Upload
others
View
11
Download
0
Embed Size (px)
Citation preview
General Election and Political Uncertainty in the Malaysia Stock Market:
Evidence from Stock Market Returns and their Volatility
Ricky Chia Chee Jiun
Academic Advisor:
Professor Kobayashi Masahito
This dissertation has been submitted in partial fulfilment of the requirements for the Degree of Doctor of Philosophy in Economics
Graduate School of International Social Sciences
Yokohama National University
September 2018
ii
Copyright by
Ricky Chia Chee Jiun
2018
iii
ACKNOWLEDGEMENTS
With great pleasure, I would like to express my deep and sincere gratitude to my research
supervisor, Professor Kobayashi Masahito, who gave me his support, encouragement and
advice. Professor Kobayashi Masahito provided me great ideas through discussion, excellent
error checking, and added clear instructions to improve the quality of my research. The
research facilities and equipment provided by him and the faculty are acknowledged. It was a
great privilege and honor to study under Professor Kobayashi Masahito guidance.
Next, I would like to express my heartfelt gratitude to Graduate School of Social Sciences,
Yokohama National University, for giving me the opportunity to further my doctoral study
and providing all the facilities throughout my study. I also extend my special thanks to the
Japanese government, the Ministry of Education, Culture, Sports, Science and Technology
(MEXT) for offer me the Monbukagakusho scholarship throughout my study, which enable
me to focus more clearly and aggressively on my academics rather than my finances. Their
kindness and generosity are greatly appreciated. Besides, I would like to thanks my faculty,
Labuan Faculty of International Finance, Universiti Malaysia Sabah, for allowing my study
leave and supporting me to further my doctoral study in Yokohama National University.
Completion of this doctoral dissertation was not possible without the academic’s support of
the academic committee members (Professor Parsons Craig and Professor Suzuki Masataka),
lecturers, university officers, dissertation reviewers and friends. Their continuous
encouragement patience and forbearance have been valuable for the successful completion of
this research. I would like to express my sincere gratitude to all of them.
iv
Lastly, I would like to express my love and thanks to my family members, especially my
lovely wife and daughter for their understanding and encouragement in my study.
v
ABSTRACT
This thesis consists of three research papers that thoroughly examine the effect of the
Malaysian general elections on its stock market volatility from the year 1994 to 2015. The
link of the general election and stock market volatility is crucial for market participants. If the
market is subject to investor sentiment during the times of political election, investors who
got the right direction could make profitable trading around general election date. Malaysia
has a unique empirical setting in investigating the impact of political uncertainty on stock
market performance. In term of market behaviour, Malaysia is one of the Asia countries with
higher behavioural risk and higher returns compared to developed financial markets. In term
of political, Malaysia has been long enjoying stable political condition where the incumbent
won all the general elections with a two-thirds majority in the Parliament. However, during
the 12th and 13th Malaysian general elections, the incumbent faced challenges from increasing
pressure for electoral reform and lost the two-thirds majority in parliament which is never
happened in the political history of Malaysia. The presence of a political shock in the 12th and
13th Malaysia general election provides a great opportunity to investigate how investors react
to the market uncertainty. Chapter 1 presents an overview of Malaysian general election
which briefly describes the Parliamentary system in Malaysia and the condition of general
elections held in years 1995-2013.
This study contributes to the literature on a few grounds. First, the election window is
set in an extraordinary way which is in line with the Malaysian electoral process. In particular,
the pre-general election period is defined as the duration from the day of dissolution of the
parliament until the day before voting, while the post-general election period refers to the
duration from the day after voting until the first parliament assembly. Second, this study
examines the election effect in the Malaysian stock market level by level. The examination
vi
starts with the big picture by using seven benchmark indices, including the Shariah-compliant
indices, where each of the benchmark indices represents a different level of market
capitalization. Next, the examination is extended to the sectoral indices of the Malaysian
stock market. By breaking down into industry type, the findings illustrate the sensitivity of
each sector to the market condition during general elections. Lastly, the examination is
conducted at the firm level to complete the understanding of election effect on the stock
market. So far, the election effect at the industry level and firm level remain an unexplored
issue in the literature.
Third, in the asymmetric GARCH models, control variables of MSCI World Index
and MSCI Emerging Market Index are added into both the Exponential Generalized
Autoregressive Conditional Heteroskedasticity (EGARCH) and the Threshold Generalized
Autoregressive Conditional Heteroskedasticity (Threshold GARCH / GJR GARCH) model to
account for external effects. Moreover, this study also conducts an array of robustness checks
by considering the Chicago Board Options Exchange (CBOE) Volatility Index (VIX) as one
of the market uncertainty indicators for global risk, and controlling the US Federal Fund Rate
for interest rate differentials. Fourth, this study contributes to the literature by formally
investigating whether the high volatility is associated with high trading volume as suggested
by Admati and Pfleiderer (1988) or low trading volume as proposed by Foster and
Viswanathan (1990). The trading volume analysis is not included in previous studies in
Malaysia, nonetheless, it reveals an important trading pattern that investors could not miss.
Overall, the findings of this study indicate that the Malaysian stock market volatility
is associated with investors' behaviour during the period of the general elections. Chapter 2
shows the relevance of market capitalization to stock market volatility when there is political
vii
uncertainty surrounding elections. Companies with small capital experienced higher stock
volatility prior to general election. Conversely, the stock volatility is lower for larger
companies' stock. Furthermore, lower stock volatility is observed in Shariah-compliant stock
indices which suggest that Shariah-compliant companies have a lower risk during the pre-
general election periods. Further examination on sectoral indices is presented in Chapter 3.
The finding in Chapter 3 shows that volatility on the stock return is lower during the pre-
general election periods of 1994-2005, conversely, the stock volatility is higher in the pre-
general election periods of 2006-2015. Thus, the finding sheds light on the importance of
breaking down the full sample period into two sub-samples in order to address the difference
of political condition. Besides, the finding also indicates that the sectors of Construction,
Finance, Mining, and Property are more sensitive to the market condition with significant
result found in stock volatility, while Consumer Product is a defensive sector where the
estimated results are mostly insignificant.
Since evidence of election effect is found on the main stock indices and sectoral
indices, Chapter 4 further explores the reaction of stock returns and volatility in the firm level.
The finding in this chapter also shows that the pattern of the stock volatility in GLCs and
Non-GLCs is clearly different in the two sub-samples, and thus, lends support to the
observation in Chapter 3. As well, lower volatility of returns is found before the general
elections in years 1994-2005, for both the GLCs and Non-GLCs stock indices. In the general
election years of 2006-2015, most of the GLCs and Non-GLCs stock prices were highly
volatile before the general elections. Additionally, analysis on trading volume shows that the
high volatility found in the GLCs and Non-GLCs is associated with high trading volume.
Another interesting point found in the finance sector is that investors are still willing to
viii
actively trade the GLCs stock despite market uncertainties. Nonetheless, this trading pattern
appears only in the finance sector, but not in other sectors.
This study provides a complete understanding of election effects on the Malaysian
stock market. The market uncertainty induced by the political shock in the recent general
elections has changed the trading pattern in the market. Therefore, this study is of great
importance to market participants to understand the pattern of volatility in the Malaysian
stock market during general election years, and perhaps provide an insight for investors in
adjusting their portfolio around the next general election.
ix
TABLE OF CONTENTS
Page
Acknowledgements iii
Abstract v
Table of Contents ix
List of Tables xi
List of Figures xiv
List of Appendixes xv
Abbreviations xvi
1 General Introduction
1.1 Political Elections and Financial Markets 1
1.2 Overview of Malaysian General Election 5
1.3 Contributions of the Study 11
1.4 Contents and Organization 13
References 15
2 The Effect of Political Elections on Stock Market Volatility in Malaysia
2.1 Introduction 19
2.2 Literature Review 22
2.3 Data 27
2.4 Empirical Methodology 27
2.5 Empirical Results and Discussions 30
2.6 Conclusion 33
References 35
x
3 Stock Market Volatility in Malaysia Sectoral Indices during the General Election
3.1 Introduction 42
3.2 Literature Review 47
3.3 Data and Empirical Methodology 52
3.4 Empirical Results and Discussions 56
3.5 Extensions and Robustness 61
3.6 Conclusion 67
References 71
4 General Election and Stock Market Volatility in Malaysia: Evidence from GLCs and Non-GLCs Stock Performance
4.1 Introduction 101
4.2 Literature Review 106
4.3 Data and Empirical Methodology 113
4.4 Empirical Results and Discussions 118
4.5 Conclusion 128
References 132
5 General Conclusion 164
References 168
xi
LIST OF TABLES
Pages
Table 1.0: Malaysia General Elections Result from 1995 – 2013 18
Table 2.1: Malaysia General Election 37
Table 2.2: Descriptive Statistics for the Malaysian Stock Indices 37
Table 2.3: Summary Statistics for the Returns on Pre- and Post-General Election 38
Table 2.4: Pre-General Election and Post-General Election: EGARCH Results Controlled by World Market Effect
39
Table 2.5: Pre-General Election and Post-General Election: EGARCH Results Controlled by Emerging Market Effect
40
Table 3.1: Malaysia General Election Information 75
Table 3.2: Descriptive Statistics for the Malaysian Sectoral Indices (1994 - 2015) 75
Table 3.3: Mean Returns on Pre-General Election and Post-General Election 76
Table 3.4(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) - Controlled by World Market Effect
77
Table 3.4(b): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) - Controlled by World Market Effect
78
Table 3.5(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2005) - Controlled by World Market Effect
79
Table 3.5(b): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2005) - Controlled by World Market Effect
80
Table 3.6(a): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) - Controlled by World Market Effect
81
Table 3.6(b): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) - Controlled by World Market Effect
82
Table 3.7(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) - Controlled by Emerging Market Effect
83
Table 3.7(b): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) - Controlled by Emerging Market Effect
84
Table 3.8(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2005) - Controlled by Emerging Market Effect
85
xii
Table 3.8(b): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2005) - Controlled by Emerging Market Effect
86
Table 3.9(a): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) - Controlled by Emerging Market Effect
87
Table 3.9(b): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) - Controlled by Emerging Market Effect
88
Table 4.1: Malaysia General Election 135
Table 4.2: Descriptive Statistics for GLCs Stock Return (1994 – 2015) 136
Table 4.3: Descriptive Statistics for Non-GLCs Stock Return (1994 – 2015) 136
Table 4.4: Sub-Sample Mean Return for GLCs 137
Table 4.5: Sub-Sample Mean Return for Non-GLCs 137
Table 4.6: Selected Sample of Malaysian Government-Link Companies (GLCs) 138
Table 4.7: Selected Sample of Malaysian Non-Government-Link Companies (Non-GLCs)
139
Table 4.8(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) – Government Link Companies (GLCs)
140
Table 4.8(b): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) – Government Link Companies (GLCs)
141
Table 4.9(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) – Non-Government Link Companies (Non-GLCs)
142
Table 4.9(b): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) – Non-Government Link Companies (Non-GLCs)
143
Table 4.10(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2005) – Government Link Companies (GLCs)
144
Table 4.10(b): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2005) – Government Link Companies (GLCs)
145
Table 4.11(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2005) – Non-Government Link Companies (Non-GLCs)
146
Table 4.11(b): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2005) – Non-Government Link Companies (Non-GLCs)
147
Table 4.12(a): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) – Government Link Companies (GLCs)
148
xiii
Table 4.12(b): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) – Government Link Companies (GLCs)
149
Table 4.13(a): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) – Non-Government Link Companies (Non-GLCs)
150
Table 4.13(b): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) – Non-Government Link Companies (Non-GLCs)
151
Table 4.14(a): Threshold GARCH Results for Trading Volume during Pre- and Post-General Election (1994 - 2005) – Government Link Companies (GLCs)
152
Table 4.14(b): Threshold GARCH Results for Trading Volume during Pre- and Post-General Election (1994 - 2005) – Government Link Companies (GLCs)
153
Table 4.15(a): Threshold GARCH Results for Trading Volume during Pre- and Post-General Election (1994 - 2005) – Non-Government Link Companies (Non-GLCs)
154
Table 4.15(b): Threshold GARCH Results for Trading Volume during Pre- and Post-General Election (1994 - 2005) – Non-Government Link Companies (Non-GLCs)
155
Table 4.16(a): Threshold GARCH Results for Trading Volume during Pre- and Post-General Election (2006 - 2015) – Government Link Companies (GLCs)
156
Table 4.16(b): Threshold GARCH Results for Trading Volume during Pre- and Post-General Election (2006 - 2015) – Government Link Companies (GLCs)
157
Table 4.17(a): Threshold GARCH Results for Trading Volume during Pre- and Post-General Election (2006 - 2015) – Non-Government Link Companies (Non-GLCs)
158
Table 4.17(b): Threshold GARCH Results for Trading Volume during Pre- and Post-General Election (2006 - 2015) – Non-Government Link Companies (Non-GLCs)
159
xiv
LIST OF FIGURES
Pages
Figure 2.1: Volatility during the Pre-General Election for the Selected Stock Indices
41
Figure 4.1: Relationship between Trading Volume and Stock Price Volatility for GLCs and Non-GLCs from 1994 – 2005
160
Figure 4.2: Relationship between Trading Volume and Stock Price Volatility for GLCs and Non-GLCs from 2006 – 2015
161
Figure 4.3: Trading Volume and Stock Prices for the Highly Traded GLCs before the General Elections
162
Figure 4.4: Trading Volume and Stock Prices for the Highly Traded Non-GLCs before the General Elections
163
xv
LIST OF APPENDIXES
Pages
Appendix 2.1: Details of the Selected Indices in this Study 41
Appendix 3.1(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) - Controlled by Volatility Index (VIX)
89
Appendix 3.1(b): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) - Controlled by Volatility Index (VIX)
90
Appendix 3.2(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2005) - Controlled by Volatility Index (VIX)
91
Appendix 3.2(b): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2005) - Controlled by Volatility Index (VIX)
92
Appendix 3.3(a): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) - Controlled by Volatility Index (VIX)
93
Appendix 3.3(b): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) - Controlled by Volatility Index (VIX)
94
Appendix 3.4(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) - Controlled by Federal Fund Rate
95
Appendix 3.4(b): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) - Controlled by Federal Fund Rate
96
Appendix 3.5(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2005) - Controlled by Federal Fund Rate
97
Appendix 3.5(b): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2005) - Controlled by Federal Fund Rate
98
Appendix 3.6(a): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) - Controlled by Federal Fund Rate
99
Appendix 3.6(b): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) - Controlled by Federal Fund Rate
100
xvi
ABBREVIATIONS
ASEAN Association of Southeast Asian Nations
BERSIH The Coalition for Clean and Fair Elections
BN Barisan National
CBOE Chicago Board Options Exchange
DAP Democratic Action Party
EGARCH Exponential Generalized Autoregressive Conditional Heteroskedasticity
GARCH Generalized Autoregressive Conditional Heteroskedasticity
Gerakan Gerakan Rakyat Malaysia
GLCs Government-Linked Companies
KLCI Kuala Lumpur Composite Index
MCA Malaysian Chinese Association
MIC Malaysian Indian Congress
MENA Middle East and North Africa
MSCI Morgan Stanley Capital International
NGOs Non-Governmental Organisations
Non-GLCs Non-Government-Linked Companies
OECD The Organisation for Economic Co-operation and Development
PAS Pan-Malaysian Islamic Party
PBB Parti Pesaka Bumiputera Bersatu
PKR People’s Justice Party
SUPP Sarawak United Peoples' Party
TGARCH Threshold Generalized Autoregressive Conditional Heteroskedasticity
UMNO United Malays National Organization
VIX Volatility Index
1
CHAPTER 1
GENERAL INTRODUCTION
1.1 Political Elections and Financial Markets The history of the stock market suggests that stock index is one of the most sensitive
indicators of business cycle. In the annals of cyclical analysis, the most popular cycle is the
four years Presidential Election Cycle (Wong and McAleer, 2009). The existence of the
Presidential Election Cycle in stock market is simply due to investors' sentiment. In
behavioral finance, investor sentiment is broadly defined as a belief about future cash flows
and investment risks that is not justified by the facts at hand (Baker and Wurgler, 2007).
Specifically, investor sentiment on political election could exhibit optimism or pessimism
which induce underreaction or overreaction in the market. Whenever investors are optimistic
about the future of the economy, they are more inclined to invest in stock market. On the
contrary, whenever investors feel unsecured with the future or policy of the country, they are
more likely to withdraw from the market (Bialkowski, Gottachalk, & Wisniewski, 2008). The
response of investor, thus, causes changes in trading volume, volatility and stock prices in the
market. (Tuyon et al., 2016).
In investment practice, investor sentiment plays an important role on the stock market
activity because stock prices are affected by both the fundamental and behavioural forces
(Akerlof and Shiller, 2009). During the times of political elections, negative news could
destabilize investor trading, thereby creating disarray and possible panic in the markets.
Specifically, in the pre-election periods, election campaign rhetoric may cause bounded
rationality in market players. On the other hand, post-election shock can be caused by several
factors such as a narrow margin of victory, lack of compulsory voting laws, change in the
2
political orientation of the government, or the failure to form a government with
parliamentary majority. Evident from empirical study found support that investors are
affected by sentiment in their investment decision making (Statman, 2008). Furthermore,
Tuyon et al. (2016) also highlighted that investor sentiment risks could influence stock prices
regardless of size and industry groups. Therefore, it is crucial for market participants to know
whether the market is subject to investor sentiment during the times of political election.
Empirical literature on the Presidential Election Cycle dates back to Allivine and
O’Neill (1980). The four years pattern started with a relatively weak performance in the stock
market during the first year of a presidency. Gradually, the stock market's performance
improves in the second and three years. Lastly, the stock market tends to prosper in the
Presidential election year. Nordhaus (1975) explained the causes of the Presidential Election
Cycle through the predictable pattern of the government policies during a term of office. In
order to gain voters support and win the election, the decision made by incumbent political
parties tends to stimulate the economy prior to elections. Hence, political factors are
conjectured to influence the economy through government policies, which affect the timing
and severity of the business conditions.
In developed countries, election effect in the stock market is shown by a number of
studies, among them are Allivine and O’Neill (1980) and Huang (1985), and Gemmill (1992).
Generally, evidence from the U.S. stock markets showed that the market made larger gains in
the third and fourth years of a presidential term. In Britain general election, Gemmill (1992)
also found an extremely close relationship between opinion polls and the FTSE 100 share
index during the 1987 election. Beside that, Pantzalis et al. (2000) conducted a study on an
international scale which covered stock market indices across 33 countries around political
3
election dates during the sample period 1974 - 1995. They found a positive abnormal return
in two weeks prior to the election week.
When the existence of abnormal return is confirmed during the election period,
researchers started to consider the magnitude of market volatility. The study of Białkowski et
al. (2008) aimed to test whether national elections induce higher stock market volatility in a
sample of 27 OECD countries. The finding from the GARCH model indicated that the
country-specific component of index return variance can easily double during the week
around an election. By using a single country testing case, Wang and Lin (2009) also found
that democratically presidential elections negatively impacted stock returns and induced
higher volatility in the Taiwanese stock market. Lean and Yeap (2017) also found similar
pattern in Malaysian stock market where the key index of FTSE Bursa Malaysia KLCI
experienced significant volatility during the general election years.
The evidence found in previous studies is mostly based on the examination of main
composite indices which provided a big picture about the stock market. Nonetheless, stock
market information from the top to the bottom is precious for investor. Sector-specific
information could be useful for investors to narrow down their investments option in the
financial market. Moreover, volatility of the sectoral stock may evolve differently from the
composite indices. The existence of election effect in firm level is equally important. Study
on listed company index could be useful for investors because the individual firm may react
differently to election effect due to the nature of the business industry. However, the aspect of
the influence of political events on the movement of sectoral indices and firm index has not
been thoroughly explored in the literature. Hence, a comprehensive analysis of stock market
4
performance by breaking down into smaller segments should be conducted in order to gain
insight into the sectoral market as well as individual firm index.
Among the emerging markets, Asian markets are interesting in examining the
influence of election on the stock market performance. Although Asia suffers from higher
risk of behavioural biases than other developed markets (Ritter, 2003; Schmeling, 2009), the
markets are still attractive to investors because of their relatively higher returns compared to
developed financial markets (Kearney, 2012). Malaysian stock market is quite a developed
capital market among Asian markets (Mohamad et al. 2007). The Bursa Malaysia has
steadily emerged as one of the top-performing markets in Asia. Its capitalisation has reached
USD 382 billion in December 2015 and the market ranked the second highest in ASEAN
markets after the Singapore Exchange. Moreover, Malaysia has a unique empirical setting in
investigating the impact of political uncertainty on stock market performance. Since the
independence of Malaysia, the country has been enjoying stable political condition where the
incumbent won all the general elections with a two-thirds majority in the Parliament.
However, during the 12th and 13th Malaysian general elections, the incumbent faced
challenges from increasing pressure for electoral reform. Also, in these two general elections,
the incumbent lost the two-thirds majority in parliament which is never happened in the
political history of Malaysia. Without the two-thirds majority in parliament, the incumbent
may face difficulty in amending the constitution in the new cabinet. The unexpected election
outcomes induced an unusual high spike in the stock market as investors were worried about
the danger of unrest and instability.
Hence, the Malaysian stock market is chosen as a single country testing case to see
the influence of political events on the movement of stock prices during the years 1994-2015.
5
The Exponential Generalized Autoregressive Conditional Heteroskedasticity (EGARCH)
model developed by Nelson (1991) and the Threshold Generalized Autoregressive
Conditional Heteroskedasticity (Threshold GARCH / GJR GARCH) model developed by
Glosten et al. (1993) suit the objective of this study in examining the time-varying of stock
volatility.
The remainder of the chapter is organized as follows. The next section presents an
overview of Malaysian general election which briefly describes the Parliamentary system in
Malaysia and the condition of general election held in years 1995-2013. The subsequent
section discusses the contribution of this study. The last section gives a brief description of
each chapter.
1.2 Overview of Malaysian General Election
Malaysia is a country with a Parliamentary system where the Parliament of Malaysia is the
national legislature of Malaysia. The Malaysian electoral system follows the British
parliamentary, as opposed to the US presidential system. The country is geographically
divided into constituencies. Each constituency is contested by candidates who stand either as
nominees of political coalitions or as independents. The elected candidate will become the
Member of Parliament of the constituency and sit in the Lower House of Parliament (the
Dewan Rakyat). As in all parliamentary systems, the leader of the political coalition with a
majority of seats in parliament or a state legislative assembly would be appointed as the
Prime Minister to form the government.
The legislature at the state level is called the State Legislative Assembly. Each of the
13 states has a State Legislative Assembly (Dewan Undangan Negeri), while the federal
6
parliament has two houses, namely the Senate (Dewan Negara) and the House of
Representatives (Dewan Rakyat). The Senate consists of 70 senators of whom 26 are elected
by the state legislative assemblies, with two senators for each state, while the rest are
appointed by the King (Yang di-Pertuan Agong). The Senate, which represents the interests
of the various states, reviews legislation that has been passed by the lower house. The
Parliament, or a State Legislature, has a term of five years, unless dissolved sooner. Thus,
elections are held for the Parliament and the state legislatures at intervals not exceeding five
years (Lim, 2002). When the Parliament is dissolved, a general election shall be held within
two months in West Malaysia and three months for East Malaysia (Sabah and Sarawak) from
the date of dissolution. Within this period, the Election Commission fixes the nomination
day, followed by a formal period of campaigning before the polling day.
Since the first election of the Federal Legislative Council of the Malaya in 1955, the
Alliance Party or Parti Perikatan coalition is the incumbent ruling coalition, and from 1973
onwards, its successor, the Barisan Nasional (BN) coalition. The coalition of BN consists of
more than ten component parties and the three main component parties are racially based
parties. They are the United Malays National Organization (UMNO) represents the Malay
which is the largest ethnic group in Malaysia, the Malaysian Chinese Association (MCA)
represents the Chinese, and the Malaysian Indian Congress (MIC) represents the Indians. The
regional parties from Sabah and Sarawak, namely Sabah Alliance Party, Sarawak United
Peoples' Party (SUPP), and Parti Pesaka Bumiputera Bersatu (PBB) also joined the BN to
form a grand coalition.
Malaysia has an interesting political background for the testing of election effect. It is
therefore essential to begin with a brief summary of the general elections in Malaysia during
7
the sample period from 1995 to 2015. This period covers five general elections, the 9th to the
13th Malaysian general election. Although BN coalition won all the five general elections, the
general elections were accompanied by a certain level of political uncertainty. Table 1.0
summarizes the percentage of votes and seats won by BN and the opposition for the five
general elections.
In early 1990s, Malaysia has a booming economy with nearly zero unemployment,
low inflation and superior organization and finance under the lead of the fourth Prime
Minister, Mahathir Mohamad. In conjunction with the strong economy, the United Malays
National Organisation (UMNO) has grown stronger and united under Mahathir's leadership.
Given the buoyant economy and stable political condition, Malaysia went to the 9th general
election on April 25, 1995. The election was gone through without a hitch and BN recorded
the greatest victory in the election history. The BN won 162 of the 192 parliamentary seats
and secured an the highest percentage of votes of 65.2%. The victory of BN coalition
demonstrated the strengths of Mahathir's administration and the economic boom. The BN
coalition successfully gained support not only from the rural Malay voters, but most
importantly from the Chinese in urban centres. The 9th general elections provided a clear
mandate for the BN coalition to form the government with multiethnic support from urban as
well as rural constituencies (Moten and Mokhtar, 1995).
Two years after the 1997 Asian financial crisis, the 10th general election held on
November 29, 1999, witnessed the tremendous change of voters' behaviour due to dramatic
political and economic changes in Malaysia. The two important events that influenced the
election was the dismissal of Anwar Ibrahim and the 1997 financial crisis. The issue of
Anwar Ibrahim dominated the election campaigns and manipulated to gain voters support.
8
The 1997 financial crisis also caused Malaysia in an unstable political climate. The opposing
views in handling the crisis had struggled both the leaders to continue leading the country
together. Mahathir bravely strategized the economy against the norm in order to protect
Malaysia. Its effect was clearly reflected in the 10th general election when the BN coalition
managed to secure a two-third majority in parliament but with fewer majority seats (Lin,
1999). This election was mostly a two-bloc antagonism. The ruling BN coalition obtained
56.5% of the votes with a drop of 8.7% from the previous general election. The opposition
gained 43.5% of the votes, mostly contributed by the Pan-Malaysian Islamic Party (PAS).
. In year 2004, the Prime Minister, Abdullah Badawi, led the BN coalition to contest in
the 11th general election. This election was the first election led by Abdullah Badawi.
Abdullah had come into office with a distinctly different political persona with the previous
prime minister. His Islamic credentials gained him much mileage among the Muslim voters. He
also gained public support by the promise of an anticorruption platform. The rural Malay
electorate also supported Abdullah administration as he put much closer attention to
agriculture and rural development. and a moderate, progressive version of Islam (Liow,
2004). Given the Abdullah factor, the 11th general election held on March 21, 2004 was the
greatest electoral victory for the ruling BN in the history of Malaysian electoral politics. The
BN coalition garnered a total of 199 of the 219 seats (91%) and limited the opposition to 20
seats. The opposition PAS leader, Abdul Hadi Awang, lost his parliamentary seat in his own
state of Terengganu. The failure of the PAS coalition was largely attributed to the party’s
inability to convince the moderate Muslim and non-Muslim electorate. The non-religious
Democratic Action Party (DAP) was the only winner among the opposition parties by
securing 12 seats in the Parliament.
9
The 12th general election in Malaysia was the second election held under the Prime
Minister, Abdullah Ahmad Badawi. The nomination day is set by the Election Commission
on February 24, 2008, and the general election is set on March 8, 2008. The campaign period
of 13 days was the longest in Malaysian electoral history since 1969. The opposition parties,
DAP, PAS and the People’s Justice Party (PKR), formed an alliance to contest against BN.
The election was conducted with a low expectation for a repeat of the fruitful result of the
11th general election. The environment was surrounded by worries over rising oil prices in the
global market, the increase in consumer price index, perceptions of ethnic inequality
especially among the Indian ethnic group, concerns over the independence of the judiciary,
and a revived opposition under the leadership of former Deputy Prime Minister, Anwar
Ibrahim. Besides, electoral reform has been demanded and officially started with the forming
of The Coalition for Clean and Fair Elections or BERSIH by non-governmental organisations
(NGOs) in November 2006. The first ever mass rally called by BERSIH happened before the
12th general election, and ended with detention of a group of more than two hundred people.
The 12th general election were held on March 8, 2008 and the outcome was a great
shock to BN coalition. It is quoted as ‘Political Tsunami’ because the result was the worst
performance ever for BN coalition. BN won 140 of the 222 seats in the federal parliament,
where 55 of the 57 seats in East Malaysia, and 85 of the 165 seats in the peninsula.
Obviously, the BN’s margin of victory was helped by its performance in East Malaysia.
Overall, BN won just 51.4% of the votes and 63% of parliamentary seats. It was the first time
the BN lost the two-thirds majority in parliament which might affect the new cabinet in
amending the constitution in future. The failure to secure a two-thirds majority was partly due
to the poor performance of BN’s non-Malay components, the Malaysian Indian Congress
(MIC), Malaysian Chinese Association (MCA), and Gerakan Rakyat Malaysia (Malaysian
10
People Movement, Gerakan), in securing the votes of the non-Malays in the peninsula. The
electoral outcomes also imply that BN has significantly lost support among the non-Malay
constituencies, due in large part of its failure to address the economic factors such as rising
fuel and consumer prices and the issue of ethnic inequality among the Chinese and Indian
ethnic minorities (Mokhtar, 2008)
The 13th general election in 2013 is quoted as the most tensely contested general
election in the political history of Malaysia. The complexity of the 13th general election
begun as soon as the 12th general election was over. The gains of the PR coalition in the 12th
general election raised worries to the BN coalition, about the strength of PR and the ability of
BN to retain power in the 13th general election, and the changes on the political system if the
election result were to be a very close between the two coalitions. Moreover, in the five years
period, the most spectacular event were the mass demonstrations in 2011 and 2012 called by
BERSIH 2.0 and BERSIH 3.0 (Coalition for Clean and Fair Elections). The demonstrations
were joined by small political coalitions, non-governmental organizations, and mainly
supported by the opposition coalitions. The demonstrations demanded for a free and fair
electoral process and also called for voters to show up in large numbers to negate illegal
voting by non-citizens.
On May 5, 2013, Malaysia went to the 13th general election with the highest record of
voter turnout in the history of Malaysia. The 13th general election was the first election held
under the Prime Minister, Najib Razak. Despite the uncertainty, BN won against the
opposition and formed the government at the federal level. BN won 133 of the 222 seats
while the opposition won 89 seats. This was the best performance shown by the opposition
coalition, but the worst for BN coalition with only 47.4% of the popular vote. The result of
11
the 13th general election altered the trends of earlier general elections. The success of the
opposition coalition contributed to the progress towards two-coalition system in Malaysia
(Khoo, 2013).
1.3 Contributions of the Study
The examination of asymmetric effect in the Malaysian stock market around elections
contributes to the literature on a few grounds. First, the election window is precisely set in
accord with the Malaysian electoral process in order to capture the full impact of the general
election. The selection of the election window in this study is different with previous studies
that focused on the event day (Wang and Lin, 2009) or fix event windows before and after the
election, for example, 1 week, 2 weeks and 1 month (Nippani and Arize, 2005; Chuang and
Wang, 2010; Lean and Yeap, 2017). In this study, the pre-general election period is defined
as the duration from the day of dissolution of the parliament until the day before voting,
while the post-general election period refers to the duration from the day after voting until the
first parliament assembly. Furthermore, to test whether the pattern of the stock volatility
changes according to the political condition, the full sample period is divided into early years
(1994-2005), and later years (2006-2015). The year 2006 has been chosen as the cut-off date
because the 12th general election (2008) and the 13th general election (2013) induced election
uncertainty with a strong expectation of political changeover and eventually the incumbent
lost its two-thirds majority in the Parliament. Thus, the first sub-sample period covers the
general election years of 1995, 1999 and 2005 where general ups and downs happened in the
market during general elections, while the second sub-sample covers the general election
years of 2008 and 2013 where the market experiences drastic ups and downs.
12
Second, this study examines the election effect in the Malaysian stock market by
using the top-down approach. The examination starts with the big picture by using seven
benchmark indices, including the Shariah-compliant indices. Each of the benchmark indices
represents a different level of market capitalization and the findings show the pattern of stock
movement during election periods in each level of market capitalization. The selection of data
provides a clearer picture than previous studies (Lean, 2010; Lean & Yeap, 2017) that only
focused on the FTSE KLCI index. Next, the examination breaks down into smaller segments
by investigating the election effect on ten sectoral indices of the Malaysian stock market. By
breaking down into industry type, the findings illustrate the sensitivity of each sector to the
market condition during general elections. Lastly, the examination is reduced to the base
elements by conducting analysis on the firm level to complete the understanding of election
effect on the stock market. So far, this remains an unexplored issue in the literature.
Therefore, this study attempts to uncover evidence of election effect in firm-level by selecting
GLC firms and non-GLC firms as the sample.
Third, in the asymmetric GARCH models, control variables are added into both the
Exponential Generalized Autoregressive Conditional Heteroskedasticity (EGARCH) and the
Threshold Generalized Autoregressive Conditional Heteroskedasticity (Threshold GARCH /
GJR GARCH) model to account for external effects. Specifically, the MSCI World Index is
to capture the global market effect to Malaysian stock market, and the MSCI Emerging
Market Index are included to control for emerging market effect. Moreover, this study also
conducts an array of robustness checks by considering the Chicago Board Options Exchange
(CBOE) Volatility Index (VIX) as one of the market uncertainty indicator for global risk, and
controlling the US Federal Fund Rate for interest rate differentials.
13
High volatility in stock returns is always a concern for market participants. Fourth,
this study contributes to the literature by formally investigating whether the high volatility is
associated with high trading volume as suggested by Admati and Pfleiderer (1988) or low
trading volume as proposed by Foster and Viswanathan (1990). The trading volume analysis
is not included in previous studies in Malaysia, nonetheless, it reveals an important trading
pattern that investors could not miss.
Overall, this study may be of interest to investors as the results will reveal the
Malaysian stock market information, from the top level of benchmark indices to the middle
level of sectoral indices, and further to the base level of firm indices. This precious
information is useful for investors to construct an effective equity portfolio investment,
especially during the times of election.
1.4 Contents and Organization
The rest of the thesis is organized as follows:
Chapter 2 examines the election effect on FTSE Bursa Malaysia KLCI Index and
selected main stock indices in the Malaysian stock market that represent large, medium, and
small market capitalization, including the Shariah-compliant indices. The scope of the
examination only covers the 12th and 13th general election, which are the most two recent
general elections with turbulent political condition. This chapter shows the relevance of
market capitalization to stock market volatility when there is political uncertainty
surrounding elections.
Chapter 3 attempts to identify the influence of general elections on the movement of
ten selected sectoral indices in the Malaysian stock market. Further examination on sector-
14
specific information is useful for investors to narrow down their investments option. This
chapter sheds light on the importance of addressing the difference of political condition when
testing for asymmetry effect during election periods.
Since evidence of election effect is found on the main stock indices and sectoral
indices, Chapter 4 further explores the reaction of stock returns and volatility in the firm
level. Additionally, analysis on trading volume is performed to see whether the observed high
volatilities are associated with low or high trading volume. This chapter highlights that the
high volatility found in the GLCs and Non-GLCs is associated with high trading volume,
which lend support to the argument of Admati and Pfleiderer (1988).
Finally, Chapter 5 provides a general conclusion on the election effect found in the
Malaysian stock market.
15
References
Admati, A. and Pfleiderer, P. (1988). “A theory of intraday patterns: Volume and price variability”, Review of Financial Studies, Vol. 1, pp. 3 – 40. Akerlof, G. A. and Shiller, R. J. (2009). “Animal spirits: How human psychology drives the economy, and why it matters for global capitalism”, New Jersey: Princeton University Press. Allivine, F. D. and O’Neil, D. D. (1980). “Stock Market Returns and the Presidential Election Cycle”, Financial Analysts Journal, Vol. 36, No. 5, pp. 49 - 56. Baker, M. and Wurgler, J. (2007). “Investor Sentiment in the Stock Market”, Journal of Economic Perspectives, Vol. 21, No. 2, pp. 129 – 151. Bialkowski, J., Gottachalk, K. and Wisniewski, T. P. (2008). “Stock Market Volatility around National Elections”, Journal of Banking and Finance, Vol. 32, No. 9, pp. 1941 - 1953. Chuang, C. and Wang, Y. (2010). “Electoral Information in Develop Stock Market: Testing Conditional Heteroscedasticity in the Market Model”, Applied Economics, Vol. 42, No. 9, pp. 1125-1131. Foster, F. D. and Viswanathan, S. (1990). “A theory of the interday variations in volume, variances, and trading cost in securities market”, Review of Financial Studies, Vol. 3, pp. 593 – 624. Gemmill, G. (1992). “Political Risk and Market Efficiency: Tests Based in British Stock and Options Markets in the 1987 Election”, Journal of Banking and Finance, Vol. 16, No. 1, pp. 211 - 231. Glosten, L. R., Jagannathan, R. and Runkle, D. E. (1993). “Relationship between the expected value and the volatility of the nominal excess return on stocks”, The Journal of Finance, Vol. 48, No. 5, pp. 1779 - 1801. Huang, R. D. (1985). “Common Stock Returns and Presidential Elections”, Financial Analysts Journal, Vol. 41, No. 2, pp. 58 - 65. Kearney, C. (2012). “Emerging markets research: Trends, issues and future directions”, Emerging Markets Review, Vol. 13, pp. 159 - 183. Khoo, B. T. (2013). “13th General Election in Malaysia: Overview and Summary”, in Khoo, B. T. (Ed.), 13th General Election in Malaysia: Issues, Outcomes and Implications, Institute of Developing Economies (IDE-JETRO) Interim Report 2013, pp. 1 – 8. Retrieved from http://www.ide.go.jp/English/Publish/Download/Report/2013/2013_malaysia.html. Lean, H. H. (2010). “Political General Election and Stock Performance: The Malaysian Evidence”, in Ismail, M. T. and Mustafa, A. (Ed.), Research in Mathematics and Economics, Penang: Universiti Sains Malaysia, Malaysia, pp. 111 - 120. Lean, H. H. and Yeap, G. P. (2017). “Asymmetric Effect of Political Elections on Stock
16
Returns and Volatility in Malaysia”, in Munir, Q. and Kok, S. C. (Ed.), Information Efficiency and Anomalies in Asian Equity Markets, Routledge, Taylor and Francis Group, pp. 228 - 245. Lim, H. H. (2002). “Electoral Politics in Malaysia: ‘Managing’ Elections in a Plural Society”, in Croissant, A. et al. (Ed.), Electoral Politics in Southeast and East Asia, Friedrich Ebert Stiftung, Singapore, pp. 101 – 148. Lin, J. Y. (2002). “A Structural Analysis of the 1999 Malaysian General Election: Changing Voting Preference of Ethnic Chinese and Malay Groups and Party”, Taipei: Tamkang Journal of International Affairs, Vol. 6, No. 3. Retrieved from https://www.brookings.edu/articles/a-structural-analysis-of-the-1999-malaysian-general-election-changing-voting-preference-of-ethnic-chinese-and-malay-groups-and-party/. Liow, J. C. Y. (2004). “A Brief Analysis of Malaysia’s Eleventh General Election”, UNISCI Discussion Papers. Retrieved from https://www.ucm.es/data/cont/media/www/pag-72535/Liow.pdf. Mohamad, S., Hassan T. and Ariff, M. (2007). “Research in an emerging Malaysian capital market: A guide to future direction”, International Journal of Economics and Management, Vol. 1, No. 2, pp. 173 – 202. Mokhtar, T. M. (2008). “The Twelfth General Elections in Malaysia”, Intellectual Discourse, Vol. 16, No. 1, pp. 89 – 100. Moten, A. R. and Mokhtar, T. M. (1995). “The 1995 Parliamentary Elections in Malaysia”, Intellectual Discourse, Vol. 3, No. 1, pp. 77 – 93. Nelson, D. (1991). “Conditional Heteroskedasticity in Asset Returns: A New Approach”, Econometrica, Vol. 59, pp. 347-370. Nippani, S. and Arize, A. C. (2005). “U.S. Presidential Election Impact on Canadian and Mexican Stock Markets”, Journal of Economics and Finance, Vol. 29, No. 2, pp. 271 - 279. Nordhaus, W. (1975). “The Political Business Cycle”, Review of Economic Studies, Vol. 42, No. 2, pp. 169 - 190. Pantzalis, C., Stangeland, D. A. and Turtle, H. J. (2000). “Political Elections and the Resolution of Uncertainty: The International Evidence”, Journal of Banking and Finance, Vol. 24, pp. 1575 – 1604. Ritter, J. R. (2003). “Behavioral finance”, Pacific-Basin Finance Journal, Vol. 11, pp. 429 - 437. Schmeling, M. (2009). “Investor sentiment and stock returns: Some international evidence”, Journal of Empirical Finance, Vol. 16, No. 3, pp. 394 - 408. Statman, M., Fisher, K. L. and Anginer, D. (2008). “Affect in a behavioural asset-pricing model”, Financial Analysts Journal, Vol. 64, No. 2, pp. 20 - 29.
