2 Autocorrelation Test . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.3.3 Test for ARCH Effect . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.3.4 Model Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

5 Conclusion and Recommendations . . . . . . . . . . . . . . . . . . . . . . 37

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
5.2 Summary of Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
5.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
5.4 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

vi
List of Figures

4.2.1 Time series plot of the closing prices of MSCI World index for the period
01/06/2015 till 30/05/2025 . . . . . . . . . . . . . . . . . . . . . . . . . . 24
4.2.2 Time series plot of the closing prices of Unilever for the period 01/06/2015
till 30/05/2025 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
4.2.3 Time series plot of the closing prices of Standard Chartered for the period
01/06/2015 till 30/05/2025 . . . . . . . . . . . . . . . . . . . . . . . . . . 24
4.2.4 Time series plot of the closing prices of Tullow oil for the period 01/06/2015
till 30/05/2025 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
4.2.5 Daily returns on the MSCI World index for the period 01/06/2015 till
30/05/2025 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.2.6 Daily returns on Standard Chartered for the period 01/06/2015 till 30/05/2025 25
4.2.7 Daily returns on Unilever for the period 01/06/2015 till 30/05/2025 . . . 25
4.2.8 Daily returns on Tullow oil for the period 01/06/2015 till 30/05/2025 . . 25
4.3.1 Forecasted conditional variance for the MSCI for a period of 30 days using
the GARCH(1,1) model . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.3.2 Forecasted conditional variance for Standard Chartered for a period of 30
days using the GARCH(1,1) model . . . . . . . . . . . . . . . . . . . . . 32
4.3.3 Forecasted conditional variance for Unilever for a period of 30 days using
the GARCH(1,1) model . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.3.4 Forecasted conditional variance for Tullow oil for a period of 30 days using
the GARCH(1,1) model . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

vii
List of Tables

4.2.1 Descriptive statistics of the daily log returns of the equities . . . . . . . . 23

4.3.1 Stationarity test for MSCI, Unilever, Standard Chartered and Tullow oil 27
4.3.2 Ljung Box test for Autocorrelation . . . . . . . . . . . . . . . . . . . . . 28
4.3.3 ARCH effect test for the data. . . . . . . . . . . . . . . . . . . . . . . . . 28
4.3.4 GARCH model for the MSCI World index . . . . . . . . . . . . . . . . . 29
4.3.5 GARCH model for Standard Chartered . . . . . . . . . . . . . . . . . . . 31
4.3.6 GARCH model for Unilever . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.3.7 GARCH model for Tullow oil . . . . . . . . . . . . . . . . . . . . . . . . 35

1
 Chapter 1

Introduction
1.1 Background of Study
[1] . One of the primary concerns of the stock market is the risk associated with high
fluctuations in the stock prices, which are much beyond the probable changes in the real
value of companies representing the stock (Omari-Sasu et al, 2015). Volatility has proven
to be a very important concept in Finance and Econometrics as it measures how much
and how quickly the value of an asset changes. Investors also define volatility as upswings
or downswings or rapid price movement within a short period of time (Agarwal, 2017).
High volatility in the stock market makes investment riskier. Extreme volatility is a dan-
gerous signal as it ruins the smooth working of the financial system and has a negative
impact on economic performance (Agarwal, 2017).
The openness of a country’s economy is recognized as a cause of volatility of its mar-
kets(Adjasi et al, 2008). The increasing importance of international trade has resulted
in the need for an accurate forecast of volatility in the global markets. Forecasting stock
market volatility plays a crucial role in financial risk assessment, influencing investment
strategies, portfolio management and regulatory decisions(Poon and Granger, 2003). By
forecasting volatility, investors and financial institutions can enhance their risk manage-
ment capabilities, making informed decisions to mitigate potential losses. They are also
able to develop strategies that align with their risk tolerance.
Various models, particularly those from the GARCH family, have been developed to
analyze and forecast market fluctuations using historical price data. While extensive re-
search has been conducted on global markets, including those represented by the Morgan
Stanley Capital International(MSCI) Index, relatively little attention has been given to
emerging markets such as the Ghana Stock Market (GSM).
The MSCI World Index is widely regarded as a global benchmark for equity market per-

2
formance, representing developed economies across North America, Europe, and Asia. It
serves as a reference point for global investors and is often used in portfolio allocation,
risk assessment, and financial modeling. By contrast, the Ghana Stock Market (GSM)
represents a much smaller and less liquid market, with different structural characteristics,
regulatory environments, and economic conditions. Unlike developed markets, emerging
markets like Ghana often experience higher volatility due to factors such as currency
fluctuations, political instability, and lower investor participation.
As African financial markets continue to evolve, a comparative analysis of volatility be-
havior between GSM and MSCI can provide valuable insights for risk management and
investment planning.

1.2 Problem Statement

Globally, indices like MSCI provides insight into major financial markets and serve as an
essential benchmark for assessing global investment [Link] index is often referenced in
financial modeling, economic research, and investment decision-making. However, while
the MSCI World Index captures volatility trends in developed markets, the volatility
patterns of smaller, less liquid markets such as the Ghana Stock Market remains under-
explored.
Emerging markets often exhibit different volatility characteristics due to factors such as
liquidity constraints, capital flow restrictions, and macroeconomic instability. Without
a clear understanding of the volatility dynamics of these markets,investors may struggle
to make informed decisions, leading to increased financial uncertainties. This lack of
analysis makes it difficult for investors to guage the risk associated with Ghana’s stock
market in relation to international benchmarks.
Given the growing integration of African economies into the global financial system, it
is essential to analyze how local markets like Ghana respond to international economic
shocks, financial crises, and macroeconomic policies. By employing comparative GARCH
volatility forecasting, this study aims to bridge the research gap, offering a comprehensive
risk assessment that will benefit policymakers, portfolio managers, and investors seeking

3
to navigate the complexities of both local and global markets.

1.3 Objectives of the Study

The main objective of this study is to analyze and compare the volatility patterns of the
MSCI Index and some equities on the Ghana Stock Exchange using GARCH models.
The specific objectives of this study include;

1. To determine the effectiveness of different GARCH model variations in predicting

market fluctuations for the MSCI Index and some equities on the Ghana Stock
Exchange.

2. To explore how volatility forecasting can contribute to financial stability and en-
hance investor confidence in Ghana’s stock market.

3. To assess the risk implications of volatility trends in these markets and their impact
on investment decisions.

1.4 Justification of the Study

Accurate volatility forecasting is essential for managing financial risks and making in-
formed investment decisions. In financial markets, volatility plays a critical role in shap-
ing investment strategies, asset pricing, and regulatory policies. While extensive research
exists on major global markets, African stock exchanges, particularly the Ghana Stock
Market, have received limited attention. This study addresses this gap by comparing
the volatility characteristics of the GSM and MSCI, offering insights into how market
fluctuations differ between a global benchmark and an emerging market. The findings
will be valuable for investors, portfolio managers, and policymakers seeking to improve
risk management strategies.
By utilizing comparative GARCH modeling techniques, this study will contribute to the

4
growing body of financial literature on emerging market volatility and offer practical rec-
ommendations for improving risk management frameworks in Ghana’s financial sector.

1.5 Limitations of the Study

This study has a few limitations which include;
Lack of Macroeconomic Variables: The dataset does not include key macroeconomic
indicators such as inflation rates, interest rates, GDP growth, or monetary policy changes,
which are known to influence stock price fluctuations. As a result, the study cannot an-
alyze the impact of macroeconomic conditions on stock market volatility.
Currency Differences and Exchange Rate Risks: The MSCI World Index is de-
nominated in USD, while equities on the Ghana Stock Exchange are denominated in
Ghanaian Cedi (GHS). If the difference in currencies remains the same, exchange rate
fluctuations may introduce additional volatility unrelated to actual stock market move-
ments. Converting all data into a single currency may mitigate this issue but introduces
reliance on exchange rate data accuracy.
Absence of Sectoral or Firm-Level Analysis: The data consists of aggregate index
values rather than individual stock data, limiting insights into sector-specific or firm-level
volatility drivers. Sectoral shocks, company-specific events, or industry trends cannot be
analyzed in this study.
While the study applies GARCH models, alternative volatility forecasting approaches,
such as machine learning techniques, could provide additional insights but are not ex-
plored in this research.