17
Tuyon, J., Ahamd, Z. and Matahir, H. (2016). “The role of investor sentiment in malaysian stock market”, Asian Academy of Management Journal of Accounting and Finance, Vol. 12 Suppl. 1, pp. 43 - 75. Wang, Y. H. and Lin, C. T. (2009). “The Political Uncertainty and Stock Market Behavior in Emerging Democracy: The Case of Taiwan”, Quality and Quantity, Vol. 43, No. 2, pp. 237 - 248. Wong, W. K. and McAleer, M. (2009). “Mapping the Presidential Election Cycle in U.S. Stock Market”, Mathematics and Computers in Simulation, Vol. 79, No. 11, pp. 3267 - 3277.
18
Table 1.0: Malaysia General Elections Result from 1995 – 2013
Election Year
Alliance / Barisan Nasional / National Front All Opposition Parties Total No. of Seats
Contested
No. of seat won % Total Vote
% Seats
No. of seat won % Total Vote
% Seats UMNO MCA MIC Others Total
Seat PAS PKR DAP Others Total Seat
1995 89 30 7 36 162 65.2 84 7 n.a. 9 14 30 34.8 16 192 1999 71 29 7 41 148 56.5 77 27 5 10 3 45 43.5 23 193 2004 109 31 9 49 198 63.8 91 7 1 12 1 21* 36.2 9 219 2008 79 15 3 43 140 51.4 63 23 31 28 0 82 48.6 37 222 2013 88 7 4 34 133 47.4 60 21 30 38 0 89 50.8 40 222
Note: *Figure includes one independent candidate. UMNO: United Malays National Organisation, MCA: Malaysian Chinese Organisation, MIC: Malaysian Indian Congress, PAS: Islamic Party of Malaysia, PKR: People’s Justice Party, and DAP: Democratic Action Party. Sources: Suruhanjaya Pilihan Raya, Election Report, various years.
19
CHAPTER 2 THE EFFECT OF POLITICAL ELECTIONS ON STOCK MARKET VOLATILITY
IN MALAYSIA 2.1 Introduction
History suggests that the stock market has a role as one of the most sensitive indicators of the
business cycle and the most popular cycle is the four years Presidential Election Cycle (Wong
and McAleer, 2009). The four years pattern started with a relatively weak performance in the
stock market during the first year of a presidency. Gradually, the stock market's performance
improves in the second and three years. Lastly, the stock market tends to prosper in the
Presidential election year. Nordhaus (1975) explained the causes of the Presidential Election
Cycle through the predictable pattern of the government policies during a term of office. In
order to gain voters support and win the election, the decision made by incumbent political
parties tends to stimulate the economy prior to elections. Hence, political factors are
conjectured to influence the economy through government policies, which affect the timing
and severity of the business conditions.
Moreover, investors' expectation also explains the political election effect on stock
market performance (Bialkowski et al., 2008). Whenever investors are optimistic about the
future of the economy, they are more inclined to invest in stock market. On the contrary,
whenever investors feel unsecured with the future or policy of the country, they are more
likely to withdraw from the market. Specifically, in the pre-election periods, general public
and investors may be affected by the election campaign rhetoric and the promises made by
the candidates, which can cause dramatic changes in stock prices. On the other hand, post-
election shock can be caused by several factors such as a narrow margin of victory, lack of
compulsory voting laws, change in the political orientation of the government, or the failure
20
to form a government with parliamentary majority significantly. Hence, investors’ sentiment
around election could induce under-reaction or over-reaction in the market and consequently
influence changes in trading volume, volatility, prices and accordingly determine stock
returns (Tuyon et al., 2016).
In the literature, there has been a constant stream of work analyzing the impact of the
political factor on stock market performance and the empirical evidence suggests that stock
market is significantly affected by the election. In developed countries, election effect in the
stock market is shown by number of studies, among them are Peel and Pope (1983), Gemmill
(1992), Lobo (1999), Nippani and Arize (2005), and Wong and McAleer (2009). However,
not many studies have been done in the emerging market, except for Wang and Lin (2009) on
Taiwanese stock market, Lean (2010) and Lean and Yeap (2017) on Malaysian stock market.
Actually, emerging market like Asian market is an interesting case study to investigate the
influence of election on the stock market performance. Asian are more socially collective in
decision-making (Kim and Nofsinger, 2008), collectively, political events such as national
election could cause investors to react irrationally.
Among the Asian countries, Malaysia has a unique empirical setting in investigating the
impact of political uncertainty on stock market performance. Since the independence of
Malaysia, the country has been enjoying stable political condition where the incumbent won
all the general elections with a two-thirds majority in the Parliament. However, during the
12th and 13th Malaysian general elections, the incumbent faced challenges from increasing
pressure for electoral reform. Also, in these two general elections, the incumbent lost the two-
thirds majority in parliament which is never happened in the political history of Malaysia.
Without the two-thirds majority in parliament, the incumbent may face difficulty in amending
21
the constitution in the new cabinet. The unexpected election outcomes induced an unusual
high spike in the stock market as investors were worried about the danger of unrest and
instability. Nonetheless, the two exogenous events provide a great opportunity to investigate
the impact of general elections on the stock market in Malaysia.
Hence, the Malaysian stock market is chosen as a single country testing case to see the
influence of political events on the movement of stock prices. From the perspective of
statistical analysis, selection of a single country is preferred to eliminate the heterogeneous
effect of multiple country characteristics such as differences in economics, political,
institutional, demographics and culture (Bekaert and Harvey, 2002; Statman et al., 2008).
Moreover, when performing standard time series analysis, breaking the series into similar
condition is recommended to avoid erroneous inferences. Therefore, this study only focuses
on the drastic shock periods during the 12th and 13th general elections because the market
condition during these two general elections is clearly different with earlier general elections.
The aim of this study is to examine the asymmetric effect of Malaysian general elections held
in the year 2008 and 2013 on the performance of Malaysian stock market. The Exponential
Generalized Autoregressive Conditional Heteroskedasticity (EGARCH) model developed by
Nelson (1991) suits the objective of this study in examining the time-varying of stock
volatility. Furthermore, the data is segregated into pre-general election period and post-
general election period. The pre-general election period starts from the day of dissolution of
the Parliament until the day before voting, while the post-general election period starts from
the day after voting until the first Parliament assembly. The selected time frames include all
important events before and after the general election which possibly affect the investors'
confidence in making their investment decision.
22
The examination of stock volatility around elections in this paper contributes to the
literature on a few grounds. First, the election window in this study is designed in accord with
the Malaysian electoral process, which is different with previous studies that focused on the
event day (Wang and Lin, 2009) or fix event windows (Nippani and Arize, 2005; Chuang and
Wang, 2010; Lean and Yeap, 2017). The pre-general election and post-general election
sample period are set according to the important date of the election in order to capture the
full impact of the general election. Second, this study examines seven benchmark indices in
Malaysia, including Shariah-compliant indices, to determine the impact of the election on
stock indices in different market capitalization. This approach enables us to relate the stock
volatility with market capitalization during the general election. Third, the MSCI World
Index and MSCI Emerging Market Index are included to control for global and emerging
market effect. This study has significant implication for investors as the findings can be of
interest to adjust their portfolio during the general election.
This paper is organized as follows. The next section presents the review of the impact
of political events on stock markets. Section 2.2 presents the data and Section 2.3 presents the
methodology and preliminary analysis. Next, Section 2.4 reports the estimated results and
Section 2.5 provides a brief conclusion.
2.2 Literature Review
The pattern in stock market prices related to the four years of Presidential term has been the
focus of study for some time. The empirical study of stock market behaviour and the US
elections was initiated by Niederhoffer et al. (1970). Further studies by Allivine and O’Neill
(1980) and Huang (1985) also supported the theory of presidential election cycle, in which
23
the US stock market has a weak performance in the first year of a presidential term and then
turn to a larger gain in the third and fourth years. On the other hand, the study by Jones and
Banning (2009) did not support the theory. They found little relationship between stock
market performance and US elections and the market returns do not appear to vary based on
partisan control of the government from 1896 through 2000. Nevertheless, recent study by
Wong and McAleer (2009) applying spectral analysis reconfirmed that US stock prices
closely followed the four-year presidential election cycle and the cyclical trend existed for the
last ten administrations from 1965 through 2003, particularly when the incumbent is a
Republican.
A notable study by Leblang and Mukherjee (2005) provided a deep understanding on
how political information affects market participants. By examining the US and British
equity markets between 1930 and 2000, they found statistical supports for their model which
predict that rational expectations of higher (lower) inflation under left-wing (right-wing)
administrations leads to lower (higher) trading volume of stocks, in turn, leads to a decrease
(increase) in the mean and volatility of stock prices. The finding also showed that the model
not only applicable on government partisanship, but also traders’ expectations of electoral
victory. Thus, their study explained the sensitivity of stock prices to elections and partisan
politics.
Beside the Presidential election cycle in the US, other studies have been done in other
countries to show abnormal return around election dates. The study of Pantzalis et al. (2000)
covered stock indices across 33 countries including major OECD countries of the US, UK,
and emerging countries. Their analysis showed that politics does matter in emerging markets
by comparing the mean and median around election with a comparison period mean.
24
Abnormal returns were detected during the two-week period prior to the election week.
Furthermore, there is also evidence that one-time occurrence of political event impacted
financial markets. Studies conducted by Nippani and Medlin (2002) and Nippani and Arize
(2005) examined the impact of the delay in the 2000 presidential election results on stock
market performance. Conventional heteroskedastic t-tests and binary variable regression were
applied to show that the delay had negatively impacted stock prices. The negative impact not
only appeared in the US stock market, but also the Mexican and Canadian stock markets. The
findings also implied that elections are wide watched events by local and foreign market
participants and the regional contagion risk of election shock to international markets.
Earlier studies mentioned above showed that national elections have significant
impact on stock return by comparing mean returns. Without considering the magnitude of
market volatility, the evidences are quite limited and arguable. In the study of Mei and Guo
(2004), increased market volatility was discovered during political election and transition
periods across 22 countries. Furthermore, over the sample period, financial crises mostly
happened during the periods of political election and transition and led to the conclusion that
political uncertainty could be a major contributing factor to financial crisis. After that, the
linkage between national elections and stock market volatility was formally investigated by
Białkowski et al. (2008). The study covered a sample of 27 OECD countries where majority
of the countries operate are under the parliamentary systems and aimed to test whether
national elections induce higher stock market volatility. The finding from the GARCH model
indicated that the country-specific component of index return variance can easily double
during the week around an election.
25
After the study of Białkowski et al. (2008), academic studies started to center in
testing the relationship between political events and stock market volatility by employing the
GARCH model (Wang and Lin, 2009; Chau et al., 2014; among others). Wang and Lin
(2009) found that democratically presidential elections negatively impacted stock returns and
induced higher volatility in the Taiwanese stock market. Unlike others, Chau et al. (2014)
studied the impact of political uncertainty caused by Arab Spring which started in December
2010 on conventional and Islamic stock market indices in the MENA region. Despite
heterogeneous reaction of the conventional and Islamic stock indices to that political turmoil,
only significant increase in the volatility of Islamic indices is detected during the period of
June 2009 to June 2012, whereas there is no significant effect on the volatility in
conventional markets.
Malaysia is an ideal test case because it is a country in which political situation and
market development have attracted recent theoretical interest. The Malaysian stock market
has steadily emerged as one of the top-performing markets in Asia. Its capitalisation has
reached USD 382 billion in December 2015 and it ranked the second highest in ASEAN
market after the Singapore exchange. In term of political situation, Malaysia is well-know as
a politically stable country. However, in term of electoral process, Malaysia has been quoted
with a long history of vote buying, vote stealing, and campaign media blitzes (Pepinsky,
2007). Electoral reform has been demanded and officially started with the forming of The
Coalition for Clean and Fair Elections or BERSIH by non-governmental organisations
(NGOs) in November 2006. This issue marked the start of Malaysian political tense. Due to
the significance of the Malaysian stock market, political stability will be an important factor
for local and foreign market participant in adjusting portfolio distribution.
26
From the literature of Malaysia stock market, only few studies had been done in investigating
the impact of general election on stock market volatility. A related study by Ali et al. (2010)
revealed that Malaysian stock market was affected by political events. They found significant
over-reaction behaviour existed in the Malaysian market upon announcement of the removal
of the deputy prime minister and announcement of the resignation of the prime minister. In
contrast, evidence of under-reaction was detected upon announcement of the national
election. Furthermore, Lean (2010) showed that general election in Malaysia significantly
affected the stock market performance. The stock returns reacted positively before election
and negatively after election. However, both studies did not focus on the possible impact of
general election on stock volatility. The study of Lean and Yeap (2017) circumvented the
limitation of previous studies and examined stock volatility during election periods. They
found significant election effect in stock volatility but not in the stock returns. Lean and Yeap
(2017) covered six general elections in their study, where the sample period is from the 8th to
13th general election. However, the market condition during the 12th and 13th general election
are clearly different with previous general election. The significant break point in the 12th
general election may have an effect on the inference of the analysis.
From the perspective of policy makers or government, this study might provide some
explanation about the reaction of the market in return and volatility. Policy makers and
government can use the information from this study to analyze the determinants that cause
the volatility of the stock market performance. Various policies either in fiscal or monetary
can be implemented to improve and stabilize the stock market in Malaysia. For investors or
investment institutions, they can use the information to get a better understanding on the
effect of general election. This will be increasing their awareness toward the government
action on policies and be able for them to adjust portfolio accordingly. For fund managers,
27
they can use this result to analyze the relationship of government policy on the stock return in
either a positive effect or negative way. With this, they can manage fund in a more effective
and efficient way.
2.3 Data
This study uses daily closing values of seven selected indices in Bursa Malaysia, namely
FTSE Bursa Malaysia Hijrah Shariah Index, FTSE Bursa Malaysia KLCI Index, FTSE Bursa
Malaysia Top 100 Index, FTSE Bursa Malaysia EMAS Shariah Index, FTSE Bursa Malaysia
EMAS Index, FTSE Bursa Malaysia Mid 70 Index and FTSE Bursa Malaysia Small Cap
Index1. The sample period covers the 12th and 13th Malaysian general election (21 May 2007
to 31 December 2015), with a total of 2,248 observations. All data are collected from Bursa
Malaysia (http://www.bursamalaysia.com). Table 2.1 shows the date of dissolution of
Parliament, election date or voting date and the date of 1st Parliament assembly after the
election for the 12th and 13th Malaysian general election.
[Insert Table 2.1: Malaysia General Election]
2.4 Empirical Methodology
Daily returns are calculated as the first difference in the natural logarithms of the stock
market index, )/ln(100 1 ttt IIR where tI and 1tI are the values for each index for
1 See Appendix 2.1: Details of the selected indices in this study.
28
periods t and 1t , respectively. In the case of a day following a non-trading day, the return
is calculated using the closing price of the latest trading day.
For an overview of the sample period, Table 2.2 presents the descriptive statistics of
daily stock returns for the selected stock indices. All the stock indices have a positive average
return over the sample period. The standard deviation and kurtosis are all positive, while the
skewness for all the series is negative. The null hypothesis of normally distributed daily
returns is rejected by the highly significant Jarque-Bera normality test. This finding is in line
with most of the previous findings which found that stock return series is non-normally
distributed. In addition, Table 2.3 shows the summary statistics of daily stock returns on the
pre-general election and post-general election periods. Interestingly, the mean returns for the
indices are all negative prior general election, while positive mean returns are recorded after
the general election.
[Insert Table 2.2: Descriptive Statistics for the Malaysian Stock Indices]
[Insert Table 2.3: Summary Statistics for the Returns on Pre- and Post-General
Election]
The Exponential Generalized Autoregressive Conditional Heteroscedastic (EGARCH)
(p, q)2 model with dummy variables is applied to examine the general election effect and its
volatility. The mean equation and variance equation of the Exponential GARCH model are
expressed as:
2 According to Bolerslev et al. (1992), in testing the GARCH models, p = q = 1 is sufficient for most financial series.
29
tttttt RWMRPtGEPGER 1113210 (1)
t
q
it
it
iti
it
itijt
p
jjt PtGEPGE 2
11
2
10
2 2loglog
(2)
tttttt REMRPtGEPGER 1113210 (3)
t
q
it
it
iti
it
itijt
p
jjt PtGEPGE 2
11
2
10
2 2loglog
(4)
where tR is the logarithmic return of the market index at day t ; tPGE and tPtGE are dummy
variables; t is the error term. tPGE takes a value of one if the corresponding return for the
day t is pre-general election period, while tPtGE takes a value of one if the corresponding
return for the day t is post-general election period, and 0 otherwise. Meanwhile, in the mean
equations of Equation (1) and Equation (3), the 30 ,..., are parameters to be estimated.
Among them, 0 measures the mean return (in percentage) on other trading days; whereas
1 and 2 capture the average return of the stock index for the pre-general election period
and post-general election period. The null hypothesis of this test is 0: 210 H , which
implies that average daily returns (volatility) for the period of pre-general election and post-
general election have no different. If the null hypothesis does not hold, then it can be
concluded that the market index is characterized by statistically different on average returns
(volatility) for the period of pre-general election and post-general election. In another word,
this would imply that general election effect is indeed present in the market.
30
For the Equation (2) and (4), the left-hand side of the equations is the logarithm of the
conditional variance. This implies that the leverage effect is exponential, rather than
quadratic, and that forecasts of the conditional variance are guaranteed to be non-negative. In
this case, the presence of leverage effects can be tested by the hypothesis that i > 0, whereas
the impact is asymmetric if 0i . Furthermore, a lagged value of the return variable was
introduced in the equations to avoid serial correlation error terms in the model, which may
yield misleading inferences.
Besides, the return variables for MSCI World Index ( 1tRWM ) and MSCI Emerging
Market Index ( 1tREM ) are introduced into the mean equation. For the Equation (1), return
variable of MSCI World Index ( 1tRWM ) is added to examine whether the returns during the
election are associated with the MSCI World Index returns. While return variable of MSCI
Emerging Market Index ( 1tREM ) is added to Equation (3) as the control variables for
emerging market effect. Both the MSCI World Index and MSCI Emerging Market Index are
obtained from S&P Capital IQ. If the parameter of 1 is insignificant, then it can be
concluded that the returns during general election are not influenced by the MSCI World
Index ( 1tRWM ) and MSCI Emerging Market Index ( 1tREM ) returns.
2.5 Empirical Results and Discussions
Firstly, we examine the presence of pre-general election and post-general election effect in
the series of FTSE Bursa Malaysia Index by controlling the global effect. Table 2.4 reports
the estimation results of the mean equation and variance equation of the EGARCH (1, 1)
model based on Equation (1) and (2). Under the mean equation, the dummy coefficients of
the pre-general election are positive only for BMT100, BMKLCI and Shariah-compliant
31
stock indices (BMES and BMHS). Conversely, the pre-general election dummy coefficients
for the other stock indices are all negative. On the other hand, for post-general election, the
dummy coefficients are all positive. Nevertheless, the high p-value of dummy coefficient
indicates insignificant stock returns among all the series of indices, for both the pre-general
election and post-general election periods. Thus, there is no significant impact of general
elections on average stock market returns in Malaysia. This finding with an insignificant
abnormal return around election period is consistent with the studies of Lean (2010) and Lean
& Yeap (2017).
The estimation of the variance equation is presented in the second part of Table 2.4.
The results show that the Malaysian stock market encountered significant higher volatility in
pre-general election periods during the 12th and 13th General Election. This finding is evident
by the positive and highly significant pre-general election dummy coefficients for all the
stock indices model estimation. Meanwhile, the post-general election dummy coefficients in
the variance equation are all negative. Particularly, only the FTSE Bursa Malaysia Small Cap
Index showed insignificant negative volatility during post-general election.
By controlling the emerging market effect, the results of the mean equation and
variance equation of the EGARCH (1, 1) model based on Equation (3) and (4) are presented
in Table 2.5. The estimated results are similar to the first model which controlled for global
market effect. In term of control variables, the dummy coefficients of the MSCI World Index
and MSCI Emerging Index for the mean equations, as shown in Table 2.4 and Table 2.5, are
positive and significant at 1% for all the series of FTSE Bursa Malaysia Index. The positive
sign indicates that the Malaysian stock market returns are positively affected by an increase
in the return in global market and emerging market.
32
The asymmetric effect of the general election is reported in Table 2.4 and Table 2.5.
The significant asymmetry coefficient ( i ) strongly supports the asymmetric effect in most
of the indices. Moreover, the negative sign of the asymmetry coefficient means that volatility
decreases more when returns shocks are positive. Besides, the validity of the model is
supported by the diagnostic test with no remaining ARCH effect and serial correlation in all
of the estimated models.
[Insert Table 2.4: Pre-General Election and Post-General Election: EGARCH Results
Controlled by World Market Effect]
[Insert Table 2.5: Pre-General Election and Post-General Election: EGARCH Results
Controlled by Emerging Market Effect]
Overall, the examination of Malaysian stock market performance by large, medium and small
companies’ capitalization enable us to observe the impact of general election more precisely.
Moreover, we also examine the general election effect on Shariah-compliant stocks. As
shown in Figure 2.1, the index of the FTSE Bursa Malaysia Hijrah Shariah has the lowest
volatility for pre-general election, followed by the FTSE Bursa Malaysia KLCI Index and the
FTSE Bursa Malaysia Top 100 Index. Meanwhile, the FTSE Bursa Malaysia Small
Capitalization Index has the highest volatility during the pre-general election periods. The
result indicates that, in term of volatility, companies stock with larger market capitalization
and fulfilling the Shariah-compliant requirement are less affected by the election shock prior
general election.
33
[Insert Figure 2.1: Volatility during the Pre-General Election for the Selected Stock
Indices]
2.6 Conclusion
This study investigates the effect of the Malaysian general elections on its stock market
volatility from the year 2007 to 2015. Using the EGARCH model, we find significant
election effect in stock volatility but not in stock returns. Specifically, the stock volatility for
all selected stock indices is significantly higher during pre-general election periods but only
six stock indices recorded lower stock volatility in the post-general election periods. Notably,
political uncertainty due to the close fight between major parties during the 2008 and 2013
general election had a significant role in influencing the stock volatility prior to the election.
Furthermore, this study also finds that Shariah-compliant indices have lower stock volatility
compare to other indices.
The value added of this paper is we provide a detailed examination of Malaysian stock
market performance around general election by dividing into large, mid, small cap, and
Shariah-compliant indices. The findings show the relevance of market capitalization to stock
market volatility. Companies with small capital experienced higher stock volatility prior to
general election. The stock market volatility is indeed lower for larger companies stock. We
also observe relatively lower stock volatility in Shariah-compliant indices which suggest that
Shariah-compliant companies stock have a lower risk during pre-general election periods.
This study contributed to the evidence of general election influences stock market volatility
and made an effort to investigate the general election effect on stock indices with different
size of market capitalization. The implications of this study for investors are important. Risk-
34
averse investors could mitigate the political risk by diversifying their portfolio in large
companies stock and Shariah-compliant companies stock. Furthermore, an investor should be
vigilant during pre-general election periods as their profits are underlying high volatility and
compensation for abnormal high returns is negligible.
35
References Ali, N., Nassir, A. M., Hassan, T. and Abidin, S. Z. (2010). “Short Run Stock Overreaction: Evidence from Bursa Malaysia”, International Journal of Economics and Management, Vol. 4, pp. 319 – 333. Allivine, F. D. and O’Neil, D. D. (1980). “Stock Market Returns and the Presidential Election Cycle”, Financial Analysts Journal, Vol. 36, pp. 49 - 56. Bekaert, G. and Harvey, C. R. (2002). “Research in emerging markets finance: Looking to the future”, Emerging Markets Review, Vol. 3, pp. 429 - 448. Bialkowski, J., Gottachalk, K. and Wisniewski, T. (2008). “Stock Market Volatility around National Elections”, Journal of Banking and Finance, Vol. 32, pp. 1941-1953. Bolerslev, T., Chou, R. and Kroner, K. (1992). “ARCH Modelling in Finance: A Selective Review of the Theory and Empirical Evidence”, Journal of Econometrics, Vol. 52, pp. 5-59. Chau, F., Deesomsak, R. and Wang, J. (2014). “Political Uncertainty and Stock Market Volatility in the Middle East and North African (MENA) Countries”, Journal of International Financial Markets, Institutions & Money, Vol. 28, pp. 1 – 19. Chuang, C. and Wang, Y. (2010). “Electoral Information in Develop Stock Market: Testing Conditional Heteroscedasticity in the Market Model”, Applied Economics, Vol. 42, No. 9, pp. 1125-1131. Gemmill, G. (1992). “Political Risk and Market Efficiency: Tests based in British Stock and Options Markets in the 1987 Election”, Journal of Banking and Finance, Vol. 16, pp. 211-231. Huang, R. D. (1985). “Common Stock Returns and Presidential Elections”, Financial Analysts Journal, Vol. 41, pp. 58 - 65. Jones, S. T. and Banning, K. (2009). “US Elections and Monthly Stock Market Returns”, Journal of Economics and Finance, Vol. 33, pp. 273 - 287. Kim, K. A. and Nofsinger, J. R. (2008). “Behavioral finance in Asia”, Pacific-Basin Finance Journal, Vol. 16, pp. 1 - 7. Leblang, D. and Mukherjee, B. (2005). “Government Partisanship, Elections and the Stock Market: Examining Americana and British Stock Returns 1930-2000”, American Journal of Political Science, Vol. 49, pp. 780 – 802. Lean, H. H. (2010). “PoliticalGeneral Election and Stock Perfomance: The Malaysia Evidence”, In M. Ismail, & A. Mustafa, Research in Mathematics and Economics (pp.111 -120). Penang: Universiti Sains Malaysia.
36
Lean, H. and Yeap, G. (2017). “Asymmetric Effect of Political Elections on Stock Returns and Volatility in Malaysia”, In Q. Munir, & S. C. Kok, Information Efficiency and Anomalies in Asian Equity Markets (pp. 228-245). UK: Routledge, Taylor and Francis Group. Lobo, B. (1999). “Jump Risk in the US Stock Market: Evidence using Political Information”, Review of Financial Economics, Vol. 8, pp. 149-163. Mei, J. P. and Guo, L. M. (2004). “Political Uncertainty, Financial Crisis and Market Volatility”, European Financial Management, Vol. 10, pp. 639 – 657. Nelson, D. (1991). “Conditional Heteroskedasticity in Asset Returns: A New Approach”, Econometrica, Vol. 59, pp. 347-370. Niederhoffer, V., Gibbs, S. and Bullock, J. (1970). “Presidential Elections and the Stock Market”, Financial Analysts Journal, Vol. 26, pp. 111 - 113. Nippani, S. and Arize, A. (2005). “US Presidential Election Impact on Cadian and Mexican Stock Markets”, Journal of Economics and Finance, Vol. 29, pp. 271-279. Nippani, S. and Medlin, W. B. (2002). “The 2000 Presidential Election and the Stock Market”, Journal of Economics and Finance, Vol. 26 No. 2, pp. 162 - 169. Nordhaus, W. (1975). “The Political Business Cycle”, Review of Economic Studies, Vol. 42, pp. 169-190. Pantzalis, C., Stangeland, D. A. and Turtle, H. J. (2000). “Political Elections and the Resolution of Uncertainty: The International Evidence”, Journal of Banking and Finance, Vol. 24, pp. 1575 - 1604. Peel, D. and Pope, P. (1983). “General Election in the UK in the post-1950 period and the Behavior of the Stock Market”, Investment Analysis, Vol. 67, pp. 4-10. Pepinsky, T. (2007). “Autocracy, Elections and Fiscal Policy: Evidence from Malaysia”, Studies in Comparative International Development, Vol. 42, No. 1, pp. 136 - 163. Statman, M., Fisher, K. L. and Anginer, D. (2008). “Affect in a behavioural asset-pricing model”, Financial Analysts Journal, Vol. 64, No. 2, pp. 20 - 29. Tuyon, J., Ahamd, Z. and Matahir, H. (2016). “The role of investor sentiment in Malaysian stock market”, Asian Academy of Management Journal of Accounting and Finance, Vol. 12, Suppl. 1, pp. 43 - 75. Wang, Y. H. and Lin, C. T. (2009). “The Political Uncertainty and Stock Market Behavior in Emerging Democracy: The Case of Taiwan”, Quality and Quantity, Vol. 43, pp. 237-248. Wong, W. and McAleer, M. (2009). “Mapping the Presidential Election Cycle in US Stock Market”, Mathematics and Computers in Simulation, Vol. 79, pp. 3267-3277.
37
Table 2.1: Malaysia General Election
Dissolution of
Parliament Election Day 1st Parliament Assembly after Election
12th General Election
13 February 2008 (Wednesday)
8 March 2008 (Saturday)
28 April 2008 (Monday)
13th General Election
3 April 2013 (Wednesday)
5 May 2013 (Sunday)
24 June 2013 (Monday)
Source: Authors' compilation based on information from Election Commission of Malaysia and Parliament of Malaysia websites.
Table 2.2: Descriptive Statistics for the Malaysian Stock Indices
BMEMAS BMT100 BMM70 BMKLCI BMSC BMES BMHS Mean 0.0119 0.0115 0.0135 0.0102 0.0188 0.0142 0.0191 Max 4.4184 4.1961 5.2661 4.2587 6.7322 4.0747 4.5368 Min -9.9494 -10.0817 -9.9045 -9.9785 -9.0170 -11.3205 -11.0873 Std. Dev. 0.7769 0.7746 0.8634 0.7622 1.0333 0.8061 0.8245 Skewness -1.1855 -1.1571 -1.1950 -1.1581 -0.8067 -1.5094 -1.1982 Kurtosis 18.8855 19.4337 15.7521 19.2995 11.2360 23.8308 21.7728 Jarque- Bera 24163.18 25797.76 15766.73 25387.37 6597.39 41497.55 33547.62
Probability 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Note: BMEMAS: FTSE Bursa Malaysia EMAS Index, BMT100: FTSE Bursa Malaysia Top 100 Index, BMM70: FTSE Bursa Malaysia Mid 70 Index, BMKLCI: FTSE Bursa Malaysia KLCI, BMSC: FTSE Bursa Malaysia Small Cap Index, BMES: FTSE Bursa Malaysia EMAS Shariah Index, and BMHS: FTSE Bursa Malaysia Hijrah Shariah Index.
38
Table 2.3: Summary Statistics for the Returns on Pre- and Post-General Election
BMEMAS BMT100 BMM70 BMKLCI BMSC BMES BMHS PreGE- Mean -0.2062 -0.2021 -0.2687 -0.2039 -0.2928 -0.1982 -0.1892
PostGE- Mean 0.0456 0.0344 0.1007 0.0340 0.2369 0.0355 0.0533
PreGE- Max 1.3967 1.5471 1.3502 1.5086 1.1529 1.1519 1.4903
PostGE- Max 3.5972 3.5318 4.3072 3.3222 5.1983 3.2472 3.3339
PreGE- Min -2.4440 -2.5106 -2.4713 -2.6051 -1.9755 -2.7384 -3.2789
PostGE- Min -9.9494 -10.0817 -9.9045 -9.9785 -9.0170 -11.3205 -11.0873
Note: Pre-General Election: Dissolution of Parliament to the day before General Election; and Post-General Election: Day after the General Election to the first day of the Parliament Assembly.
39
Table 2.4: Pre-General Election and Post-General Election: EGARCH Results Controlled by World Market Effect
Variables BMEMAS BMT100 BMM70 BMKLCI BMSC BMES BMHS
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) (2, 1) (1, 1) Mean Equation
0 0.0104 (0.3474)
0.0107 (0.3315)
0.0047 (0.6942)
0.0119 (0.2907)
0.0315 (0.0300)**
0.0200 (0.0898)*
0.0268 (0.0291)**
PGE -0.0018 (0.9948)
0.0051 (0.9849)
-0.1081 (0.6830)
0.0111 (0.9674)
-0.2385 (0.4578)
0.0129 (0.9575)
0.0099 (0.9598)
PtGE 0.0692 (0.4752)
0.0595 (0.5396)
0.1264 (0.2394)
0.0261 (0.7846)
0.1553 (0.2624)
0.0465 (0.6349)
0.0648 (0.4950)
1tR 0.0450 (0.0183)**
0.0397 (0.0406)**
0.0602 (0.0013)***
0.0348 (0.0868)*
0.0996 (0.0000)***
0.0264 (0.2178)
0.0310 (0.0942)*
1tRWM 0.2162 (0.0000)***
0.2220 (0.0000)***
0.2382 (0.0000)***
0.2194 (0.0000)***
0.1791 (0.0000)***
0.2043 (0.0000)***
0.1980 (0.0000)***
Variance Equation
0 -0.1485 (0.0000)***
-0.1395 (0.0000)***
-0.1520 (0.0000)***
-0.1275 (0.0000)***
-0.2412 (0.0000)***
-0.1400 (0.0000)***
-0.1191 (0.0000)***
1 0.1698 (0.0000)***
0.1616 (0.0000)***
0.1806 (0.0000)***
0.1449 (0.0000)***
0.2952 (0.0000)***
0.2643 (0.0000)***
0.1483 (0.0000)***
2 -- --
-- --
-- --
-- --
-- --
-0.0973 (0.0000)***
-- --
i -0.0791 (0.0000)***
-0.0730 (0.0000)***
-0.0666 (0.0000)***
-0.0688 (0.0000)***
-0.0561 (0.0000)***
-0.0641 (0.0000)***
-0.0641 (0.0000)***
1 0.9797 (0.0000)***
0.9826 (0.0000)***
0.9785 (0.0000)***
0.9825 (0.0000)***
0.9409 (0.0000)***
0.9827 (0.0000)***
0.9894 (0.0000)***
PGE 0.1514 (0.0000)***
0.1411 (0.0000)***
0.1688 (0.0000)***
0.1314 (0.0000)***
0.2798 (0.0000)***
0.1438 (0.0000)***
0.1094 (0.0000)***
PtGE -0.0517 (0.0009)***
-0.0511 (0.0003)***
-0.0553 (0.0059)***
-0.0455 (0.0000)***
-0.0201 (0.5818)
-0.0443 (0.0034)***
-0.0467 (0.0010)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.5321 0.5960 0.3584 0.3131 0.6421 0.6534 0.7076 10 lags 0.3713 0.3218 0.5014 0.1941 0.3161 0.3387 0.4458
Ljung-Box Q2 Statistic (p-value) 5 lags 0.5190 0.5870 0.3560 0.2960 0.6490 0.6510 0.7150
10 lags 0.3800 0.3300 0.5090 0.1970 0.3070 0.3330 0.4290 Return Equation: Wald Test (p-value)
F-stat 0.7745 0.8256 0.4699 0.9628 0.4192 0.8905 0.7872 Chi-Square 0.7745 0.8256 0.4698 0.9628 0.4190 0.8905 0.7871
Variance Equation: Wald Test (p-value) F-stat 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
Chi-Square 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Note: BMEMAS: FTSE Bursa Malaysia EMAS Index, BMT100: FTSE Bursa Malaysia Top 100 Index, BMM70: FTSE Bursa Malaysia Mid 70 Index, BMKLCI: FTSE Bursa Malaysia KLCI, BMSC: FTSE Bursa Malaysia Small Cap Index, BMES: FTSE Bursa Malaysia EMAS Shariah Index, and BMHS: FTSE Bursa Malaysia Hijrah Shariah Index. ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (same average return / volatility for pre- and post-General Election).
40
Table 2.5: Pre-General Election and Post-General Election: EGARCH Results Controlled by Emerging Market Effect
Variables BMEMAS BMT100 BMM70 BMKLCI BMSC BMES BMHS
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0255 (0.0235)**
0.0256 (0.0211)**
0.0162 (0.2279)
0.0211 (0.0721)*
0.0329 (0.0380)**
0.0247 (0.0355)**
0.0252 (0.0413)**
PGE 0.0429 (0.8610)
0.0486 (0.8390)
-0.0329 (0.9066)
0.0551 (0.8244)
-0.0794 (0.8253)
0.0138 (0.9505)
0.0555 (0.7576)
PtGE 0.0952 (0.3211)
0.0905 (0.3450)
0.1534 (0.1365)
0.1002 (0.3046)
0.1696 (0.2094)
0.0975 (0.3165)
0.0987 (0.2831)
1tR 0.0069 (0.7820)
-0.0137 (0.5704)
0.0392 (0.1039)
-0.0195 (0.4224)
0.1066 (0.0000)***
-0.0219 (0.3341)
-0.0217 (0.3037)
1tREM 0.1394 (0.0000)***
0.1491 (0.0000)***
0.1333 (0.0000)***
0.1496 (0.0000)***
0.0718 (0.0000)***
0.1420 (0.0000)***
0.1530 (0.0000)***
Variance Equation
0 -0.1487 (0.0000)***
-0.1358 (0.0000)***
-0.1573 (0.0000)***
-0.1220 (0.0000)***
-0.2405 (0.0000)***
-0.1592 (0.0000)***
-0.1183 (0.0000)***
1 0.1687 (0.0000)***
0.1567 (0.0000)***
0.1843 (0.0000)***
0.1389 (0.0000)***
0.2975 (0.0000)***
0.1892 (0.0000)***
0.1486 (0.0000)***
i -0.0774 (0.0000)***
-0.0693 (0.0000)***
-0.0662 (0.0000)***
-0.0655 (0.0000)***
-0.0589 (0.0000)***
-0.0627 (0.0000)***
-0.0596 (0.0000)***
1 0.9791 (0.0000)***
0.9830 (0.0000)***
0.9748 (0.0000)***
0.9831 (0.0000)***
0.9408 (0.0000)***
0.9792 (0.0000)***
0.9897 (0.0000)***
PGE 0.1427 (0.0000)***
0.1317 (0.0000)***
0.1707 (0.0000)***
0.1220 (0.0000)***
0.2611 (0.0000)***
0.1490 (0.0000)***
0.1048 (0.0000)***
PtGE -0.0507 (0.0011)***
-0.0506 (0.0002)***
-0.0489 (0.0214)**
-0.0478 (0.0000)***
-0.0188 (0.6071)
-0.0430 (0.0120)**
-0.0469 (0.0002)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.2413 0.2413 0.4836 0.0702 0.6652 0.3359 0.5219 10 lags 0.1065 0.0840 0.5644 0.0276 0.2773 0.0970 0.1887
Ljung-Box Q2 Statistic (p-value) 5 lags 0.2410 0.2410 0.4700 0.0650 0.6610 0.3270 0.5100
10 lags 0.1110 0.0870 0.5690 0.0280 0.2640 0.1030 0.1760 Return Equation: Wald Test (p-value)
F-stat 0.5845 0.6066 0.3298 0.5664 0.4402 0.5956 0.5104 Chi-Square 0.5845 0.6065 0.3296 0.5663 0.4400 0.5955 0.5103
Variance Equation: Wald Test (p-value) F-stat 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
Chi-Square 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Note: BMEMAS: FTSE Bursa Malaysia EMAS Index, BMT100: FTSE Bursa Malaysia Top 100 Index, BMM70: FTSE Bursa Malaysia Mid 70 Index, BMKLCI: FTSE Bursa Malaysia KLCI, BMSC: FTSE Bursa Malaysia Small Cap Index, BMES: FTSE Bursa Malaysia EMAS Shariah Index, and BMHS: FTSE Bursa Malaysia Hijrah Shariah Index. ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (same average return / volatility for pre- and post-General Election).
41
Figure 2.1: Volatility during the Pre-General Election for the Selected Stock Indices
Appendix 2.1: Details of the Selected Indices in this Study
Selected Indices in this Study Details of the Indices BMEMAS FTSE Bursa Malaysia
EMAS Index Constituents of the FTSE Bursa Malaysia Top 100 Index and FTSE Bursa Malaysia Small Cap Index
BMT100 FTSE Bursa Malaysia Top 100 Index
Sum of constituents in FTSE Bursa Malaysia Mid 70 Index and FTSE Bursa Malaysia KLCI
BMM70 FTSE Bursa Malaysia Mid 70 Index
Constituents of the next 70 companies in FBMEMAS
BMKLCI FTSE Bursa Malaysia KLCI
Constituents of 30 largest companies in FBMEMAS by full market capitalization
BMSC FTSE Bursa Malaysia Small Cap Index
Constituents of top 98% of the Bursa Malaysia Main Market excluding FTSE Bursa Malaysia Top 100 Index constituents
BMES FTSE Bursa Malaysia EMAS Shariah Index
Shariah-compliant constituents of the FBMEMAS that meet the screening requirement of the SAC
BMHS FTSE Bursa Malaysia Hijrah Shariah Index
Constituents of 30 largest Shariah-compliant companies in FBMEMAS screened by Yasaar Ltd and the Securities Commission's Shariah Advisory Council (SAC)
Source: Authors' compilation based on information from Bursa Malaysia’s website. For further information, please visit: http://www.bursamalaysia.com.
42
CHAPTER 3
STOCK MARKET VOLATILITY IN MALAYSIA SECTORAL INDICES DURING THE GENERAL ELECTION
3.1 Introduction Political factor that exerts influence on investors' decision-making is one of the possible
causes of market sentiment in the stock market. Specifically, investor sentiment or
expectations on major political events could exhibit optimism or pessimism. The stage of
sentiment will induce underreaction or overreaction in the market which influence changes in
trading volume, volatility, prices and accordingly determine stock returns (Tuyon et al.,
2016). Hence, sentiment risk could be deemed as a systematic behavioural risk. In investment
practice, the role of investor sentiment on the stock market activity is important because the
stock prices are affected by both the fundamental and behavioural forces (Akerlof and Shiller,
2009). During major political events, the combination of fundamental and behavioural forces
in decision-making cause bounded rationality in market players which could induce
uncertainty in the stock market.
In the literature, it is evident from several studies that the occurrences of major
political events induced higher market volatility. The recent empirical evidence is found on
the national election (Mei and Guo, 2004; Białkowski et al., 2008; Jones and Banning, 2009;
Lean and Yeap, 2017; among others), delay in election results (Nippani and Arize, 2005),
change of ruling party (Lin and Wang, 2007), as well as the political scandal (Lobo, 1999).
Previous studies on the relationship between political events and stock market performance
are largely centered on elections. Earlier studies of Niederhoffer et al. (1970), Peel and Pope
(1983) and Gemmill (1992) have examined the stock price behaviours during national
43
elections in developed countries. These studies found that changes in government
administration after elections tend to affect financial policies or legislation, thereby stock
prices were significantly impacted. However, the study on election effect in emerging market
only started in recent years, for instance, Wang and Lin (2009) and Hung (2011) on
Taiwanese stock market, Lean (2010) and Lean and Yeap (2017) on Malaysian stock market.
The evidence found in previous studies is mostly based on the examination of main
composite indices, such as the Toronto 300 Composite and the I.P.C. All-Share in Nippani
and Arize (2005), the Taiwan Stock Exchange Value Weighted Index (TAIEX) in Wang and
Lin (2009), and the FTSE Bursa Malaysia KLCI Index in Lean and Yeap (2017). Besides
information from the composite index, sector-specific information could be useful for
investors to narrow down their investments option in the financial market. Nevertheless, the
stock return volatility due to changes in political may evolve differently in sectoral indices.
Therefore, the evidence found based on composite indices need not be applicable to the
individual sectors.
In addition, there are recent studies on sector-specific analysis of the stock market in
the Asian region (Cao et al., 2013; Lakshmi, 2013). The main focus of their studies is to
investigate the sensitivity of the sectoral indices to market fluctuation and the performance of
the sectoral indices. Nevertheless, the aspect of the influence of political events on the
movement of sectoral indices has not been thoroughly discussed. Moreover, recent research
provided evidence that firms in different sectors are reported to have different sentiment
effect (Kaplanski and Levy, 2010; Chou et al. 2012; Chen et al., 2013; Dash and Mahakud,
2013). Hence, a comprehensive analysis of stock market performance based on sectoral
44
indices should be addressed in order to have a better understanding of political changes in
relation to fluctuation in sectoral indices.
In behavioural finance, Asia suffers from higher risk of behavioural biases than other
developed markets (Ritter, 2003; Schmeling, 2009). Nevertheless, emerging financial
markets are still attractive to investors because of their relatively higher returns compared to
developed financial markets (Kearney, 2012). Among the emerging markets, the Malaysian
stock market is quite a developed capital market (Mohamad et al. 2007). The Bursa Malaysia
has steadily emerged as one of the top-performing markets in Asia. Its capitalisation has
reached USD 382 billion in December 2015 and the market ranked the second highest in
ASEAN markets after the Singapore Exchange. In terms of behavioural risk, empirical
studies of Statman (2008) and Tuyon et al. (2016) found that Malaysian investors are affected
by sentiment in their investment decision making. The finding of Tuyon et al. (2016) further
highlighted that investor sentiment risks influence stock prices regardless of size and industry
groups.
From the perspective of statistical analysis, single country data analysis is preferred to
mitigate the heterogeneous effect of multiple country characteristics such as differences in
economics, political, institutional, demographics and culture (Bekaert and Harvey, 2002;
Statman, 2008). Hence, taken all these facts, the Malaysian stock market is chosen as a single
country testing case to see the influence of political events on the movement of stock prices
and this study could be of interest to international investors. Evidently, as a proxy of the
Malaysian stock market, the key index of FTSE Bursa Malaysia KLCI experienced
significant volatility during the general election years (Lean and Yeap, 2017). Prior to the
year 2005, the 9th, 10th, and 11th Malaysian general election have not resulted in unexpected
45
outcomes as the coalition Barisan Nasional (BN) won and continued ruling with a stable two-
thirds majority. Hence, general ups and downs in the stock market are well-anticipated by
investors. On the other hand, the coalition BN experienced close fight in the 12th and 13th
general election and consecutively lost the two-thirds majority in parliament, which is never
happened in political history since Malaysia independence. Besides, the total percentage vote
for BN experienced significant drop from 63.8% in year 2004, to 51.4% in year 2008 and
47.4% in year 2013. Due to political uncertainty, a sharp decline in the key indices of FTSE
Bursa Malaysia was recorded prior to the general election and investors' confidence was
badly shaken due to the potential shift of ruling party.
Therefore, in order to examine the election effect, the focus of this study is on the
Malaysian sectoral indices for the past general election years of 1995, 1999, 2004, 2008 and
2013. The sectoral index provides a value for the aggregate performance of a number of
companies of a particular sector and it serves as an indirect measure of the performance of the
economy. There are ten main indices based on sectors or industries at the Bursa Malaysia,
each represents the sector of Construction, Consumer Product, Finance, Industrial, Industrial
Product, Mining, Plantation, Property, Trading and Services, Technology. A benchmark
index of FBMKLCI also included in the analysis for comparison purpose.
In general, using a long history of aggregate stock returns that incorporates a sharp
decline may produce erroneous inferences due to model misspecification. However, previous
studies on the Malaysia election effect did not address this issue. For example, Lean and
Yeap (2017) covered six general elections (the 8th to 13th general elections) under the same
sample period. In fact, the market condition during the general election years of 2008 and
2013 (the 12th and 13th general elections) is clearly different with previous general elections.
46
In concern of the different effect of the general election on market volatility, this study
divides the general election periods into two stages. One stage represents the general ups and
downs periods from 1994 to 2005, and the other represents drastic shock periods from 2006
to 2015.
In brief, the contributions of this study are, first, the Threshold Generalized
Autoregressive Conditional Heteroskedasticity (Threshold GARCH / GJR GARCH) model
developed by Glosten et al. (1993) is applied to investigate the pre-general election and post-
general election effect on sectoral indices of the Malaysian stock market. Previous studies
examined the impact of the election on the composite index, while this study attempts to see
the election effect on the ten sectoral indices. Second, the selection of the event window in
this study is in line with the Malaysian general election process. Relevant studies normally
used trading day windows before and after the election, for example, 1 week, 2 weeks and 1
month, to see the different effect of election. This study precisely defines the pre-general
election period as the trading days from the day of dissolution of the parliament until the day
before voting, while the post-general election period covers the trading days from the day
after voting until the day of first parliament assembly.
Third, this study enhances the knowledge in the case of Malaysia by investigating the
election effect in two different stages which represent the general up and down and the
drastic rise and fall period. Fourth, the MSCI World Index is included as a control variable in
the model to account for the global market effect. Moreover, this study also conducts an array
of robustness checks, including analyzing the model with the MSCI Emerging Market Index
to control for emerging market effect, considering Chicago Board Options Exchange (CBOE)
Volatility Index (VIX) as one of the market uncertainty indicator for global risk, and
47
controlling the US Federal Fund Rate for interest rate differentials. This study may be of
interest to investors as the results will contribute the information that most investors require
particularly in constructing an effective equity portfolio investment during the times of
election.
The rest of the paper is organized as follows. Section 3.1 summarizes the literature of
related studies. Section 3.2 describes the nature of the data sets and the methodology adopted
in this study. Section 3.3 reports the estimation results. Next, Section 3.4 reports the
Extensions and Robustness results in this study. Lastly, Section 3.5 concludes by highlighting
some implications of the findings.
3.2 Literature Review
In modern finance, investors are assumed to be rational in making decisions on portfolio
investment (Lawrence et al., 2007). Nevertheless, behavioural finance research has confirmed
bounded rationality of investors that drives to market inefficiency. Irrational behaviours in
decision-making including considered gossip, rumors and tips as an information (Bauman,
1989). Moreover, this behavioural risk is critical for Asia emerging financial markets because
Asian are more socially collective in decision-making (Kim and Nofsinger, 2008).
Collectively, political events such as national election could be possibly one of the drivers of
bounded rationality that influence the performance of a stock market. Specifically, during the
election period, the rumors circulated by the media and campaign strategists may induce
irrational behaviours in investors decision-making. Investors who confronted with political
risk are likely to have an extreme response and their reaction will differ substantially from the
optimal forecast. Thus, if the outcome of the election does not allow investors to immediately
assess the effect on the country's future, then this will induce surprises in the markets.
48
In the literature, the mounting evidence of political election effect lends support to
bounded rational of investors' behaviour. The pioneering study by Nordhaus (1975) discussed
the relationship between economic performance and political business cycle. Other studies
such as Allivine and O’Neill (1980) and Huang (1985) concentrated on the U.S. presidential
election cycle, in which the U.S. stock markets make larger gains in the third and fourth years
of a presidential term. Recent evidence in the area includes the study by Wong and McAleer
(2009) indicating the impact of U.S. presidential elections on its stock market. They found
that the U.S. stock prices closely followed the four-year presidential election cycle and the
cyclical trend existed for the last ten administrations from the year 1965 through 2003,
particularly when the incumbent is Republican.
Besides the U.S. presidential election cycle, additional studies including Foerster and Schmitz
(1997), Leblang and Mukherjee (2005), Wang and Lin (2009) and Hung (2013) investigated
the impact of presidential election results on international stock markets. Finding of these
studies indicated that stock markets are affected by the presidential election. Moreover, there
are several studies in the empirical literature that examined the impact of one-time occurrence
of the political event on financial markets. Lobo (1999) found that there was a great deal of
insecurity amongst investors in the U.S. stock market after a political scandal had been
revealed. Other related studies of Nippani and Medlin (2002) and Nippani and Arize (2005)
examined the impact of the delay in the 2000 presidential election results to stock market
performance. Using the conventional heteroskedastic t-tests and binary variable regression,
they discovered that the delay had negatively impacted stock prices. The negative impact not
only appeared in U.S. stock market, but also Mexican and Canadian stock market. It is
49
evident that election results are a widely watched event by local and foreign market
participants.
In earlier literature, the conventional t-test is used to test the hypothesis of presidential
election effect in the stock market, for example, Allivine and O’Neill (1980), Nippani and
Medlin (2002) and Hung (2011). The limitations are that these studies do not consider the
effect of time dependence and conditional heteroskedasticity or the GARCH effect in stock
returns. Campbell and Hentschel (1992) presented the volatility feedback hypothesis, where
any innovations to volatility (especially positive ones) lead to a decrease in returns. Hence,
volatility is a fundamental issue in testing the impact of the political election on stock market
performance. There is another strand of interesting research on political elections and stock
market volatility (Białkowski et al., 2008; Wang and Lin, 2009; Chau et al., 2014; Smales,
2016; among others). In particular, Białkowski et al. (2008) revealed that the index return
variance can easily double during the week around an election in a sample of 27 OECD
countries. Wang and Lin (2009) also found that presidential elections negatively impacted
stock returns and induced higher volatility in the Taiwanese stock market. In the case of Arab
Spring which started in December 2010, Chau et al. (2014) studied the impact of political
uncertainty on conventional and Islamic stock market indices in the MENA region. Despite
the heterogeneous reaction of the conventional stock indices and Islamic stock indices to that
political turmoil, volatility of the Islamic indices significantly increased during the period of
June 2009 to June 2012, whereas there is no significant effect on the volatility of
conventional markets.
On top of that, previous studies as mentioned above mostly focused on the relationship
between political election and the stock market in a Presidential system country. However,
50
less attention is given in country with a Parliamentary system which held elections for the
national parliament. The major difference between these two systems is that in a Presidential
system, the President is the executive leader directly voted upon by the people, while in a
Parliamentary system, the Prime Minister is the executive leader elected from the legislative
branch directly. For that reason, the sensitiveness of stock market to the election may vary
between Presidential system country and Parliamentary system country.
Malaysia is a country with a Parliamentary system where the Parliament of Malaysia is the
national legislature of Malaysia. Moreover, Malaysia has an interesting political background
for the testing of election effect. The recent five general elections held during 1994 - 2015
were accompanied by a political uncertainty due to fierce challenge between the opposition
and the incumbent. Moreover, there have been increasing concerns about the change of
government in the recent two general elections in the year 2008 and 2013. The key index of
FTSE Bursa Malaysia KLCI has undergone significant volatility during the general election
years. Notably, the KLCI shows an unusually high spike before the election during the
general election years of 2008 and 2013 (Lean and Yeap, 2017). As documented by Smales
(2016), the implied volatility of financial markets increases in line with uncertainty about the
election outcome. Therefore, uncertainty arising from political elections in Malaysia can
cause a drastic effect on the performance of the stock market.
In the Malaysian stock market, only a few studies had been done in relation to the general
election. The study by Ali et al. (2010) found significant over-reaction behaviour existed
upon the announcement of the removal of the deputy prime minister and announcement of the
resignation of the prime minister. In contrast, evidence of under-reaction was detected upon
the announcement of the national election. Ali et al. (2010) explained that investors are well
51
predicted with the election outcome, hence the finding of under-reaction is in line with the
political condition at that times of election. Furthermore, Lean (2010) showed that general
election in Malaysia significantly affected the stock market performance, where stock returns
react positively before the election and negatively after the election. However, both studies
did not focus on the possible impact of the general election on stock volatility. The study of
Lean and Yeap (2017) circumvented the limitation of previous studies and examined stock
volatility during election periods. They found significant election effect in stock volatility but
not in the stock returns.
The study of Lean and Yeap (2017) used a long history of stock returns that covered six
general elections from the 8th to 13th general election. However, the Malaysian stock market
experienced a sharp decline prior to the 12th and 13th general election due to the close fight
between major coalitions. The market condition during the 12th and 13th general election are
clearly different from the previous general election. It is noteworthy that there is a significant
research showed that stock volatility varies according to market condition. The study of Cao
et al. (2013) examined the stock market performance in China during the bull and bear
market happened in the year 2007 and 2008. The study is specially designed to capture the
movement of stock during the stage of drastic shock periods in 2007 and 2008, and the
general ups and downs periods. The study concluded that the movement of stock indices are
different in the two stages. When the market experiences drastic ups and downs, the sectoral
indices exhibit very close correlations between each other. However, much smaller
correlations appear in the general ups and downs period. Hence, taken together with all these
considerations, this present study divides the sample period into two sub-samples to avoid
erroneous inferences, and to reflect the real market volatility under different political tense. In
52
brief, statistical results of this study are able to fill in the research gap by showing a
significant difference of market volatility in the two sub-sample periods.
Despite different industry characteristic, previous studies conclude the findings on election
effect by observing the stock market aggregately. In fact, firms in a different industry are
expected to have a different characteristic. Based on the industry type definition used by
(Becher et al., 2008; Held, 2009; Nagy and Ruban, 2011), there are two industry groups in
which the defensive industry is expected to be less sensitive to macroeconomic and market
fluctuations, and the cyclical industry is more sensitive to the macroeconomic and market
developments. Moreover, the study of Lakshmi (2013) found volatility patterns are not the
same across the eleven selected sectoral indices in Indian Stock Market. They found out that
the realty sector has witnessed higher volatility than any other sector during the period of the
global meltdown in the year 2008 to 2013. Thus, there are possibilities that the impact of the
general election on the industry may vary among each other, and this present study attempts
to fill in the research gap by examining the sectoral stock returns and volatility in the recent
five Malaysian general elections.
3.3 Data and Empirical Methodology
This study uses daily closing values of the FTSE Bursa Malaysia KLCI Index and ten
selected main sectors indices (Construction, Consumer Product, Finance, Industrial,
Industrial Product, Mining, Plantation, Property, Trade and Services, and Technology). The
full sample period covers from 4 January 1994 to 31 December 2015, with a total of 5,738
observations, which covers the recent five Malaysia general elections. All data are collected
from the Bursa Malaysia (http://www.bursamalaysia.com). For control variable, the MSCI
World Index and MSCI Emerging Index, obtained from the S&P Capital IQ, are used to
53
control for world market and emerging market effect, respectively. Besides, the Chicago
Board Options Exchange (CBOE) Volatility Index (VIX) is used as an indicator of global
risk, and the US Federal Fund Rate is used for interest rate differentials. The important dates
of general elections are summarized in Table 3.1, which are the date of dissolution of
parliament, election date or voting date and the 1st parliament assembly after election. The
pre-general election period refers to the duration from the day of dissolution of the parliament
until the day before voting, while the post-general election period refers to the duration from
the day after voting until the day of first parliament assembly.
[Insert Table 3.1: Malaysia General Election Information]
Table 3.2 presents the descriptive statistics for daily returns series for the full sample period.
Daily returns are calculated as the first difference in the natural logarithms of the stock
market index, )/ln(100 1 ttt IIR where tI and 1tI are the values of each index for
periods t and 1t , respectively. In the case of a trading day following a non-trading day, the
return is calculated using the closing price of the last trading day. From the descriptive
statistics, the null hypothesis of normally distributed daily returns is rejected by the Jarque-
Bera normality test. This finding is in line with most of the previous findings, saying that
daily stock returns are not normally distributed.
[Insert Table 3.2: Descriptive Statistics for the Malaysian Stock Indices]
Furthermore, mean returns for the periods of pre-general election and post-general
election are presented in Table 3.3. It is observed that the mean returns prior to general
election are mostly positive for the sub-sample period of 1994-2005. However, for the sub-
54
sample period of 2006-2015, the mean returns are all negative prior to general election. On
the other hand, for the period of post-general election, the mean returns for the indices are all
negative for the sub-sample period of 1994-2005, except for the sectoral indices of Consumer
Product and Industrial. For the period of 2006-2015, all the mean returns are positive after
general election. From the descriptive statistics and mean returns for the two sub-sample
periods, it is notable that there could be different election effects on the stock market for the
general elections in year 1994 to 2005 and 2006 to 2015. The preliminary statistics justify the
aim of this study in dividing the full sample period into two sub-samples in order to study the
election effects under different political condition.
[Insert Table 3.3: Mean Returns on Pre-General Election and Post-General Election]
In this study, the test for market volatility during general elections is carried out by
using the Threshold Generalized Autoregressive Conditional Heteroskedasticity (Threshold
GARCH / GJR GARCH) model developed by Glosten et al. (1993), Threshold GARCH3
model with dummy variables:
tttttt RWMRPtGEPGER 1113210 (1)
ttttttt PtGEPGEN 212
1112
12
1102
(2)
where tR is the logarithmic return of the market index at day t ; tPGE and tPtGE are dummy
variables which take on value 1 if the corresponding return for day t is the period of the pre-
3 According to Bollerslev et al. (1992), in testing the GARCH models, p = q = 1 is sufficient for most financial series. Hence, the sufficient order of p and q considered in this study for the Threshold GARCH model, is (1, 1).
55
general election, and the period of the post-general election respectively, and 0 otherwise; t
is the error term. Meanwhile, 30 ,..., are the parameters to be estimated. Among them, 0
measures the mean return (in percentage) on other trading days; whereas 1 and 2 capture
the average return of the stock index for the period of pre-general election and post-general
election. At the later part of the estimation, a lagged value return variable for the MSCI
World Index ( 1tRWM ) is introduced into the mean equation and variance equation as control
variables to examine whether the returns of the general election years are associated with the
MSCI World Index lagged return.
The null hypothesis of the test is 0: 210 H , which implies that average daily
returns (volatility) for the period of pre-general election and post-general election are
significantly different from zero. If the null hypothesis does not hold, then it can be
concluded that the market index is characterized by statistically different on average returns
(volatility) for the period of pre-general election and post-general election. In another word,
this would imply that general election effect is indeed present in the market. Besides, if the
parameter of 3 is insignificant, then it can be concluded that the stock returns of the general
election years are not influenced by the MSCI World Index lagged return.
In the Equation (2), tN takes on value 1 when the stock quote falls in a period and 0
for increments of the stock quotation. Besides, the parameter is used to capture the
asymmetrical effect of bad news (decrease in stock indices, hence negative tR ) and good
news (increase stock indices, hence positive tR ). If 0 by the t-test of significance, then it
can be concluded that the impact of news is asymmetric. If the parameter is positive, then
56
good news has an impact of i on volatility while bad news has an impact of ( i ) on
volatility. Thus, the positive value of indicates the existence of a leverage effect in that bad
news increases volatility. The additional parameters, t , which makes this specification
different from the original Threshold GARCH model, are employed to capture the daily
effect. Furthermore, a lagged value of the return variable is introduced in the equations to
avoid serial correlation error terms in the model, which may yield misleading inferences.
3.4 Empirical Results and Discussions
Firstly, the results of the pre-general election and post-general election effect on the sectoral
indices for the full-sample period of 1994-2015 are presented in Table 3.4(a) and Table
3.4(b). Table 3.4(a) reports the results of the mean equation and variance equation of the
Threshold GARCH (1, 1) model for the FTSE Bursa Malaysia KLCI index and the sectoral
indices of Construction, Consumer Product, Finance, and Industrial. Meanwhile, Table 3.4(b)
reports the estimation results for the sectoral indices of Industrial Product, Mining,
Plantation, Property, Trade and Services, and Technology. The diagnostic test result is
included in the lower part of the tables to support the validity of the models.
Under the mean equation, the dummy coefficients are all insignificant. The high p-
value of dummy coefficient indicates insignificant stock returns for both the pre-general
election and post-general election periods. The finding of insignificant abnormal return
around election period is consistent with the studies of Lean (2010) and Lean and Yeap
(2017). In term of control variables, the dummy coefficients of the MSCI World Index for the
mean equation are all positive and significant at 1%. The results indicate that the Malaysian
stock market returns are strongly affected by global market environment.
57
The estimation results of the variance equations are also presented in Table 3.4(a) and
Table 3.4(b). For the variance equation, the pre-general election dummy coefficients for eight
out of ten sectoral indices are positive and highly significant. These eight sectoral indices of
Construction, Consumer Product, Finance, Industrial, Industrial Product, Property, Trade and
Services, and Technology experienced significant high volatility in pre-general election
periods. Besides, significant low volatility is found in the sectoral index of Mining during the
pre-general election periods. The plantation is the only sector with the insignificant result.
Thus, the results of Threshold GARCH estimation on the pre-general election period show
the existence of significant pre-general election effect in stock volatility in eight out of ten
sectoral indices in the Malaysian stock market. Meanwhile, for the period of the post-general
election, the dummy coefficients of the variance equations are positive and significant for the
Construction, Plantation, and Technology sectoral indices.
The leverage effect term, , in the variance equation is positive and statistically
different from zero for all the sectoral indices. The positive value of indicates that the
leverage effect in bad news increases the volatility. In particular, the bad news has an impact
of ( i ), while good news has an impact of ( i ) only. For example, refer to Table 3.4(a),
bad news in the Construction sectoral index has an impact of 0.9682 (0.8926 + 0.0756), while
good news has an impact of 0.8926 only. Hence, the results indicate the existence of the
asymmetric effect on stock volatility in all ten sectoral stock indices of Malaysian stock
market.
[Insert Table 3.4(a): Threshold GARCH Results for Pre-General Election and Post-
General Election (1994 - 2015) - Controlled by World Market Effect]
58
[Insert Table 3.4(b): Threshold GARCH Results for Pre-General Election and Post-
General Election (1994 - 2015) - Controlled by World Market Effect]
Next, this study examines the presence of pre-general election effect and post-general
election effect in the sectoral indices for the first sub-sample period of 1994-2005 and the
results are presented in Table 3.5(a) and Table 3.5(b). For the stock return, the dummy
coefficients for the mean equations of the pre-general election are significantly positive for
two out of ten sectoral indices, which are Construction and Industrial Product. These two
sectoral indices recorded significant positive return before the general election. On the other
hand, for post-general election, the dummy coefficients are significantly negative for
Technology sectoral index. The general election has negatively impacted this sector. Overall,
the results indicate that the election effect in stock return only exists in certain sectors in the
Malaysian stock market. From the dummy coefficients of the control variables, it is evident
by the positive and significant coefficients that the Malaysian stock market returns are
positively impacted by the MSCI stock return.
Furthermore, the estimation results of the variance equations with control variables
are also presented in Table 3.5(a) and Table 3.5(b). For the sub-sample period of 1994-2005,
the results are consistent among the sectoral indices, compare to the results of the full-sample
period. Among the ten sectoral indices, eight of them experienced significant volatility
change before and after the general election. In particular, the sectoral indices of
Construction, Finance, Industrial Product, Mining, Plantation, Property, and Trade and
Services experienced significant low volatility before the general election. However, after the
announcement of the election result, the stock volatility increased significantly in these seven
sectoral indices. For the sector of Construction, this sector recorded significant low volatility
59
after the general election. Thus, it is evident that most of the sectoral indices in the Malaysian
stock market experienced significant volatility change due to the general election.
Meanwhile, no significant result is found for the sectoral indices of Consumer Product.
The results of variance equations also confirm that there is an asymmetric effect of
political elections on stock volatility for the sub-sample period of 1994-2005. The positive
value of the leverage effect term is statistically significant, and this indicates the existence of
asymmetrical effect in the Malaysian stock market. This finding implies that negative shocks
or bad news from the election have larger impact on stock volatility than good news in the
sub-sample period of 1994-2005. Lastly, the validity of the model is checked by the
diagnostic tests. No remaining ARCH effect and serial correlation are found in most of the
estimated models.
[Insert Table 3.5(a): Threshold GARCH Results for Pre-General Election and Post-
General Election (1994 - 2005) - Controlled by World Market Effect]
[Insert Table 3.5(b): Threshold GARCH Results for Pre-General Election and Post-
General Election (1994 - 2005) - Controlled by World Market Effect]
For the second sub-sample period of 2006-2015, Table 3.6(a) present the results of the
pre-general election and post-general election effect for the FTSE Bursa Malaysia KLCI
index and the sectoral indices of Construction, Consumer Product, Finance, and Industrial,
while Table 3.6(b) reports the estimation results for the sectoral indices of Industrial Product,
Mining, Plantation, Property, Trade and Services, and Technology. From the estimations of
mean equations, the sectoral index of Consumer Product and Mining are the only two indices
60
that show the significant result for the period of pre-general election and post-general
election. The Mining index has a negative and significant return during the period of the pre-
general election, while the Consumer Product sectoral index has a positive and significant
return during the period of post-general election. The finding indicates that the general
election result brought a negative impact to the Mining sector and a positive impact on the
Consumer Product sector. Besides, the dummy coefficients of the MSCI World Index for the
mean equations are positive and significant at 1% for all the sectoral indices. The positive
sign of the dummy coefficient indicates that the global index has a positive impact on the
Malaysian sectoral indices.
As explained earlier, the political condition in the 12th and 13th Malaysia general
elections were different with previous general elections due to the close fight between the two
major coalition. Prior to the general election, the market condition experienced significant
volatility change as supported by the empirical results of this study. From the estimation
results of the Threshold GARCH variance equations, six out of ten of the sectoral indices
encountered significant high volatility in pre-general election periods. The Mining sectoral
index is the only one which recorded significant low volatility during the period. On the other
hand, this study also finds evidence on post-general election effect in stock volatility. The
results of post-general election show insignificant low volatility in the sectoral indices of
Construction, Consumer Product, Industrial, Mining, Plantation, Property, and Trade and
Services. Meanwhile, the Technology sectoral index is the only sector with significant high
volatility in the post-general election period. The result on the second sub-sample period of
2006-2015 is clearly different between the first sub-sample period which covers the 9th, 10th
and 11th Malaysia general elections, where most of the sectoral indices recorded significant
low volatility before general elections and significant high volatility after general elections.
61
The asymmetric effect of the general election is also reported in Table 3.6(a) and
Table 3.6(b). The significant asymmetry coefficient ( ) strongly supports the asymmetric
effect in most of the indices. The leverage effect term, , is statistically different from zero
for all the indices, indicating the existence of the asymmetrical stock returns in the
Malaysian. Besides, the validity of the model is supported by the diagnostic test with no
remaining ARCH effect and serial correlation in all of the estimated models.
[Insert Table 3.6(a): Threshold GARCH Results for Pre-General Election and Post-
General Election (2006 - 2015) - Controlled by World Market Effect]
[Insert Table 3.6(b): Threshold GARCH Results for Pre-General Election and Post-
General Election (2006 - 2015) - Controlled by World Market Effect]
3.5 Extensions and Robustness
In order to test the robustness of the model, this study extends the analysis by using the
lagged value of MSCI Emerging Market Index ( 1tREM ) return as an alternative control
variable to test the impact of emerging market on Malaysian stock market returns for all the
three sample periods. The full-sample period results from the model with control variable
( 1tREM ) are presented in Table 3.7(a) and Table 3.7(b). The results for the first sub-sample
period of 1994-2005 and the second sub-sample period of 2006-2015 are shown in Table
3.8(a) and Table 3.8(b), and Table 3.9(a) and Table 3.9(b), respectively. Similar to the
previous models with World Market Index as the control variable, the dummy coefficients of
the MSCI Emerging Market Index in the mean equation are all positive and significant at 1%
for all the three sample periods. The results indicate that the Malaysian stock market returns
62
are strongly affected by emerging market environment as well. It also implies the robustness
of the models after control for external factor, either for world market effect or emerging
market effect. Nevertheless, in term of stock market volatility, there are a few differences in
the estimation result by comparing the models with control variable of World Market Index
to the models with MSCI Emerging Market Index. The differences mainly occur in the first
sample period of 1994-2005.
In particular, for the full sample period, the pre-general election dummy coefficient
for the sector of Consumer Product becomes insignificant after controlling for the emerging
market effect. Previously, this dummy coefficient is significant at 10% for the model
controlled by World Market Index as presented in Table 3.4(a). On the other hand, the stock
index of the Property sector showed a significant higher volatility in post-general election
periods. Similar to the result presented in Table 3.4(b), the Property stock index also has a
positive and significant dummy coefficient in the pre-general election periods. This implies
that the Property sector is sensitive to the influence of general election after controlled for
external factor. The rest of the sectors have similar results for the mean equation and variance
equation.
Overall, as presented in Table 3.7(a) and 3.7(b), the dummy coefficients for the mean
equation are all insignificant for both the pre-general election and post-general election
periods from the year 1994-2015. For the variance equation, the pre-general election dummy
coefficients for seven out of ten sectoral indices are positive and highly significant. This
implies that the seven sectoral indices of Construction, Finance, Industrial, Industrial Product,
Property, Trade and Services, and Technology showed significant high volatility in pre-
general election periods. Besides, significant low volatility is found in the sectoral index of
63
Mining during the pre-general election periods. The Plantation is the only sector with an
insignificant result. Thus, the results of Threshold GARCH estimation on the pre-general
election period show the existence of significant pre-general election effect in stock volatility
in seven out of ten sectoral indices in the Malaysian stock market. Meanwhile, for the period
of the post-general election, the dummy coefficients of the variance equations are positive
and significant for the Construction, Plantation, Property, and Technology sectoral indices.
The leverage effect term, , in the variance equation is positive and statistically different
from zero for all the sectoral indices. The positive value of indicates that the leverage
effect in bad news increases the volatility. In particular, the bad news has an impact of
( i ), while good news has an impact of ( i ) only. Hence, the results indicate the
existence of the asymmetric effect on stock volatility in all ten sectoral stock indices of
Malaysian stock market.
[Insert Table 3.7(a): Threshold GARCH Results for Pre-General Election and Post-
General Election (1994 - 2015) - Controlled by Emerging Market Effect]
[Insert Table 3.7(b): Threshold GARCH Results for Pre-General Election and Post-
General Election (1994 - 2015) - Controlled by Emerging Market Effect]
Turning to the first sub-sample period of 1994-2005, changes happened in the sector
of Property, Construction, and Industrial Product after controlling for emerging market effect.
As shown in Table 3.8(b), a significant result is found in the mean equation where the
Property sectoral stock index has a significant positive return before the general election and
significant negative return after the general election. While controlling for world market
effect before this, there is no significant result found in the Property sector mean return. This
64
result further implies that the stock index of Property sector is sensitive to the influence of
general election after controlled for emerging market effect. On the other hand, the pre-
general election dummy coefficient for the sector of Construction and Industrial Production
becomes insignificant after controlling for the emerging market effect. Previously, with the
control variable of World Market Index, the estimated model has a significant dummy
coefficient on the pre-general election for the sector of Construction (significant level at 5%)
and Industrial Product (significant level at 10%) as presented in Table 3.5(a) & 3.5(b). Also,
in the variance equation of Industrial Product, the dummy coefficient for the pre-general
election remains significant at 10%, but the significant high volatility in the post-general
election no longer exist in the model controlled by emerging market effect.
The full result for the first sub-sample period of 1994-2005, Table 3.8(a) and Table
3.8(b) are described as follow. For the stock return, significant results are only found in the
sectoral indices of Property and Technology. The Property sectoral index has a significant
positive dummy coefficient for the pre-general election period and a significant negative
dummy coefficient for the post-general election period. The Technology sectoral index also
has a significant negative dummy coefficient for the post-general election period. Hence, it is
evident that the outcome of the general election has negatively impacted these two sectors.
Furthermore, the estimation results of the variance equations are consistent among the
sectoral indices, compared to the results of the full-sample period. Among the ten sectoral
indices, six of them experienced significant volatility change before and after the general
election. In particular, before the general election, the sectoral indices of Construction,
Finance, Mining, Plantation, Property, and Trade and Services experienced significant low
volatility. However, after the announcement of the election result, the stock volatility
increased significantly in the six sectoral indices. The sector of the Industrial product also has
65
a significant negative dummy coefficient in the pre-general election period but no significant
result is found in the post-general election period. Moreover, the outcome of the general
election also led to a high volatility in the Technology sector. The Industrial sector is the only
sector that recorded significant high volatility before the general election. Meanwhile, no
significant result is found for the sectoral indices of Consumer Product. The results of the
variance equations also confirm that there is an asymmetric effect of political elections on
stock volatility. The positive value of the leverage effect term is statistically significant and
this indicates the existence of asymmetrical effect in the Malaysian stock market. This
finding implies that negative shocks or bad news from the election have larger impact on
stock volatility than good news. Lastly, the validity of the model is checked by the diagnostic
tests. No remaining ARCH effect and serial correlation are found in most of the estimated
models.