1.6 Organization of the Study

This study is structured into five main chapters;
Chapter One provides an introduction to the research, including the background of the
study, problem statement, justification, objectives, scope, and limitations of the study.
Chapter Two presents a review of relevant literature on stock market volatility, GARCH
modeling, and comparative studies on emerging and global markets. It also discusses

5
previous studies on volatility forecasting and financial risk assessment.
Chapter Three outlines the research methodology, including data sources, the GARCH
modeling framework, and the statistical techniques used for volatility forecasting and risk
assessment.
Chapter Four presents the empirical analysis, comparing the volatility patterns of the
Ghana Stock Market (GSM) and the MSCI World Index, discussing key findings, trends,
and their implications.
Chapter Five concludes the study by summarizing the key findings, discussing policy
implications, and providing recommendations for investors, policymakers, and future re-
searchers interested in financial risk management in emerging markets.

6
 Chapter 2

Literature Review
2.1 Introduction
This chapter provides an overview of existing research on volatility modelling and fore-
casting. By examining existing knowledge on volatility, we can identify patterns, trends
and gaps in literature and provide a framework for future research

2.2 Conceptual Framework

Over the past few decades, numerous papers have attempted to find the features and
underlying factors contributing to volatility and how each factor actually affect volatility.
Rashid Sabri(2004) explored the new features of emerging stock markets, in order to point
out the most associated indicators of increasing stock return volatility, which may lead to
instability of emerging markets. His study covered a sample of five geographical areas of
emerging economies, including Mexico, Korea, South Africa, Turkey and [Link]
backward multiple-regression technique was used to examine the relationship between
monthly changes of stock price indices as dependent variable and the associated predict-
ing local as well as international variables, which represent possible causes of increasing
price volatility and initiating crises in emerging stock markets. The result revealed that
stock trading volume and currency exchange rate respectively represent the most pre-
dicting variables of increasing price volatility. International stock price index, deposit
interest rate, and bond trading volume were moderate predicting-variables for emerging
stock price volatility. While changes in inflation rate were the least predicting variable
Futhermore, Mamtha and Srinivasan (2016) in their paper tried to understand the char-
acteristics of volatility with the evidence of surveys of literatures. They found out that
information flow, trading volume, economical aspects and investor’s behavior contribute

7
to volatility in the stock market. They also observed that stock markets around the world
have asymmetry response and spillover effects on volatility. That is , the stock market
responds differently to positive and negative news and volatility in one market affects
others as well.
Bhatta and Duwal (2021) analyzed the impact of dividend policy on stock price volatility.
They investigated the relationship between the dividend policy and stock price volatility.
They discovered that in most of the cases, dividend policy has a significant negative rela-
tionship with the stock price volatility. This implies that stock price volatility decreaces
as divident is paid out.
Bui and kusuma (2023) summarized the main findings in the literature related to stock
market volatility during the global financial crisis. They also went futher to present a
deeper understanding of the factors that influence stock market volatility in the context
of financial [Link] explored various theories and models used to understand stock
market volatility, which include ARCH/GARCH models, market efficiency, behavioral
finance theory, and behavioral economics theory. They also explored literature that ex-
amines the impact of macroeconomic factors, monetary policy, investor sentiment, and
other factors on stock market volatility during the global financial crisis. The results
showed that the global financial crisis had a significant impact on stock market volatility.
They also discovered that economic uncertainty, changes in monetary policy, and nega-
tive investor sentiment all contributed to the increase in stock market volatility.
Dhingra et al. (2024) investigated the major factors impacting stock market volatility.
The results showed that there was an increasing interest among researchers in exploring
the relationship between stock markets and “cryptocurrencies” and “bitcoin,” particu-
larly in the context of the “COVID-19” pandemic. It also revealed that oil prices, policy
uncertainty and investor sentiments have a significant impact on stock market volatility.
Various studies have investigated the relationship between exchange rate volatility and
stock market volatility using different methods and [Link] et al.(2008) inves-
tigated the relationship between exchange rate volatility and stock market volatility in
Ghana using the Exponential Generalized Autoregressive Conditional Heteroskedasticity
(EGARCH) model. They observed that there existed a negative relationship between

8
exchange rate and stock market volatility. This implies that a decrease in the local cur-
rency leads to an increase in stock market returns in the long run.
Diamandis and Drakos (2011) analysed the long-run relationships and short-run dynamics
between stock prices and exchange rates as well as the channels through which exogenous
shocks influence these markets. They did this by using cointegration analysis and multi-
variate Granger causality tests. They observed that stock and foreign exchange markets
in the economies cosidered are positively related and that the U.S. stock market acts as
a channel for these links
Mlambo et al.(2013) studied the effects of currency volatility on the Johannesburg Stock
[Link] Generalised Autoregressive Conditional Heteroskedascity (1.1) (GARCH)
model was used in establishing the relationship between exchange rate volatility and stock
market performance. The results showed that there existed a very weak relationship be-
tween currency volatility and the stock [Link] to them, the weak relationship
between currency volatility and the stock market suggests that the JSE can be marketed
as a safe market for foreign investors. However, investors, bankers and portfolio managers
still need to be vigilant in regard to the spillovers from the foreign exchange rate into the
stock market.
Aslam (2014) investigated the impact of exchange rate Pakistan rupee in terms of US
Dollar and KSE 100 index. He went futhur to analyze the causal relationship between
both the time series. The results showed that there existed a very weak negative corre-
lation between KSE100 indices and USD-PKR exchange rate. The result also showed a
bidirectional relationship between KSE 100 indices and Exchange Rate.
Mburu (2015) investigated the relatioship between exchange rate volatility and stock
market performance in the Nairobi Securities Exchange. He used regression and corre-
lation analysis to determine the effects of exchange rate volatility on the performance of
the stock a market. Other macro-economic factors like Inflation volatility, Money supply
volatility and interest rate volatility were also included to determine the impact it had
on the performance in the stock market The result showed that exchange rate volatility
was among the determinants of stock market performance though not a very significant
one.

9
There are other research that have been made which are related to the study of volatility.
Gokcan (2000) examined the effectiveness of linear (GARCH (1,1)) and non-linear (EGARCH)
models in forecasting volatility in emerging stock markets. While GARCH models (Engle,
1982; Bollerslev, 1986) are widely used for financial time series forecasting, their ability
to capture asymmetric volatility and skewness in emerging markets remains uncertain.
Research by Nelson (1991) and Glosten et al. (1993) suggests that non-linear models
are better suited to handling negative market shocks, which tend to amplify volatility
more than positive ones. Analyzing monthly stock returns from seven emerging markets
(1988–1997), Gokcan finds that GARCH (1,1) outperforms EGARCH, despite the pres-
ence of skewed distributions and excess kurtosis. The findings challenge the assumption
that non-linear models always improve forecasting accuracy, underscoring the need for
tailored risk management strategies in volatile markets.
Yang et al.(2001) analyzed the impact of the 1996 FAIR Act on agricultural price volatil-
ity using GARCH models. Their findings show higher volatility for corn, soybeans, and
wheat, little change for oats, and reduced volatility for cotton, contradicting earlier stud-
ies (Crain and Lee, 1996). This research provides empirical insights into how liberalization
influences commodity markets, informing policymakers and agribusinesses on risk man-
agement.
Li et al. (2005) examined the relationship between expected stock returns and volatility
in the 12 largest international stock markets from January 1980 to December 2001. They
discovered a very weak positive relationship during the sample period for the majority of
the markets based on parametric EGARCH-M [Link], they found a significant
negative relationship between expected returns and volatility in 6 out of the 12 markets
when using a flexible semiparametric specification of conditional variance.
Abdalla and Winker (2012) investigated stock market volatility in Sudan and Egypt
through the application of GARCH models, drawing on the foundational work of Engle
(1982) and Bollerslev (1986). Examining daily stock returns from the Khartoum Stock
Exchange (KSE) and the Cairo and Alexandria Stock Exchange (CASE) between 2006
and 2010, they used symmetric and asymmetric GARCH models to evaluate volatility
patterns. Their findings suggest that KSE experiences higher volatility, signaling market

10
instability, while CASE demonstrates greater stability and persistence. Additionally, the
study identifies a positive risk premium, reinforcing the established relationship between
risk and expected returns. These insights contribute to improved volatility management
strategies for investors and policymakers in emerging markets.
Agarwal (2017) tried to model the volatility of two indices: MSCI emerging markets index
and MSCI world index using ARCH and GARCH models. Both the ARCH and GARCH
terms were found to be significant in both the market indices. However, in emerging
markets, yesterday’s volatility had greater influence in explaining today’s volatility while
in case of developed markets, both yesterday’s volatility and information had immense
influence in explaining today’s volatility.