[Insert Table 3.8(a): Threshold GARCH Results for Pre-General Election and Post-
General Election (1994 - 2005) - Controlled by Emerging Market Effect]
[Insert Table 3.8(b): Threshold GARCH Results for Pre-General Election and Post-
General Election (1994 - 2005) - Controlled by Emerging Market Effect]
For the second sub-sample period of 2006-2015, the only change in the result comes
from the sector of Consumer Product in the variance equation. The post-general election
dummy coefficient for the sector becomes significant after controlled for the emerging
market effect as presented in Table 3.9(a). This implies that the Consumer Product sector has
a significant low volatility after controlling for external factor. The results for the rest of the
66
sectors are consistent with the main analysis, for both the mean equation and variance
equation.
Table 3.9(a) and 3.9(b) present the full results for the second sub-sample period of
2006-2015. From the estimation of mean equations, the sectoral index of Consumer Product
and Mining are the only two indices that show a significant result. The Mining index has a
negative and significant return in the period of the pre-general election, while the Consumer
Product sectoral index has a positive and significant return in the period of post-general
election. As explained earlier, the political condition in the 12th and 13th Malaysia general
elections were different with previous general elections due to the fierce challenge between
the opposition and the incumbent. Prior to the general election, the market condition
experienced significant volatility change as supported by the empirical results of this study.
From the estimation results of the Threshold GARCH variance equations, six out of ten of the
sectoral indices encountered significant high volatility in pre-general election periods. The
Mining sectoral index is the only one which recorded significant low volatility during the
period. On the other hand, this study also finds evidence on post-general election effect in
stock volatility. The results of post-general election show insignificant low volatility in the
sectoral indices of Construction, Industrial, Mining, Property, and Trade and Services.
Meanwhile, the Technology sectoral index is the only sector with significant high volatility in
the post-general election period. On the other hand, Consumer Product recorded significant
low volatility during the post-general election period. The asymmetric effect of the general
election is also reported in Table 3.9(a) and Table 3.9(b). The significant asymmetry
coefficient ( ) strongly supports the asymmetric effect in most of the indices. The leverage
effect term, , is statistically different from zero for all the indices, indicating the existence
of the asymmetrical stock returns in the Malaysian. Besides, the validity of the model is
67
supported by the diagnostic test with no remaining ARCH effect and serial correlation in all
of the estimated models.
[Insert Table 3.9(a): Threshold GARCH Results for Pre-General Election and Post-
General Election (2006 - 2015) - Controlled by Emerging Market Effect]
[Insert Table 3.9(b): Threshold GARCH Results for Pre-General Election and Post-
General Election (2006 - 2015) - Controlled by Emerging Market Effect]
Next, to consider the possibility of other effects, this study extends the analysis by
using the control variables of VIX ( 1 tVIX ) to measure the market uncertainty and U.S.
Federal Fund Rate ( 1 tFFR ) for interest rate differentials. Higher volatility in the U.S. stocks
could affect the expectations about the future monetary policy stances of major central banks,
resulting in shifts of capital out from the U.S. and into Malaysia stock market. Furthermore,
international investors might take the interest rate differentials opportunity, to borrow in
currencies with low-interest rates and invest in a potential growth market, such as Malaysia,
to gain some better returns. From the findings, the VIX exhibits some degree of predictability
in the sense that the lagged variable of VIX is statistically significant in the empirical
analyses. However, both the control variables do not qualitatively change the main results.4
3.6 Conclusion
This study empirically examines the behavior of the Malaysian stock return and volatility
using the Threshold GARCH model for the period of 4 January 1994 to 31 December 2015.
4 Results of the additional control variables (VIX and US Federal Fund Rate) are not included for brevity. However, all results pertaining to this section are available at the Appendix 3.1(a) and (b) to 3.6(a) and (b).
68
From the perspective of behavioral finance, it is worthwhile to analyze the investor's behavior
before and after the general election in a socially collective market. Beside the full sample
period, this study divides the five general election periods into two stages. The first sub-
sample covers the 9th, 10th and 11th Malaysia general election from 1994 to 2005. This period
represents the general ups and downs periods where the existing parties continued to win 2/3
majority seats. The second sub-sample period represents drastic shock periods during the 12th
and 13th Malaysia general election, from 2006 to 2015. Interestingly, the finding of the first
sub-sample period is obviously different from the second sub-sample period.
For the first sub-sample period of 1994 to 2005, there is an asymmetric effect of
political elections on stock volatility. Moreover, there is a significant pre-general election
effect in the sectoral indices of Construction and Industrial Product. These two sectoral
indices had a significant positive return associated with low volatility before the general
election. Another five sectoral indices also recorded significant low stock volatility prior
general election, but no significant election effect in term of stock returns. The low volatility
in the market before the election is a good sign to indicate that there is no uncertainty due to
the general election. After the general election, there are seven sectoral indices encountered
significant high volatility. Even though there were no unexpected outcomes as the coalition
of Barisan Nasional won in the general elections, the stock volatility increased significantly
during the period of post-general election. Looking at the stable political condition at that
election year, the high volatility is not induced by the uncertainty of the general elections.
Nevertheless, it is possibly due to active trading activity in the market right after the election.
Investors were highly confident with the stable political condition in the country and they
started to trade actively after the market reopened after election dates.
69
For the period of 2006 to 2015, the results of the second sub-sample confirm the
asymmetric effect of pre-general election and post-general election periods on stock
volatility. Prior to the general election, most of the sectoral indices were highly volatile,
except for the Mining sectoral index with low volatility. The pre-general election results are
consistent with Lean and Yeap (2017), who found that volatility of the FTSE KLCI index
reacts positively before the election. According to the political condition during that period,
the high volatility in the market was due to uncertainties associated with the general election.
However, after the election, most of the sectoral indices results are insignificant. The sectoral
index of Technology is the only one that influences by the political uncertainties and shows
significant high volatility in the post-general election periods.
The examination of the Malaysian stock market performance by sector illustrates the impact
of general elections more precisely. Generally, the results of the selected sectoral indices are
in line with the sensitivity of industry type as mentioned in Tuyon and Ahmad (2016). The
cyclical sector of Construction, Finance, Mining, and Property are more sensitive to the
market condition with significant result found in stock volatility. While Consumer Product is
a defensive sector where the estimated results are mostly insignificant. Moreover, the results
also show that the volatility of the Malaysian stock market during the 12th and 13th general
election are different from the previous general election. Notably, while volatility on the
stock market return is low during the pre-general election periods of 1994-2005, it did show
its negative and significant influence in the 2008 and 2013 election years. The results of this
study clearly show that the election effect is different in the two sub-sample periods.
Therefore, future studies in this area should be caution in grouping the general election
periods. Furthermore, the results of the extension by using the emerging market index as an
alternative control variable, however, are very similar to the results of the main analysis.
70
Hence, the findings imply that the Threshold GARCH model used in this study is completely
robust after the model taking into consideration for few external factors.
Overall, the analysis results indicate that the Malaysian stock market volatility is
associated with the investors' behaviour during the periods of the general election. The
possible rationale is that whenever the political condition is stable in a country and investors
feel optimistic about the future of the economy under the ruling politic party, willingness to
trade in the stock market is higher. On the contrary, whenever there is political uncertainty,
interest to trade is much lower in the market. Therefore, this study is of great importance to
risk managers, portfolio managers, policymakers, and market participants to understand the
volatility in the Malaysian stock market during general election years. Thus, the results of this
study perhaps provide an insight for investors in adjusting their portfolio around the next
general election. Future work in this area can proceed in several directions. First, microdata
on investors' personal investment choices can be used to study their influence on stock
market performance during the general election. Second, future study can be conducted to
compare the market performance of different stocks characteristics to evaluate the volatility
during the general election.
71
References
Akerlof, G. A. and Shiller, R. J. (2009). “Animal spirits: How human psychology drives the economy, and why it matters for global capitalism”, New Jersey: Princeton University Press. Ali, N., Nassir, A. M., Hassan, T. and Abidin, S. Z. (2010). “Short Run Stock Overreaction: Evidence from Bursa Malaysia”, International Journal of Economics and Management, Vol. 4, No. 2, pp. 319 – 333. Allivine, F. D. and O’Neil, D. D. (1980). “Stock Market Returns and the Presidential Election Cycle”, Financial Analysts Journal, Vol. 36, No. 5, pp. 49 - 56. Bauman, W. (1989). “Investment research analysis in an emerging market: Singapore and Malaysia”, Financial Analysts Journal, pp. 60 - 67. Becher, D. A., Jensen, G. R. and Mercer, J. M. (2008). “Monetary policy indicators as predictors of stock returns”, Journal of Financial Research, Vol. 31, No.4, pp. 357–379. Bekaert, G. and Harvey, C. R. (2002). “Research in emerging markets finance: Looking to the future”, Emerging Markets Review, Vol. 3, pp. 429 - 448. Bialkowski, J., Gottachalk, K. and Wisniewski, T. P. (2008). “Stock Market Volatility around National Elections”, Journal of Banking and Finance, Vol. 32, No. 9, pp. 1941 - 1953. Bollerslev, T., Chou, R. and Kroner, K. (1992). “ARCH Modelling in Finance: A Selective Review of the Theory and Empirical Evidence”, Journal of Econometrics, Vol. 52, pp. 5 - 59. Campbell, J.Y. and Hentschel, L. (1992). “No News is Good News: An Asymmetric Model of Changing Volatility in Stock Returns”, Journal of Financial Economics, Vol. 31, No. 3, pp. 281 - 318. Cao, D. M., Long, W. and Yang, W. N. (2013). “Sector Indices Correlation Analysis in China’s Stock Market”, Procedia Computer Science, Vol. 17, pp. 1241 - 1249. Chau, F., Deesomsak, R. and Wang, J. (2014). “Political Uncertainty and Stock Market Volatility in the Middle East and North African (MENA) Countries”, Journal of International Financial Markets, Institutions & Money, Vol. 28, pp. 1 - 19. Chen, M. P., Chen, P. F. and Lee, C. C. (2013). “Asymmetric effects of investor sentiment on industry stock returns: Panel data evidence”, Emerging Markets Review, Vol 14, pp. 35 - 54. Chou, P. H., Ho, P. H. and Ko, K. C. (2012). “Do industries matter in explaining stock returns and asset-pricing anomalies?” Journal of Banking & Finance, Vol. 36, No. 2, 355 - 370. Dash, S. R. and Mahakud, J. (2013). “Investor sentiment and stock return: Do industries matter? ” Journal of Applied Economic Research, Vol. 7, No 3, 315 - 349. Foerster, S. R. and Schmitz, J. J. (1997). “The Transmission of U.S. Election Cycles to
72
International Stock Returns”, Journal of International Business, Vol. 28, No. 1, pp. 1 - 27. Gemmill, G. (1992). “Political Risk and Market Efficiency: Tests Based in British Stock and Options Markets in the 1987 Election”, Journal of Banking and Finance, Vol. 16, No. 1, pp. 211 - 231. Glosten, L. R., Jagannathan, R. and Runkle, D. E. (1993). “Relationship between the expected value and the volatility of the nominal excess return on stocks”, The Journal of Finance, Vol. 48, No. 5, pp. 1779 - 1801. Held, J. (2009). “Why it is (still) all about sectors – Sectors as a tool for asset allocation”, Journal of Indexes, Vol. 12, No. 5, pp. 10 - 56. Huang, R. D. (1985). “Common Stock Returns and Presidential Elections”, Financial Analysts Journal, Vol. 41, No. 2, pp. 58 - 65. Hung, L. C. (2011). “The Presidential Election and the Stock Market in Taiwan”, Journal of Business and Policy Research, Vol. 6, No. 2, pp. 36 - 48. Hung, L. C. (2013). “U.S. Presidential Elections and the Taiwanese Stock Market”, Issues & Studies, Vol. 49, No. 1, pp. 71 - 97. Jones, S. T. and Banning, K. (2009). “U.S. Elections and Monthly Stock Market Returns”, Journal of Economics and Finance, Vol. 33 No. 3, pp. 273 - 287. Kaplanski, G. and Levy, H. (2010). “Sentiment and stock prices: The case of aviation disaster”, Journal of Financial Economics, Vol. 95, pp. 174 - 201. Kearney, C. (2012). “Emerging markets research: Trends, issues and future directions”, Emerging Markets Review, Vol. 13, pp. 159 - 183. Kim, K. A. and Nofsinger, J. R. (2008). “Behavioral finance in Asia”, Pacific-Basin Finance Journal, Vol. 16, pp. 1 - 7. Lakshmi, S. P. (2013). “Volatility Patterns in Various Sectoral Indices in Indian Stock Market”, Global Journal of Management and Business Studies, Vol. 3, No. 8, pp. 879 - 886. Lawrence, E.R., McCabe, G. and Prakash, A.J. (2007). “Answering Financial Anomalies: Sentiment-based Stock Pricing”, Journal of Behavioural Finance, Vol. 8, No.3, pp.161 - 171. Leblang, D. and Mukherjee, B. (2005). “Government Partisanship, Elections and the Stock Market: Examining Americana and British Stock Returns 1930-2000”, American Journal of Political Science, Vol. 49, No. 4, pp. 780 - 802. Lean, H. H. (2010). “Political General Election and Stock Performance: The Malaysian Evidence”, in Ismail, M. T. and Mustafa, A. (Ed.), Research in Mathematics and Economics, Penang: Universiti Sains Malaysia, Malaysia, pp. 111 - 120. Lean, H. H. and Yeap, G. P. (2017). “Asymmetric Effect of Political Elections on Stock
73
Returns and Volatility in Malaysia”, in Munir, Q. and Kok, S. C. (Ed.), Information Efficiency and Anomalies in Asian Equity Markets, Routledge, Taylor and Francis Group, pp. 228 - 245. Lin, C. T. and Wang, Y.H. (2007). “The Impact of Party Alternative on the Stock Market: The Case of Japan”, Applied Economics, Vol. 39, No. 1, pp. 79 - 85. Lobo, B. J. (1999). “Jump Risk in the U.S. Stock Market: Evidence Using Political Information”, Review of Financial Economics, Vol. 8, No.2, pp. 149 - 163. Mei, J. P. and Guo, L. M. (2004). “Political Uncertainty, Financial Crisis and Market Volatility”, European Financial Management, Vol. 10, No. 4, pp. 639 - 657. Mohamad, S., Hassan T. and Ariff, M. (2007). “Research in an emerging Malaysian capital market: A guide to future direction” International Journal of Economics and Management, Vol. 1, No. 2, pp. 173 - 202. Nagy, Z. and Ruban, O. (2011). “Does style make the sector”, MSCI Applied Research, pp. 1 - 17. Niederhoffer, V., Gibbs, S. and Bullock, J. (1970). “Presidential Elections and the Stock Market”, Financial Analysts Journal, Vol. 26, No. 2, pp. 111 - 113. Nippani, S. and Arize, A. C. (2005). “U.S. Presidential Election Impact on Canadian and Mexican Stock Markets”, Journal of Economics and Finance, Vol. 29, No. 2, pp. 271 - 279. Nippani, S. and Medlin, W. B. (2002). “The 2000 Presidential Election and the Stock Market”, Journal of Economics and Finance, Vol. 26, No. 2, pp. 162 - 169. Nordhaus, W. (1975). “The Political Business Cycle”, Review of Economic Studies, Vol. 42, No. 2, pp. 169 - 190. Peel, D. and Pope, P. (1983). “General Election in the U.K. in the Post-1950 Period and the Behavior of the Stock Market”, Investment Analysis, Vol. 67, pp. 4 - 10. Ritter, J. R. (2003). “Behavioral finance”, Pacific-Basin Finance Journal, Vol. 11, pp. 429 - 437. Schmeling, M. (2009). “Investor sentiment and stock returns: Some international evidence”, Journal of Empirical Finance, Vol. 16, No. 3, pp. 394 - 408. Smales, L. A. (2016). “The role of political uncertainty in Australian financial markets”, Accounting and Finance, Vol. 56, No. 2, pp. 545–575. Statman, M., Fisher, K. L. and Anginer, D. (2008). “Affect in a behavioural asset-pricing model”, Financial Analysts Journal, Vol. 64, No. 2, pp. 20 - 29. Tuyon, J., Ahamd, Z. and Matahir, H. (2016). “The role of investor sentiment in malaysian stock market”, Asian Academy of Management Journal of Accounting and Finance, Vol. 12
74
Suppl. 1, pp. 43 - 75. Tuyon, J. and Ahamd, Z., (2016). “Behavioural finance perspectives on Malaysian stock market efficiency”, Borsa Istanbul Review "Vol.16, No.1, pp. 43 - 61. Wang, Y. H. and Lin, C. T. (2009). “The Political Uncertainty and Stock Market Behavior in Emerging Democracy: The Case of Taiwan”, Quality and Quantity, Vol. 43, No. 2, pp. 237 - 248. Wong, W. K. and McAleer, M. (2009). “Mapping the Presidential Election Cycle in U.S. Stock Market”, Mathematics and Computers in Simulation, Vol. 79, No. 11, pp. 3267 - 3277.
75
Table 3.1: Malaysia General Election Information
Dissolution of Parliament Election Date and Day 1st Parliament Assembly
after Election 9th General
Election 6 April 1995 (Thursday)
25 April 1995 (Tuesday)
7 June 1995 (Wednesday)
10th General Election
11 November 1999 (Wednesday)
29 November 1999 (Monday)
20 December 1999 (Monday)
11th General Election
4 March 2004 (Thursday)
21 March 2004 (Sunday)
17 May 2004 (Monday)
12th General Election
13 February 2008 (Wednesday)
8 March 2008 (Saturday)
28 April 2008 (Monday)
13th General Election
3 April 2013 (Wednesday)
5 May 2013 (Sunday)
24 June 2013 (Monday)
Sources: Suruhanjaya Pilihan Raya, Election Report, various years.
Table 3.2: Descriptive Statistics for the Malaysian Sectoral Indices (1994 - 2015)
KLCI CONST CONPR FIN IND INDPRO MNG PLANT PROP TRAD TECH
Mean 0.0047 -0.0075 0.0167 0.0084 0.0085 -0.0062 -0.0096 0.0120 -0.0185 -0.0003 -0.0446
Max 20.8174 23.9197 16.1281 22.6276 17.2483 18.9714 52.0143 16.9362 20.9022 22.3703 11.3668
Min -24.1534 -22.7828 -16.4773 -20.5651 -22.6965 -24.7880 -42.0379 -16.6592 -18.9174 -21.0987 -13.3861
Std. Dev. 1.3097 1.7787 1.0439 1.4683 1.2145 1.3035 2.9459 1.3692 1.5963 1.3945 1.5378
Skewness 0.4731 0.6526 0.1895 1.2226 -0.1577 -0.7173 0.7910 -0.2772 0.5177 0.8819 -0.0574
Kurtosis 58.5326 33.3929 51.6275 39.4080 54.3015 49.8949 46.6704 29.3345 24.8775 43.0376 11.2884
Jarque-Bera 737515.40 221254.80 565378.40 318344.50 629254.30 526268.20 456555.10 165878.50 114687.10 383997.20 11678.11
Probability 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
Note: KLCI: FTSE Bursa Malaysia KLCI Index, CONST: Construction, CONPR: Consumer Product, FIN: Finance, IND: Industrial, INDPRO: Industrial Product, MNG: Mining, PLANT: Plantation, PROP: Property, TRAD: Trade and Services, TECH: Technology (TECH data only available since May 15, 2000).
76
Table 3.3: Mean Returns on Pre-General Election and Post-General Election
1994 - 2005 KLCI CONST CONPR FIN IND INDPRO MNG PLANT PROP TRAD TECH
PreGE-Mean 0.0762 0.0167 0.0157 0.0311 -0.0297 0.0261 0.1972 0.0663 -0.0425 0.0586 0.1484
Observations 39.0000 39.0000 39.0000 39.0000 39.0000 39.0000 39.0000 39.0000 39.0000 39.0000 12.0000
PostGE-
Mean -0.0123 -0.1253 0.0617 -0.0205 0.0363 -0.0250 -0.0380 -0.0899 -0.1742 -0.0055 -0.5105
Observations 87.0000 87.0000 87.0000 87.0000 87.0000 87.0000 87.0000 87.0000 87.0000 87.0000 41.0000
2006 - 2015 KLCI CONST CONPR FIN IND INDPRO MNG PLANT PROP TRADSER TECH
PreGE-Mean -0.2039 -0.4886 -0.1524 -0.1920 -0.2428 -0.1513 -0.5291 -0.1223 -0.3323 -0.2182 -0.1881
Observations 41.0000 41.0000 41.0000 41.0000 41.0000 41.0000 41.0000 41.0000 41.0000 41.0000 41.0000
PostGE-
Mean 0.0340 0.0688 0.1195 0.0787 0.0311 0.1137 0.1909 0.0700 0.0765 0.0406 0.2037
Observations 72.0000 72.0000 72.0000 72.0000 72.0000 72.0000 72.0000 72.0000 72.0000 72.0000 72.0000
Notes: KLCI: FTSE Bursa Malaysia KLCI Index, CONST: Construction, CONPR: Consumer Product, FIN: Finance, IND: Industrial, INDPRO: Industrial Product, MNG: Mining, PLANT: Plantation, PROP: Property, TRAD: Trade and Services, TECH: Technology. Pre-General Election: start from Dissolution of Parliament to the day before General Election, and Post-General Election: start from Day after the General Election to the first day of the Parliament Assembly.
77
Table 3.4(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) - Controlled by World Market Effect
Variables KLCI Construction Consumer
Product Finance Industrial
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0081 (0.3549)
-0.0027 (0.8461)
0.0205 (0.0080)***
0.0153 (0.1446)
0.0141 (0.1412)
PGE 0.1226 (0.2740)
0.1401 (0.5256)
-0.0157 (0.8686)
0.0571 (0.7352)
-0.0269 (0.7994)
PtGE -0.0783 (0.2195)
-0.2170 (0.1263)
0.0083 (0.8624)
-0.0704 (0.4133)
-0.0658 (0.3164)
1tR 0.0925 (0.0000)***
0.0721 (0.0000)***
0.0645 (0.0000)***
0.1139 (0.0000)***
0.0354 (0.0079)***
1tRWM 0.2244 (0.0000)***
0.2774 (0.0000)***
0.1603 (0.0000)***
0.2396 (0.0000)***
0.1942 (0.0000)***
Variance Equation
0 0.0057 (0.0000)***
0.0218 (0.0000)***
0.0040 (0.0000)***
0.0068 (0.0000)***
0.0069 (0.0000)***
1 0.0538 (0.0000)***
0.0663 (0.0000)***
0.0468 (0.0000)***
0.0624 (0.0000)***
0.0446 (0.0000)***
i 0.0665 (0.0000)***
0.0756 (0.0000)***
0.0377 (0.0000)***
0.0458 (0.0000)***
0.0512 (0.0000)***
1 0.9106 (0.0000)***
0.8926 (0.0000)***
0.9299 (0.0000)***
0.9132 (0.0000)***
0.9240 (0.0000)***
PGE 0.0479 (0.0041)***
0.2568 (0.0000)***
0.0090 (0.0801)*
0.0746 (0.0000)***
0.0529 (0.0001)***
PtGE 0.0006 (0.9093)
0.0224 (0.0615)*
0.0005 (0.8230)
0.0011 (0.8850)
-0.0006 (0.9013)
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.2865 0.5556 0.2209 0.0023 0.9160 10 lags 0.3826 0.4533 0.3500 0.0057 0.9617
Ljung-Box Q2 Statistic (p-value) 5 lags 0.2840 0.5500 0.2070 0.0020 0.9170
10 lags 0.3370 0.4010 0.3030 0.0030 0.9610 Return Equation: Wald Test (p-value)
F-stat 0.1405 0.1426 0.9704 0.6149 0.6031 Chi-Square 0.1404 0.1425 0.9704 0.6149 0.6030
Variance Equation: Wald Test (p-value) F-stat 0.0108 0.0000 0.1791 0.0001 0.0003
Chi-Square 0.0108 0.0000 0.1790 0.0001 0.0002 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
78
Table 3.4(b): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) - Controlled by World Market Effect
Variables Industrial
Product Mining Plantation Property Trade and
Services Technology
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0039 (0.6898)
0.0378 (0.2308)
0.0142 (0.2255)
-0.0172 (0.1422)
0.0044 (0.6346)
-0.0380 (0.0350)**
PGE 0.0798 (0.5014)
-0.2898 (0.2649)
0.0701 (0.5091)
0.1208 (0.5663)
0.1329 (0.2989)
0.0072 (0.9684)
PtGE -0.0117 (0.8482)
-0.1277 (0.4857)
-0.0055 (0.9471)
-0.1226 (0.2179)
-0.0915 (0.2112)
-0.0982 (0.4705)
1tR 0.0696 (0.0000)***
-0.0612 (0.0000)***
0.1047 (0.0000)***
0.1242 (0.0000)***
0.0572 (0.0000)***
0.1189 (0.0000)***
1tRWM 0.2068 (0.0000)***
0.3195 (0.0000)***
0.1948 (0.0000)***
0.2213 (0.0000)***
0.2173 (0.0000)***
0.2330 (0.0000)***
Variance Equation
0 0.0139 (0.0000)***
0.4406 (0.0000)***
0.0208 (0.0000)***
0.0157 (0.0000)***
0.0048 (0.0000)***
0.0343 (0.0000)***
1 0.0798 (0.0000)***
0.1244 (0.0000)***
0.0894 (0.0000)***
0.1172 (0.0000)***
0.0483 (0.0000)***
0.0794 (0.0000)***
i 0.0790 (0.0000)***
0.1302 (0.0000)***
0.0450 (0.0000)***
0.0285 (0.0000)***
0.0758 (0.0000)***
0.0206 (0.0046)***
1 0.8742 (0.0000)***
0.7855 (0.0000)***
0.8765 (0.0000)***
0.8698 (0.0000)***
0.9144 (0.0000)***
0.8964 (0.0000)***
PGE 0.0486 (0.0143)**
-0.2188 (0.0012)***
0.0002 (0.9893)
0.2440 (0.0000)***
0.0645 (0.0020)***
0.0631 (0.0071)***
PtGE 0.0135 (0.1264)
0.0462 (0.6804)
0.0312 (0.0143)**
0.0170 (0.1021)
-0.0034 (0.5671)
0.0698 (0.0000)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.5444 0.9999 0.0154 0.4935 0.4044 0.0120 10 lags 0.6203 0.9996 0.0252 0.6618 0.4799 0.0565
Ljung-Box Q2 Statistic (p-value) 5 lags 0.5430 1.0000 0.0160 0.4810 0.4140 0.0120 10 lags 0.5900 1.0000 0.0190 0.6290 0.4540 0.0480
Return Equation: Wald Test (p-value) F-stat 0.7502 0.4159 0.8025 0.2827 0.1349 0.7707
Chi-Square 0.7502 0.4158 0.8025 0.2826 0.1348 0.7707 Variance Equation: Wald Test (p-value)
F-stat 0.0037 0.0053 0.0433 0.0000 0.0083 0.0000 Chi-Square 0.0037 0.0053 0.0432 0.0000 0.0083 0.0000
Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
79
Table 3.5(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2005) - Controlled by World Market Effect
Variables KLCI Construction Consumer
Product Finance Industrial
(p, q) (1, 1) (1, 0) (1, 1) (1, 1) (0, 1) Mean Equation
0 -0.0089 (0.5590)
-0.0849 (0.0118)**
0.0075 (0.5510)
0.0032 (0.8662)
0.0149 (0.3229)
PGE 0.1812 (0.3122)
0.2991 (0.0267)**
0.0973 (0.5496)
0.1125 (0.5832)
-0.0169 (0.9482)
PtGE -0.1225 (0.2268)
-0.0545 (0.7923)
-0.0566 (0.3920)
-0.1914 (0.1898)
-0.1189 (0.2530)
1tR 0.1274 (0.0000)***
0.0093 (0.3114)
0.1221 (0.0000)***
0.1534 (0.0000)***
0.0585 (0.0001)***
1tRWM 0.2353 (0.0000)***
0.4677 (0.0000)***
0.1448 (0.0000)***
0.2879 (0.0000)***
0.1723 (0.0000)***
Variance Equation
0 0.0088 (0.0000)***
2.4997 (0.0000)***
0.0042 (0.0000)***
0.0156 (0.0000)***
0.0057 (0.0000)***
1 0.0464 (0.0000)***
0.3265 (0.0000)***
0.0306 (0.0000)***
0.0584 (0.0000)***
-- --
i 0.0752 (0.0000)***
0.1947 (0.0000)***
0.0438 (0.0000)***
0.0559 (0.0000)***
0.0745 (0.0000)***
1 0.9152 (0.0000)***
-- --
0.9448 (0.0000)***
0.9104 (0.0000)***
0.9583 (0.0000)***
PGE -0.0585 (0.0503)*
-2.0862 (0.0000)***
-0.0035 (0.7338)
-0.0785 (0.0751)*
0.0885 (0.0034)***
PtGE 0.0437 (0.0075)***
-0.6282 (0.0000)***
0.0030 (0.6081)
0.0900 (0.0000)***
-0.0134 (0.1883)
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.1877 0.0000 0.2209 0.0206 0.0000 10 lags 0.4197 0.0000 0.5794 0.0355 0.0000
Ljung-Box Q2 Statistic (p-value) 5 lags 0.1890 0.0000 0.2090 0.0180 0.0000
10 lags 0.3920 0.0000 0.5250 0.0290 0.0000 Return Equation: Wald Test (p-value)
F-stat 0.3209 0.0806 0.6099 0.4005 0.5157 Chi-Square 0.3208 0.0805 0.6099 0.4004 0.5156
Variance Equation: Wald Test (p-value) F-stat 0.0258 0.0000 0.8671 0.0001 0.0098
Chi-Square 0.0256 0.0000 0.8671 0.0001 0.0097 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
80
Table 3.5(b): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2005) - Controlled by World Market Effect
Variables Industrial
Product Mining Plantation Property Trade and
Services Technology
(p, q) (1, 1) (1, 1) (0, 1) (1, 0) (1, 1) (1, 1) Mean Equation
0 -0.0333 (0.0275)**
-0.0161 (0.6811)
0.0095 (0.6016)
-0.0716 (0.0039)***
-0.0179 (0.3017)
-0.1027 (0.0015)***
PGE 0.2455 (0.0063)***
0.2758 (0.3284)
0.1677 (0.2937)
0.3339 (0.1204)
0.2351 (0.1899)
0.2444 (0.1025)
PtGE -0.0793 (0.4388)
-0.3505 (0.1631)
-0.0831 (0.3978)
-0.0752 (0.7756)
-0.1272 (0.1648)
-0.3136 (0.0075)***
1tR 0.0922 (0.0000)***
0.0362 (0.0476)**
0.1282 (0.0000)***
0.2193 (0.0000)***
0.0921 (0.0000)***
0.1384 (0.0000)***
1tRWM 0.2050 (0.0000)***
0.2401 (0.0000)***
0.1575 (0.0000)***
0.3476 (0.0000)***
0.2465 (0.0000)***
0.3884 (0.0000)***
Variance Equation
0 0.0136 (0.0000)***
0.1604 (0.0000)***
0.0199 (0.0000)***
1.6640 (0.0000)***
0.0071 (0.0000)***
-0.0003 (0.7331)
1 0.0788 (0.0000)***
0.0953 (0.0000)***
-- --
0.5432 (0.0000)***
0.0428 (0.0000)***
0.0045 (0.0826)*
i 0.1046 (0.0000)***
0.0845 (0.0000)***
0.1075 (0.0000)***
0.2151 (0.0002)***
0.0797 (0.0000)***
0.0233 (0.0000)***
1 0.8742 (0.0000)***
0.8548 (0.0000)***
0.9320 (0.0000)***
-- --
0.9204 (0.0000)***
0.9833 (0.0000)***
PGE -0.0630 (0.0000)***
-0.3728 (0.0001)***
-0.0409 (0.0556)*
-0.9755 (0.0000)***
-0.0896 (0.0002)***
-0.0296 (0.1211)
PtGE 0.0646 (0.0021)***
0.2623 (0.0214)**
0.0297 (0.0171)**
0.9104 (0.0000)***
0.0332 (0.0152)**
0.0174 (0.0017)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.5556 0.3317 0.0000 0.0000 0.0869 0.0000 10 lags 0.8104 0.5875 0.0000 0.0000 0.2191 0.0000
Ljung-Box Q2 Statistic (p-value) 5 lags 0.5390 0.3300 0.0000 0.0000 0.0940 0.0000 10 lags 0.7900 0.5930 0.0000 0.0000 0.2200 0.0000
Return Equation: Wald Test (p-value) F-stat 0.0096 0.2499 0.3937 0.2861 0.1627 0.0067
Chi-Square 0.0096 0.2497 0.3936 0.2859 0.1625 0.0066 Variance Equation: Wald Test (p-value)
F-stat 0.0000 0.0005 0.0352 0.0000 0.0004 0.0061 Chi-Square 0.0000 0.0005 0.0350 0.0000 0.0004 0.0060
Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
81
Table 3.6(a): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) - Controlled by World Market Effect
Variables KLCI Construction Consumer
Product Finance Industrial
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0216 (0.0429)**
0.0227 (0.1886)
0.0364 (0.0004)***
0.0282 (0.0232)**
0.0145 (0.2494)
PGE 0.1135 (0.5588)
-0.1558 (0.8012)
-0.0750 (0.4723)
0.0221 (0.9340)
-0.0107 (0.9455)
PtGE -0.0542 (0.6658)
-0.1594 (0.3982)
0.1355 (0.0356)**
-0.0131 (0.9162)
-0.0477 (0.7123)
1tR 0.0567 (0.0052)***
0.0406 (0.0584)*
-0.0016 (0.9402)
0.0683 (0.0011)***
0.0244 (0.2295)
1tRWM 0.2230 (0.0000)***
0.2583 (0.0000)***
0.1620 (0.0000)***
0.2280 (0.0000)***
0.2046 (0.0000)***
Variance Equation
0 0.0120 (0.0000)***
0.0324 (0.0000)***
0.0209 (0.0000)***
0.0205 (0.0000)***
0.0141 (0.0000)***
1 0.0723 (0.0000)***
0.0977 (0.0000)***
0.0966 (0.0000)***
0.1086 (0.0000)***
0.0349 (0.0001)***
i 0.0748 (0.0000)***
0.0697 (0.0000)***
0.0657 (0.0000)***
0.0652 (0.0001)***
0.0745 (0.0000)***
1 0.8627 (0.0000)***
0.8454 (0.0000)***
0.8077 (0.0000)***
0.8233 (0.0000)***
0.9015 (0.0000)***
PGE 0.1139 (0.0092)***
0.9667 (0.0000)***
0.0179 (0.1869)
0.1745 (0.0004)***
0.0898 (0.0015)***
PtGE 0.0008 (0.9040)
-0.0365 (0.2144)
-0.0073 (0.1172)
0.0048 (0.6069)
-0.0026 (0.6623)
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.9859 0.9655 0.7149 0.7217 0.9640 10 lags 0.5712 0.9110 0.4356 0.7667 0.9958
Ljung-Box Q2 Statistic (p-value) 5 lags 0.9870 0.9630 0.7010 0.7130 0.9640
10 lags 0.5590 0.9060 0.4290 0.7370 0.9960 Return Equation: Wald Test (p-value)
F-stat 0.6819 0.6947 0.0813 0.9870 0.9342 Chi-Square 0.6818 0.6947 0.0811 0.9870 0.9342
Variance Equation: Wald Test (p-value) F-stat 0.0300 0.0000 0.1166 0.0012 0.0064
Chi-Square 0.0298 0.0000 0.1164 0.0012 0.0063 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
82
Table 3.6(b): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) - Controlled by World Market Effect
Variables Industrial
Product Mining Plantation Property Trade and
Services Technology
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0355 (0.0071)***
0.0867 (0.0719)*
0.0199 (0.2303)
0.0225 (0.1431)
0.0170 (0.1217)
0.0055 (0.7993)
PGE -0.1005 (0.6673)
-0.8343 (0.0067)***
0.0062 (0.9701)
-0.3028 (0.4974)
0.0897 (0.7309)
-0.2408 (0.3007)
PtGE 0.0651 (0.5884)
0.0319 (0.9111)
0.0137 (0.9108)
0.0947 (0.6387)
-0.0687 (0.5838)
0.2014 (0.2995)
1tR 0.0395 (0.0607)*
-0.1900 (0.0000)***
0.0906 (0.0000)***
0.0922 (0.0000)***
0.0197 (0.2973)
0.0869 (0.0000)***
1tRWM 0.2099 (0.0000)***
0.3426 (0.0000)***
0.2357 (0.0000)***
0.2151 (0.0000)***
0.2120 (0.0000)***
0.1726 (0.0000)***
Variance Equation
0 0.0230 (0.0000)***
0.8847 (0.0000)***
0.0155 (0.0000)***
0.0215 (0.0000)***
0.0085 (0.0000)***
0.1596 (0.0000)***
1 0.0890 (0.0000)***
0.1666 (0.0000)***
0.0624 (0.0000)***
0.1260 (0.0000)***
0.0551 (0.0000)***
0.1349 (0.0000)***
i 0.0488 (0.0000)***
0.2124 (0.0000)***
0.0301 (0.0001)***
-0.0073 (0.5003)
0.0726 (0.0000)***
0.0341 (0.0402)*
1 0.8459 (0.0000)***
0.6787 (0.0000)***
0.9084 (0.0000)***
0.8561 (0.0000)***
0.8907 (0.0000)***
0.7536 (0.0000)***
PGE 0.1322 (0.0000)***
-0.3065 (0.0561)*
0.0413 (0.1107)
0.5704 (0.0000)***
0.1571 (0.0020)***
0.0502 (0.4033)
PtGE 0.0064 (0.4738)
-0.1674 (0.4616)
-0.0036 (0.8040)
-0.0255 (0.3699)
-0.0105 (0.1309)
0.4001 (0.0000)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.8076 0.9999 0.1851 0.7353 0.9263 0.8261 10 lags 0.7050 1.0000 0.1459 0.8317 0.6053 0.9279
Ljung-Box Q2 Statistic (p-value) 5 lags 0.8080 1.0000 0.1990 0.7340 0.9320 0.8230 10 lags 0.7030 1.0000 0.1410 0.8270 0.6000 0.9200
Return Equation: Wald Test (p-value) F-stat 0.7755 0.0254 0.9933 0.6707 0.7595 0.4092
Chi-Square 0.7755 0.0252 0.9933 0.6706 0.7595 0.4091 Variance Equation: Wald Test (p-value)
F-stat 0.0000 0.1144 0.2801 0.0000 0.0054 0.0000 Chi-Square 0.0000 0.1142 0.2799 0.0000 0.0053 0.0000
Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
83
Table 3.7(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) - Controlled by Emerging Market Effect
Variables KLCI Construction Consumer
Product Finance Industrial
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0097 (0.2845)
0.0004 (0.9750)
0.0226 (0.0046)***
0.0196 (0.0707)*
0.0126 (0.1965)
PGE 0.1435 (0.1704)
0.1480 (0.4842)
-0.0035 (0.9712)
0.0856 (0.5882)
-0.0150 (0.8794)
PtGE -0.0524 (0.4159)
-0.1957 (0.1763)
0.0229 (0.6326)
-0.0519 (0.5517)
-0.0417 (0.5188)
1tR 0.0663 (0.0000)***
0.0609 (0.0000)***
0.0354 (0.0099)***
0.1011 (0.0000)***
0.0056 (0.6947)
1tEWM 0.1295 (0.0000)***
0.1451 (0.0000)***
0.1048 (0.0000)***
0.1222 (0.0000)***
0.1345 (0.0000)***
Variance Equation
0 0.0064 (0.0000)***
0.0246 (0.0000)***
0.0045 (0.0000)***
0.0075 (0.0000)***
0.0072 (0.0000)***
1 0.0543 (0.0000)***
0.0650 (0.0000)***
0.0492 (0.0000)***
0.0617 (0.0000)***
0.0459 (0.0000)***
i 0.0692 (0.0000)***
0.0778 (0.0000)***
0.0402 (0.0000)***
0.0510 (0.0000)***
0.0534 (0.0000)***
1 0.9082 (0.0000)***
0.8911 (0.0000)***
0.9259 (0.0000)***
0.9112 (0.0000)***
0.9217 (0.0000)***
PGE 0.0400 (0.0083)***
0.2480 (0.0000)***
0.0096 (0.1007)
0.0690 (0.0000)***
0.0495 (0.0002)***
PtGE 0.0047 (0.4197)
0.0286 (0.0210)**
-0.0001 (0.9741)
0.0046 (0.5675)
-0.0007 (0.8982)
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.3082 0.4561 0.4210 0.0019 0.9449 10 lags 0.4629 0.4345 0.4863 0.0047 0.9621
Ljung-Box Q2 Statistic (p-value) 5 lags 0.3040 0.4500 0.4050 0.0020 0.9450
10 lags 0.4140 0.3820 0.4420 0.0030 0.9600 Return Equation: Wald Test (p-value)
F-stat 0.1769 0.1787 0.8904 0.6510 0.8104 Chi-Square 0.1768 0.1786 0.8904 0.6510 0.8104
Variance Equation: Wald Test (p-value) F-stat 0.0122 0.0000 0.2485 0.0001 0.0005
Chi-Square 0.0122 0.0000 0.2484 0.0001 0.0005 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
84
Table 3.7(b): Threshold GARCH Results for Pre-General Election and Post-General
Election (1994 - 2015) - Controlled by Emerging Market Effect
Variables Industrial Product
Mining Plantation Property Trade and Services
Technology
(p, q) (1, 2) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0038 (0.7017)
0.0300 (0.3351)
0.0146 (0.2053)
-0.0129 (0.2822)
0.0059 (0.5325)
-0.0355 (0.0487)**
PGE 0.1135 (0.3057)
-0.2391 (0.3551)
0.1110 (0.2847)
0.1315 (0.5327)
0.1537 (0.1848)
0.0131 (0.9483)
PtGE -0.0054 (0.9282)
-0.1113 (0.5441)
-0.0018 (0.9823)
-0.1177 (0.2554)
-0.0621 (0.4017)
-0.0830 (0.5359)
1tR 0.0427 (0.0061)***
-0.0722 (0.0000)***
0.0726 (0.0000)***
0.1186 (0.0000)***
0.0378 (0.0089)***
0.1133 (0.0000)***
1tEWM 0.1304 (0.0000)***
0.2293 (0.0000)***
0.1561 (0.0000)***
0.1048 (0.0000)***
0.1204 (0.0000)***
0.1001 (0.0000)***
Variance Equation
0 0.0172 (0.0000)***
0.4381 (0.0000)***
0.0219 (0.0000)***
0.0170 (0.0000)***
0.0055 (0.0000)***
0.0376 (0.0000)***
1 0.0925 (0.0000)***
0.1213 (0.0000)***
0.0968 (0.0000)***
0.1155 (0.0000)***
0.0516 (0.0000)***
0.0865 (0.0000)***
i 0.0940 (0.0000)***
0.1238 (0.0000)***
0.0501 (0.0000)***
0.0289 (0.0000)***
0.0779 (0.0000)***
0.0226 (0.0032)***
1 0.6506 (0.0000)***
0.7902 (0.0000)***
0.8673 (0.0000)***
0.8699 (0.0000)***
0.9097 (0.0000)***
0.8881 (0.0000)***
2 0.2012 (0.0320)**
-- --
-- --
-- --
-- --
-- --
PGE 0.0453 (0.0660)*
-0.2445 (0.0003)***
-0.0041 (0.8246)
0.2456 (0.0000)***
0.0567 (0.0033)***
0.0736 (0.0050)***
PtGE 0.0165 (0.1247)
0.0496 (0.6514)
0.0379 (0.0078)***
0.0203 (0.0630)*
0.0002 (0.9715)
0.0693 (0.0000)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.9847 0.9999 0.0177 0.5028 0.4586 0.0146 10 lags 0.9584 0.9996 0.0271 0.5687 0.5757 0.0763
Ljung-Box Q2 Statistic (p-value) 5 lags 0.9850 1.0000 0.0180 0.4920 0.4650 0.0140 10 lags 0.9540 1.0000 0.0210 0.5410 0.5490 0.0660
Return Equation: Wald Test (p-value) F-stat 0.5578 0.5307 0.5641 0.2920 0.1551 0.8248
Chi-Square 0.5578 0.5307 0.5640 0.2919 0.1550 0.8248 Variance Equation: Wald Test (p-value)
F-stat 0.0280 0.0014 0.0286 0.0000 0.0122 0.0000 Chi-Square 0.0280 0.0014 0.0286 0.0000 0.0121 0.0000
Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
85
Table 3.8(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2005) - Controlled by Emerging Market Effect
Variables KLCI Construction Consumer
Product Finance Industrial
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (0, 1) Mean Equation
0 -0.0107 (0.5032)
-0.0383 (0.1127)
0.0065 (0.6157)
0.0042 (0.8303)
0.0118 (0.4422)
PGE 0.1759 (0.3341)
0.1348 (0.3223)
0.1019 (0.5300)
0.1192 (0.5280)
-0.0107 (0.9648)
PtGE -0.1300 (0.2418)
-0.2250 (0.1747)
-0.0489 (0.4619)
-0.2013 (0.1824)
-0.0950 (0.3466)
1tR 0.1127 (0.0000)***
0.0957 (0.0000)***
0.1017 (0.0000)***
0.1386 (0.0000)***
0.0404 (0.0110)**
1tEWM 0.1119 (0.0000)***
0.1349 (0.0000)***
0.0979 (0.0000)***
0.1353 (0.0000)***
0.1086 (0.0000)***
Variance Equation
0 0.0101 (0.0000)***
0.0328 (0.0000)***
0.0043 (0.0000)***
0.0178 (0.0000)***
0.0061 (0.0000)***
1 0.0443 (0.0000)***
0.0439 (0.0000)***
0.0305 (0.0000)***
0.0568 (0.0000)***
-- --
i 0.0812 (0.0000)***
0.0803 (0.0000)***
0.0452 (0.0000)***
0.0616 (0.0000)***
0.0798 (0.0000)***
1 0.9135 (0.0000)***
0.9100 (0.0000)***
0.9443 (0.0000)***
0.9085 (0.0000)***
0.9559 (0.0000)***
PGE -0.0571 (0.0631)*
-0.1004 (0.0000)***
-0.0033 (0.7836)
-0.0866 (0.0414)**
0.0639 (0.0341)**
PtGE 0.0509 (0.0049)***
0.1132 (0.0000)***
0.0038 (0.5524)
0.0992 (0.0000)***
-0.0072 (0.5604)
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.2852 0.0904 0.3208 0.0156 0.0000 10 lags 0.6075 0.2865 0.6987 0.0323 0.0000
Ljung-Box Q2 Statistic (p-value) 5 lags 0.2840 0.0860 0.3100 0.0130 0.0000
10 lags 0.5920 0.2300 0.6540 0.0280 0.0000 Return Equation: Wald Test (p-value)
F-stat 0.3558 0.2483 0.6480 0.3705 0.6372 Chi-Square 0.3556 0.2482 0.6480 0.3703 0.6371
Variance Equation: Wald Test (p-value) F-stat 0.0191 0.0000 0.8381 0.0001 0.0546
Chi-Square 0.0190 0.0000 0.8381 0.0001 0.0545 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
86
Table 3.8(b): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2005) - Controlled by Emerging Market Effect
Variables Industrial
Product Mining Plantation Property Trade and
Services Technology
(p, q) (1, 0) (2, 1) (0, 1) (2, 1) (1, 1) (1, 1) Mean Equation
0 -0.0629 (0.0003)***
-0.0272 (0.4882)
0.0072 (0.6929)
-0.0726 (0.0003)***
-0.0196 (0.2708)
-0.1035 (0.0014)***
PGE 0.1215 (0.3925)
0.3327 (0.2292)
0.1816 (0.2190)
0.3627 (0.0139)**
0.2239 (0.2199)
0.2348 (0.1676)
PtGE 0.0308 (0.8414)
-0.3968 (0.1418)
-0.0811 (0.3829)
-0.1582 (0.0826)*
-0.1283 (0.1853)
-0.3085 (0.0206)**
1tR 0.0933 (0.0000)***
0.0240 (0.2422)
0.1188 (0.0000)***
0.1288 (0.0000)***
0.0779 (0.0000)***
0.1335 (0.0000)***
1tEWM 0.2724 (0.0000)***
0.1459 (0.0000)***
0.0989 (0.0000)***
0.1247 (0.0000)***
0.1209 (0.0000)***
0.1380 (0.0000)***
Variance Equation
0 1.1561 (0.0000)***
0.1163 (0.0000)***
0.0209 (0.0000)***
0.0153 (0.0000)***
0.0082 (0.0000)***
-0.0004 (0.6803)
1 0.3568 (0.0000)***
0.1743 (0.0000)***
-- --
0.1117 (0.0000)***
0.0441 (0.0000)***
0.0050 (0.0587)*
2 -- --
-0.1024 (0.0000)***
-- --
0.0498 (0.0000)***
-- --
-- --
i 0.3589 (0.0000)***
0.0742 (0.0000)***
0.1135 (0.0000)***
0.8747 (0.0000)***
0.0842 (0.0000)***
0.0245 (0.0000)***
1 -- --
0.8858 (0.0000)***
0.9289 (0.0000)***
0.8747 (0.0000)***
0.9167 (0.0000)***
0.9822 (0.0000)***
PGE -0.7475 (0.0000)***
-0.3286 (0.0000)***
-0.0529 (0.0050)***
-0.1001 (0.0001)***
-0.0949 (0.0014)***
-0.0191 (0.3395)
PtGE 0.2915 (0.1050)
0.3250 (0.0008)***
0.0378 (0.0027)***
0.0332 (0.0203)**
0.0445 (0.0072)***
0.0164 (0.0077)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.0000 0.8807 0.0000 0.0976 0.1348 0.0000 10 lags 0.0000 0.9518 0.0000 0.2458 0.3195 0.0001
Ljung-Box Q2 Statistic (p-value) 5 lags 0.0000 0.8800 0.0000 0.0820 0.1420 0.0000 10 lags 0.0000 0.9520 0.0000 0.2060 0.3330 0.0000
Return Equation: Wald Test (p-value) F-stat 0.6825 0.1650 0.3073 0.0135 0.2119 0.0249
Chi-Square 0.6824 0.1649 0.3072 0.0134 0.2118 0.0247 Variance Equation: Wald Test (p-value)
F-stat 0.0000 0.0000 0.0018 0.0002 0.0022 0.0186 Chi-Square 0.0000 0.0000 0.0018 0.0002 0.0022 0.0184
Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
87
Table 3.9(a): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) - Controlled by Emerging Market Effect
Variables KLCI Construction Consumer
Product Finance Industrial
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0257 (0.0201)**
0.0297 (0.0889)*
0.0390 (0.0002)***
0.0319 (0.0146)**
0.0138 (0.2867)
PGE 0.1500 (0.3667)
-0.1126 (0.8526)
-0.0533 (0.6090)
0.0674 (0.7731)
-0.0030 (0.9836)
PtGE -0.0263 (0.8270)
-0.1134 (0.5282)
0.1518 (0.0182)**
-0.0093 (0.9424)
-0.0099 (0.9363)
1tR 0.0073 (0.7565)
0.0214 (0.3695)
-0.0468 (0.0404)**
0.0477 (0.0422)**
-0.0235 (0.2804)
1tEWM 0.1466 (0.0000)***
0.1545 (0.0000)***
0.1129 (0.0000)***
0.1283 (0.0000)***
0.1509 (0.0000)***
Variance Equation
0 0.0119 (0.0000)***
0.0347 (0.0000)***
0.0237 (0.0000)***
0.0175 (0.0000)***
0.0133 (0.0000)***
1 0.0726 (0.0000)***
0.1040 (0.0000)***
0.0922 (0.0000)***
0.0919 (0.0000)***
0.0376 (0.0000)***
i 0.0682 (0.0000)***
0.0661 (0.0000)***
0.0752 (0.0000)***
0.0640 (0.0000)***
0.0687 (0.0000)***
1 0.8674 (0.0000)***
0.8404 (0.0000)***
0.8007 (0.0000)***
0.8474 (0.0000)***
0.9039 (0.0000)***
PGE 0.0875 (0.0163)**
0.9875 (0.0000)***
0.0165 (0.2918)
0.1397 (0.0008)***
0.0825 (0.0011)***
PtGE 0.0019 (0.7944)
-0.0431 (0.1994)
-0.0102 (0.0699)*
0.0034 (0.7248)
-0.0038 (0.5531)
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.8866 0.9446 0.7148 0.3830 0.9516 10 lags 0.4240 0.7896 0.1422 0.4259 0.9745
Ljung-Box Q2 Statistic (p-value) 5 lags 0.8890 0.9410 0.7050 0.3780 0.9510
10 lags 0.4030 0.7810 0.1340 0.3590 0.9710 Return Equation: Wald Test (p-value)
F-stat 0.5654 0.8167 0.0522 0.9405 0.9968 Chi-Square 0.5653 0.8167 0.0520 0.9405 0.9968
Variance Equation: Wald Test (p-value) F-stat 0.0416 0.0000 0.1063 0.0020 0.0047
Chi-Square 0.0414 0.0000 0.1061 0.0020 0.0046 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
88
Table 3.9(b): Threshold GARCH Results for Pre-General Election and Post-General
Election (2006 - 2015) - Controlled by Emerging Market Effect
Variables Industrial Product
Mining Plantation Property Trade and Services
Technology
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0387 (0.0040)***
0.0766 (0.1107)
0.0248 (0.1230)
0.0273 (0.0847)*
0.0205 (0.0690)*
0.0072 (0.7401)
PGE -0.0490 (0.8271)
-0.7956 (0.0160)**
0.0522 (0.7557)
-0.2578 (0.5716)
0.1367 (0.5332)
-0.2316 (0.3433)
PtGE 0.0830 (0.4813)
0.1037 (0.7090)
0.0690 (0.5866)
0.1000 (0.6126)
-0.0289 (0.8137)
0.2073 (0.2732)
1tR 0.0102 (0.6667)
-0.1988 (0.0000)***
0.0342 (0.0815)*
0.0865 (0.0003)***
-0.0164 (0.4638)
0.0778 (0.0004)***
1tEWM 0.1282 (0.0000)***
0.2417 (0.0000)***
0.1975 (0.0000)***
0.1047 (0.0000)***
0.1306 (0.0000)***
0.0911 (0.0000)***
Variance Equation
0 0.0251 (0.0000)***
0.8586 (0.0000)***
0.0174 (0.0000)***
0.0240 (0.0000)***
0.0093 (0.0000)***
0.1587 (0.0000)***
1 0.0860 (0.0000)***
0.1603 (0.0000)***
0.0774 (0.0000)***
0.1228 (0.0000)***
0.0602 (0.0000)***
0.1419 (0.0000)***
i 0.0519 (0.0000)***
0.1862 (0.0000)***
0.0415 (0.0000)***
-0.0005 (0.9646)
0.0716 (0.0000)***
0.0365 (0.0347)**
1 0.8449 (0.0000)***
0.6948 (0.0000)***
0.8887 (0.0000)***
0.8533 (0.0000)***
0.8859 (0.0000)***
0.7490 (0.0000)***
PGE 0.1203 (0.0002)***
-0.3375 (0.0227)**
0.0418 (0.1933)
0.5797 (0.0000)***
0.1366 (0.0044)***
0.0538 (0.3727)
PtGE 0.0074 (0.4373)
-0.1946 (0.3538)
0.0047 (0.8124)
-0.0220 (0.4888)
-0.0097 (0.2209)
0.4124 (0.0000)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.8945 0.9999 0.2149 0.5430 0.9688 0.7742 10 lags 0.6578 1.0000 0.1351 0.5663 0.5163 0.8774
Ljung-Box Q2 Statistic (p-value) 5 lags 0.8950 1.0000 0.2270 0.5440 0.9710 0.7710 10 lags 0.6580 1.0000 0.1220 0.5630 0.5150 0.8630
Return Equation: Wald Test (p-value) F-stat 0.7497 0.0536 0.8318 0.7009 0.7445 0.4100
Chi-Square 0.7496 0.0534 0.8318 0.7009 0.7445 0.4099 Variance Equation: Wald Test (p-value)
F-stat 0.0008 0.0436 0.4050 0.0000 0.0131 0.0000 Chi-Square 0.0008 0.0434 0.4049 0.0000 0.0130 0.0000
Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
89
Appendix 3.1(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) - Controlled by Volatility Index (VIX)
Variables KLCI Construction Consumer
Product Finance Industrial
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0138 (0.1233)
0.0039 (0.7791)
0.0224 (0.0042)***
0.0221 (0.0363)**
0.0177 (0.0657)*
PGE 0.1287 (0.2590)
0.1503 (0.4875)
-0.0014 (0.9882)
0.0786 (0.6451)
-0.0138 (0.8965)
PtGE -0.0821 (0.2254)
-0.2285 (0.1231)
-0.0002 (0.9962)
-0.0771 (0.3936)
-0.0824 (0.2394)
1tR 0.1337 (0.0000)***
0.0972 (0.0000)***
0.0916 (0.0000)***
0.1500 (0.0000)***
0.0629 (0.0000)***
1 tVIX -0.1158 (0.0000)***
-0.1454 (0.0000)***
-0.0869 (0.0000)***
-0.1266 (0.0000)***
-0.1000 (0.0000)***
Variance Equation
0 0.0062 (0.0000)***
0.0242 (0.0000)***
0.0042 (0.0000)***
0.0067 (0.0000)***
0.0072 (0.0000)***
1 0.0541 (0.0000)***
0.0682 (0.0000)***
0.0474 (0.0000)***
0.0635 (0.0000)***
0.0437 (0.0000)***
i 0.0684 (0.0000)***
0.0741 (0.0000)***
0.0368 (0.0000)***
0.0467 (0.0000)***
0.0542 (0.0000)***
1 0.9090 (0.0000)***
0.8898 (0.0000)***
0.9294 (0.0000)***
0.9122 (0.0000)***
0.9232 (0.0000)***
PGE 0.0455 (0.0056)***
0.2468 (0.0000)***
0.0092 (0.0858)*
0.0711 (0.0000)***
0.0523 (0.0002)***
PtGE 0.0031 (0.6106)
0.0289 (0.0205)**
0.0012 (0.6139)
0.0034 (0.6734)
0.0012 (0.8131)
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.3783 0.4573 0.2411 0.0040 0.9491 10 lags 0.4933 0.3520 0.4488 0.0055 0.9794
Ljung-Box Q2 Statistic (p-value) 5 lags 0.3750 0.4530 0.2290 0.0030 0.9490
10 lags 0.4470 0.2990 0.4050 0.0040 0.9790 Return Equation: Wald Test (p-value)
F-stat 0.1345 0.1267 0.9999 0.5398 0.4992 Chi-Square 0.1344 0.1266 0.9999 0.5398 0.4991
Variance Equation: Wald Test (p-value) F-stat 0.0099 0.0000 0.1489 0.0001 0.0002
Chi-Square 0.0099 0.0000 0.1489 0.0001 0.0002 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
90
Appendix 3.1(b): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) - Controlled by Volatility Index (VIX)
Variables Industrial
Product Mining Plantation Property Trade and
Services Technology
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0097 (0.3295)
0.0367 (0.2476)
0.0195 (0.1018)
-0.0100 (0.3987)
0.0101 (0.2797)
-0.0308 (0.0889)*
PGE 0.0904 (0.4159)
-0.2323 (0.4009)
0.1022 (0.3370)
0.1351 (0.5271)
0.1399 (0.2712)
0.0190 (0.9219)
PtGE -0.0175 (0.7849)
-0.1029 (0.5820)
-0.0140 (0.8700)
-0.1321 (0.2214)
-0.0939 (0.2062)
-0.1125 (0.4030)
1tR 0.0988 (0.0000)***
-0.0515 (0.0003)***
0.1262 (0.0000)***
0.1488 (0.0000)***
0.0924 (0.0000)***
0.1377 (0.0000)***
1 tVIX -0.1158 (0.0000)***
-0.1734 (0.0000)***
-0.0991 (0.0000)***
-0.1262 (0.0000)***
-0.1152 (0.0000)***
-0.1253 (0.0000)***
Variance Equation
0 0.0150 (0.0000)***
0.4571 (0.0000)***
0.0215 (0.0000)***
0.0164 (0.0000)***
0.0053 (0.0000)***
0.0344 (0.0000)***
1 0.0834 (0.0000)***
0.1280 (0.0000)***
0.0928 (0.0000)***
0.1194 (0.0000)***
0.0498 (0.0000)***
0.0841 (0.0000)***
i 0.0825 (0.0000)***
0.1199 (0.0000)***
0.0422 (0.0000)***
0.0277 (0.0001)***
0.0761 (0.0000)***
0.0227 (0.0021)***
1 0.8688 (0.0000)***
0.7837 (0.0000)***
0.8741 (0.0000)***
0.8678 (0.0000)***
0.9122 (0.0000)***
0.8918 (0.0000)***
PGE 0.0434 (0.0283)**
-0.2043 (0.0005)***
-0.0025 (0.8814)
0.2471 (0.0000)***
0.0634 (0.0020)***
0.0714 (0.0038)***
PtGE 0.0170 (0.0863)*
0.0294 (0.7963)
0.0383 (0.0050)***
0.0218 (0.0440)**
-0.0019 (0.7783)
0.0684 (0.0000)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.7659 0.9998 0.0138 0.4778 0.6242 0.0081 10 lags 0.7291 0.9994 0.0261 0.5591 0.5878 0.0347
Ljung-Box Q2 Statistic (p-value) 5 lags 0.7640 1.0000 0.0140 0.4680 0.6310 0.0080 10 lags 0.7000 0.9990 0.0210 0.5300 0.5590 0.0260
Return Equation: Wald Test (p-value) F-stat 0.6384 0.5953 0.6222 0.2483 0.1149 0.7034
Chi-Square 0.6384 0.5953 0.6222 0.2482 0.1148 0.7034 Variance Equation: Wald Test (p-value)
F-stat 0.0044 0.0023 0.0183 0.0000 0.0081 0.0000 Chi-Square 0.0044 0.0023 0.0182 0.0000 0.0080 0.0000
Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
91
Appendix 3.2(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2005) - Controlled by Volatility Index (VIX)
Variables KLCI Construction Consumer
Product Finance Industrial
(p, q) (1, 1) (1, 0) (1, 1) (1, 1) (1, 0) Mean Equation
0 -0.0034 (0.8301)
-0.0606 (0.0725)*
0.0096 (0.4549)
0.0155 (0.5558)
-0.0191 (0.3648)
PGE 0.2101 (0.2041)
0.3738 (0.0053)***
0.1116 (0.5125)
0.1345 (0.4779)
0.0653 (0.5484)
PtGE -0.1493 (0.1659)
-0.0627 (0.7635)
-0.0673 (0.3148)
-0.2184 (0.1297)
0.0631 (0.6818)
1tR 0.1476 (0.0000)***
0.0120 (0.1983)
0.1346 (0.0000)***
0.1731 (0.0000)***
0.0929 (0.0000)***
1 tVIX -0.1382 (0.0000)***
-0.2683 (0.0000)***
-0.0906 (0.0000)***
-0.1669 (0.0000)***
-0.2060 (0.0000)***
Variance Equation
0 0.0100 (0.0000)***
2.5180 (0.0000)***
0.0044 (0.0000)***
0.0176 (0.0000)***
1.1623 (0.0000)***
1 0.0477 (0.0000)***
0.3155 (0.0000)***
0.0309 (0.0000)***
0.0588 (0.0000)***
0.3014 (0.0000)***
i 0.0769 (0.0000)***
0.2315 (0.0000)***
0.0426 (0.0000)***
0.0574 (0.0000)***
0.2521 (0.0000)***
1 0.9121 (0.0000)***
-- --
0.9449 (0.0000)***
0.9082 (0.0000)***
-- --
PGE -0.0680 (0.0043)***
-2.1606 (0.0000)***
-0.0046 (0.6325)
-0.0873 (0.0104)**
-0.8753 (0.0000)***
PtGE 0.0602 (0.0002)***
-0.5709 (0.0001)***
0.0038 (0.5245)
0.1071 (0.0000)***
-0.0614 (0.6494)
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.3171 0.0000 0.1997 0.0210 0.0002 10 lags 0.6205 0.0000 0.5731 0.0303 0.0000
Ljung-Box Q2 Statistic (p-value) 5 lags 0.3140 0.0000 0.1920 0.0180 0.0000
10 lags 0.5990 0.0000 0.5270 0.0260 0.0000 Return Equation: Wald Test (p-value)
F-stat 0.1827 0.0191 0.5178 0.2792 0.7724 Chi-Square 0.1825 0.0190 0.5177 0.2790 0.7724
Variance Equation: Wald Test (p-value) F-stat 0.0005 0.0000 0.7946 0.0000 0.0000
Chi-Square 0.0005 0.0000 0.7946 0.0000 0.0000 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
92
Appendix 3.2(b): Threshold GARCH Results for Pre-General Election and Post-
General Election (1994 - 2005) - Controlled by Volatility Index (VIX)
Variables Industrial Product
Mining Plantation Property Trade and Services
Technology
(p, q) (1, 1) (1, 0) (1, 0) (1, 0) (1, 1) (1, 1) Mean Equation
0 -0.0263 (0.0908)*
-0.0207 (0.6309)
-0.0069 (0.7371)
-0.0588 (0.0186)**
-0.0104 (0.5520)
-0.0899 (0.0056)***
PGE 0.2240 (0.0446)**
0.3746 (0.1651)
0.1085 (0.4666)
0.3718 (0.0631)*
0.2583 (0.1356)
0.2403 (0.1348)
PtGE -0.1054 (0.3367)
-0.3056 (0.2312)
-0.0776 (0.6252)
-0.0927 (0.7312)
-0.1359 (0.1520)
-0.3544 (0.0046)***
1tR 0.1041 (0.0000)***
0.0350 (0.0023)***
0.1319 (0.0000)***
0.2255 (0.0000)***
0.1091 (0.0000)***
0.1592 (0.0000)***
1 tVIX -0.1154 (0.0000)***
-0.0970 (0.0004)***
-0.1467 (0.0000)***
-0.2164 (0.0000)***
-0.1445 (0.0000)***
-0.2178 (0.0000)***
Variance Equation
0 0.0166 (0.0000)***
0.1485 (0.0000)***
1.0964 (0.0000)***
1.7002 (0.0000)***
0.0078 (0.0000)***
-0.0002 (0.8518)
1 0.0940 (0.0000)***
0.1359 (0.0000)***
0.4689 (0.0000)***
0.5290 (0.0000)***
0.0459 (0.0000)***
0.0060 (0.0374)**
i 0.1181 (0.0000)***
0.9201 (0.0000)***
0.0284 (0.5487)
0.2025 (0.0002)***
0.0783 (0.0000)***
0.0262 (0.0000)***
1 0.6186 (0.0000)***
-- --
-- --
-- --
0.9175 (0.0000)***
0.9804 (0.0000)***
2 0.2326 (0.0565)*
-- --
-- --
-- --
-- --
-- --
PGE -0.0687 (0.0042)***
-0.3921 (0.0000)***
-0.5374 (0.0000)***
-1.0678 (0.0000)***
-0.0981 (0.0000)***
-0.0234 (0.2644)
PtGE 0.0814 (0.0029)***
0.4124 (0.0000)***
0.5489 (0.0005)***
0.9331 (0.0000)***
0.0470 (0.0019)***
0.0169 (0.0069)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.8461 0.0000 0.0000 0.0000 0.1751 0.0000 10 lags 0.9673 0.0000 0.0000 0.0000 0.3666 0.0000
Ljung-Box Q2 Statistic (p-value) 5 lags 0.8420 0.0000 0.0000 0.0000 0.1820 0.0000 10 lags 0.9670 0.0000 0.0000 0.0000 0.3690 0.0000
Return Equation: Wald Test (p-value) F-stat 0.0730 0.2136 0.6771 0.1666 0.1151 0.0051
Chi-Square 0.0728 0.2134 0.6771 0.1664 0.1149 0.0050 Variance Equation: Wald Test (p-value)
F-stat 0.0034 0.0000 0.0000 0.0000 0.0000 0.0203 Chi-Square 0.0034 0.0000 0.0000 0.0000 0.0000 0.0201
Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
93
Appendix 3.3(a): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) - Controlled by Volatility Index (VIX)
Variables KLCI Construction Consumer
Product Finance Industrial
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0257 (0.0182)**
0.0290 (0.0966)*
0.0370 (0.0004)***
0.0320 (0.0101)**
0.0172 (0.1764)
PGE 0.1267 (0.4908)
-0.1299 (0.8380)
-0.0638 (0.5592)
0.0511 (0.8428)
0.0073 (0.9604)
PtGE -0.0625 (0.6041)
-0.1444 (0.4379)
0.1396 (0.0238)**
-0.0235 (0.8493)
-0.0621 (0.6382)
1tR 0.1229 (0.0000)***
0.0786 (0.0003)***
0.0417 (0.0523)*
0.1253 (0.0000)***
0.0684 (0.0005)***
1 tVIX -0.1110 (0.0000)***
-0.1372 (0.0000)***
-0.0835 (0.0000)***
-0.1198 (0.0000)***
-0.0969 (0.0000)***
Variance Equation
0 0.0122 (0.0000)***
0.0370 (0.0000)***
0.0220 (0.0000)***
0.0183 (0.0000)***
0.0140 (0.0000)***
1 0.0690 (0.0000)***
0.1065 (0.0000)***
0.0953 (0.0000)***
0.1014 (0.0000)***
0.0278 (0.0004)***
i 0.0824 (0.0000)***
0.0667 (0.0000)***
0.0721 (0.0000)***
0.0675 (0.0000)***
0.0871 (0.0000)***
1 0.8627 (0.0000)***
0.8339 (0.0000)***
0.8038 (0.0000)***
0.8341 (0.0000)***
0.9036 (0.0000)***
PGE 0.1038 (0.0120)**
1.0508 (0.0000)***
0.0234 (0.1387)
0.1587 (0.0006)***
0.0847 (0.0023)***
PtGE 0.0002 (0.9826)
-0.0306 (0.2938)
-0.0102 (0.0597)*
0.0035 (0.7197)
-0.0011 (0.8521)
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.9782 0.9699 0.4830 0.6130 0.9670 10 lags 0.5677 0.7271 0.3056 0.6123 0.9973
Ljung-Box Q2 Statistic (p-value) 5 lags 0.9790 0.9660 0.4750 0.6010 0.9660
10 lags 0.5540 0.7180 0.2910 0.5620 0.9970 Return Equation: Wald Test (p-value)
F-stat 0.5779 0.7374 0.0629 0.9447 0.8868 Chi-Square 0.5779 0.7374 0.0627 0.9447 0.8868
Variance Equation: Wald Test (p-value) F-stat 0.0394 0.0000 0.0549 0.0020 0.0094
Chi-Square 0.0392 0.0000 0.0547 0.0020 0.0094 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
94
Appendix 3.3(b): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) - Controlled by Volatility Index (VIX)
Variables Industrial
Product Mining Plantation Property Trade and
Services Technology
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0386 (0.0037)***
0.0809 (0.0929)*
0.0278 (0.1042)
0.0267 (0.0810)*
0.0215 (0.0535)*
0.0086 (0.6908)
PGE -0.0689 (0.7629)
-0.7665 (0.0238)**
0.0226 (0.8953)
-0.2625 (0.5723)
0.1127 (0.6466)
-0.2123 (0.3963)
PtGE 0.0628 (0.5908)
0.0576 (0.8379)
0.0078 (0.9528)
0.0888 (0.6640)
-0.0725 (0.5510)
0.2047 (0.2842)
1tR 0.0896 (0.0000)***
-0.1725 (0.0000)***
0.1266 (0.0000)***
0.1279 (0.0000)***
0.0763 (0.0001)***
0.1039 (0.0000)***
1 tVIX -0.1173 (0.0000)***
-0.1810 (0.0000)***
-0.1041 (0.0000)***
-0.1218 (0.0000)***
-0.1097 (0.0000)***
-0.1035 (0.0000)***
Variance Equation
0 0.0277 (0.0000)***
0.9811 (0.0000)***
0.0166 (0.0000)***
0.0224 (0.0000)***
0.0093 (0.0000)***
0.1465 (0.0000)***
1 0.0958 (0.0000)***
0.1730 (0.0000)***
0.0700 (0.0000)***
0.1302 (0.0000)***
0.0532 (0.0000)***
0.1384 (0.0000)***
i 0.0617 (0.0000)***
0.2002 (0.0000)***
0.0261 (0.0010)***
-0.0039 (0.7253)
0.0807 (0.0000)***
0.0348 (0.0309)**
1 0.8259 (0.0000)***
0.6622 (0.0000)***
0.9022 (0.0000)***
0.8497 (0.0000)***
0.8874 (0.0000)***
0.7599 (0.0000)***
PGE 0.1408 (0.0000)***
-0.2748 (0.0875)*
0.0480 (0.1050)
0.5952 (0.0000)***
0.1519 (0.0027)***
0.0901 (0.1373)
PtGE 0.0080 (0.4581)
-0.2439 (0.3020)
0.0028 (0.8465)
-0.0160 (0.6114)
-0.0112 (0.1678)
0.3531 (0.0000)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.8575 0.9999 0.1740 0.6677 0.9938 0.7437 10 lags 0.5366 0.9999 0.1450 0.6734 0.6766 0.8887
Ljung-Box Q2 Statistic (p-value) 5 lags 0.8580 1.0000 0.1860 0.6620 0.9950 0.7410 10 lags 0.5390 1.0000 0.1440 0.6680 0.6670 0.8810
Return Equation: Wald Test (p-value) F-stat 0.8162 0.0771 0.9902 0.7352 0.6789 0.4495
Chi-Square 0.8162 0.0769 0.9902 0.7352 0.6789 0.4494 Variance Equation: Wald Test (p-value)
F-stat 0.0001 0.1284 0.2537 0.0000 0.0073 0.0000 Chi-Square 0.0001 0.1281 0.2535 0.0000 0.0073 0.0000
Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
95
Appendix 3.4(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) - Controlled by Federal Fund Rate
Variables KLCI Construction Consumer
Product Finance Industrial
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0136 (0.1470)
0.0038 (0.7858)
0.0242 (0.0036)***
0.0225 (0.0404)**
0.0161 (0.1033)
PGE 0.1220 (0.2840)
0.1358 (0.5287)
-0.0099 (0.9226)
0.0703 (0.6729)
-0.0160 (0.8803)
PtGE -0.0815 (0.2493)
-0.2224 (0.1392)
0.0004 (0.9936)
-0.0763 (0.4112)
-0.0758 (0.2788)
1tR 0.1489 (0.0000)***
0.1166 (0.0000)***
0.1018 (0.0000)***
0.1622 (0.0000)***
0.0711 (0.0000)***
1 tFFR 0.0604 (0.2092)
0.1003 (0.1329)
-0.0038 (0.9479)
0.0368 (0.5780)
0.0468 (0.4175)
Variance Equation
0 0.0065 (0.0000)***
0.0253 (0.0000)***
0.0052 (0.0000)***
0.0067 (0.0000)***
0.0075 (0.0000)***
1 0.0526 (0.0000)***
0.0644 (0.0000)***
0.0474 (0.0000)***
0.0592 (0.0000)***
0.0434 (0.0000)***
i 0.0726 (0.0000)***
0.0809 (0.0000)***
0.0435 (0.0000)***
0.0512 (0.0000)***
0.0593 (0.0000)***
1 0.9084 (0.0000)***
0.8899 (0.0000)***
0.9251 (0.0000)***
0.9143 (0.0000)***
0.9212 (0.0000)***
PGE 0.0466 (0.0036)***
0.2566 (0.0000)***
0.0116 (0.0577)*
0.0726 (0.0000)***
0.0473 (0.0010)***
PtGE 0.0039 (0.5330)
0.0289 (0.0277)**
-0.0011 (0.7435)
0.0036 (0.6607)
0.0005 (0.9375)
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.2811 0.4417 0.2896 0.0029 0.9410 10 lags 0.4024 0.3932 0.4820 0.0052 0.9720
Ljung-Box Q2 Statistic (p-value) 5 lags 0.2770 0.4360 0.2830 0.0020 0.9410
10 lags 0.3560 0.3470 0.4480 0.0030 0.9690 Return Equation: Wald Test (p-value)
F-stat 0.1644 0.1509 0.9952 0.5711 0.5561 Chi-Square 0.1643 0.1508 0.9952 0.5711 0.5560
Variance Equation: Wald Test (p-value) F-stat 0.0070 0.0000 0.1647 0.0000 0.0018
Chi-Square 0.0070 0.0000 0.1646 0.0000 0.0018 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
96
Appendix 3.4(b): Threshold GARCH Results for Pre-General Election and Post-
General Election (1994 - 2015) - Controlled by Federal Fund Rate
Variables Industrial Product
Mining Plantation Property Trade and Services
Technology
(p, q) (1, 2) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0079 (0.4410)
0.0304 (0.3025)
0.0183 (0.1237)
-0.0071 (0.5619)
0.0090 (0.3499)
-0.0297 (0.1026)
PGE 0.1036 (0.3757)
-0.2079 (0.4482)
0.1005 (0.3463)
0.1218 (0.5769)
0.1316 (0.2873)
0.0105 (0.9589)
PtGE -0.0202 (0.7579)
-0.0811 (0.6647)
-0.0149 (0.8646)
-0.1280 (0.2447)
-0.0899 (0.2449)
-0.0958 (0.4618)
1tR 0.1156 (0.0000)***
-0.0516 (0.0002)***
0.1325 (0.0000)***
0.1652 (0.0000)***
0.1107 (0.0000)***
0.1462 (0.0000)***
1 tFFR 0.0875 (0.1303)
-0.1760 (0.1916)
0.0771 (0.2736)
0.0599 (0.3897)
-0.0324 (0.6087)
0.2015 (0.4135)
Variance Equation
0 0.0167 (0.0000)***
0.4570 (0.0000)***
0.0206 (0.0000)***
0.0165 (0.0000)***
0.0057 (0.0000)***
0.0367 (0.0000)***
1 0.0827 (0.0000)***
0.1217 (0.0000)***
0.0913 (0.0000)***
0.1112 (0.0000)***
0.0508 (0.0000)***
0.0842 (0.0000)***
i 0.0890 (0.0000)***
0.1168 (0.0000)***
0.0448 (0.0000)***
0.0298 (0.0000)***
0.0807 (0.0000)***
0.0234 (0.0016)***
1 0.7389 (0.0000)***
0.7897 (0.0000)***
0.8754 (0.0000)***
0.8735 (0.0000)***
0.9092 (0.0000)***
0.8904 (0.0000)***
2 0.1247 (0.2055)
-- --
-- --
-- --
-- --
-- --
PGE 0.0486 (0.0336)**
-0.2139 (0.0008)***
-0.0019 (0.9198)
0.2446 (0.0000)***
0.0626 (0.0012)***
0.0771 (0.0026)***
PtGE 0.0163 (0.1382)
0.0320 (0.7742)
0.0411 (0.0033)***
0.0189 (0.0688)*
-0.0003 (0.9674)
0.0603 (0.0000)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.9705 0.9998 0.0034 0.5873 0.5782 0.0065 10 lags 0.9229 0.9994 0.0074 0.6673 0.6070 0.0365
Ljung-Box Q2 Statistic (p-value) 5 lags 0.9710 1.0000 0.0030 0.5820 0.5830 0.0060 10 lags 0.9150 0.9990 0.0050 0.6470 0.5810 0.0300
Return Equation: Wald Test (p-value) F-stat 0.5801 0.6714 0.6323 0.2943 0.1435 0.7623
Chi-Square 0.5801 0.6714 0.6323 0.2942 0.1434 0.7622 Variance Equation: Wald Test (p-value)
F-stat 0.0122 0.0038 0.0120 0.0000 0.0048 0.0000 Chi-Square 0.0121 0.0038 0.0119 0.0000 0.0047 0.0000
Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
97
Appendix 3.5(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2005) - Controlled by Federal Fund Rate
Variables KLCI Construction Consumer
Product Finance Industrial
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 -0.0021 (0.8975)
-0.0298 (0.2146)
0.0117 (0.3795)
0.0127 (0.5278)
0.0172 (0.2660)
PGE 0.1512 (0.4201)
0.1344 (0.2884)
0.0795 (0.6450)
0.0988 (0.6147)
-0.0206 (0.9397)
PtGE -0.1408 (0.2299)
-0.2411 (0.1401)
-0.0679 (0.3123)
-0.2049 (0.1899)
-0.1333 (0.2224)
1tR 0.1522 (0.0000)***
0.1245 (0.0000)***
0.1358 (0.0000)***
0.1754 (0.0000)***
0.0745 (0.0000)***
1 tFFR 0.0581 (0.3513)
0.1098 (0.2016)
0.0381 (0.5597)
0.0268 (0.7521)
0.0589 (0.3677)
Variance Equation
0 0.0102 (0.0000)***
0.0322 (0.0000)***
0.0047 (0.0000)***
0.0168 (0.0000)***
0.0060 (0.0000)***
1 0.0438 (0.0000)***
0.0422 (0.0000)***
0.0305 (0.0000)***
0.0521 (0.0000)***
0.0775 (0.0000)***
i 0.0800 (0.0000)***
0.0793 (0.0000)***
0.0447 (0.0000)***
0.0585 (0.0000)***
0.9567 (0.0000)***
1 0.9142 (0.0000)***
0.9119 (0.0000)***
0.9439 (0.0000)***
0.9143 (0.0000)***
-- --
PGE -0.0504 (0.1387)
-0.1010 (0.0000)***
0.0003 (0.9809)
-0.0709 (0.0764)*
0.0868 (0.0076)***
PtGE 0.0520 (0.0067)***
0.1183 (0.0000)***
0.0023 (0.7304)
0.0937 (0.0000)***
-0.0114 (0.3143)
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.3148 0.0851 0.2493 0.0250 0.0000 10 lags 0.6526 0.2729 0.6480 0.0473 0.0000
Ljung-Box Q2 Statistic (p-value) 5 lags 0.3140 0.0810 0.2410 0.0220 0.0000
10 lags 0.6380 0.2200 0.6040 0.0440 0.0000 Return Equation: Wald Test (p-value)
F-stat 0.3859 0.1987 0.5706 0.3968 0.4705 Chi-Square 0.3857 0.1985 0.5705 0.3967 0.4704
Variance Equation: Wald Test (p-value) F-stat 0.0230 0.0000 0.9197 0.0001 0.0150
Chi-Square 0.0229 0.0000 0.9197 0.0001 0.0149 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
98
Appendix 3.5(b): Threshold GARCH Results for Pre-General Election and Post-
General Election (1994 - 2005) - Controlled by Federal Fund Rate
Variables Industrial Product
Mining Plantation Property Trade and Services
Technology
(p, q) (0, 1) (1, 1) (0, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 -0.0208 (0.2003)
-0.0122 (0.7588)
0.0139 (0.4487)
-0.0569 (0.0054)***
-0.0103 (0.5673)
-0.0894 (0.0066)***
PGE 0.2015 (0.0344)**
0.2856 (0.3088)
0.1681 (0.2707)
0.3376 (0.0321)**
0.1969 (0.3005)
0.1917 (0.2638)
PtGE -0.1667 (0.1338)
-0.3655 (0.1564)
-0.0859 (0.3712)
-0.1770 (0.0781)*
-0.1405 (0.1770)
-0.3381 (0.0138)**
1tR 0.1733 (0.0000)***
0.0440 (0.0135)**
0.1452 (0.0000)***
0.1633 (0.0000)***
0.1157 (0.0000)***
0.1735 (0.0000)***
1 tFFR 0.1448 (0.0274)**
-0.0675 (0.6576)
0.0650 (0.4295)
0.0537 (0.5442)
-0.0182 (0.8056)
0.6944 (0.0491)**
Variance Equation
0 0.0145 (0.0000)***
0.1552 (0.0000)***
0.0204 (0.0000)***
0.0137 (0.0000)***
0.0081 (0.0000)***
-0.0003 (0.7616)
1 -- --
0.0944 (0.0000)***
-- --
0.1025 (0.0000)***
0.0448 (0.0000)***
0.0058 (0.0406)**
i 0.1504 (0.0000)***
0.0870 (0.0000)***
0.1107 (0.0000)***
0.0459 (0.0000)***
0.0810 (0.0000)***
0.0264 (0.0000)***
1 0.9224 (0.0000)***
0.8557 (0.0000)***
0.9305 (0.0000)***
0.8840 (0.0000)***
0.9171 (0.0000)***
0.9805 (0.0000)***
PGE -0.0567 (0.0000)***
-0.3762 (0.0000)***
-0.0501 (0.0107)**
-0.0886 (0.0009)***
-0.0864 (0.0125)**
-0.0166 (0.4353)
PtGE 0.0834 (0.0000)***
0.2854 (0.0137)**
0.0421 (0.0015)***
0.0310 (0.0182)**
0.0460 (0.0106)**
0.0156 (0.0186)**
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.0000 0.2714 0.0000 0.0848 0.1817 0.0000 10 lags 0.0001 0.5306 0.0000 0.2359 0.3943 0.0000
Ljung-Box Q2 Statistic (p-value) 5 lags 0.0000 0.2670 0.0000 0.0720 0.1860 0.0000 10 lags 0.0000 0.5330 0.0000 0.2000 0.4000 0.0000
Return Equation: Wald Test (p-value) F-stat 0.0244 0.2305 0.3520 0.0306 0.2455 0.0230
Chi-Square 0.0242 0.2303 0.3518 0.0305 0.2454 0.0227 Variance Equation: Wald Test (p-value)
F-stat 0.0000 0.0001 0.0022 0.0010 0.0145 0.0422 Chi-Square 0.0000 0.0001 0.0022 0.0010 0.0145 0.0420
Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
99
Appendix 3.6(a): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) - Controlled by Federal Fund Rate
Variables KLCI Construction Consumer
Product Finance Industrial
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0242 (0.0336)**
0.0282 (0.1132)
0.0372 (0.0006)***
0.0310 (0.0181)**
0.0140 (0.2885)
PGE 0.1404 (0.4033)
-0.0636 (0.9148)
-0.0405 (0.7049)
0.0795 (0.7342)
0.0215 (0.8761)
PtGE -0.0533 (0.6538)
-0.1295 (0.4484)
0.1310 (0.0387)**
-0.0180 (0.8923)
-0.0519 (0.6560)
1tR 0.1492 (0.0000)***
0.1139 (0.0000)***
0.0592 (0.0083)***
0.1452 (0.0000)***
0.0791 (0.0002)***
1 tFFR -0.2808 (0.0508)*
-0.0637 (0.7079)
-0.2961 (0.0207)**
-0.0136 (0.9335)
0.0510 (0.7144)
Variance Equation
0 0.0116 (0.0000)***
0.0356 (0.0000)***
0.0243 (0.0000)***
0.0150 (0.0000)***
0.0135 (0.0000)***
1 0.0684 (0.0000)***
0.1039 (0.0000)***
0.0827 (0.0000)***
0.0871 (0.0000)***
0.0276 (0.0006)***
i 0.0846 (0.0000)***
0.0803 (0.0000)***
0.0902 (0.0000)***
0.0724 (0.0000)***
0.0912 (0.0000)***
1 0.8667 (0.0000)***
0.8348 (0.0000)***
0.8047 (0.0000)***
0.8544 (0.0000)***
0.9044 (0.0000)***
PGE 0.0881 (0.0148)**
0.9985 (0.0000)***
0.0187 (0.2614)
0.1336 (0.0009)***
0.0668 (0.0083)***
PtGE 0.0001 (0.9915)
-0.0482 (0.1705)
-0.0143 (0.0274)**
0.0009 (0.9173)
-0.0039 (0.6168)
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.8837 0.9275 0.3531 0.2487 0.9439 10 lags 0.3749 0.7928 0.0790 0.2354 0.9928
Ljung-Box Q2 Statistic (p-value) 5 lags 0.8850 0.9200 0.3450 0.2380 0.9420
10 lags 0.3400 0.7830 0.0710 0.1770 0.9920 Return Equation: Wald Test (p-value)
F-stat 0.5063 0.7500 0.1071 0.9016 0.8775 Chi-Square 0.5063 0.7500 0.1069 0.9016 0.8775
Variance Equation: Wald Test (p-value) F-stat 0.0493 0.0000 0.0511 0.0036 0.0301
Chi-Square 0.0491 0.0000 0.0510 0.0035 0.0300 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
100
Appendix 3.6(b): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) - Controlled by Federal Fund Rate