2.3 Theoretical Framework

With volatility forecasting being a long-standing concern in finance, various models have
been proposed or introduced to capture the dynamics of volatility. Hwang and Satchell
(2005) introduced the GARCHX model, an extension of the standard GARCH frame-
work (Bollerslev, 1986; Engle, 1982), incorporating cross-sectional market volatility as
an explanatory variable. By integrating cross-sectional volatility, the GARCHX model
captures common market risk factors, differentiating it from traditional GARCH models,
which primarily rely on past variances. Empirical results using UK and US stock market
data reveal that GARCHX models better specify individual return volatility, though they
do not consistently outperform standard GARCH models in forecasting accuracy. This
study contributes to volatility modeling by demonstrating that market-wide fluctuations
are significant in explaining asset price movements, with implications for risk assessment
and portfolio management.
In addition, Lanne and Saikkonen (2005) proposed the Smooth Transition GARCH
(STGARCH) model to better capture persistent volatility in financial markets. Unlike
standard GARCH models, which tend to overestimate volatility persistence following ex-
treme events (Engle and Mustafa, 1992; Lamoureux and Lastrapes, 1993), the STGARCH

11
model introduces lagged conditional variance as a transition variable, enhancing accuracy
in long-term volatility modeling (Hagerud, 1997; González-Rivera, 1998). The study ap-
plies Markov chain theory to ensure stationarity and geometric ergodicity, making the
model more robust for financial applications. Empirical testing on DEM/USD exchange
rates confirms the model’s superiority over traditional GARCH models, highlighting its
potential in volatility forecasting and risk assessment.
Monfared and Enke (2014) developed a hybrid GJR-GARCH Neural Network model to
enhance volatility forecasting. Building on GARCH (Bollerslev, 1986) and GJR-GARCH
(Glosten et al., 1993), the study accounts for asymmetric volatility effects (Nelson, 1991).
To improve accuracy, neural networks—Feed-Forward Back Propagation (FFBP), Gen-
eralized Regression (GR), and Radial Basis Function (RBF) are integrated with GJR-
GARCH and tested across four economic cycles (1997–2011). Results show that RBF
networks outperform traditional models, particularly during financial crises, highlighting
the role of AI in risk management and financial forecasting.
With the existence of various models for volatility forecasting, researchers have been able
to forecast volatility and have therefore, also made comparisms to find out the best per-
forming model among the GARCH model and its modifications. Chong et al. (1999)
studied the performance of GARCH model and it’s modifications, using the rate of re-
turns from the daily stock market indices of the Kuala Lumpur Stock Exchange (KLSE).
They observed that, although the exponential GARCH was not the best model in the
goodness-of-fit statistics, it outperformed all the other models in describing the often-
observed skewness in stock market indices and in out-of-sample forecasting. In contrast,
the integrated GARCH is the poorest performing model in both respect.
Similarly, Hamadu and Ibiwoye (2010) examined the volatility of the daily returns of
Nigerian insurance stocks. The result also revealed the Exponential Generalized Au-
toregressive Conditional Heteroskedasticity (EGARCH) model to be the best model for
modelling stock price return as it outperformed the other competing models in both the
model-estimation evaluation and out-of-sample volatility forecasting.
Ou and Wang (2011) studied how the Gaussian processes are applied, to model and pre-
dict financial volatility based on GARCH, EGARCH and GJR to capture the symmetric

12
and asymmetric effects. The outcome, by using five different kernels to train each of
the proposed volatility model, show that the non-linear hybrid models can capture well
the symmetric and asymmetric effects of news on volatility and yield better predictive
performance than the GARCH, EGARCH and GJR approaches
Kosapattarapim et al. (2011) investigated the volatility forecasting capability of GARCH(p,q)
models with six different type of error distributions and apply them to three South East
Asian emerging stock markets. Their results showed that a GARCH(p,q) model with non-
normal error distributions tends to provide better out-of-sample forecast performance
than a GARCH(p,q) model with normal error distribution. Simulation and empirical
studies show that MSE(MAE) given by the best fitted model is insignificantly different
from that given by the best forecast performance model since it is not practicable to
identify the best performance model in practice.
Babikir et al. (2012) investigated the empirical relevance of structural breaks in fore-
casting stock return volatility using both in-sample and out-of-sample tests applied to
daily returns of the Johannesburg Stock Exchange(JSE) All Share Index. The results for
the period July 2, 1995 to August 25, 2010 revealed the evidence of structural breaks in
the unconditional variance of the stock returns series over the period, with high levels of
persistence and variability in the parameter estimates of the GARCH(1,1) model across
the sub-samples defined by the structural breaks. They also observed that, for shorter
horizons, the MS-GARCH model better captures asymmetry in stock return volatility
than the GJR-GARCH (1, 1) model, which is better suited to longer horizons, but in
general, the asymmetric models fail to outperform the GARCH (1,1) model.
Lim and Seek (2013) modelled the volatility of the stock market of Malaysia using sym-
metric and asymmetric GARCH-type models. The results showed that symmetric and
asymmetric GARCH models have different performances in different time frames. For
normal periods (pre and post-crisis), symmetric GARCH model outperforms asymmetric
GARCH but for fluctuation period (crisis period), asymmetric GARCH model is pre-
ferred.
Lin (2018) studied the econometric features of the SSE Composite Index, using GARCH
type models, and compares the adaptability of these models in fitting Chinese stock

13
market. It was observed that the SSE Composite Index possesses significant proper-
ties of time-varying and clustering and its series distribution presents leptokurtosis with
significant ARCH and GARCH effects. By comparing the fitting and the forecast per-
formance of GARCH(1,1)(symmetric) and TARCH(1,1), EGARCH(1,1)(asymmetric), it
turned out that EGARCH(1,1) generally out performed the others, which follows the
same conclusion as Chong et al. (1999) and Hamadu and Ibiwoye (2010).
Almisshal and Emir (2021) in their paper, explored the volatility of exchange rates using
GARCH models, specifically analyzing the USD/TRY and EUR/TRY pairs between 2005
and 2019. Their research builds on the ARCH model (Engle, 1982) and its extension,
GARCH (Bollerslev, 1986), applying both symmetric (GARCH (1,1)) and asymmetric
variants (EGARCH, GJR-GARCH, PGARCH) to capture volatility clustering and lever-
age effects (Nelson, 1991). Utilizing ARMA models for mean estimation and assessing
forecast accuracy with RMSE, MAE, and MAPE, their results indicate that GARCH
(1,1) and GJR-GARCH (1,1) best model USD/TRY volatility, while PGARCH (1,1)
performs better for EUR/TRY. The study provides insights into risk management and
foreign exchange policy decisions.
Agyarko et al. (2023) analyzed the volatility of the Ghana stock market using three mod-
els, GARCH,TGARCH AND EGARCH .The results showed that the TGARCH(1,1) was
the model that suited the data best based on the Akaike Information Criterion (AIC), The
Root Mean Square Error (RMSE) and the Mean Absolute Percentage Error (MAPE).
A lot of research has also been made to find the complexities of volatility in option
pricing, stock market analysis etc including linear and non linear approaches. Gokcan
(2000) examines volatility forecasting in emerging stock markets using linear GARCH
(1,1) and non-linear EGARCH models. The study builds on the GARCH framework
(Engle, 1982; Bollerslev, 1986), widely used in financial time series analysis, and extends
prior research on asymmetry in volatility modeling (Nelson, 1991; Glosten et al., 1993).
Emerging markets, known for higher risk and return volatility, often exhibit non-normal
return distributions with skewness and leptokurtosis. While non-linear GARCH mod-
els address these issues, Gokcan finds that linear GARCH (1,1) performs better than
EGARCH in capturing volatility patterns across seven emerging markets (1988–1997).

14
The study highlights the importance of selecting appropriate volatility models for risk
assessment and portfolio management in less mature financial markets.
Ritchken and Trevor (2003) explore option pricing under Generalized GARCH and stochas-
tic volatility models, developing an efficient lattice algorithm that integrates both frame-
works. The study aligns with prior work by Bollerslev (1986) and Engle (1982) in volatil-
ity modeling, while also linking GARCH processes to bivariate diffusion models (Nelson,
1990). The proposed algorithm efficiently handles path dependency in GARCH models
and extends to generalized settings, allowing for pricing European and American options.
By incorporating stochastic volatility models like those of Hull and White (1987), Heston
(1993), and Stein and Stein (1991), the study unifies different pricing models under a
single computational framework. This enhances empirical testing and improves the prac-
tical implementation of option pricing methodologies.
Mala and Reddy (2007) investigated the presence of stock market volatility on Fiji’s stock
market, an emerging economy using ARCH and GARCH models. The results showed that
seven out of the sixteen firms listed on Fiji’s stock market is volatile. They subsequently
regressed the volatility of stock returns and the interest rate which revealed that interest
rate changes have a significant effect on stock market volatility.

15
 Chapter 3

Methodology
3.1 Overview of Study
This chapter provides the methodological framework used to compare the volatility of the
MSCI World Index and equities on the Ghana Stock Exchange. The study employs the
Generalized Autoregressive Conditional Heteroskedasticity (GARCH) model to evaluate
and forecast volatility for financial risk assessment. This chapter details the data sources,
preprocessing steps, and statistical tests conducted to validate the applicability of the
model. Finally, it outlines the modeling procedure used to assess the volatility dynamics
of the two indices.

3.2 Data

3.2.1 Source and Type of Data

The analysis is based on secondary quantitative time series data for the MSCI world
index, which represents large and mid-cap equity performance in 23 developed countries
and some equities on the Ghana Stock Exchange.
The dataset spans from 1st June 2015 to 30th May 2025. Due to discrepancies in the
trading calendars between global markets and the Ghanaian stock market, the data was
aligned to retain only common trading dates, ensuring consistency in comparative anal-
ysis.
The primary variables for this study were the daily closing prices for the MSCI World
Index and Closing Price - VWAP(GHC) variable for all the Ghana Stock Exchange data.
The key variable used for modeling volatility is the daily simple return derived from
closing prices. Equities considerd under the ghana stock exchange are that of Standard
Chartered Bank in the finance industry, Unilever in the consumer goods industry and

16
Tullow oil in the oil and gas industry.

3.2.2 Method of Collection

Data Sources:
The MSCI World Index data was retrieved using the Alpha Vantage API at [Link]
and stored in a structured SQL database whiles the data of each equity analyzed in this
work was manually downloaded from the official GSE website([Link]) in excel
format. An exchange rate data was also downloaded from [Link].

Cleaning and Preprocessing steps

Dates were converted to DatetimeIndex format to facilitate time-series operations. Du-
plicated observations in each data were removed. The MSCI data was converted from
USD to GHC using an exchange rate data. Both datasets were merged using an inner
join on common dates to remove missing or unmatched entries. All non-numeric strings
(e.g., volume, capitalization) were cleaned and converted to float for computation. All
duplicated entries based on dates or index values were eliminated. Daily log returns were
calculated using the formula:
Pt
rt = ln (3.2.1)
Pt−1
Where: rt represents return at time t,
Pt and Pt−1 represents closing prices at time t and t − 1, respectively.

3.3 Model Description

3.3.1 Test for Stationarity

To test whether the return series is stationary, the Augmented Dickey-Fuller (ADF) test
was used. The null hypothesis for this test is that the return series has a unit root
(non-stationary). The equation below is an expression of the ADF test.

∆yt = α + βt + γyt−1 + δ1 ∆yt−1 + ... + δp ∆yt−p + ϵt (3.3.1)

17
where ∆yt is the first difference of the series
α is a constant
βt is the trend term (optional)
γ coefficient on the lagged level of the series.
The null hypothesis is rejected if the test stastistic is less than the critical value.
To complement the ADF test, the Kwiatkowski–Phillips–Schmidt–Shin (KPSS) test was
also used. Unlike the ADF test, which assumes non-stationarity, the KPSS reverses the
assumption, allowing for a balanced diagnosis. The null hypothesis is that the series is
stationary. The KPSS test decomposes the series yt as:

yt = rt + δt + ϵt (3.3.2)

where rt is the random walk (non-stationary component)

δt is the deterministic trend
ϵt is the stationary error term.

The test statistic is derived from the residuals obtained after regressing the se-
ries on a constant or a trend, using the Lagrange Multiplier (LM) principle. If the test
statistic exceeds the critical value, we reject the null hypothesis.

3.3.2 Testing for ARCH Effect

The Lagrange Multiplier (LM) test was used to To test for the presence of ARCH effects,
which signify time-varying volatility, justifying the use of GARCH models. The null
hypothesis is that there is no ARCH effects (homoskedasticity). The Lagarange Multiplier
test procedure involves;

1. Estimate a constant mean model and extract residuals ϵ̂t

2. Square the residuals and regress ϵ̂2t on p lags of itself.

18
 3. Compute the LM statistic:
LM = nR2 (3.3.3)

where n and R2 represents sample size and coefficient of determination from the auxiliary
regression respectively.
We reject the null hypothesis if the p-value is less than the critical value.

3.3.3 Test for Autocorrelation

The Ljung-Box Q test to detect autocorrelation in the return series or residuals at multiple
lags. The null hypothesis is that there is no autocorrelation at lag k. The test statistic
is:
h
X ρˆk 2
Q = n(n + 2) (3.3.4)
k=1
n−k

Where n represents the number of observations

ρˆk represents sample autocorrelation at lag k
h represents the number of lags.

We reject the null hypothesis if the p-value is less than the critical value.

3.3.4 The ARCH model

The Autoregressive Conditional Heteroskedasticity(ARCH) model, which was introduced

by Robert Engle (1982), describes how volatility clusters over time, with the current
variance being influenced by past error terms. The ARCH(q) model is specified as:

q
X
σt2 =ω+ αϵ2t−i (3.3.5)
i=i

where σt2 represents the conditional variance

ω is the constant term

19
α represents the ARCH effect, measuring the impact of past shocks on current volatility
ϵ2t−i represents the squared error terms

3.3.5 The Generalized ARCH model

The GARCH model, which was introduced by Tim Bollerslev, analyzes and forecasts
financial time series volatility. It is an extention of the ARCH model and therefore,
overcomes the weakness of the ARCH model. The GARCH(p,q) model is specified as:

q p
X X
σt2 =ω+ αi ϵ2t−i + 2
βj σt−i (3.3.6)
i=1 j=1

where σt2 represents the conditional variance,

ω, represents the constant term
αi and βj are non-negative coefficients.
ϵ2t−i represents the past squared error terms
2
σt−j represents the past volatility.

The reduced form of the GARCH model is the GARCH (1, 1) which is represented
as:
σt2 = ω + α1 ϵ2t−1 + β1 σt−1
2
(3.3.7)

where ω is the constant term, α1 and β1 are nonnegative coefficients and α1 + β1 < 1 in
order to achieve stationarity.

3.3.6 Information Criterion

Information criteria are statistical tools commonly employed in statistics and machine
learning to facilitate the comparison and selection of competing models, particularly dur-
ing the model fitting process. They are especially useful in areas like regression analysis,
time series modeling, and machine learning, where several models may provide different

20
levels of explanatory power for the same [Link] are designed to achieve a goodness
of fit, how accurately the model captures the underlying patterns in the observed data
and model complexity, the number of parameters or the flexibility of the model. The two
information criteria used in this work were the Akaike Information Criterion (AIC) and
the Bayesian Information Criterion (BIC).

Akaike Information Criterion (AIC)

The Akaike Information Criterion (AIC) is a commonly used statistical method for as-
sessing and comparing the performance of multiple models, aiming to identify the one
that offers the optimal balance between goodness of fit and model simplicity. Introduced
in 1974 by Japanese statistician Hirotugu Akaike, it has since become a fundamental
tool in the field of model selection, widely applied across various statistical and machine
learning contexts. The best competing model is selected by choosing model with the least
AIC values. The Akaike Information Criterion (AIC) is represented as,

AIC = 2k − 2 ln L (3.3.8)

where ln L is the maximized value of the likelihood function for the model, and k is the
number of estimated parameters in the model.

Bayesian Information Criterion (BIC)

The Bayesian Information Criterion (BIC) is a statistical tool used for model comparison
and selection, aimed at identifying the model that offers the most appropriate balance
between goodness of fit and model simplicity. Like the Akaike Information Criterion
(AIC), BIC evaluates how well a model fits the data, but it differs in the strength of its
penalty for complexity, especially as the sample size [Link] best competing model
is selected by choosing model with the least BIC values. The Bayesian Information
Criterion (BIC) is represented as,

BIC = k ln n − 2 ln L (3.3.9)

21
where k is the number of estimated parameters in the model, n is the number of obser-
vations in the data and ln L is the maximized value of the likelihood function for the
model.

Log-Likelihood

The log-likelihood is a measure of how well a statistical model explains the observed data.
Although log-likelihood is not an information criterion, it is a core measure of model fit.
It cannot be used alone as it does not count for model complexity. The log-likelihood is
represented as,
n
X
ln L(θ|y) = ln P (yi |θ) (3.3.10)
i=1

where P (yi |θ) is the probability (or probability density) of the i-th data point under the
model.