Variables Industrial
Product Mining Plantation Property Trade and
Services Technology
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0361 (0.0093)***
0.0784 (0.0936)*
0.0247 (0.1482)
0.0277 (0.0865)*
0.0195 (0.0886)*
0.0099 (0.6498)
PGE -0.0259 (0.9107)
-0.7563 (0.0300)**
0.0335 (0.8458)
-0.2110 (0.6588)
0.1365 (0.5110)
-0.1884 (0.4974)
PtGE 0.0677 (0.5631)
0.1047 (0.7054)
0.0108 (0.9370)
0.0949 (0.6378)
-0.0552 (0.6405)
0.2058 (0.2652)
1tR 0.1189 (0.0000)***
-0.1826 (0.0000)***
0.1312 (0.0000)***
0.1556 (0.0000)***
0.1069 (0.0000)***
0.1107 (0.0000)***
1 tFFR -0.2630 (0.0681)*
-1.7293 (0.0000)***
-0.0668 (0.7742)
0.1236 (0.3025)
-0.2271 (0.1268)
-0.0557 (0.8681)
Variance Equation
0 0.0273 (0.0000)***
0.9459 (0.0000)***
0.0163 (0.0000)***
0.0246 (0.0000)***
0.0098 (0.0000)***
0.1484 (0.0000)***
1 0.0853 (0.0000)***
0.1680 (0.0000)***
0.0726 (0.0000)***
0.1196 (0.0000)***
0.0559 (0.0000)***
0.1366 (0.0000)***
i 0.0583 (0.0000)***
0.1898 (0.0000)***
0.0348 (0.0000)***
0.0065 (0.5480)
0.0886 (0.0000)***
0.0319 (0.0466)**
1 0.8401 (0.0000)***
0.6759 (0.0000)***
0.8970 (0.0000)***
0.8529 (0.0000)***
0.8822 (0.0000)***
0.7629 (0.0000)***
PGE 0.1266 (0.0002)***
-0.2200 (0.1600)
0.0429 (0.1811)
0.5780 (0.0000)***
0.1319 (0.0039)***
0.1268 (0.0432)**
PtGE 0.0030 (0.7654)
-0.2472 (0.2884)
0.0107 (0.5233)
-0.0219 (0.5087)
-0.0113 (0.2439)
0.2975 (0.0000)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.8074 0.9999 0.0451 0.3452 0.9924 0.6740 10 lags 0.6310 0.9999 0.0327 0.5340 0.6823 0.7415
Ljung-Box Q2 Statistic (p-value) 5 lags 0.8050 1.0000 0.0490 0.3520 0.9930 0.6800 10 lags 0.6460 1.0000 0.0300 0.5380 0.6780 0.7260
Return Equation: Wald Test (p-value) F-stat 0.8333 0.0914 0.9792 0.7686 0.6120 0.4625
Chi-Square 0.8333 0.0912 0.9792 0.7685 0.6120 0.4624 Variance Equation: Wald Test (p-value)
F-stat 0.0010 0.2006 0.3105 0.0000 0.0108 0.0000 Chi-Square 0.0010 0.2004 0.3103 0.0000 0.0107 0.0000
Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (average daily returns (volatility) for the period of pre-general election and post-general election are significant different from zero).
101
CHAPTER 4
GENERAL ELECTION AND STOCK MARKET VOLATILITY IN MALAYSIA: EVIDENCE FROM GLCs AND NON-GLCs STOCK PERFORMANCE
4.1 Introduction
Domestic political events have a major influence on financial markets, especially national
election. Investors need to actively keep updated with market movement to rearrange their
investment strategies in times of political uncertainty. Negative news could destabilize
investor trading, thereby creating disarray and possible panic in the markets. It is, therefore,
crucial to examine whether this is the national election induces financial uncertainty.
In literature, Nordhaus (1975) is the pioneer in analysing the intimate link between
economics and political business cycles. The empirical literature on the relationship between
stock market performance and political elections dates back to Allivine and O’Neill (1980)
which focused on market behaviours at the time of U.S. elections. Motivated by the evidence
found in the U.S. market, academic studies commenced to investigate the impact of national
election on financial market in other regions, among them are Foerster and Schmitz (1997),
Pantzalis et al. (2000), Chiu et al. (2005), Wang and Lin (2009) and Hung (2013).
In the examination of election effect, previous studies largely focus on composite indices
which represent respective stock markets or top companies of the stock market division. For
example, Chiu et al. (2005) investigated the movement of KOSPI 200 index to see the effect
of presidential and parliamentary elections on South Korea financial market. For the same
purpose, Wang and Lin (2009) studied the Taiwan Stock Exchange Value Weighted Index
(TAIEX) and Lean and Yeap (2017) studied the FTSE Bursa Malaysia KLCI Index. From the
102
perspective of an investor, the existence of election effect in firm level is equally important.
Study on listed company index could be useful for investors because the individual firm may
react differently to election effect due to the nature of the business industry. However, this
remains an unexplored issue in the literature. Therefore, this study attempts to uncover
evidence of election effect in firm level. Selection of company index to perform the study on
election effect is another interesting point of this research. Since election effect is a political
factor to explain the fluctuation of stock returns, this study selects politically connected firm
as the sample to see the impact of the general election in stock returns of these firms.
Political connection and firm value are intimately linked. The incentives of political
connectedness have probably been recognized by citizens and investors. Investors generally
believe that political favours granted to a firm will enhance a firm’s market value. In
academic research, the positive relationship between political connectedness and firm value
has been a topic of intense debate. Evidence has been mounting over the years to suggest that
political connections allow politically connected firms to benefit from various governmental
interventions, including preferential access to credit (Dinç, 2005), lighter taxation (De Soto,
1989), reduced regulatory enforcement (Kroszner and Stratmann, 1998; Agrawal and
Knoeber, 2001), improved financial and accounting performance (Fisman, 2001; Johnson and
Mitton, 2003; Cooper et al., 2010), and many other forms.
Meanwhile, the impact of political connectedness on firm value is observable by its
stock returns. This is explained by Civilize et al. (2015) that in an informationally efficient
market if political connections create firm value, the existence of the value should be
systematically reflected in stock returns. In the finance literature, recent empirical studies
found evidence in supporting this view. A study by Niessen and Reunzi (2009) showed that
103
politically connected firms were significantly outperformed unconnected firms on Germany
stock market. Moreover, it is also evident from the study of Civilize et al. (2015) that
politically connected firms systematically enjoy higher realized stock returns over a long
period of time in Thailand. These studies suggest that the impact of political connections on
firm value is evident in stock returns.
Attempting to examine the link between political connections and stock returns is
fraught with difficulties. First, a researcher needs to be thorough in identifying the right
measure of political connections of a firm (Hassan et al., 2012). Different definitions of
political connections may lead to conflicting research results. It is possible to overcome this
problem in Malaysia by studying the Government-Link-Companies (GLCs) and Non-
Government-Link-Companies (Non-GLCs). The GLCs are controlled by the Malaysian
government and the GLCs play a significant role in the development of the country’s
economy. Second, a study would ideally be carried out in an environment with a political
shift to see the effect of the change in political leadership on politically connected firms
(Civilize et al., 2015). In Malaysia, potential political change happened during the 12th and
13th Malaysian General Election in the year 2008 and 2013. In view of that, Malaysian stock
market has a unique empirical setting in investigating the impact of political uncertainty on
politically connected firms in years of general election.
Given the significant role of GLCs firms in the development of Malaysia economy,
the effect of government intervention on companies’ performance has been empirically
assessed in previous studies (Razak et al., 2011; Lau and Tong, 2008). Evidence from
previous studies confirmed the impact of political connectedness on firm value. However, the
effect of political uncertainty is not considered in their study and the impact could be the
104
other way round. There is some evidence of adverse effects on firms with political
connections in Malaysia following the removal of regulation and handover of political power.
For example, the study by Mitchell and Joseph (2010) found adverse effects on political
connections firms during the removal of capital controls. Moreover, Mitchell and Joseph
(2010) also found that GLCs firms are more badly affected than non-GLCs firms during the
first month of the resignation of Tun Dr. Mahathir Mohammad as prime minister and the
handover of control to Datuk Seri Abdullah Ahmad Badawi. The study evidently showed that
political uncertainty has a negative impact on politically connected firms.
Even though Malaysia is politically stable in recent years, uncertainty still exists especially
during the election period. Fluctuation of stock market around election dates was clearly
observed in Malaysia especially during the 12th and 13th Malaysian General Election in recent
years. Evidently, as a proxy of the Malaysian stock market, the key index of FTSE Bursa
Malaysia KLCI experienced significant volatility during the general election years (Lean and
Yeap, 2017). Prior to the year 2005, the coalition Barisan Nasional won in the Malaysian
general elections in the year 1995, 1999 and 2004 and continued ruling with a stable two-
thirds majority. However, there was close fight between the two major coalitions in the 12th
and 13th general election. The coalition Barisan Nasional consecutively lost the two-thirds
majority in parliament, which is never happened in political history since Malaysia
independence. The political uncertainty due to potential shift of ruling party provides an
opportunity to conduct a research on political uncertainty and stock market performance.
Therefore, this paper aims to contribute to the current literature by empirically investigating
the performance of GLCs and Non-GLCs over the period of Malaysia general elections from
the year 1995 to 2013.
105
In brief, the contributions of this study are, first, the Threshold Generalized
Autoregressive Conditional Heteroskedasticity (Threshold GARCH / GJR GARCH) model
developed by Glosten et al. (1993) is applied to investigate the election effect on the stock
returns of GLCs firms and non-GLCs firms. Since empirical works have produced substantial
evidence on the influence of political outcomes to the main stock index, this present study
attempts to further examine the reaction of companies' stock returns and volatility in the
recent five Malaysian general elections in Malaysia. Second, this study enhances the
knowledge in the case of Malaysia by investigating the election effect in two different stages
which represent the general up and down and the drastic rise and fall. In concern of the
different effect of the general election on market volatility, the general election years of 1995,
1999 and 2005 are classified as general ups and downs periods, while the general election
years of 2008 and 2013 is drastic shock periods. Third, trading volume is also included in this
study. The theoretical model of Foster and Viswanathan (1990) suggests that low volume
comes with high volatility. Conversely, Admati and Pfleiderer (1988) speculated that trading
volume would be high when price volatility is high. Hence, this study further analyze the
observed volatilities to see whether they are related to trading volume. This study may be of
interest to investors as the results will contribute the information that most investors require
particularly in constructing an effective equity portfolio investment during the times of
election.
The rest of the paper is organized as follows. The next section summarizes the
literature of related studies. Section 4.3 outlines the research design and data used in this
study. Section 4.4 reports the empirical results, followed by the conclusion in Section 4.5.
106
4.2 Literature Review
The study on election effect started with the relationship between economic performance and
political business cycle in the United States. With evidence from the literature, it is
undeniable that government policies have great influence on all aspects of economic life. In
the literature, Nordhaus (1975) mentioned that the pattern of government policies are
predictable within an incumbent's term in office because the government actively manage the
economy to gain public support. Since then, studies such as Allivine and O’Neill (1980) and
Huang (1985) concentrated on the stock market performance during the U.S. presidential
election cycle to see the impact of government policies on the economy. Generally, evidence
from these studies indicated the impact of U.S. presidential elections on its stock market, in
which the U.S. stock markets make larger gains in the third and fourth years of a presidential
term. The political business cycle is not a short-term trend as shown by the study of Wong
and McAleer (2009). They found that the cyclical trend existed for the last ten
administrations from the year 1965 through 2003, particularly when the incumbent is
Republican.
Other than the US presidential election, empirical investigations have examined
financial market movements in relation to political events. Earlier studies presented evidence
indicating that elections do affect financial markets across their sample countries (Gemmill,
1992; Pantzalis et al., 2000; Smales, 2016). In Britain general election, Gemmill (1992)
found an extremely close relationship between opinion polls and the FTSE 100 share index
during the 1987 election. The analysis by Pantzalis et al. (2000) is the first of its type on an
international scale which covered stock market indices across 33 countries around political
election dates during the sample period 1974 - 1995. They found a positive abnormal return
in two weeks prior to the election week. The result was generally in line with the models of
107
Harrington (1993) which implies that the existence of a higher degree of uncertainty before
an election process will lead to a corresponding increase in equity prices as the uncertainty is
resolved. In a recent study, Smales (2016) provided evidence that the turbulent Australian
federal elections induce higher levels of uncertainty in financial markets. Moreover, the effect
is more pronounced as election-day draws near. In view of the role of political uncertainty in
financial markets around election dates, the study in this field has crucial implications for
investors looking to make investment decision before the election day and after the election
day.
In the Malaysian stock market, Ali et al. (2010) have attempted to show the
relationship between political uncertainty and market uncertainty. Ali et al. (2010) discovered
a significant over-reaction behaviour in the Malaysian market upon the announcement of the
removal of the deputy prime minister and announcement of the resignation of the prime
minister. Moreover, Ali et al. (2010) also investigated the market behaviour during general
election and evidence of under-reaction was found upon the announcement of the election
result. They interpreted that investors are well predicted with the election outcome and the
finding of under-reaction is in line with the stable political condition in the year 1987 to 2006.
However, the study of Ali et al. (2010) was limited to gauging the market reaction to political
uncertainty and general election, abnormal stock returns and volatility were not investigated
in the study.
Then, the study of Lean (2010) and Lean and Yeap (2017) circumvented the
limitation and examined the performance of Kuala Lumpur Composite Index (KLCI) on
stock returns and volatility during general election periods. Lean (2010) showed that there are
different effect for the pre-general election and post-general election over the period of 1994-
108
2009, where stock returns react positively before the election and negatively after the election.
Moreover, Lean (2010) also found strong evidence of GARCH effect in the model. Lean and
Yeap (2017) extended the period of study from 1989 to 2014, which covers six Malaysian
general elections. They found significant election effect in stock volatility but not in the stock
returns, where stock volatility is higher during the pre-general election but lower in the post-
general election. Moreover, the post-election effect only last for one month as there is no
unexpected outcome from the general elections. The changes of high volatility before the
election to low volatility after the election are not consistent with the theory that market is
nervous and downward-bound due to the political uncertainty. In fact, Lean and Yeap (2017)
explained the results based on the real situation of Malaysian general election. The ruling
coalition of Barisan Nasional (BN) had been in office since Malaysia independence, therefore,
investors in the Malaysian market do not expect any major changes of the election outcome
and stay confident with the market.
Despite the different level of election tension in the past six general elections, Lean
and Yeap (2017) employed a long history of stock returns that incorporates an unusually high
spike during the 12th and 13th general elections. During the 8th, 9th, and 10th general elections,
there was significant victory for BN that signaled voters' confidence in the governance.
However, the election outcome of the 12th general election was described as a political
tsunami where the incumbent lost its two-thirds majority in the Parliament. The 13th general
election created election uncertainty with a strong expectation of change from the ruling
coalition. Although BN retained its power, the worst results of BN induced surprise to the
investors that affected the stock volatility in the market. Due to the significant difference of
political condition, the analysis that covers all the previous Malaysian general elections under
one sample period may produce erroneous inferences. For that reason, this study intends to
109
divide the general election periods into two stages. One stage represents the general ups and
downs periods in election years of 1995, 1999, and 2004. The other stage represents drastic
shock periods in election years of 2008 and 2013.
Prior empirical work as discussed earlier has provided some insights into the election effect
in Malaysia. This study extends the scope of research to cover the election effect in politically
connected firms. As a primer, the Malaysian government exerts a significant influence over
the corporate sector through direct equity ownership of listed firms, namely government-
linked companies (GLCs). In addition, the Malaysian government also has influence in the
appointment of members of the GLCs board of directors and senior management positions,
including making major decisions such as contract awards, strategy, restructuring, and
financing, acquisition and divestments. Hence, the political patron in the corporate sector and
the political uncertainty due to the general election in Malaysia produce an interesting and
important case study.
Empirical evidence of election effect on the politically connected firm is lacking in Malaysia.
Nevertheless, previous literature has shown the close link between political connection and
firm value (Lau and Tong, 2008; Razak et al., 2011; Poon et al., 2013). Lau and Tong (2008)
revealed a significant positive relationship between the degree of government ownership and
firm value on 15 GLCs over the years 2000-2005. The results implied that Malaysia
government intervention improves the firm value of GLCs. Meanwhile, based on a sample of
210 GLCs and Non-GLCs firms for the period 1995-2005, Razak et al. (2011) used a panel
based regression approach to determine the impact of ownership mechanism on firm’s
performance. Results from their study appear to support the findings of Lau and Tong (2008),
where GLCs outperformed Non-GLCs in term of market-based valuation measures and
110
accounting-based measures. On the other hand, when the sample is matched for comparison
for small size companies in the sectors of Trading, Production, Plantation, Properties, and
Consumer Product, findings highlighted that Non-GLCs performance is better GLCs.
With a similar objective, Poon et al. (2013) narrowed down the research scope to Malaysian
commercial bank industry. Specifically, Poon et al. (2013) examined the effect of politically
connected boards on commercial bank performance by considering the interaction effect
between age, ethnicity, and the political connections of board directors. Generally, the result
of the study revealed that bank performance of politically connected firms improves
compared to the non-connected firms. Other than that, Poon et al. (2013) also investigated the
performance of political-connected banks in the years (2005-2009) after the 4th Prime
Minister, Tun Dr. Mahathir Mohammad, officially left the office. From the results, Poon et al.
(2013) concluded that high political-connected banks performance were poorer than high
political-connected banks in the later five years due to the handover of control among
politicians. The negative effect on the changing hand is consistent with an earlier study of
Mitchell and Joseph (2010). The resignation was genuinely unexpected and it created a shock
to the market. Mitchell and Joseph (2010) examined the immediate effect of the resignation
on political-connected firms and found that GLCs were more badly affected than Non-GLCs
during the first month of resignation in the year 2002.
Furthermore, remarkable studies of Johnson and Mitton (2003) and Mitchell and Joseph
(2010) focused on regulation changes that have occurred in Malaysia and discuss their likely
effect on the performance of individual Malaysian firms. In particular, Johnson and Mitton
(2003) analyzed the performance of the politically connected firm over the Asian Financial
crisis period, including the period of the imposition of capital control, while Mitchell and
111
Joseph (2010) focused on the effect of the removal of regulation. Johnson and Mitton (2003)
found evidence that politically favoured firms outperformed unconnected firms in the period
immediately after the imposition of capital controls during the 1997 Asian currency crisis in
Malaysia. However, when the capital controls were gradually reduced and the exchange rate
peg was removed, financial firms with political connections have not performed as well as
others since the measures set up to support them have been removed (Mitchell and Joseph,
2010).
A number of studies have investigated the link between political connections and firm
performance in countries in which relationship is required in doing business effectively, such
as China. In China, the local economy is less market-oriented or the government has more
discretion in allocating economic resources, and hence political connections are important for
Chinese firms (Chen et al., 2011). Therefore, it is intuitive to expect that political
connections have a positive effect on Chinese firms. But, empirical evidence in the literature
related to political connections in China is mixed. Negative effects of political connections on
firm performance were demonstrated by Fan et al. (2007) and Wu et al. (2012). Both the
studies found that firms with connected CEOs and chairman have lower values, whereas the
private firms have higher values. In this case, Su and Fung (2013) argued that definition of
political connections is problematic and lead to conflicting results. Hence, they compiled a
new database of political connections to include the firm’s top management team or board
members that have a close relationship to the Chinese government. By using the new data set,
a robust result confirmed the positive relationship between political connections and firm
performance. Hence, the China case study clearly indicates the importance of the right
measurement for political connections.
112
Besides the close link between political connections and firm value, a growing body
of evidence also supports the notion that share prices of politically connected firms are
strongly related to political news. In Malaysia, Hassan et al. (2012) examined share price
reaction to political events by using a sample of 138 firms for the period of 1990 - 2001. The
two important political events included in the study were the rumour of ill health of Prime
Minister Tun Dr. Mahathir Mohamad, and the sacking of Deputy Prime Minister Anwar
Ibrahim. They found that share prices of the politically connected firms increase with the
announcement of favourable political events, which is the sacking of the deputy prime. On
the other hand, the politically connected firms reacted negatively to the news of ill health of
the prime minister. Similarly, in the Philippines, the stock price of connected firms was badly
impacted by the negative news on the health of President Suharto during the final years of his
life (Fisman, 2001).
Turning to election effects on politically connected firms, only a couple of studies
have been done in other countries (Goldman et al., 2009; Yeh et al., 2013), but none of them
include Malaysia. Goldman et al. (2009) showed that political connections are important in
US company that connected to the parties. The company's stock return appeared to be
abnormally high following the announcement of the nomination of a politically connected
individual to the board. Moreover, in the 2000 presidential election, the Republican-
connected firm increased in value while Democratic-connected firms decreased in value
when the Republican candidate won the presidential election. A similar pattern was found in
Taiwan during the 2004 presidential election when there was an unexpected shock of the
election outcome (Yeh et al., 2013). Before the election, the Kuomintang was expected to
win and Kuomintang-connected firms were associated with higher abnormal returns.
Eventually, the incumbent President Chen from Democratic Progressive Party was re-elected
113
by a slim margin. The investors were shocked with the election result and led to a reversal in
stock returns for the Kuomintang-connected firm.
In summary, empirical results have shown that stock markets responded to the elections in
Presidential system countries and Parliamentary system countries. Moreover, studies on other
countries provide confidence results that the stock price of politically connected firm reacted
to political events as well as the election. The studies in Malaysian market have proved that
there is a link between political connections and firm value, and the share price of connected
firm react to political events. Based on the above literature review, none of the studies in
Malaysia has attempted to explore the election effect on politically connected firm or the
GLCs. Therefore, this present study would like to fill in the research gap in the literature by
examining the election effect on GLCs and Non-GLCs in Malaysia.
4.3 Data and Empirical Methodology
The daily closing stock price of 11 selected Government-Link-Companies (GLCs) and Non-
Government-Link-Companies (Non-GLCs) employed in this study was collected from Bursa
Malaysia, and the election dates were obtained from the Electoral Commission of Malaysia.
The selection of the sample companies in this study is based on the following sample
selection method which fulfilling two basics condition: (1) most active traded stock at the end
of the year 2015, and (2) must be listed in Bursa Malaysia since 4 January 1994. The full
sample period covers from 4 January 1994 to 31 December 2015, with a total of 5,738
observations, which covers the recent five Malaysia general elections. All data are collected
from the Bursa Malaysia (http://www.bursamalaysia.com). The important dates of general
elections are summarized in Table 4.1, which are the date of dissolution of parliament,
114
election date or voting date and the 1st parliament assembly after the election. The pre-general
election period refers to the duration from the day of dissolution of the parliament until the
day before voting, while the post-general election period refers to the duration from the day
after voting until the first parliament assembly.
[Insert Table 4.1: Malaysia General Election]
Table 4.2 and 4.3 presents the descriptive statistics for daily returns series for the full sample
(1994 – 2015) period of GLCs and Non-GLCs stock return. Daily returns are calculated as
the first difference in the natural logarithms of the stock market index, )/ln(100 1 ttt IIR
where tI and 1tI are the values of each index for periods t and 1t , respectively. In the
case of a trading day following a non-trading day, the return is calculated using the closing
price of the last trading day. From the descriptive statistics, the null hypothesis of normally
distributed daily returns is rejected by the Jarque-Bera normality test. This finding is in line
with most of the previous findings, saying that daily stock returns are not normally
distributed.
[Insert Table 4.2: Descriptive Statistics for GLCs Stock Return (1994 – 2015)]
[Insert Table 4.3: Descriptive Statistics for Non-GLCs Stock Return (1994 – 2015)]
Furthermore, mean returns for the sub-sample periods of pre-general election and
post-general election for GLCs and Non-GLCs are presented in Table 4.4 and 4.5. It is
observed that the mean returns prior to the pre-general election are mixed for the sub-sample
period of 1994-2005. However, for the sub-sample period of 2006-2015, the mean returns
115
(GLCs and Non-GLCs) are mostly negative (9 out of 11) prior to the pre-general election. On
the other hand, for the period of the post-general election, the mean returns for the indices are
mixed for the sub-sample period of 1994-2005. For the period of 2006-2015, most of the
mean returns (GLCs and Non-GLCs) are positive (8 out of 11) after the general election.
From the descriptive statistics and mean returns for the two sub-sample periods, it is notable
that there could be different election effects on the stock market for the general elections in
the year 1994 to 2005 and 2006 to 2015. The preliminary statistics justify the aim of this
study in dividing the full sample period into two sub-samples in order to study the election
effects under different political condition.
[Insert Table 4.4: Sub-sample Mean Return for GLCs]
[Insert Table 4.5: Sub-sample Mean Return for Non-GLCs]
Table 4.6 and 4.7 present the selected sample of Malaysian Government-Link
Companies and Non-Government-Link Companies information in this study. Information
included in the tables are company name, industry or sector, symbol, market capitalization,
number of shares, earning per share, revenue, profit before tax and net profit.
[Insert Table 4.6: Selected Sample of Malaysian Government-Link Companies (GLCs)]
[Insert Table 4.7: Selected Sample of Malaysian Non-Government-Link Companies
(Non-GLCs)]
116
In this study, the test for market volatility during general elections is carried out by using the
Threshold Generalized Autoregressive Conditional Heteroskedasticity (Threshold GARCH /
GJR GARCH) model developed by Glosten et al. (1993), Threshold GARCH / GJR GARCH
(1, 1)5 model with dummy variables:
ttttt RPtGEPGER 13210 (1)
ttttttt PtGEPGEN 212
1112
12
1102
(2)
where tR is the logarithmic return of the market index at day t ; tPGE and tPtGE are dummy
variables which take on value 1 if the corresponding return for day t is the period of pre-
general election, and the period of post-general election respectively, and 0 otherwise; t is
the error term. Meanwhile, 30 ,..., are the parameters to be estimated. Among them, 0
measures the mean return (in percentage) on other trading days; whereas 1 and 2 capture
the average return of the stock index for the period of pre-general election and post-general
election.
The null hypothesis of the test is 0: 210 H , which implies that average daily
returns (volatility) for the period of pre-general election and post-general election are
significant different from zero. If the null hypothesis does not hold, then it can be concluded
that the market index is characterized by statistically different on average returns (volatility)
for the period of pre-general election and post-general election. In another word, this would
imply that general election effect is indeed present in the market.
5 According to Bollerslev et al. (1992), in testing the GARCH models, p = q = 1 is sufficient for most financial series. Hence, the highest order of p and q considered in this study for the Threshold GARCH model, is (1, 1).
117
In the Equation (2), tN takes on value 1 when the stock quote falls in a period and 0
for increments of the stock quotation. Besides, the parameter is used to capture the
asymmetrical effect of bad news (decrease in stock indices, hence negative tR ) and good
news (increase stock indices, hence positive tR ). If 0 by the t test of significance, then it
can be concluded that the impact of news is asymmetric. If the parameter is positive, then
good news has an impact of i on volatility while bad news has an impact of ( i ) on
volatility. Thus, positive value of indicates the existence of a leverage effect in that bad
news increases volatility. The additional parameters, t , which makes this specification
different from the original Threshold GARCH model, are employed to capture the daily
effect. Furthermore, a lagged value of the return variable is introduced in the equations to
avoid serial correlation error terms in the model, which may yield misleading inferences.