22
 Chapter 4

Results and Analysis

4.1 Introduction
In pursuit of our research objectives, this section presents the findings of our study,
accompanied by an analysis that assesses the extent to which our results meet these
objectives.

4.2 Preliminary Results

4.2.1 Descriptive Analysis of Data

The table below shows the daily log return of the MSCI data and all equities considered
in this research. It was revealed that Unilever had the highest maximum log return
(0.737263) followed by the MSCI (0.185098) and the Standard Chartered (0.139762) with
Tullow oil having the least mamximum log return(0.007380) .

Table 4.2.1: Descriptive statistics of the daily log returns of the equities

STATISTIC MSCI UNILEVER STANDARD TULLOW

Minimum −0.184703 −0.840335 −0.162519 −0.162553
Maximum 0.185098 0.737263 0.139762 0.007380
Median 0.002109 0.000000 0.000000 0.000000
Mean 0.001306 0.000419 0.000110 −0.000426
Standard Deviation 0.023157 0.028622 0.023157 0.007580
Range 0.369801 1.577598 0.302281 0.169933
Skewness −0.298271 −3.441114 0.333178 −20.335486
Kurtosis 10.213882 497.240474 40.99155 421.742336

23
Figure 4.2.1: Time series plot of the closing Figure 4.2.2: Time series plot of the closing
prices of MSCI World index for the period prices of Unilever for the period 01/06/2015
01/06/2015 till 30/05/2025 till 30/05/2025

Figure 4.2.3: Time series plot of the closing Figure 4.2.4: Time series plot of the clos-
prices of Standard Chartered for the period ing prices of Tullow oil for the period
01/06/2015 till 30/05/2025 01/06/2015 till 30/05/2025

24
Figure 4.2.5: Daily returns on the MSCI Figure 4.2.6: Daily returns on Standard
World index for the period 01/06/2015 till Chartered for the period 01/06/2015 till
30/05/2025 30/05/2025

Figure 4.2.7: Daily returns on Unilever for Figure 4.2.8: Daily returns on Tullow oil for
the period 01/06/2015 till 30/05/2025 the period 01/06/2015 till 30/05/2025

Figure 4.1 illustrates the time series of the closing prices of the MSCI World index.
From the plot, it is obvious that there is an increasing trend throughout the period. Since

25
there is trend, we can also conclude that the series is non-stationary. Figure 4.2 illustrates
the time series of the closing prices of Unilever. It is obvious from the graph that trend
exists in the series. There is a gradually increasing trend in prices from 2015 to 2018
followed by a rapid rise from 2018 to 2020. There is a steady decline in prices from
2020 to 2022. There is an irregular spike at the beginning of the year 2023 follow by an
increasing trend. The series appears to be non-stationary. Figure 4.3 illustrates the time
series of the closing prices for Standard Chartered. It can be seen that trend exists in
the data. Since there is trend, this implies that the data is non-stationary. The plot also
shows regular price movement, implying active trading. Figure 4.4 illustrates the closing
prices of Tullow oil. It can be seen from the graph that there is a decreasing trend in the
data indicating that the data is non stationary.
Figure 4.5 illustrates the daily returns on the MSCI World index. It can be seen that
majority of the returns are clustered around 0 suggesting efficient market behavior. There
appears to be presence of volatility clustering which makes it suitable for modeling with
the GARCH model. Figure 4.6 illustrates the daily returns on Standard Chartered Bank.
It can be seen that the stock is thinly traded which is common in emerging markets.
There seems to be no visible volatility clustering in the [Link] 4.7 illustrates the
daily returns on Unilever. The plot shows limited market activity and an increased
trading volatility from 2023 to 2025. Figure 4.8 illustrates the daily returns on Tullow
oil. It can be seen that there is volatility from 2015 to 2018. From 2019 onward, daily
returns are almost zero every day indicating extremely low liquidity.

4.3 Main Results

4.3.1 Stationarity Test

We used both the KPSS test and the ADF test for stationarity with the following hypoth-
esis; For KPSS, For ADF,
H0 : Level stationary H0 :Non-stationary
H1 : Non-stationary H1 :Stationary

26
A significance level(α) of 0.05 was used for the test. The results of the tests conducted
on the data is shown in Table 4.2 below. Since the p-value of all the data is below 0.05
in the KPSS test (0.01), we rejected the null-hypothesis and concluded that the MSCI,
Unilever, Standard Chartered and Tullow oil data was non-stationary. The results also
showed that in the ADF test, the p-values of MSCI, Unilever and Standard Chartered
were all above the significance value, therefore we failed to reject the null hypothesis and
concluded that they were non-stationary. However, Tullow had a p-value less than the
significance level(0.0269775) suggesting stationarity. Since this contradicts with the kpss,
we used the result from the kpss and concluded that Tullow was non-stationary.

Table 4.3.1: Stationarity test for MSCI, Unilever, Standard Chartered and Tullow oil

TEST DATA STATISTIC P-VALUE

MSCI 0.6307776 0.01
KPSS Unilever 0.868557 0.01
Standard Chartered 0.3933321 0.01
Tullow 1.7756924 0.01
MSCI −0.9454258 0.772645
ADF Unilever −0.0012887 0.9584254
Standard Chartered −2.5403517 0.105962
Tullow −3.0943043 0.0269775

4.3.2 Autocorrelation Test

We employed the Ljung-Box test to test for the significant autocorrelaton present in the
data. The following hypothesis was made;
H0 : There is no significant autocorrelation.
H1 : There is a significant autocorreletion.
A 5% level of significance (α = 0.05) was used for this test. The result for the test is
shown in Table 4.3 below. The results showed that the p-value for all the data was below
the significance value. Therefore, the null hypothesis was rejected and we concluded that
there was autocorrelation in the data.

27
 Table 4.3.2: Ljung Box test for Autocorrelation

DATA STATISTIC P-VALUE

MSCI 107.81946 1.467481e − 18
Standard Chartered 100.110871 5.178243e − 17
Unilever 181.29283 1.262780e − 33
Tullow 85.975612 3.350714e − 14

4.3.3 Test for ARCH Effect

The ARCH-LM test was conducted to determine the presence of ARCH effect in the data.
The following hypothesis was made;
H0 : There is no ARCH effect. H1 : There is an ARCH effect.

A 5% level of significance(α = 0.05) was used for the test. The result for the test is
shown in Table 4.4 below. It was revealed that all the data had a p-value less than
the significance value. Therefore the null-hypothesis was rejected and we concluded that
there was presence of an ARCH effect.

Table 4.3.3: ARCH effect test for the data.

DATA STATISTIC P-VALUE

MSCI 272.785630 8.6524127e − 53
Standard Chartered 187.83962 5.503682e − 35
Unilever 858.358470 5.810683e − 178
Tullow 98.581095 1.0475586e − 16

4.3.4 Model Selection

Model Selection for MSCI

The results for modelling the MSCI index data is shown in Table 4.5 below. From the
table, it can be seen that only GARCH(0,1) and GARCH(1,1) have all its parameters
being statistically significant at a 5% level of significance. Therefore, they are candi-

28
dates for being the best model. The best model was chosen by comparing the AIC,
BIC and the log-likelihood of the two models. From the table, it can be seen that
the GARCH(1,1) model had the least AIC(−9387.59), the least BIC(−9365.36) and the
highest log-likelihood(4697.79). Therefore, GARCH(1,1) is the best model among all the
models. The GARCH(1,1) model is therefore, represented as,

2
σt2 = 0.00001169 + 0.1Yt−1 2
+ 0.88σt−1 (4.3.1)

Table 4.3.4: GARCH model for the MSCI World index

MODEL COEF STD ERR P > |t| AICs BICs LOG-LIK

GARCH(0, 1) ω 2.928e − 04 2.246e − 05 7.839e − 39 −8917.37 −8900.70 4461.68
α1 0.2001 4.091e − 02 1.006e − 06
ω 1.169e − 05 6.371e − 12 0.000
GARCH(1, 1) α1 0.1000 1.847e − 02 6.199e − 08 −9387.59 −9365.36 4697.79
β1 0.8800 1.568e − 02 0.000
ω 1.169e − 05 2.070e − 11 0.000
GARCH(2, 1) α1 0.1000 2.009e − 02 6.199e − 08 −9377.70 −9349.92 4693.85
β1 0.4400 0.958 0.000
β2 0.4400 0.947 0.642
ω 1.169e − 05 6.461e − 12 0.000
α1 0.0500 2.164e − 02 2.086e − 02
GARCH(1, 2) α2 0.0500 3.030e − 02 9.894e − 02 −9376.32 −9348.53 4693.16
β1 0.8800 1.755e − 02 0.000
ω 1.1693e − 05 2.298e − 11 0.000
α1 0.0500 3.079e − 02 0.104
GARCH(2, 2) α2 0.0500 3.230e − 02 0.122 −9366.17 −9332.83 4689.09
β1 0.4400 1.403 0.754
β2 0.4400 1.363 0.747

29
Figure 4.3.1: Forecasted conditional variance for the MSCI for a period of 30 days using
the GARCH(1,1) model

The forecasted conditional variances for the MSCI Index, estimated using the
GARCH(1,1) model, exhibit a steady downward trend from June 2, 2025, to July 11,
2025. Volatility decreases from approximately 0.0331 at the start of the forecast to
0.0291 by the end, reflecting a 12.03% decline. This trend suggests a reduction in market
uncertainty and a potential stabilization of the global financial environment.
The smooth decline in volatility implies a resilient market with a low likelihood of sudden
shocks. This may be attributed to factors such as reduced macroeconomic disruptions,
stabilizing inflation and interest rates, and a relatively calm geopolitical landscape.
Overall, the results indicate an improving risk outlook for the global market. The MSCI
Index appears to be entering a more stable phase, offering investors greater predictability
for risk management and investment decisions.

Model Selection for Standard Chartered

The results for modelling the Standard Chartered data is shown in Table 4.6 below. From
the table, it can be seen that only GARCH(0,1) and GARCH(1,1) have all its parameters
being statistically significant at a 5% level of significance. Therefore, they are candidates
for being the best model.

30
The best model was chosen by comparing the AIC, BIC and the log-likelihood of the
two models. From the table, it can be seen that the GARCH(1,1) model had the least
AIC(−10655.0), the least BIC(−10632.7) and the highest log-likelihood(5331.48). There-
fore, GARCH(1,1) is the best model among all the models.
The GARCH(1,1) model is therefore, represented as,

σt2 = 0.0000062445 + 0.05Yt−1

2 2
+ 0.93σt−1 (4.3.2)

Table 4.3.5: GARCH model for Standard Chartered

MODEL COEF STD ERR P > |t| AICs BICs LOG-LIK

GARCH(0, 1) ω 1.5611e − 04 1.951e − 05 1.225e − 159 −10337.2 −10320.5 5171.60
α1 0.2000 5.954e − 02 7.819e − 04
ω 6.2445e − 06 3.085e − 11 0.000
GARCH(1, 1) α1 0.0500 2.412e − 02 3.822e − 02 −10655.0 −10632.7 5331.48
β1 0.9300 2.205e − 02 0.000
ω 3.1223e − 05 3.740e − 06 6.980e − 17
GARCH(2, 1) α1 0.1000 3.573e − 02 5.134e − 03 −10622.6 −10594.8 5316.28
β1 0.4400 0.472 0.397
β2 0.4000 0.479 0.404
ω 6.2445e − 06 6.757e − 12 0.000
α1 0.0250 1.005e − 02 1.283e − 02
GARCH(1, 2) α2 0.0250 4.377e − 02 0.568 −10587.7 −10559.9 5298.86
β1 0.9300 2.989e − 02 1.685e − 212
ω 3.1223e − 05 3.572e − 06 2.301e − 18
α1 0.0500 1.693e − 02 3.143e − 03
GARCH(2, 2) α2 0.0500 5.502e − 02 0.363 −10588.3 −10555.0 5300.15
β1 0.4000 0.479 0.403
β2 0.4000 0.473 0.398

31
Figure 4.3.2: Forecasted conditional variance for Standard Chartered for a period of 30
days using the GARCH(1,1) model

The GARCH(1,1) forecasted volatility for Standard Chartered, representing the

Ghanaian equity market, reveals an upward trend followed by stabilization over the fore-
cast period from June 2, 2025, to July 11, 2025. Volatility rises from approximately
0.0130 to 0.01673, reflecting a 29.1% increase. This indicates a short-term rise in market
uncertainty before reaching a steady state.
The initial surge in volatility may be driven by factors such as anticipated earnings re-
leases, domestic inflation concerns, currency fluctuations, or limited market liquidity.
These elements often contribute to heightened sensitivity in frontier markets like Ghana.
Despite the early increase, the convergence toward a stable volatility level suggests that
market participants expect conditions to normalize in the medium term. This pattern re-
flects a temporary phase of elevated risk, followed by greater predictability and potential
recovery in investor confidence.

Model Selection for Unilever

The results for modelling the Unilever data is shown in Table 4.7 below. From the ta-
ble, it can be seen that only GARCH(0,1) and GARCH(1,1) have all its parameters
being statistically significant at a 5% level of significance. Therefore, they are candi-

32
dates for being the best model. The best model was chosen by comparing the AIC,
BIC and the log-likelihood of the two models. From the table, it can be seen that
the GARCH(1,1) model had the least AIC(−9764.46), the least BIC(−9742.24) and the
highest log-likelihood(4886.23). Therefore, GARCH(1,1) is the best model among all the
models. The GARCH(1,1) model is therefore, represented as,

2
σt2 = 0.000019319 + 0.1Yt−1 2
+ 0.88σt−1 (4.3.3)

Table 4.3.6: GARCH model for Unilever

MODEL COEF STD ERR P > |t| AICs BICs LOG-LIK

GARCH(0, 1) ω 4.8292e − 04 2.510e − 04 5.434e − 02 −8881.67 −8865.00 4443.84
α1 0.2000 5.045e − 02 7.367e − 05
ω 1.9319e − 05 2.398e − 12 0.000
GARCH(1, 1) α1 0.1000 2.387e − 02 2.802e − 05 −9764.46 −9742.24 4886.23
β1 0.8800 4.136e − 02 1.901e − 100
ω 1.9317e − 05 1.494e − 12 0.000
GARCH(2, 1) α1 0.1000 3.290e − 02 2.371e − 03 −10001.8 −9974.02 5005.90
β1 0.4400 0.290 0.129
β2 0.4400 0.253 8.231e − 02
ω 1.9318e − 05 2.203e − 12 0.000
α1 0.0500 2.346e − 02 3.309e − 02
GARCH(1, 2) α2 0.0500 3.153e − 02 0.113 −9796.63 −9768.85 4903.31
β1 0.8800 4.024e − 02 5.455e − 106
ω 1.9318e − 05 1.533e − 12 0.000
α1 0.0500 2.880e − 02 8.256e − 02
GARCH(2, 2) α2 0.0500 4.246e − 0 0.239 −10024.5 −9991.18 5018.26
β1 0.4400 0.462 0.341
β2 0.4400 0.413 0.286

33
Figure 4.3.3: Forecasted conditional variance for Unilever for a period of 30 days using
the GARCH(1,1) model

The GARCH(1,1) forecasted volatility for Unilever Ghana, representing another

key stock within the Ghanaian equity market, displays a steady upward trend throughout
the forecast period from June 2, 2025, to July 11, 2025. Volatility rises from approxi-
mately 0.01715 to 0.02296, marking a 33.8% increase. This persistent climb suggests a
continued rise in market uncertainty with no signs of short-term stabilization.
The continuous growth in volatility may be attributed to factors such as economic policy
shifts, inflationary pressures, reduced investor confidence, or external shocks impacting
multinational firms operating in Ghana. As Unilever Ghana operates within both local
and global economic frameworks, fluctuations in exchange rates, import costs, and supply
chain dynamics may contribute to this sustained volatility.
Unlike Standard Chartered, which shows a plateauing behavior, the forecasted volatility
for Unilever Ghana does not converge within the analyzed window. This indicates that
investors may be bracing for ongoing risk and instability in the near term. The absence
of volatility stabilization highlights a less predictable market environment, emphasizing
the need for cautious investment decisions and robust risk management strategies.