Next, this paper also investigates whether the observed return volatilities on the
various pre-general election and post-general election are related to trading volume, indirectly
testing the Admati and Pfleiderer (1988) and Foster and Viswanathan (1990) models. This
paper model the natural logarithm of the volume as the following Threshold GARCH (1,1)
process.
ttttt VolPtGEPGEVol 13210 lnln (3)
2111
21
2110
2 ttttt N (4)
where tVolln is the natural logarithmic trading volume at day t ; tPGE and tPtGE are dummy
variables which take on value 1 if the corresponding trading volume for day t is the period of
118
pre-general election, and the period of post-general election respectively, and 0 otherwise; t
is the error term. Meanwhile, 30 ,..., are the parameters to be estimated. For example, if the
result showed high volatility (Equation 2) and low trading volume (Equation 3), then this
result is concluded to support the theoretical model of Foster and Viswanathan (1990) which
stated that liquidity traders being unwilling to trade in the periods where the prices are more
volatile. However, if the result showed high volatility and high trading volume, then this
result is concluded to support Admati and Pfleiderer (1988) theoretical argument which stated
that liquidity traders are willing to trade in the periods where the prices are more volatile.
4.4 Empirical Results and Discussions
Firstly, the results of pre-general election effect and post-general election effect for the full-
sample period of 1994-2015 are presented in Table 4.8(a), 4.8(b), 4.9(a) and 4.9(b). Table
4.8(a) and 4.8(b) report the results of the mean equation and variance equation of the
Threshold GARCH (1, 1) model for Government-Link Companies (GLCs). Meanwhile,
Table 4.9(a) and 4.9(b) report the estimation results for Non-Government-Link Companies
(Non-GLCs). The diagnostic test result is included in the lower part of the tables to support
the validity of the models. Under the mean equation, the dummy coefficients of the GLCs
and Non-GLCs are all insignificant in the pre-general election period. For the post-general
election, there are only three out of eleven GLCs recorded significant returns in the full
sample period. Specifically, the stock returns of the TNB and MRC are negatively significant,
while the CCM is positively significant after the general elections. For Non-GLCs, the NESZ
is the only positive and significant stocks return in the post-general election.
119
The estimation results of the variance equations are also presented in Table 4.8(a),
4.8(b), 4.9(a) and 4.9(b). For the variance equation, the pre-general election dummy
coefficients are positively significant for seven GLCs and six Non-GLCs. This result
indicates that the share prices of GLCs (AHB, CIMB, MAY, BOUS, TNB, UMWH, and
MRC) and Non-GLCs (GENM, GENT, MISC, YTL, IOI, and KLK) experienced significant
high volatility in the pre-general election periods. Besides, significant low volatility is found
in a GLCs (TMK) and a Non-GLCs (RHBC) in the pre-general election periods. Meanwhile,
for the post-general election periods, the dummy coefficients of the variance equations are
positive and significant for MBS (GLCs), BOUS (GLCs) and PEP (Non-GLCs). Thus, it is
evident that general election result increases the volatility in these three companies. On the
other hand, significant low volatility is found in BIMB, CIMB, TMK, and MRC for GLCs,
and RHBC, MISC, YTL, NESZ, ROTH, and KLK for Non-GLCs.
In the variance equations, the leverage effect term, , in the variance equation is
positive and statistically different from zero in seven GLCs and nine Non-GLCs. The positive
value of indicates that the leverage effect in bad news increased the volatility. In particular,
the bad news has an impact of ( i ), while good news has an impact of ( i ) only. For
example, refer to Table 4.8(a), bad news in the AHB (GLCs) has an impact of 0.9085
(0.8864+0.0221), while good news only has an impact of 0.8864. Hence, the results indicate
the existence of the asymmetric effect on stock volatility in seven GLCs (AHB, CIMB, MAY,
BOUS, TNB, UMWH, CCM, and MRC) and nine Non-GLCs (PBK, RHBC, GENM, GENT,
MISC, YTL, PEP, ROTH, and IOI).
[Insert Table 4.8(a): Threshold GARCH Results for Pre-General Election and Post-
General Election (1994 - 2015) – Government Link Companies (GLCs)]
120
[Insert Table 4.8(b): Threshold GARCH Results for Pre-General Election and Post-
General Election (1994 - 2015) – Government Link Companies (GLCs)]
[Insert Table 4.9(a): Threshold GARCH Results for Pre-General Election and Post-
General Election (1994 - 2015) – Non-Government Link Companies (Non-GLCs)]
[Insert Table 4.9(b): Threshold GARCH Results for Pre-General Election and Post-
General Election (1994 - 2015) – Non-Government Link Companies (Non-GLCs)]
Next, this study examines the presence of pre-general election effect and post-general
election effect in GLCs and Non-GLCs for the sub-sample period of 1994-2005 and the
results are presented in Table 4.10(a), 4.10(b), 4.11(a) and 4.11(b). For the stock return, most
of the dummy coefficients for the mean equations of the pre-general election are positive for
GLCs (BIMB, CIMB, MAY, MBS, BOUS, TMK, TNB, CCM, and MRC), however, they are
insignificant. All the Non-GLCs are found to have insignificant dummy coefficients for stock
return, but the sign of the dummy coefficient is positive for the stock of RHBC, GENT, YTL,
NESZ, ROTH, and KLK and negative for the other five Non-GLCs. On the other hand, for
post-general election, MRC is the only GLCs in the Property sector that has a significant and
negative dummy coefficient. This indicates that the outcome of the general election has
negatively impacted the share price of this company.
Furthermore, the estimation results of the variance equations are presented in Table
4.10(a), 4.10(b), 4.11(a) and 4.11(b). For the sub-sample period of 1994-2005, the results
show that the GLCs and Non-GLCs react differently as compared with the results of the full-
121
sample period. Among the GLCs, six of them experienced significant volatility changed
before the general elections. In particular, BIMB, MBS, BOUS, UMWH, CCM, and MRC
experienced significant low volatility before the general elections. On the other hand, six of
the Non-GLCs (PBK, MISC, YTL, PEP, IOI, and KLK) also experienced significant low
volatility during the period of pre-general election. However, after the announcement of the
election result, the stock volatility increased significantly in CIMB, BOUS, TNB, and
UMWH for GLCs, while PBK, MISC, and PEP for Non-GLCs. Thus, it is evident that most
of the GLCs and Non-GLCs in this study experienced significant volatility change due to the
general election and the finding is contradicted with the full sample period.
The results of variance equations also confirm that there is an asymmetric effect of
political elections on stock volatility in most of the GLCs and Non-GLCs for the sub-sample
period of 1994-2005. The positive value of the leverage effect term is statistically significant,
and this indicates the existence of asymmetrical effect in the sample of this study. This
finding implies that negative shocks or bad news from election may have larger impact on
stock volatility than good news in the sub-sample period of 1994-2005. Lastly, the validity of
the model is checked by the diagnostic tests. No remaining ARCH effect and serial
correlation are found in the estimated models.
[Insert Table 4.10(a): Threshold GARCH Results for Pre-General Election and Post-
General Election (1994 - 2005) – Government Link Companies (GLCs)]
[Insert Table 4.10(b): Threshold GARCH Results for Pre-General Election and Post-
General Election (1994 - 2005) – Government Link Companies (GLCs)]
122
[Insert Table 4.11(a): Threshold GARCH Results for Pre-General Election and Post-
General Election (1994 - 2005) – Non-Government Link Companies (Non-GLCs)]
[Insert Table 4.11(b): Threshold GARCH Results for Pre-General Election and Post-
General Election (1994 - 2005) – Non-Government Link Companies (Non-GLCs)]
For the second sub-sample period of 2006-2015, Table 4.12(a) and 4.12(b) present the
results of pre-general election effect and post-general election effect for the GLCs, while
Table 4.13(a) and 4.13(b) report the estimation results for the Non-GLCs. From the
estimation of mean equations, TMK (GLCs) is the only company from the sector of Trade
and Services, that showed significant negative stock returns after the general elections. On
the other hand, NESZ (Non-GLCs) has a positive and significant return during the period of
the post-general election, which indicates that the general election result brought a positive
impact to this company’s share price.
As mentioned earlier, the turbulent political condition in the 12th and 13th Malaysia
due to the fierce competition between the two major coalition induced market uncertainty.
Prior to the general elections, the market condition experienced significant volatility changes,
and the empirical results of this study are in line with the situation. From the estimation
results of the TGARCH variance equations, seven out of eleven (CIMB, MAY, MBS, BOUS,
TNB, UMWH, and MRC) of the GLCs encountered significant high volatility in the pre-
general election periods, and six of the Non-GLCs (GENM, GENT, MISC, YTL, IOI, and
KLK) also recorded the same pattern of stock volatility. Nevertheless, two Non-GLCs,
namely PBK and NESZ, recorded significant low volatility during the pre-general election
periods. For the post-general election periods, this study also finds some evidence of election
123
effect in stock volatility. The results of the post-general election show significant low
volatility in three GLCs (CIMB, TMK, and MRC). Notably, ten of the Non-GLCs has
negative volatility during the post-general election and four of them are statistically
significant (RHBC, MISC, YTL, and KLK). The result on the second sub-sample period of
2006-2015 is evidently different compared to the first sub-sample period in early years,
where most of the GLCs and Non-GLCs recorded significant low volatility before general
elections and significant high volatility after general elections.
The asymmetric effect of the general elections is also reported in Table 4.12(a),
4.12(b), 4.13(a), and 4.13(b). The asymmetric effect appeared only on a few of the GLCs and
Non-GLCs. The leverage effect term, , is statistically different from zero for BIMB, CIMB,
MAY, BOUS, and TMK (GLCs), indicating the existence of the asymmetrical effect in the
stock index. Unlike the results of the first sub-sample, some of the company stock has a
significant negative leverage effect term, which indicates that good news increased the
volatility. In particular, the good news has an impact of ( i ), while bad news has an impact
of ( i – ). For the case of BIMB, good news has an impact of 0.7231, while bad news has a
lower impact of 0.5917 (0.7231–0.1314). Therefore, during the sub-sample period of 2006-
2015, positive shocks create a greater impact on the conditional volatility of the GLCs
(BIMB, BOUS, and TMK) than negative shocks. Besides, the validity of the model is
supported by the diagnostic test with no remaining ARCH effect and serial correlation in
most of the estimated models.
[Insert Table 4.12(a): Threshold GARCH Results for Pre-General Election and Post-
General Election (2006 - 2015) – Government Link Companies (GLCs)]
124
[Insert Table 4.12(b): Threshold GARCH Results for Pre-General Election and Post-
General Election (2006 - 2015) – Government Link Companies (GLCs)]
[Insert Table 4.13(a): Threshold GARCH Results for Pre-General Election and Post-
General Election (2006 - 2015) – Non-Government Link Companies (Non-GLCs)]
[Insert Table 4.13(b): Threshold GARCH Results for Pre-General Election and Post-
General Election (2006 - 2015) – Non-Government Link Companies (Non-GLCs)]
High volatility in stock returns is always a concern for market participants, therefore
this study also investigates whether the high volatility is associated with high trading volume
as proposed by Admati and Pfleiderer (1988), or low trading volume as proposed by Foster
and Viswanathan (1990). Preliminarily, the relationship between trading volume and stock
price volatility before and after the election dates is briefly explained in Figure 4.1 and Figure
4.2. For GLCs, weaker relationship is found in both the pre-general election period and post-
general election period. On the other hand, the relationship between Non-GLCs trading
volume and stock volatility appears to be weaker before general elections. Then, after general
elections, stronger relationship is found on Non-GLCs share prices. For both the GLCs and
Non-GLCs stock prices, although the pattern is not quite as tight, the relationship between
trading volume and stock volatility remain the same pattern in the two sub-sample of 1994-
2005 and 2006-2015.
[Insert Figure 4.1: Relationship between Trading Volume and Stock Price Volatility for
GLCs and Non-GLCs from 1994 - 2005]
125
[Insert Figure 4.2: Relationship between Trading Volume and Stock Price Volatility for
GLCs and Non-GLCs from 2006 - 2015]
Further, the examination of election effect in trading volume is done by modeling the
logarithm of trading volume in the Threshold GARCH (1,1) process. The results of the pre-
general election and post-general election on trading volume for GLCs and Non-GLCs during
the period of 1994-2005 are reported in Table 4.14(a), 4.14(b), 4.15(a) and 4.15(b). In early
years of 1994-2005, the empirical results show that NESZ (Non-GLCs) is the only company
with a low volume of trading during the pre-general election. For the post-general election
periods, three of the GLCs (BOUS, TMK, and CCM) and four of the Non-GLCs (GENM,
MISC, PEP, and IOI) have high trading volume.
In summary, the volatility of returns and trading volume findings for GLCs and Non-
GLCs during the 1994 – 2005 period is as follows: High volatility of returns and high trading
volume are observed during the post-general election for BOUS (GLCs) and PEP (Non-GLCs)
stocks. Investors were confident and started to trade in the market after the coalition of
Barisan Nasional won in the general elections. In other words, the high volatility after the
general election is not induced by uncertainties of the general election. In fact, the high
volatility could be induced by the active trading activity in the market right after the election.
These findings support the argument of Admati and Pfleiderer (1988), which speculate that
trading volume would be high when price volatility is high because of the willingness of
liquidity traders to trade in the periods where the prices are more volatile.
[Insert Table 4.14a: Threshold GARCH Results for Trading Volume during Pre- and
Post-General Election (1994 - 2005) – Government Link Companies (GLCs)]
126
[Insert Table 4.14b: Threshold GARCH Results for Trading Volume during Pre- and
Post-General Election (1994 - 2005) – Government Link Companies (GLCs)]
[Insert Table 4.15a: Threshold GARCH Results for Trading Volume during Pre- and
Post-General Election (1994 - 2005) – Non-Government Link Companies (Non-GLCs)]
[Insert Table 4.15b: Threshold GARCH Results for Trading Volume during Pre- and
Post-General Election (1994 - 2005) – Non-Government Link Companies (Non-GLCs)]
For the second sub-sample period of 2006-2015, the results of the pre-general election
and post-general election in trading volume for GLCs and Non-GLCs are reported in Table
4.16(a), 4.16(b), 4.17(a), and 4.17(b). The empirical results show that the significant low
volume of trading during the pre-general election is recorded by ROTH (Non-GLCs) only.
Meanwhile, significant high trading volume occurs on the pre-general election periods in four
of the GLCs (CIMB, MAY, TNB, and UMWH) and two of the Non-GLCs (MISC and
NESZ). For the post-general election periods, all the significant trading volume are with a
positive sign, which indicates a high trading volume in four of the GLCs (CIMB, MAY,
UMWH, and CCM) and three of the Non-GLCs (NESZ, ROTH, and IOI).
By combining the finding on the returns volatility and trading volume for the pre-
general election periods in 2006-2015, there are high volatility of returns and high trading
volume for GLCs (CIMB, MAY, TNB, and UMWH) and Non-GLCs (MISC). For post-
general election periods, the findings show that high volatility of returns of NESZ (Non-
GLCs) is also associated with a high trading volume. All these findings are in line with the
127
predictions of Admati and Pfleiderer (1988). In addition, a plot of trading volume and stock
prices as shown in Figure 4.3 and Figure 4.4 would further explain the relationship between
the volume of the highly traded stock and its stock volatility. Figure 3 shows the trading
volume and stock prices for the four highly traded GLCs before general elections in the years
2008 and 2013. In year 2008, high trading volume of the four GLCs stock is associated with a
sharp price drop before the general election. In another words, these GLCs stock is sold in
large quantities which caused a sharp decrease of stock price before the 12th general election.
The same case happened on the highly trade Non-GLCs stock of MISC as shown in Figure 4.
Nevertheless, before the 13th general election in year 2013, the market is very volatility and
there is a mixture of buying and selling activity in the market. For example, the stock price of
CIMB is highly volatile before general election. In some of the days, the stock is purchased in
large quantities and the stock price went up sharply. But selling activity also existed as shown
by the high trading volume with sharp decrease in price. For the Non-GLCs of MISC, the
stock price was not as volatile as the GLCs, it decreased on a particular day and then stabilize.
Selling activity is shown by the high trading volume and low stock price of the MISC.
[Insert Table 4.16a: Threshold GARCH Results for Trading Volume during Pre- and
Post-General Election (2006 - 2015) – Government Link Companies (GLCs)]
[Insert Table 4.16b: Threshold GARCH Results for Trading Volume during Pre- and
Post-General Election (2006 - 2015) – Government Link Companies (GLCs)]
[Insert Table 4.17a: Threshold GARCH Results for Trading Volume during Pre- and
Post-General Election (2006 - 2015) – Non-Government Link Companies (Non-GLCs)]
128
[Insert Table 4.17b: Threshold GARCH Results for Trading Volume during Pre- and
Post-General Election (2006 - 2015) – Non-Government Link Companies (Non-GLCs)]
[Insert Figure 4.3: Trading Volume and Stock Prices for the Highly Traded GLCs
before the General Elections]
[Insert Figure 4.4: Trading Volume and Stock Prices for the Highly Traded Non-GLCs
before the General Elections]
4.5 Conclusion
This study investigates the election effects on stock market volatility for Malaysian
Government Link Companies (GLCs) and Non-Government Link Companies (Non-GLCs)
for the period of January 4, 1994, through December 31, 2015. Using the Threshold GARCH
or GJR GARCH (1,1) model, this study analyzes the daily stock prices of eleven GLCs and
eleven Non-GLCs to see the pre-general election effect and post-general election effect
specifically. Furthermore, to test whether the pattern of the stock volatility changes according
to the political condition, the full sample period is divided into early years (1994-2005), and
later years (2006-2015). The year 2006 has been chosen as the cut-off date because the 12th
general election (2008) and the 13th general election (2013) induced election uncertainty with
a strong expectation of political changeover and eventually the incumbent lost its two-thirds
majority in the Parliament. Interestingly, the findings of this study show that the full sample
period result is likely dominated by the results of the second sub-sample period of (2006-
2015). By dividing the sample period, the pattern of the stock volatility in GLCs and Non-
129
GLCs is clearly different in the two sub-samples, which is in line with the political condition
in the early years of 1994-2005 and the later years of 2006-2015.
For the first sub-sample period of 1994 to 2005, there is an asymmetric effect of
political elections on stock volatility in both the GLCs and Non-GLCs. Moreover, this study
observes the low volatility of returns during the pre-general election period in both the GLCs
and Non-GLCs stock indices. No significant result is found on the analysis of trading volume,
except for one GLCs stock index (BOUS) and one Non-GLCs stock index (NESZ). The low
volatility in the market before election gives a positive signal that there is no uncertainty due
to the general election. After the general election, although there were no unexpected
outcomes as the coalition of Barisan Nasional won in the general elections, the stock
volatility increased for some of the GLCs and Non-GLCs stocks. Further analysis of trading
volume shows that there is high trading volume for a few of the GLCs and Non-GLCs stock
indices, which lend support to the argument of Admati and Pfleiderer (1988). During the
years of 1994-2005, investors were expecting positively aligned with the stable political
condition in the country. When the stock market reopened after the election day, the active
trading activity typically explained the high stock volatility in both the GLCs and Non-GLCs
stock.
In later years of 2006-2015, the findings from the second sub-sample show that most
of the GLCs and Non-GLCs stock prices were highly volatile before the general election.
Specifically, seven of the GLCs stock indices and six of the Non-GLCs stock indices
encountered significant high volatility in pre-general election periods. According to the
political condition during that period, the high volatility in the market was due to
uncertainties associated with the 12th and 13th Malaysian general election. Interestingly, those
130
GLCs and Non-GLCs with high market capitalization that encountered high volatility is also
associated with a significant high trading volume, for example, MAY, CIMB, TNB of GLCs,
and MISC, IOI of Non-GLCs. This pattern of trading was probably due to the willingness of
liquidity traders to trade in the periods where the prices are more volatile (Admati and
Pfleiderer, 1988). Nonetheless, during that period, it is not surprising that local investors took
the opportunity to trade actively before the general election, which is totally different with the
foreign investors trading strategy and expectation. Majority of the liquidity traders in
Malaysian stock market are from the local institutional investors, such as domestic pension
funds, insurance companies, and mutual funds. They would closely monitor the local market
development and trade confidently in those high market capitalized stock index before the
general election.
Another interesting point found in this study is that when there is uncertainty in the
market, the stock price of the GLCs and Non-GLCs in the finance sector react differently in
term of volatility and trading volume. High volatility and high trading volume are found in
GLCs before and after the 12th and 13th general election. Meanwhile, the Non-GLCs in the
finance sector have low volatility with insignificant trading volume around the general
election. Despite the market uncertainties, investors are still willing to actively trade the
GLCs stock. However, this trading pattern is not observed in early years with the stable
political condition. This indicates that investors are very careful during the time of market
uncertainty as it is always safer to trade GLCs stock which is more liquid than others.
Nonetheless, based on the finding of this study, this trading pattern appear only in the finance
sector, but not in other sectors.
131
Overall, the findings of this study indicate that the Malaysian stock market volatility
is associated with investors' behaviour during the periods of the general election. The
presence of a political shock during general election makes it possible to investigate how the
politically connected firms react to the market uncertainty. Our study provides an exemplar
for further studies to explore further details by employing a comprehensive disaggregated
data for different sectors and firm characteristics and sectors.
132
References
Admati, A. and Pfleiderer, P. (1988). “A theory of intraday patterns: Volume and price variability”, Review of Financial Studies, Vol. 1, pp. 3 – 40. Agrawal, A. and Knoeber, C. R. (2001), “Do Some Outside Directors Play a Political Role?”, Journal of Law and Economics, Vol. 44, pp. 179 – 98. Ali, N., Nassir, A. M., Hassan, T. and Abidin, S. Z. (2010), “Short Run Stock Overreaction: Evidence from Bursa Malaysia”, International Journal of Economics and Management, Vol. 4 No. 2, pp. 319 – 333. Allivine, F. D. and O’Neil, D. D. (1980). “Stock Market Returns and the Presidential Election Cycle”, Financial Analysts Journal, Vol. 36 No. 5, pp. 49 - 56. Bollerslev, T., Chou, R. and Kroner, K. (1992). “ARCH Modelling in Finance: A Selective Review of the Theory and Empirical Evidence”, Journal of Econometrics, Vol. 52, pp. 5 - 59. Chen, C. J. P., Li, Z., Su, X. and Sun, Z. (2011). “Rent-seeking Incentives, Corporate Political Connections, and the Control Structure of Private Firms: Chinese Evidence”, Journal of Corporate Finance, Vol. 17, pp. 229 – 243. Chiu, C. L., Chen, C. D. and Tang, W. W. (2005). “Political Elections and Foreign Investor Trading in South Korea’s Financial Markets”, Applied Economics Letters, Vol. 12, pp. 673 – 677. Civilize, S., Wongchoti, U. and Young, M. (2015). “ Political Connection and Stock Returns: A Longitudinal Study”, The Financial Review, Vol. 50, pp. 89 – 119. Cooper, M. J., Gulen, H. and Ovtchinnikov, A. V. (2010). “Corporate Political Contributions and Stock Returns”, Journal of Finance, Vol. 65, pp. 687 – 724. De Soto, Hernando. (1989). “The other path: The invisible revolution in the Third Worlds”, New York: Harper and Row Publishers. Dinç, I. Serdar. (2005). "Politicians and Banks: Political Influences on Government-Owned Banks in Emerging Countries”, Journal of Financial Economics, Vol. 77, No. 2, pp. 453 - 479. Fan, J. P. H., Wong, T. J. and Zhang, T. (2007). “Politically Connected CEOs, Corporate Governance, and Post-IPO Performance of China’s Newly Partially Privatized Firms”, Journal of Financial Economics, Vol. 84, pp. 330 – 357. Fisman, R. (2001). “Estimating the Value of Political Connections”, American Economic Review, Vol. 91, pp. 1095 – 1102. Foerster, S. R. and Schmitz, J. J. (1997). “The Transmission of U.S. Election Cycles to International Stock Returns”, Journal of International Business, Vol. 28, No. 1, pp. 1 - 27.
133
Foster, F. D. and Viswanathan, S. (1990). “A theory of the interday variations in volume, variances, and trading cost in securities market”, Review of Financial Studies, Vol. 3, pp. 593 – 624. Gemmill, G. (1992). “Political risk and market efficiency: tested based in British stock and option markets in the 1987 election”, Journal of Banking and Finance, Vol. 16, pp. 211 – 231. Glosten, L. R., Jagannathan, R. and Runkle, D. E. (1993). “Relationship between the expected value and the volatility of the nominal excess return on stocks”, The Journal of Finance, Vol. 48, No. 5, pp. 1779 - 1801. Goldman, E., Rocholl, J. and So, J., (2009). “Do political connected boards affect firm value?”, Review of Financial Studies, Vol. 22, pp. 2331–2360. Harrington, J. (1993). “Economic Policy, Economic Performance, and Elections”, American Economic Review, Vol. 83, No. 1, pp. 27 – 42. Hassan, T., Hassan, M. K., Mohamad, S. and Chen, C. M. (2012). “Political Patronage and Firm Performance: Further Evidence from Malaysia”, Thunderbird International Business Review, Vol. 54, No. 3, pp. 373 – 393. Huang, R. D. (1985). “Common Stock Returns and Presidential Elections”, Financial Analysts Journal, Vol. 41, No. 2, pp. 58 - 65. Hung, L. C. (2013). “U.S. Presidential Elections and the Taiwanese Stock Market”, Issues & Studies, Vol. 49, No. 1, pp. 71 - 97. Johnson, S. and Mitton, T. (2003). “Cronyism and Capital Controls: Evidence from Malaysia”, Journal of Financial Economics, Vol. 67, pp. 351 – 382. Kroszner, R. S. and Stratmann, T. (1998). “Interest Group Competition and the Organization of Congress: Theory and Evidence from Financial Services’ Political Action Committees”, American Economic Review, Vol. 88, pp. 1163 – 1188. Lau, Y. W. and Tong, C. Q. (2008). “Are Malaysian Government Linked Companies (GLCs) creating value?”, International Applied Economics and Management Letters, Vol. 1, No. 1, pp. 9 – 12. Lean, H. H. (2010). “Political General Election and Stock Performance: The Malaysian Evidence”, in Ismail, M. T. and Mustafa, A. (Ed.), Research in Mathematics and Economics, Penang: Universiti Sains Malaysia, Malaysia, pp. 111 - 120. Lean, H. H. and Yeap, G. P. (2017). “Asymmetric Effect of Political Elections on Stock Returns and Volatility in Malaysia”, in Munir, Q. and Kok, S. C. (Ed.), Information Efficiency and Anomalies in Asian Equity Markets, Routledge, Taylor and Francis Group, pp. 228 - 245. Mitchell, H. and Joseph, S. (2010). “Changes in Malaysia: Capital controls, prime ministers and political connections”, Pacific-Basin Finance Journal, Vol. 18, pp. 460 – 476
134
Niessen, A. and Ruenzi, S. (2010). “Political Connectedness and Firm Performance: Evidence from Germany”, German Economic Review, Vol. 11, No. 4, pp. 441 – 464. Nordhaus, W. (1975., “The Political Business Cycle”, Review of Economic Studies, Vol. 42, No. 2, pp. 169 - 190. Pantzalis, C., Stangeland, D. A. and Turtle, H. J. (2000). “Political Elections and the Resolution of Uncertainty: The International Evidence”, Journal of Banking and Finance, Vol. 24, pp. 1575 – 1604. Poon, W. C., Yap, A. K. H. and Lee, T. H. (2013). “The Outcome of Politically Connected Boards on Commercial Bank Performance in Malaysia”, Modern Applied Science, Vol. 7, No. 1, 35-50. Razak, N. H. A., Ahamd, R. and Joher, A. H. (2011). “Does Government Linked Companies (GLCs) perform better than non-GLCs? Evidence from Malaysian Listed Companies”, Journal of Applied Finance and Banking, Vol. 1, No. 1, pp. 213 – 240. Smales, L. A. (2016). “The role of political uncertainty in Australian financial markets”, Accounting and Finance, Vol. 56, No. 2, pp. 545–575. Su, Z. Q. and Fung, H. G. (2013). “Political Connections and Firm Performance in Chinese Companies”, Pacific Economic Review, Vol. 18, No. 3, pp. 283 – 317. Wang, Y. H. and Lin, C. T. (2009). “The Political Uncertainty and Stock Market Behavior in Emerging Democracy: The Case of Taiwan”, Quality and Quantity, Vol. 43, No. 2, pp. 237 - 248. Wong, W. K. and McAleer, M. (2009). “Mapping the Presidential Election Cycle in U.S. Stock Market”, Mathematics and Computers in Simulation, Vol. 79, No. 11, pp. 3267 - 3277. Wu, W., Wu, C. and Rui, O. M. (2012). “Ownership and the Value of Political Connections: Evidence from China”, European Financial Management, Vol. 18, pp. 695 – 729. Yeh, Y. H., Shu, P. G. and Chiu, S. B. (2013). “Political Connections, Corporate Governance and Preferential Bank Loans”, Pacific-Basin Finance Journal, Vol. 21, pp. 1079 – 1101.
135
Table 4.1: Malaysia General Election
Dissolution of Parliament Election Day 1st Parliament Assembly after Election 9th General Election
(1995) 6 April 1995 (Thursday)
25 April 1995 (Tuesday)
7 June 1995 (Wednesday)
10th General Election (1999)
11 November 1999 (Wednesday)
29 November 1999 (Monday)
20 December 1999 (Monday)
11th General Election (2004)
4 March 2004 (Thursday)
21 March 2004 (Sunday)
17 May 2004 (Monday)
12th General Election (2008)
13 February 2008 (Wednesday)
8 March 2008 (Saturday)
28 April 2008 (Monday)
13th General Election (2013)
3 April 2013 (Wednesday)
5 May 2013 (Sunday)
24 June 2013 (Monday)
136
Table 4.2: Descriptive Statistics for GLCs Stock Return (1994 – 2015) AHB BIMB CIMB MAY MBS BOUS TMK TNB UMWH CCM MRC Sectors Finance Finance Finance Finance Finance Trade &
Services Trade & Services
Trade & Services
Consumer Products
Industrial Products
Property
Mean -0.0111 -0.0024 0.0171 0.0112 -0.0064 0.0015 0.0088 0.0007 0.0192 -0.0087 -0.0243 Max 37.3026 29.7458 39.0198 28.2448 25.9996 19.2706 24.0617 31.7173 45.8182 15.2078 42.5611 Min -39.7003 -25.9605 -26.8912 -26.6786 -28.7955 -17.0642 -31.2550 -26.5606 -26.7595 -13.2311 -41.8419 Std. Dev. 2.5777 2.4351 2.5542 1.8591 3.0686 1.8999 1.9791 2.1068 2.3287 1.8089 3.6110 Skewness 1.0396 0.7895 1.1848 0.8466 0.6498 0.2262 -0.5603 1.0175 1.9510 0.4567 0.4850 Kurtosis 34.5299 22.3824 32.6010 33.3981 14.0606 16.8216 31.7123 32.8133 62.4940 11.0319 18.2961 Jarque-Bera
238714.40 90414.70 210832.50 221608.70 29652.64 45722.67 197398.90 213494.80 849885.40 15622.91 56163.37
Probability 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
Table 4.3: Descriptive Statistics for Non-GLCs Stock Return (1994 – 2015) PBK RHBC GENM GENT MISC YTL NESZ PEP ROTH IOI KLK Sectors Finance Finance Trade &
Services Trade & Services
Trade & Services
Trade & Services
Consumer Products
Consumer Products
Consumer Products
Plantation Plantation
Mean 0.0333 -0.0012 0.0041 0.0072 0.0168 0.0191 0.0255 0.0291 0.0168 0.0336 0.0344 Max 21.8915 45.2147 23.4994 19.9001 23.7293 28.2004 25.9958 18.0921 18.6877 25.7966 27.6331 Min -19.1699 -23.7534 -25.8235 -18.4922 -27.5225 -29.9099 -25.9511 -13.8150 -34.1749 -27.9109 -23.7400 Std. Dev. 1.7523 2.7412 2.3699 2.0926 1.8754 2.2346 1.4925 1.6799 1.6442 2.3958 1.9847 Skewness 1.1010 1.6224 0.4367 0.1034 -0.0231 0.2928 -0.4409 0.2213 -1.3616 -0.0685 0.1652 Kurtosis 30.7617 32.7859 14.1505 9.9857 31.5312 34.4055 82.0309 13.4096 44.0384 19.2672 26.8057 Jarque-Bera
185423.20 214631.80 29908.55 11677.54 194622.10 235891.00 1493473.00 25953.88 404426.10 63270.90 135517.00
Probability 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
137
Table 4.4: Sub-Sample Mean Return for GLCs
GLCs AHB BIMB CIMB MAY MBS BOUS TMK TNB UMWH CCM MRC Sectors FIN FIN FIN FIN FIN TRAD TRAD TRAD CONPR INDPRO PROP
Sub-sample 1 1994 – 2005
Pre-GE -0.4516 0.1128 0.3219 0.1348 -0.1630 0.2596 0.2042 0.3196 -0.0678 -0.0179 -0.2635 Post-GE 0.0084 -0.0343 -0.0121 -0.0038 0.0374 -0.1548 0.1105 -0.1983 0.2109 0.1962 -0.4906
Sub-sample 2 2006 - 2015
Pre-GE 0.0572 0.0001 -0.1066 -0.1449 -0.3423 -0.3282 -0.1117 -0.0900 -0.2315 -0.2659 -0.9675 Post-GE 0.0978 0.0267 0.0643 -0.0613 0.3985 0.0405 0.1867 -0.2835 0.1219 0.0868 -0.3581
Note: FIN: Finance, TRAD: Trade and Services, CONPR: Consumer Product, INDPRO: Industrial Product, PROP: Property.
Table 4.5: Sub-Sample Mean Return for Non-GLCs
Non-GLCs PBK RHBC GENM GENT MISC YTL NESZ PEP ROTH IOI KLK Sectors FIN FIN TRAD TRAD TRAD TRAD CONPR CONPR CONPR PLANT PLANT
Sub-sample 1 1994 - 2005
Pre-GE -0.0060 0.0449 -0.0766 0.0130 -0.0801 0.0398 0.2246 0.0522 0.1717 -0.2019 0.1058 Post-GE 0.0319 -0.0568 0.0502 -0.1604 -0.0114 -0.1198 0.1561 -0.1123 0.1533 0.0738 -0.1022
Sub-sample 2 2006 - 2015
Pre-GE -0.2400 -0.2356 -0.2755 -0.3407 -0.8856 -0.1532 0.0805 -0.2540 -0.0938 0.0239 -0.3134 Post-GE 0.1808 0.0898 0.0028 -0.0093 0.3637 -0.0111 0.3261 0.2371 -0.1405 0.0879 0.0995
Note: FIN: Finance, TRAD: Trade and Services, CONPR: Consumer Product, PLANT: Plantation.
138
Table 4.6: Selected Sample of Malaysian Government-Link Companies (GLCs)
No Company (Industry) Symbol Market Capitalization
(RM)*
Number of Share*
Earning Per Share
(Cent)*
Revenue (RM,000)#
Profit Before Tax
(RM,000)#
Net Profit (RM,000)#
1 Affin Holdings Berhad (Finance)
AHB 4.216b 1.943b 23.30 464,834 139,160 97,407
2 BIMB Holding Berhad (Finance)
BIMB 6.943b 1.589b 35.29 884,257 200,261 161,864
3 CIMB Group Holdings Berhad (Finance)
CIMB 44.165b 8.868b 37.39 4,0,41,563 1,132,161 825,739
4 Malayan Banking Berhad (Finance)
MAY 80.119b 10.193b 60.21 11,052,259 2,376,103 1,652,082
5 Malaysian Building Society Berhad (Finance)
MBS 5.277b 5.799b 2.51 825,687 1,312 -15,809
6 Boustead Holdings Berhad (Trading-Services)
BOUS 4.358b 2.027b 10.58 2,442,300 49,700 4,200
7 Telekom Malaysia Berhad (Trading-Services)
TMK 24.539b 3.758b 21.85 3,184,430 224,696 192,427
8 Tenaga Nasional Berhad (Trading-Services)**
TNB 81.268b 5.644b 130.55 11,744,000 1,412,500 820,900
9 UMW Holdings Berhad (Consumer Products)
UMWH 6.811b 1.168b -22.94 4,160,904 -334,250 -286,040
10 Chemical Company of Malaysia Berhad (Industrial Products)
CCM 414.16m 457.63m -10.12 160,105 19,313 -76,672
11 Malaysian Resources Corporation Berhad (Properties)
MRC 2.829b 2.080b 3.96 388,200 377 26,789
Note: *Calculated based on the net profit of the trailing twelve months and latest number of shares issued. #Calculated based on the amount declared for financial year ended quarter 4, 2015-12-31. **Calculated based on the amount declared for financial year ended 2015-08-31.
139
Table 4.7: Selected Sample of Malaysian Non-Government-Link Companies (Non-GLCs)
No Company (Industry) Symbol Market Capitalization
(RM)*
Number of Share*
Earning Per Share
(Cent)*
Revenue (RM,000)#
Profit Before Tax
(RM,000)#
Net Profit (RM,000)#
1 Public Bank Berhad (Finance)
PBK 77.643b 3.882b 134.37 4,929,046 1,857,776 1,492,428
2 RHB Bank Berhad (Finance)
RHBC 19.288b 4.010b 35.23 2,848,457 475,206 316,120
3 Genting Malaysia Berhad (Trading-Services)
GENM 28.503b 5.938b 21.94 2,291,879 357,846 338,558
4 Genting Berhad (Trading-Services)
GENT 29.175b 3.750b 30.02 4,919,421 726,691 338,946
5 MISC Berhad (Trading-Services)
MISC 33.568b 4.464b 70.65 3,312,062 763,133 752,720
6 YTL Corporation Berhad (Trading-Services)
YTL 17.443b 10.902b 8.51 4,115,753 520,117 298,928
7 NESTLE Malaysia Berhad (Consumer Products)
NESZ 18.380b 234.50m 285.70 1,198,942 118,677 99,789
8 PPB Group Berhad (Consumer Products)
PEP 18.826b 1.185b 67.76 1,090,600 378,748 341,021
9 British American Tobacco (M) (Consumer Products)
ROTH 13.882b 285.53m 220.32 1,057,992 272,557 196,121
10 IOI Corporation Berhad** (Plantation)
IOI 28.884b 6.462b 10.76 2,942,000 203,300 159,700
11 Kuala Lumpur Kepong Berhad*** (Plantation)
KLK 25.577b 1.068b 131.47 3,932,083 232,510 186,288
Note: *Calculated based on the net profit of the trailing twelve months and latest number of shares issued. #Calculated based on the amount declared for financial year ended quarter 4, 2015-12-31. **Calculated based on the amount declared for financial year ended 2015-06-30. ***Calculated based on the amount declared for financial year ended 2015-09-30.