Model Selection for Tullow Oil

The results for modelling the MSCI index data is shown in Table 4.8 below. From the

34
table, it can be seen that although GARCH(0,1) does not have the least AIC and BIC
values and also does not have the highest log-likelihood, it is the only model with all
its parameters being statistically significant at a 5% level of significance. Therefore,
GARCH(0,1) was chosen as the best model among all the models. The GARCH(0,1)
model is therefore, represented as,

2
σt2 = 0.000035943 + 0.009990Yt−1 (4.3.4)

Table 4.3.7: GARCH model for Tullow oil

MODEL COEF STD ERR P > |t| AICs BICs LOG-LIK

GARCH(0, 1) ω 3.5943e − 05 4.968e − 06 4.645e − 13 −12224.5 −12207.9 6115.27
α1 9.9990e − 03 4.885e − 03 4.067e − 02
ω 7.1750e − 06 8.621e − 13 0.000
GARCH(1, 1) α1 0.0500 7.545e − 02 0.508 −13120.5 −13098.2 6564.23
β1 0.8500 6.764e − 02 3.204e − 36
ω 7.1750e − 06 9.277e − 13 0.000
GARCH(2, 1) α1 0.0500 7.983e − 02 0.531 −13123.1 −13095.3 6566.56
β1 0.4250 0.474 0.370
β2 0.4250 0.402 0.291
ω 8.6814e − 06 1.099e − 12 0.000
α1 0.0271 1.556e − 02 8.186e − 02
GARCH(1, 2) α2 0.0271 8.697e − 02 0.756 −13132.4 −13104.6 6571.20
β1 0.8503 5.900e − 02 4.292e − 47
ω 7.2006e − 06 8.734e − 13 0.000
α1 0.0250 7.056e − 02 0.723
GARCH(2, 2) α2 0.0250 5.937e − 02 0.673 −13127.4 −13094.1 6569.72
β1 0.4250 1.187 0.720
β2 0.4250 1.102 0.700

35
Figure 4.3.4: Forecasted conditional variance for Tullow oil for a period of 30 days using
the GARCH(1,1) model

The GARCH(1,1) forecasted volatility for Tullow Oil, representing the Ghanaian
equity market, reveals a sharp decline followed by prolonged stabilization over the forecast
period from June 2, 2025, to July 11, 2025. Volatility drops steeply from approximately
0.0631 to 0.00539, marking a significant 91.5% decrease within the first few days of the
forecast window.
This early and rapid decline in volatility suggests the market may be reacting to the
resolution of recent uncertainties, such as fluctuations in global oil prices, operational
updates from the company, or broader macroeconomic developments. In commodity-
linked stocks like Tullow Oil, investor sentiment can shift quickly as new information is
absorbed.
The subsequent plateau in volatility at a low and consistent level indicates that investors
expect minimal surprises or disruptions in the near term. This sustained low volatility
suggests confidence in the stability of both company performance and broader market
conditions.
Overall, the forecast reflects a brief period of elevated risk quickly transitioning into a
phase of market calm and predictability, aligning with improved investor sentiment and
reduced uncertainty surrounding the firm or the energy sector.

36
 Chapter 5

Conclusion and Recommendations

5.1 Introduction
This chapter presents a comprehensive discussion and interpretation of the empirical re-
sults obtained in Chapter 4. It synthesizes the findings from the GARCH-based volatility
forecasting conducted on the MSCI Index and selected equities listed on the Ghana Stock
Exchange, namely Standard Chartered, Unilever Ghana, and Tullow Oil. The chapter
further evaluates how the results address the study’s objectives, considers their impli-
cations for investors and market stakeholders, and offers recommendations for future
research directions.

5.2 Summary of Results

The study employed various GARCH model specifications to estimate and forecast the
volatility of returns for both global and domestic equities over a 30-day horizon. The
selection of optimal models was informed by parameter significance and goodness-of-fit
criteria, including the Akaike Information Criterion (AIC), Bayesian Information Crite-
rion (BIC), and log-likelihood values.
The MSCI Index, serving as a proxy for the global financial market, demonstrated a con-
sistent downward trend in volatility, decreasing from 0.0331 to 0.0291 over the forecast
period representing a 12.03% decline. This trajectory suggests a relatively stable global
market environment with diminishing uncertainty.
Conversely, Standard Chartered exhibited an initial rise in volatility from 0.0130 to
0.01673 (a 29.1% increase), followed by stabilization. This pattern indicates transient
market uncertainty, potentially arising from domestic economic developments, which sub-
sequently converges toward a steady risk level.

37
Unilever Ghana displayed a continuous and unrelenting increase in forecasted volatil-
ity from 0.01715 to 0.02296 (a 33.8% increase), with no evidence of stabilization. This
persistent upward movement suggests sustained market uncertainty, possibly due to firm-
specific operational risks or broader macroeconomic volatility.
Tullow Oil experienced a sharp decline in volatility from 0.0631 to 0.00539 within the
initial days of the forecast horizon. There was a prolonged period of stability, indicating
the market’s swift assimilation of uncertainty and a subsequent return to equilibrium.
These heterogeneous volatility patterns underscore the distinct risk characteristics of the
global and Ghanaian markets, emphasizing the need for localized risk assessment frame-
works in frontier economies.

5.3 Conclusion
This section evaluates the study’s results in light of its stated objectives:
Our analysis supports the effectiveness of GARCH-type models, especially the GARCH(1,1)
and GARCH(0,1) specifications, in modeling volatility clusters and predicting conditional
variances. The varying suitability of different models for distinct assets highlights the
need for tailored model selection based on specific data characteristics. These models
not only accurately capture historical volatility patterns but also offer valuable forecasts,
confirming their predictive power in both developed and frontier markets.
The results highlight the utility of volatility forecasting as a strategic tool for market
participants and policymakers. For instance, the post-peak stabilization in Standard
Chartered and the sharp decline in Tullow Oil’s volatility signal emerging stability, which
can enhance investor confidence. Conversely, the prolonged volatility in Unilever Ghana
warns of persistent risk, advocating for informed investment decisions and policy inter-
ventions aimed at improving market transparency and efficiency.
A comparative assessment of the volatility forecasts reveals significant divergence in risk
profiles between the global market, (MSCI Index), and the selected equities from the
Ghanaian market—Standard Chartered, Unilever Ghana, and Tullow Oil.
The MSCI Index exhibited a moderate and smooth decline in volatility over the

38
forecast period, decreasing by approximately 12.03%. This behavior signals a low-risk
environment, suggesting increased investor confidence, which is supported by expec-
tations of macroeconomic stability, strong institutional frameworks, and well-developed
market mechanisms. The gradual volatility decay enhances predictability and sup-
ports strategic long-term asset allocation with reduced exposure to abrupt shocks.
In contrast, the Ghanaian equities demonstrate distinct and higher-risk profiles, with
varying implications for investor decision-making:
Standard Chartered experienced an initial surge in volatility (rising by 29.1%) followed
by a plateauing trend, indicating temporary market uncertainty. This risk pattern may
appeal to medium-term investors who are willing to tolerate short-term fluctuations in
anticipation of eventual stability. The risk implications here are moderate, requiring cau-
tious optimism and an emphasis on timing and policy monitoring.
Unilever Ghana showed a sustained and uninterrupted rise in volatility (a 33.8% increase),
with no evidence of convergence or stabilization throughout the forecast horizon. This
persistent elevation in risk levels suggests a highly uncertain market environment, poten-
tially reflective of structural weaknesses, firm-specific issues, or macroeconomic volatility.
For investors, this implies elevated risk exposure, warranting premium returns to com-
pensate for increased uncertainty and a need for robust risk mitigation strategies.
Tullow Oil demonstrated a sharp and early drop in volatility (a 91.5% decline), followed
by a prolonged phase of stability. This behavior suggests that market uncertainty, likely
associated with oil price volatility or operational developments, was rapidly resolved. The
resulting stable regime indicates a lower risk profile relative to the other Ghanaian equi-
ties and positions Tullow Oil as a potentially attractive short-to-medium-term investment
option for risk-averse investors operating in a frontier market context.
Overall, the findings highlight a clear risk gradient: from the relatively stable and low-risk
profile of the MSCI Index, to the transitional risk pattern of Standard Chartered, the
persistent risk exposure of Unilever Ghana, and the rapid risk normalization observed
in Tullow Oil. These varying risk profiles should inform investor strategies in terms of
portfolio diversification, risk-adjusted asset selection, and investment horizon alignment.
For institutional investors, the study underscores the necessity of tailoring investment

39
decisions to the unique volatility structures of each market, particularly when navigating
frontier markets characterized by limited liquidity and macroeconomic fragility.

5.4 Recommendations

Policy and Investment Recommendations

For Investors, investors seeking portfolio stability may consider exposure to global indices
such as the MSCI, particularly during periods of macroeconomic recovery. Ghanaian equi-
ties offer varying risk-reward trade-offs. Tullow Oil, post-stabilization, presents a low-risk
opportunity, while Unilever Ghana may necessitate risk premiums due to its persistent
volatility.
For Portfolio Managers, incorporating GARCH-based volatility forecasting into asset al-
location frameworks can improve portfolio resilience and strategic planning. The dif-
ferentiated behavior across equities highlights the importance of dynamic risk-adjusted
investment strategies
For Policymakers, measures to enhance market depth, liquidity, and transparency within
the Ghana Stock Exchange are imperative. Timely and standardized disclosure of fi-
nancial and macroeconomic information can reduce information asymmetry and dampen
volatility.

Future Research Directions

Future studies may explore asymmetric volatility models such as EGARCH or GJR-
GARCH to capture leverage effects prevalent in financial time series.
Integrating macroeconomic variables (e.g., interest rates and inflation) may offer deeper
insights into the determinants of volatility.
Expanding the scope to include other African frontier and emerging markets can provide
comparative insights into regional volatility behavior.
Assessing volatility over extended periods can better inform institutional investment de-
cisions and long-run risk assessments.

40
 References
1. Abdalla, S.Z.S. and Winker, P., 2012. Modelling stock market volatility using
univariate GARCH models: Evidence from Sudan and [Link] Journal
of Economics and Finance , 4(8), pp.161-176.

2. Adjasi, C., Harvey, S.K. and Agyapong, D.A., 2008. Effect of exchange rate volatil-
ity on the Ghana stock exchange. African journal of accounting, economics, finance
and banking research, 3(3).

3. Agarwal, S., 2017. Volatility in stock markets: A comparison of developed and

emerging markets of the world. Indian Journal of Commerce and Management
Studies, 8(2), pp.87-92.

4. Agyarko, K., Wiah, E.N., Frempong, N.K. and Odoi, B., 2023. Modelling the
volatility of the Ghana stock market: A comparative study. International Journal
of Statistics and Applied Mathematics, 8(3), pp.125-132.

5. Almısshal, B. and Emir, M., 2021. Modelling exchange rate volatility using GARCH
models. Gazi İktisat ve İşletme Dergisi, 7(1), pp.1-16.

6. Aslam, W., 2014. Relationship between stock market volatility and exchange rate:
a study of KSE. Journal of Public Administration, Finance and Law, (05), pp.62-72.

7. Babikir, A., Gupta, R., Mwabutwa, C. and Owusu-Sekyere, E., 2012. Structural
breaks and GARCH models of stock return volatility: The case of South Africa.
Economic Modelling, 29(6), pp.2435-2443.

8. Bhatta, S. and Duwal, B.R., 2021. A systematic review of dividend policy in rela-
tion to stock price volatility. The International Research Journal of Management
Science, 6(1), pp.92-104.

9. Biu, G.S. and Kusuma, P.K., 2023. Stock Market Volatility Analysis During the
Global Financial Crisis: Literature Review. Riwayat: Educational Journal of His-
tory and Humanities, 6(4), pp.2510-2520.

41
10. Bollerslev, T., 1986. Generalized autoregressive conditional [Link]
of econometrics,31(3), pp.307-327.

11. Chong, C.W., Ahmad, M.I. and Abdullah, M.Y., 1999. Performance of GARCH
models in forecasting stock market volatility. Journal of forecasting, 18(5), pp.333-
343

12. Crain, S.J. and Lee, J.H., 1996. Volatility in wheat spot and futures markets,
1950–1993: Government farm programs, seasonality, and [Link] journal of
finance,51(1), pp.325-343.

13. Dhingra, B., Batra, S., Aggarwal, V., Yadav, M. and Kumar, P., 2024. Stock
market volatility: a systematic review. Journal of Modelling in Management, 19(3),
pp.925-952.

14. Diamandis, P.F. and Drakos, A.A., 2011. Financial liberalization, exchange rates
and stock prices: exogenous shocks in four Latin America countries. Journal of
Policy Modeling, 33(3), pp.381-394.

15. Engle, Robert F. ”Autoregressive conditional heteroscedasticity with estimates of

the variance of United Kingdom inflation.”Econometrica: Journal of the economet-
ric society(1982): 987-1007.

16. Engle, R.F. and Mustafa, C., 1992. Implied ARCH models from options [Link]
of Econometrics,52(1-2), pp.289-311.

17. Glosten, L.R., Jagannathan, R. and Runkle, D.E., 1993. On the relation between
the expected value and the volatility of the nominal excess return on [Link]
journal of finance,48(5), pp.1779-1801.

18. Gokcan, S., 2000. Forecasting volatility of emerging stock markets: linear versus
non-linear GARCH models. Journal of forecasting, 19(6), pp.499-504.

19. González-Rivera, G., 1998. Smooth-transition GARCH [Link] in Nonlin-

ear Dynamics & Econometrics,3(2).

42
20. Hagerud, G.E., 1997.A new non-linear GARCH model. Stockholm School of Eco-
nomics.

21. Hamadu, D. and Ibiwoye, A., 2010. Modelling and forecasting the volatility of the
daily returns of Nigerian insurance stocks. International Business Research, 3(2),
pp.106-116.

22. Heston, S.L., 1993. A closed-form solution for options with stochastic volatility
with applications to bond and currency [Link] review of financial studies,6(2),
pp.327-343.

23. Hull, J. and White, A., 1987. The pricing of options on assets with stochastic
[Link] journal of finance,42(2), pp.281-300.

24. Hwang, S. and Satchell, S.E., 2005. GARCH model with cross-sectional volatility:
GARCHX models. Applied Financial Economics, 15(3), pp.203-216.

25. Kosapattarapim, C., Lin, Y.X. and McCrae, M., 2012. Evaluating the volatility
forecasting performance of best fitting GARCH models in emerging Asian stock
markets.

26. Lamoureux, C.G. and Lastrapes, W.D., 1993. Forecasting stock-return variance:
Toward an understanding of stochastic implied volatilities. The Review of Financial
Studies,6(2), pp.293-326.

27. Lanne, M. and Saikkonen, P., 2005. Non-linear GARCH models for highly persistent
volatility. The Econometrics Journal, 8(2), pp.251-276.

28. Li, Q., Yang, J., Hsiao, C. and Chang, Y.J., 2005. The relationship between stock
returns and volatility in international stock markets. Journal of Empirical Finance,
12(5), pp.650-665.

29. Lim, C.M. and Sek, S.K., 2013. Comparing the performances of GARCH-type
models in capturing the stock market volatility in Malaysia. Procedia Economics
and Finance, 5, pp.478-487.

43
30. Lin, Z., 2018. Modelling and forecasting the stock market volatility of SSE Com-
posite Index using GARCH models. Future Generation Computer Systems, 79,
pp.960-972.

31. Mala, R. and Reddy, M., 2007. Measuring stock market volatility in an emerging
economy. International research journal of finance and economics, 8, pp.126-133.

32. Mamtha, D. and Srinivasan, K.S., 2016. Stock market volatility–conceptual per-
spective through literature survey. Mediterranean Journal of Social Sciences, 7(1),
pp.208-212.

33. Mburu, D.M., 2015. Relationship between exchange rate volatility and stock market
Performance.

34. Mlambo, C., Maredza, A. and Sibanda, K., 2013. Effects of exchange rate volatility
on the stock market: A case study of South Africa (Doctoral dissertation, University
of Fort Hare).

35. Monfared, S.A. and Enke, D., 2014. Volatility forecasting using a hybrid GJR-
GARCH neural network model. Procedia Computer Science, 36, pp.246-253.

36. Nelson, D.B., 1991. Conditional heteroskedasticity in asset returns: A new ap-
[Link]: Journal of the econometric society, pp.347-370.

37. Omari-Sasu, A.Y., Frempong, N.K., Boateng, M.A. and Boadi, R.K., 2015. Model-
ing stock market volatility using GARCH approach on the Ghana stock exchange.
International Journal of Business and Management, 10(11), p.169.

38. Ou, P. and Wang, H., 2011, July. Modeling and forecasting stock market volatility
by Gaussian processes based on GARCH, EGARCH and GJR models. In Proceed-
ings of the world congress on engineering (Vol. 1, No. 1, pp. 1-5).

39. Poon, S.H. and Granger, C.W.J., 2003. Forecasting volatility in financial markets:
A review. Journal of economic literature, 41(2), pp.478-539.

44
40. Rashid Sabri, N., 2004. Stock return volatility and market crisis in emerging
economies. Review of Accounting and Finance, 3(3), pp.59-83.

41. Ritchken, P. and Trevor, R., 1999. Pricing options under generalized GARCH and
stochastic volatility processes. The Journal of Finance, 54(1), pp.377-402.

42. Stein, E.M. and Stein, J.C., 1991. Stock price distributions with stochastic volatil-
ity: an analytic [Link] review of financial studies,4(4), pp.727-752.

43. Yang, J., Haigh, M.S. and Leatham, D.J., 2001. Agricultural liberalization policy
and commodity price volatility: a GARCH application. Applied Economics Letters,
8(9), pp.593-598.

45