140
Table 4.8(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) – Government Link Companies (GLCs)
Variables AHB BIMB CIMB MAY MBS
Sector Finance Finance Finance Finance Finance (p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1)
Mean Equation
0 -0.0149 (0.4830)
-0.0202 (0.4525)
0.0149 (0.4661)
0.0107 (0.4917)
0.0172 (0.6183)
PGE 0.1183 (0.6269)
0.0373 (0.8750)
0.1322 (0.7640)
0.0809 (0.6745)
-0.1407 (0.6923)
PtGE 0.0914 (0.5682)
0.0663 (0.5722)
-0.0992 (0.4905)
-0.1271 (0.3134)
-0.1407 (0.5507)
1tR 0.0158 (0.2480)
-0.1039 (0.0000)***
0.0535 (0.0000)***
0.0044 (0.7420)
-0.0593 (0.0001)***
Variance Equation
0 0.0459 (0.0000)***
0.4124 (0.0000)***
0.0196 (0.0000)***
0.0242 (0.0000)***
0.5563 (0.0000)***
1 0.1054 (0.0000)***
0.1569 (0.0000)***
0.0450 (0.0000)***
0.0584 (0.0000)***
0.1616 (0.0000)***
i 0.0221 (0.0005)***
0.0072 (0.3593)
0.0475 (0.0000)***
0.0296 (0.0000)***
0.0022 (0.8343)
1 0.8864 (0.0000)***
0.7757 (0.0000)***
0.9312 (0.0000)***
0.9188 (0.0000)***
0.7901 (0.0000)***
PGE 0.1115 (0.0640)*
0.0149 (0.8965)
0.4018 (0.0000)***
0.0767 (0.0042)***
0.0841 (0.3964)
PtGE 0.0196 (0.5686)
-0.2277 (0.0000)***
-0.1039 (0.0000)***
-0.0002 (0.9873)
0.8614 (0.0000)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.0368 0.9252 0.0034 0.0009 0.6350 10 lags 0.0804 0.9962 0.0079 0.0108 0.7070
Ljung-Box Q2 Statistic (p-value) 5 lags 0.0360 0.9240 0.0030 0.0010 0.6290
10 lags 0.0670 0.9960 0.0040 0.0100 0.6810 Return Equation: Wald Test (p-value)
F-stat 0.7572 0.8384 0.7290 0.5299 0.7806 Chi-Square 0.7572 0.8384 0.7290 0.5298 0.7806
Variance Equation: Wald Test (p-value) F-stat 0.1255 0.0001 0.0000 0.0104 0.0000
Chi-Square 0.1254 0.0001 0.0000 0.0104 0.0000 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (same average return / volatility for pre- and post-General Election).
141
Table 4.8(b): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) – Government Link Companies (GLCs)
Variables BOUS TMK TNB UMWH CCM MRC
Sector Trade & Services
Trade & Services
Trade & Services
Consumer Products
Industrial Products
Property
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0089 (0.6511)
0.0247 (0.1780)
0.0212 (0.2812)
0.0302 (0.0999)*
-0.0113 (0.5746)
-0.0223 (0.5193)
PGE 0.0544 (0.8005)
0.0667 (0.6445)
0.2706 (0.4103)
-0.1150 (0.5328)
0.0285 (0.8474)
0.0452 (0.9191)
PtGE -0.0844 (0.5453)
-0.0925 (0.2368)
-0.3141 (0.0973)*
0.1064 (0.4128)
0.2164 (0.0808)*
-0.5235 (0.0202)**
1tR 0.0192 (0.1533)
0.0591 (0.0000)***
0.0085 (0.5112)
-0.0051 (0.6997)
-0.0896 (0.0000)***
-0.0259 (0.0494)**
Variance Equation
0 0.0844 (0.0000)***
0.0448 (0.0000)***
0.0348 (0.0000)***
0.0224 (0.0000)***
0.1793 (0.0000)***
0.1515 (0.0000)***
1 0.0957 (0.0000)***
0.1224 (0.0000)***
0.0595 (0.0000)***
0.0616 (0.0000)***
0.1411 (0.0000)***
0.0603 (0.0000)***
i 0.0060 (0.3109)
-0.0143 (0.0917)*
0.0399 (0.0000)***
0.0275 (0.0000)***
0.0376 (0.0000)***
0.0510 (0.0000)***
1 0.8812 (0.0000)***
0.8860 (0.0000)***
0.9155 (0.0000)***
0.9223 (0.0000)***
0.8036 (0.0000)***
0.9062 (0.0000)***
PGE 0.0643 (0.0725)*
-0.0288 (0.0342)**
0.3639 (0.0000)***
0.0788 (0.0160)**
-0.0762 (0.1965)
1.0793 (0.0000)***
PtGE 0.1121 (0.0000)***
-0.0267 (0.0194)**
0.0271 (0.3242)
0.0252 (0.1462)
0.0096 (0.8351)
-0.1084 (0.0420)**
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.3619 0.8070 0.1535 0.0088 0.7910 0.0110 10 lags 0.5018 0.8701 0.2008 0.0195 0.7298 0.0768
Ljung-Box Q2 Statistic (p-value) 5 lags 0.3650 0.8030 0.1560 0.0070 0.7930 0.0110
10 lags 0.4930 0.8660 0.1740 0.0150 0.7270 0.0690 Return Equation: Wald Test (p-value)
F-stat 0.8034 0.4180 0.2156 0.5940 0.2137 0.0523 Chi-Square 0.8034 0.4180 0.2155 0.5940 0.2136 0.0522
Variance Equation: Wald Test (p-value) F-stat 0.0000 0.0019 0.0000 0.0029 0.4320 0.0000
Chi-Square 0.0000 0.0019 0.0000 0.0029 0.4319 0.0000 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (same average return / volatility for pre- and post-General Election).
142
Table 4.9(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) – Non-Government Link Companies (Non-GLCs)
Variables PBK RHBC GENM GENT MISC
Sector Finance Finance Trade and Services
Trade and Services
Trade and Services
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0424 (0.0009)***
0.0264 (0.2622)
-0.0015 (0.9537)
0.0082 (0.7276)
0.0166 (0.3882)
PGE -0.0782 (0.4959)
-0.1258 (0.5892)
-0.1168 (0.6947)
-0.1560 (0.5337)
-0.3930 (0.2648)
PtGE 0.0341 (0.6294)
-0.0670 (0.5910)
-0.0664 (0.7006)
-0.2341 (0.1581)
0.0404 (0.8064)
1tR 0.0463 (0.0005)***
0.0112 (0.3528)
-0.0249 (0.0514)*
0.0032 (0.8142)
-0.0984 (0.0000)***
Variance Equation
0 0.0077 (0.0000)***
0.0176 (0.0000)***
0.0423 (0.0000)***
0.0767 (0.0000)***
0.0334 (0.0000)***
1 0.0711 (0.0000)***
0.0256 (0.0000)***
0.0505 (0.0000)***
0.0733 (0.0000)***
0.0405 (0.0000)***
i 0.0263 (0.0000)***
0.0264 (0.0000)***
0.0228 (0.0000)***
0.0177 (0.0073)***
0.0225 (0.0000)***
1 0.9222 (0.0000)***
0.9591 (0.0000)***
0.9323 (0.0000)***
0.9016 (0.0000)***
0.9378 (0.0000)***
PGE -0.0029 (0.4954)
-0.0258 (0.0171)**
0.1289 (0.0158)**
0.1361 (0.0458)**
0.3335 (0.0000)***
PtGE -0.0051 (0.3323)
-0.0179 (0.0243)**
-0.0092 (0.8078)
0.0271 (0.4602)
-0.0439 (0.0002)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.0320 0.0000 0.1918 0.2964 0.1537 10 lags 0.1592 0.0008 0.1028 0.5296 0.4830
Ljung-Box Q2 Statistic (p-value) 5 lags 0.0270 0.0000 0.1880 0.2920 0.1410
10 lags 0.1170 0.0010 0.1190 0.5120 0.4460 Return Equation: Wald Test (p-value)
F-stat 0.7122 0.7611 0.8735 0.2934 0.5307 Chi-Square 0.7122 0.7611 0.8735 0.2934 0.5306
Variance Equation: Wald Test (p-value) F-stat 0.4005 0.0009 0.0481 0.0681 0.0000
Chi-Square 0.4004 0.0009 0.0480 0.0680 0.0000 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (same average return / volatility for pre- and post-General Election).
143
Table 4.9(b): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2015) – Non-Government Link Companies (Non-GLCs)
Variables YTL NESZ PEP ROTH IOI KLK
Sector Trade and Services
Consumer Products
Consumer Products
Consumer Products
Plantation Plantation
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0050 (0.8126)
0.0281 (0.0567)*
0.0385 (0.0401)**
0.0213 (0.2129)
0.0344 (0.1165)
0.0549 (0.0144)**
PGE 0.0826 (0.7633)
0.1970 (0.2423)
-0.0847 (0.5738)
0.0531 (0.7787)
0.0260 (0.9254)
-0.0136 (0.9516)
PtGE -0.1602 (0.1528)
0.2545 (0.0027)***
0.0274 (0.8178)
-0.0579 (0.5384)
0.0454 (0.7859)
-0.0306 (0.7728)
1tR -0.0571 (0.0000)***
-0.1880 (0.0000)***
-0.0315 (0.0282)**
-0.1197 (0.0000)***
-0.0037 (0.7722)
-0.0277 (0.0454)**
Variance Equation
0 0.0482 (0.0000)***
0.0126 (0.0000)***
0.0661 (0.0000)***
0.0400 (0.0000)***
0.0444 (0.0000)***
0.0816 (0.0000)***
1 0.0545 (0.0000)***
0.0940 (0.0000)***
0.0684 (0.0000)***
0.0525 (0.0000)***
0.0602 (0.0000)***
0.0758 (0.0000)***
i 0.0431 (0.0000)***
-0.0642 (0.0000)***
0.0450 (0.0000)***
0.0124 (0.0127)**
0.0462 (0.0000)***
0.0071 (0.2191)
1 0.9164 (0.0000)***
0.9433 (0.0000)***
0.8892 (0.0000)***
0.9240 (0.0000)***
0.9109 (0.0000)***
0.8998 (0.0000)***
PGE 0.1270 (0.0020)***
0.0111 (0.2705)
-0.0308 (0.1187)
0.0309 (0.2744)
0.2412 (0.0295)**
0.0357 (0.0570)*
PtGE -0.0301 (0.0273)**
-0.0159 (0.0926)*
0.0260 (0.0893)*
-0.0289 (0.0004)***
-0.0215 (0.6584)
-0.0313 (0.0483)**
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.5202 0.9095 0.1578 0.0029 0.0743 0.1952 10 lags 0.6257 0.9977 0.2779 0.0323 0.2249 0.5975
Ljung-Box Q2 Statistic (p-value) 5 lags 0.5100 0.9090 0.1510 0.0030 0.0690 0.1930
10 lags 0.6120 0.9980 0.2720 0.0320 0.1770 0.5720 Return Equation: Wald Test (p-value)
F-stat 0.3316 0.0061 0.8382 0.7880 0.9596 0.9580 Chi-Square 0.3315 0.0061 0.8382 0.7879 0.9596 0.9580
Variance Equation: Wald Test (p-value) F-stat 0.0023 0.2025 0.0683 0.0020 0.0896 0.0317
Chi-Square 0.0023 0.2024 0.0682 0.0020 0.0895 0.0317 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (same average return / volatility for pre- and post-General Election).
144
Table 4.10(a): Threshold GARCH Results for Pre-General Election and Post-General
Election (1994 - 2005) – Government Link Companies (GLCs)
Variables AHB BIMB CIMB MAY MBS Sector Finance Finance Finance Finance Finance (p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1)
Mean Equation
0 -0.0341 (0.4168)
-0.0849 (0.0245)**
0.0152 (0.6806)
0.0229 (0.4298)
-0.0265 (0.6264)
PGE -0.2566 (0.5354)
0.0556 (0.8371)
0.2765 (0.5547)
0.1404 (0.6213)
0.2051 (0.6291)
PtGE 0.0614 (0.7724)
0.0691 (0.5973)
-0.2131 (0.4149)
-0.0867 (0.6060)
0.0601 (0.8472)
1tR 0.0336 (0.0680)*
-0.1230 (0.0000)***
0.0667 (0.0001)***
-0.0060 (0.7349)
-0.0739 (0.0015)***
Variance Equation
0 0.1484 (0.0000)***
0.3736 (0.0000)***
0.0603 (0.0000)***
0.0582 (0.0000)***
1.7300 (0.0000)***
1 0.0862 (0.0000)***
0.1073 (0.0000)***
0.0473 (0.0000)***
0.0497 (0.0000)***
0.2932 (0.0000)***
i 0.0409 (0.0000)***
0.0730 (0.0000)***
0.0523 (0.0000)***
0.0465 (0.0000)***
-0.0730 (0.0011)***
1 0.8795 (0.0000)***
0.8052 (0.0000)***
0.9213 (0.0000)***
0.9139 (0.0000)***
0.6204 (0.0000)***
PGE 0.0046 (0.9767)
-0.2311 (0.0092)***
-0.0011 (0.9911)
-0.0257 (0.7234)
-1.3497 (0.0000)***
PtGE 0.0061 (0.9528)
-0.2689 (0.0000)***
0.1212 (0.0579)*
-0.0295 (0.4275)
-0.1361 (0.4613)
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.0079 0.9996 0.0284 0.0041 0.9625 10 lags 0.0120 1.0000 0.0425 0.0332 0.9935
Ljung-Box Q2 Statistic (p-value) 5 lags 0.0070 1.0000 0.0250 0.0050 0.9610
10 lags 0.0110 1.0000 0.0360 0.0310 0.9930 Return Equation: Wald Test (p-value)
F-stat 0.7905 0.8467 0.6083 0.7736 0.8752 Chi-Square 0.7905 0.8467 0.6083 0.7735 0.8752
Variance Equation: Wald Test (p-value) F-stat 0.9968 0.0000 0.0881 0.5171 0.0000
Chi-Square 0.9968 0.0000 0.0879 0.5170 0.0000 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (same average return / volatility for pre- and post-General Election).
145
Table 4.10(b): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2005) – Government Link Companies (GLCs)
Variables BOUS TMK TNB UMWH CCM MRC
Sector Trade & Services
Trade & Services
Trade & Services
Consumer Products
Industrial Products
Property
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 -0.0128 (0.6891)
-0.0406 (0.1867)
-0.0144 (0.6546)
0.0121 (0.7153)
0.0034 (0.9065)
-0.0895 (0.0999)*
PGE 0.2402 (0.3748)
0.1684 (0.5929)
0.4230 (0.2477)
-0.2255 (0.2980)
0.1506 (0.3903)
0.1461 (0.7442)
PtGE -0.1136 (0.5128)
0.0199 (0.9208)
-0.3757 (0.2116)
0.0824 (0.6677)
0.1798 (0.1974)
-0.4796 (0.0872)*
1tR 0.0012 (0.9505)
0.0428 (0.0104)**
-0.0097 (0.5782)
0.0134 (0.4561)
-0.0317 (0.1177)
-0.0256 (0.1575)
Variance Equation
0 0.1425 (0.0000)***
0.0460 (0.0000)***
0.0778 (0.0000)***
0.0550 (0.0000)***
0.1977 (0.0000)***
0.2952 (0.0000)***
1 0.0646 (0.0000)***
0.0431 (0.0000)***
0.0524 (0.0000)***
0.0474 (0.0000)***
0.1408 (0.0000)***
0.0621 (0.0000)***
i 0.0247 (0.0006)***
0.0440 (0.0000)***
0.0637 (0.0000)***
0.0637 (0.0000)***
0.0519 (0.0000)***
0.0708 (0.0000)***
1 0.8891 (0.0000)***
0.9258 (0.0000)***
0.9018 (0.0000)***
0.9144 (0.0000)***
0.8027 (0.0000)***
0.8902 (0.0000)***
PGE -0.1576 (0.0073)***
-0.0837 (0.4251)
0.0225 (0.8506)
-0.1361 (0.0100)**
-0.2677 (0.0000)***
-0.4658 (0.0556)*
PtGE 0.1007 (0.0001)***
0.0942 (0.1145)
0.2205 (0.0004)***
0.1818 (0.0006)***
-0.0211 (0.6488)
-0.0349 (0.6900)
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.8799 0.0000 0.0047 0.0442 0.6829 0.0021 10 lags 0.9688 0.0006 0.0098 0.1035 0.5070 0.0162
Ljung-Box Q2 Statistic (p-value) 5 lags 0.8840 0.0000 0.0040 0.0390 0.6710 0.0020
10 lags 0.9660 0.0000 0.0040 0.0950 0.5070 0.0140 Return Equation: Wald Test (p-value)
F-stat 0.5850 0.8640 0.2080 0.5288 0.2834 0.2300 Chi-Square 0.5849 0.8640 0.2078 0.5288 0.2833 0.2298
Variance Equation: Wald Test (p-value) F-stat 0.0001 0.2780 0.0000 0.0011 0.0000 0.0157
Chi-Square 0.0001 0.2779 0.0000 0.0011 0.0000 0.0156 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (same average return / volatility for pre- and post-General Election).
146
Table 4.11(a): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2005) – Non-Government Link Companies (Non-GLCs)
Variables PBK RHBC GENM GENT MISC
Sector Finance Finance Trade and Services
Trade and Services
Trade and Services
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0246 (0.3898)
-0.0161 (0.7033)
-0.0188 (0.6455)
0.0168 (0.6481)
0.0352 (0.2723)
PGE -0.0838 (0.6761)
0.0009 (0.9988)
-0.0707 (0.8630)
0.0238 (0.9424)
-0.1619 (0.1473)
PtGE -0.0608 (0.7152)
-0.3541 (0.1352)
-0.1058 (0.6737)
-0.3971 (0.1005)
-0.1204 (0.4091)
1tR 0.0283 (0.1063)
0.0328 (0.0357)**
0.0310 (0.0714)*
0.0052 (0.7827)
-0.1800 (0.0000)***
Variance Equation
0 0.0583 (0.0000)***
0.0188 (0.0000)***
0.0402 (0.0000)***
0.1591 (0.0000)***
0.1923 (0.0000)***
1 0.0478 (0.0000)***
0.0147 (0.0000)***
0.0404 (0.0000)***
0.0947 (0.0000)***
0.0805 (0.0000)***
i 0.0581 (0.0000)***
0.0328 (0.0000)***
0.0329 (0.0000)***
0.0006 (0.9530)
0.0366 (0.0040)***
1 0.9109 (0.0000)***
0.9682 (0.0000)***
0.9406 (0.0000)***
0.8760 (0.0000)***
0.8507 (0.0000)***
PGE -0.1120 (0.0011)***
0.0421 (0.7397)
-0.0367 (0.7716)
-0.0370 (0.8109)
-0.2352 (0.0000)***
PtGE 0.1296 (0.0000)***
-0.0227 (0.6512)
0.0814 (0.2723)
0.1050 (0.2583)
0.1116 (0.0472)**
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.0323 0.0001 0.6665 0.8752 0.9954 10 lags 0.1723 0.0029 0.5142 0.5297 1.0000
Ljung-Box Q2 Statistic (p-value) 5 lags 0.0260 0.0000 0.6710 0.8740 0.9950
10 lags 0.1220 0.0020 0.5580 0.5150 1.0000 Return Equation: Wald Test (p-value)
F-stat 0.8630 0.3253 0.9047 0.2584 0.2792 Chi-Square 0.8630 0.3251 0.9047 0.2583 0.2790
Variance Equation: Wald Test (p-value) F-stat 0.0000 0.9025 0.4864 0.5191 0.0000
Chi-Square 0.0000 0.9025 0.4863 0.5191 0.0000 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (same average return / volatility for pre- and post-General Election).
147
Table 4.11(b): Threshold GARCH Results for Pre-General Election and Post-General Election (1994 - 2005) – Non-Government Link Companies (Non-GLCs)
Variables YTL NESZ PEP ROTH IOI KLK
Sector Trade and Services
Consumer Products
Consumer Products
Consumer Products
Plantation Plantation
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0211 (0.5526)
0.0339 (0.2346)
0.0432 (0.1122)
0.0258 (0.2943)
0.0665 (0.0778)*
0.0318 (0.2901)
PGE 0.1475 (0.6249)
0.0979 (0.5496)
-0.0599 (0.7322)
0.2192 (0.4754)
-0.2694 (0.4809)
0.1547 (0.5052)
PtGE -0.2498 (0.1032)
0.1266 (0.3959)
-0.1123 (0.4473)
0.0138 (0.9025)
0.0631 (0.7999)
-0.0930 (0.5662)
1tR -0.0193 (0.2799)
-0.1624 (0.0000)***
-0.0519 (0.0061)***
-0.1082 (0.0000)***
0.0039 (0.8266)
-0.0536 (0.0055)***
Variance Equation
0 0.0544 (0.0000)***
0.7306 (0.0000)***
0.0512 (0.0000)***
0.0263 (0.0000)***
0.0952 (0.0000)***
0.0960 (0.0000)***
1 0.0414 (0.0000)***
0.1313 (0.0000)***
0.0445 (0.0000)***
0.0441 (0.0000)***
0.0510 (0.0000)***
0.0993 (0.0000)***
i 0.0578 (0.0000)***
0.3652 (0.0000)***
0.0617 (0.0000)***
0.0132 (0.0198)**
0.0548 (0.0000)***
0.0073 (0.4303)
1 0.9258 (0.0000)***
0.4913 (0.0000)***
0.9121 (0.0000)***
0.9404 (0.0000)***
0.9079 (0.0000)***
0.8767 (0.0000)***
PGE -0.1136 (0.0848)*
-0.4460 (0.0000)***
-0.0752 (0.0006)***
0.0609 (0.1418)
0.1079 (0.4887)
-0.0840 (0.0727)*
PtGE 0.0482 (0.1507)
-0.0150 (0.9220)
0.0575 (0.0028)***
-0.0417 (0.0000)***
0.0877 (0.2966)
0.0688 (0.1030)
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.8860 0.9983 0.0669 0.0125 0.3523 0.7499 10 lags 0.9097 1.0000 0.2974 0.0852 0.7173 0.8518
Ljung-Box Q2 Statistic (p-value) 5 lags 0.8840 0.9980 0.0620 0.0150 0.3380 0.7470
10 lags 0.9110 1.0000 0.2770 0.0970 0.6500 0.7960 Return Equation: Wald Test (p-value)
F-stat 0.2393 0.5932 0.7178 0.7704 0.7576 0.6814 Chi-Square 0.2391 0.5932 0.7178 0.7703 0.7576 0.6813
Variance Equation: Wald Test (p-value) F-stat 0.1458 0.0000 0.0001 0.0000 0.3189 0.0911
Chi-Square 0.1456 0.0000 0.0001 0.0000 0.3187 0.0909 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (same average return / volatility for pre- and post-General Election).
148
Table 4.12(a): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) – Government Link Companies (GLCs)
Variables AHB BIMB CIMB MAY MBS
Sector Finance Finance Finance Finance Finance (p, q) (1, 2) (1, 1) (1, 1) (1, 1) (1, 1)
Mean Equation
0 -0.0044 (0.8562)
0.0509 (0.1999)
0.0165 (0.5140)
0.0049 (0.7773)
0.0517 (0.2247)
PGE 0.2480 (0.3363)
-0.0239 (0.9542)
0.0526 (0.9379)
0.0339 (0.8954)
-0.3488 (0.5589)
PtGE 0.1746 (0.5322)
0.1641 (0.5098)
-0.1832 (0.3725)
-0.1300 (0.4321)
-0.2758 (0.5419)
1tR -0.0139 (0.5141)
-0.0871 (0.0001)***
0.0350 (0.0764)*
0.0177 (0.4065)
-0.0271 (0.2198)
Variance Equation
0 0.0740 (0.0000)***
0.5095 (0.0000)***
0.0219 (0.0000)***
0.0302 (0.0000)***
0.3394 (0.0000)***
1 0.1608 (0.0000)***
0.2402 (0.0000)***
0.0407 (0.0000)***
0.0840 (0.0000)***
0.1422 (0.0000)***
i -0.0187 (0.1216)
-0.1314 (0.0000)***
0.0501 (0.0000)***
0.0186 (0.0584)*
-0.0006 (0.9716)
1 0.8532
(0.0000)*** 0.7231
(0.0000)*** 0.9269
(0.0000)*** 0.8847
(0.0000)*** 0.8142
(0.0000)***
2 -0.0154 (0.9009)
-- --
-- --
-- --
-- --
PGE 0.1130 (0.1944)
0.4773 (0.1736)
0.5106 (0.0000)***
0.1434 (0.0014)***
0.6348 (0.0001)***
PtGE 0.1195 (0.0446)**
0.0957 (0.5813)
-0.1131 (0.0000)***
0.0480 (0.0602)*
1.4029 (0.0009)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.5198 0.4934 0.5573 0.2690 0.8068 10 lags 0.8541 0.7372 0.7500 0.4778 0.7912
Ljung-Box Q2 Statistic (p-value) 5 lags 0.5130 0.4870 0.5580 0.2520 0.8110
10 lags 0.8500 0.7270 0.6990 0.4510 0.7880 Return Equation: Wald Test (p-value)
F-stat 0.4919 0.8048 0.6559 0.7165 0.6917 Chi-Square 0.4918 0.8048 0.6559 0.7165 0.6917
Variance Equation: Wald Test (p-value) F-stat 0.4092 0.3129 0.0000 0.0003 0.0000
Chi-Square 0.4091 0.3127 0.0000 0.0003 0.0000 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (same average return / volatility for pre- and post-General Election).
149
Table 4.12(b): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) – Government Link Companies (GLCs)
Variables BOUS TMK TNB UMWH CCM MRC
Sector Trade & Services
Trade & Services
Trade & Services
Consumer Products
Industrial Products
Property
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (2, 1) (1, 1) Mean Equation
0 0.0341 (0.1625)
0.0780 (0.0005)***
0.0447 (0.0727)*
0.0419 (0.0596)*
-0.0217 (0.4414)
0.0348 (0.4748)
PGE -0.4833 (0.2577)
-0.0933 (0.5442)
0.2852 (0.7308)
-0.0655 (0.8064)
-0.0578 (0.8484)
-0.6034 (0.6305)
PtGE -0.0119 (0.9709)
-0.1552 (0.0854)*
-0.2090 (0.2223)
0.0759 (0.7097)
0.1224 (0.6404)
-0.6795 (0.1527)
1tR 0.0342 (0.1127)
0.0657 (0.0004)***
0.0389 (0.0994)*
-0.0292 (0.1406)
-0.1496 (0.0000)***
-0.0302 (0.1478)
Variance Equation
0 0.1208 (0.0000)***
0.0559 (0.0000)***
0.0856 (0.0000)***
0.0256 (0.0000)***
0.2089 (0.0000)***
0.1276 (0.0000)***
1 0.1995 (0.0000)***
0.2925 (0.0000)***
0.1517 (0.0000)***
0.0768 (0.0000)***
0.0860 (0.0000)***
0.0689 (0.0000)***
2 -- --
-- --
-- --
-- --
0.0683 (0.0000)***
-- --
i -0.0518 (0.0045)***
-0.2329 (0.0000)***
0.0072 (0.5719)
-0.0054 (0.4702)
0.0090 (0.4532)
0.0123 (0.1385)
1 0.7848 (0.0000)***
0.8450 (0.0000)***
0.8238 (0.0000)***
0.9035 (0.0000)***
0.7726 (0.0000)***
0.9082 (0.0000)***
PGE 0.8275 (0.0000)***
-0.0351 (0.2080)
0.9302 (0.0013)***
0.1476 (0.0016)***
0.3486 (0.1012)
3.7682 (0.0000)***
PtGE 0.1871 (0.0052)***
-0.0439 (0.0255)**
-0.0234 (0.4004)
-0.0003 (0.9890)
0.3268 (0.1082)
-0.3252 (0.0206)**
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.1833 0.9841 0.8293 0.1113 0.5415 0.9398 10 lags 0.3492 0.9973 0.8289 0.2836 0.8875 0.9944
Ljung-Box Q2 Statistic (p-value) 5 lags 0.1770 0.9840 0.8290 0.1140 0.5590 0.9440
10 lags 0.3650 0.9970 0.8860 0.2720 0.8900 0.9950 Return Equation: Wald Test (p-value)
F-stat 0.5267 0.1729 0.4349 0.9033 0.8796 0.3304 Chi-Square 0.5266 0.1727 0.4348 0.9033 0.8796 0.3302
Variance Equation: Wald Test (p-value) F-stat 0.0000 0.0235 0.0042 0.0059 0.0000 0.0000
Chi-Square 0.0000 0.0234 0.0042 0.0058 0.0000 0.0000 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (same average return / volatility for pre- and post-General Election).
150
Table 4.13(a): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) – Non-Government Link Companies (Non-GLCs)
Variables PBK RHBC GENM GENT MISC
Sector Finance Finance Trade and Services
Trade and Services
Trade and Services
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 0.0436 (0.0017)***
0.0530 (0.0616)*
0.0151 (0.6412)
-0.0084 (0.7769)
-0.0084 (0.7405)
PGE -0.0422 (0.5721)
-0.1697 (0.4651)
-0.1322 (0.7536)
-0.2152 (0.5149)
-0.5580 (0.3205)
PtGE 0.0797 (0.3192)
-0.0132 (0.9217)
-0.0108 (0.9607)
-0.1194 (0.5693)
0.3456 (0.1194)
1tR 0.0446 (0.0781)*
-0.0109 (0.5992)
-0.0944 (0.0000)***
0.0001 (0.9959)
-0.0169 (0.4269)
Variance Equation
0 0.0398 (0.0000)***
0.0855 (0.0000)***
0.1148 (0.0000)***
0.0364 (0.0000)***
0.0447 (0.0000)***
1 0.2739 (0.0000)***
0.0680 (0.0000)***
0.0811 (0.0000)***
0.0515 (0.0000)***
0.0397 (0.0000)***
i -0.0157 (0.4211)
0.0367 (0.0007)***
-0.0012 (0.9025)
0.0432 (0.0000)***
0.0447 (0.0000)***
1 0.7302
(0.0000)*** 0.8804
(0.0000)*** 0.8853
(0.0000)*** 0.9191
(0.0000)*** 0.9166
(0.0000)***
PGE -0.0300 (0.0052)***
-0.0346 (0.1546)
0.3049 (0.0015)***
0.1584 (0.0305)**
0.8937 (0.0000)***
PtGE -0.0200 (0.1479)
-0.0613 (0.0000)***
-0.0753 (0.3091)
-0.0219 (0.4915)
-0.1585 (0.0002)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.4049 0.4069 0.5937 0.1550 0.7116 10 lags 0.5864 0.5941 0.6331 0.4455 0.8043
Ljung-Box Q2 Statistic (p-value) 5 lags 0.3960 0.4180 0.5710 0.1550 0.7080
10 lags 0.5480 0.5900 0.6320 0.4220 0.7820 Return Equation: Wald Test (p-value)
F-stat 0.4949 0.7657 0.9518 0.6689 0.2133 Chi-Square 0.4948 0.7657 0.9518 0.6688 0.2131
Variance Equation: Wald Test (p-value) F-stat 0.0015 0.0001 0.0051 0.0929 0.0000
Chi-Square 0.0015 0.0001 0.0050 0.0927 0.0000 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (same average return / volatility for pre- and post-General Election).
151
Table 4.13(b): Threshold GARCH Results for Pre-General Election and Post-General Election (2006 - 2015) – Non-Government Link Companies (Non-GLCs)
Variables YTL NESZ PEP ROTH IOI KLK
Sector Trade and Services
Consumer Products
Consumer Products
Consumer Products
Plantation Plantation
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) (2, 1) Mean Equation
0 0.0073 (0.7841)
0.0498 (0.0210)**
0.0395 (0.1309)
0.0216 (0.3683)
0.0199 (0.4574)
0.0814 (0.0069)***
PGE -0.1256 (0.7953)
0.0545 (0.6461)
-0.1281 (0.6035)
-0.1583 (0.5625)
0.3082 (0.4291)
-0.0745 (0.8096)
PtGE -0.1156 (0.4152)
0.3105 (0.0340)**
0.2055 (0.1887)
-0.1545 (0.2840)
0.0261 (0.9144)
-0.0709 (0.6584)
1tR -0.0994 (0.0000)***
-0.1339 (0.0000)***
-0.0151 (0.4904)
-0.1348 (0.0000)***
-0.0114 (0.5676)
-0.0307 (0.2627)
Variance Equation
0 0.1299 (0.0000)***
0.6739 (0.0000)***
0.1278 (0.0000)***
0.1570 (0.0000)***
0.0580 (0.0000)***
0.0197 (0.0000)***
1 0.0873 (0.0000)***
0.0919 (0.0000)***
0.1210 (0.0000)***
0.0826 (0.0000)***
0.0773 (0.0000)***
0.2312 (0.0000)***
2 -- --
-- --
-- --
-- --
-- --
-0.1951 (0.0000)***
i 0.0163 (0.1845)
-0.0328 (0.2226)
0.0070 (0.6560)
0.0206 (0.1988)
0.0466 (0.0000)***
-0.0056 (0.1185)
1 0.8466 (0.0000)***
0.1492 (0.0792)*
0.8213 (0.0000)***
0.8089 (0.0000)***
0.8832 (0.0000)***
0.9622 (0.0000)***
PGE 0.5143 (0.0000)***
-0.2748 (0.0000)***
0.0114 (0.8617)
0.1220 (0.1499)
0.3539 (0.0883)*
0.0531 (0.0186)**
PtGE -0.1085 (0.0043)***
0.4146 (0.0003)***
-0.0529 (0.3173)
-0.0345 (0.2672)
-0.0464 (0.6217)
-0.0414 (0.0000)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.3428 0.9993 0.9980 0.6505 0.3333 0.9989 10 lags 0.4585 1.0000 0.7519 0.8337 0.3233 1.0000
Ljung-Box Q2 Statistic (p-value) 5 lags 0.3360 0.9990 0.9980 0.6330 0.3470 0.9990
10 lags 0.4480 1.0000 0.7450 0.8280 0.3120 1.0000 Return Equation: Wald Test (p-value)
F-stat 0.6958 0.0978 0.3918 0.4783 0.7281 0.8888 Chi-Square 0.6958 0.0976 0.3917 0.4782 0.7280 0.8888
Variance Equation: Wald Test (p-value) F-stat 0.0000 0.0000 0.6066 0.2124 0.2326 0.0000
Chi-Square 0.0000 0.0000 0.6065 0.2122 0.2324 0.0000 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value. The null hypothesis of the Wald Test is 0: 210 H (same average return / volatility for pre- and post-General Election).
152
Table 4.14(a): Threshold GARCH Results for Trading Volume during Pre- and Post-General Election (1994 - 2005) – Government Link Companies (GLCs)
Variables AHB BIMB CIMB MAY MBS
Sector Finance Finance Finance Finance Finance (p, q) (2, 1) (1, 1) (1, 1) (1, 1) (1, 1)
Mean Equation
0 2.9114 (0.0000)***
3.2686 (0.0000)***
4.8322 (0.0000)***
6.1296 (0.0000)***
3.1741 (0.0000)***
PGE -0.0310 (0.7729)
0.0511 (0.7613)
0.0412 (0.6994)
-0.1373 (0.1315)
-0.1533 (0.3316)
PtGE 0.0202 (0.8111)
-0.0270 (0.8007)
0.0899 (0.2676)
0.0089 (0.9012)
0.0743 (0.4580)
1tVol 0.7718 (0.0000)***
0.6793 (0.0000)***
0.6860 (0.0000)***
0.5933 (0.0000)***
0.7038 (0.0000)***
Variance Equation
0 0.0328 (0.0000)***
0.0315 (0.0001)***
0.0346 (0.0000)***
0.1112 (0.0000)***
0.0748 (0.0000)***
1 0.0862 (0.0000)***
0.0115 (0.0365)**
-0.0238 (0.0024)***
0.0421 (0.0195)**
0.0354 (0.0000)***
2 -0.0876
(0.0000)*** -- --
-- --
-- --
-- --
i 0.0745 (0.0000)***
0.0270 (0.0009)***
0.0776 (0.0000)***
0.0753 (0.0011)***
0.0505 (0.0001)***
1 0.9135 (0.0000)***
0.9497 (0.0000)***
0.9220 (0.0000)***
0.6180 (0.0000)***
0.8719 (0.0000)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.6724 0.0052 0.0330 0.4982 0.0000 10 lags 0.8910 0.0540 0.0893 0.3662 0.0000
Ljung-Box Q2 Statistic (p-value) 5 lags 0.6720 0.0050 0.0300 0.5210 0.0000
10 lags 0.8930 0.0510 0.0840 0.3560 0.0000 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value.
153
Table 4.14(b): Threshold GARCH Results for Trading Volume during Pre- and Post-
General Election (1994 - 2005) – Government Link Companies (GLCs)
Variables BOUS TMK TNB UMWH CCM MRC
Sector Trade & Services
Trade & Services
Trade & Services
Consumer Products
Industrial Products
Property
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 3.9284 (0.0000)***
5.4750 (0.0000)***
4.8288 (0.0000)***
6.1811 (0.0000)***
2.8946 (0.0000)***
3.0601 (0.0000)***
PGE 0.2900 (0.0348)**
0.0789 (0.5084)
-0.0636 (0.5591)
-0.1464 (0.3665)
-0.1518 (0.4487)
0.0539 (0.6230)
PtGE 0.3339 (0.0021)***
0.1667 (0.0248)**
-0.0275 (0.7688)
0.0331 (0.7845)
0.2124 (0.0928)*
-0.0765 (0.1902)
1tVol 0.6540 (0.0000)***
0.6132 (0.0000)***
0.6701 (0.0000)***
0.5387 (0.0000)***
0.7304 (0.0000)***
0.7854 (0.0000)***
Variance Equation
0 0.0278 (0.0000)***
0.1197 (0.0001)***
0.0044 (0.0008)***
0.0845 (0.0000)***
0.8075 (0.0000)***
0.0234 (0.0000)***
1 0.0123 (0.0543)*
0.0333 (0.0154)**
0.0162 (0.0000)***
-0.0134 (0.1974)
0.0876 (0.0005)***
0.0163 (0.0331)**
i 0.0520 (0.0000)***
0.0789 (0.0026)***
0.0096 (0.0129)**
0.1067 (0.0000)***
0.1208 (0.0066)***
0.0543 (0.0000)***
1 0.9357 (0.0000)***
0.6528 (0.0000)***
0.9695 (0.0000)***
0.8869 (0.0000)***
0.2279 (0.0116)**
0.8993 (0.0000)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.0761 0.3555 0.0006 0.1133 0.7451 0.6811 10 lags 0.1978 0.7890 0.0045 0.4198 0.8350 0.6927
Ljung-Box Q2 Statistic (p-value) 5 lags 0.0750 0.3580 0.0000 0.1000 0.7450 0.6810
10 lags 0.1970 0.7720 0.0040 0.3820 0.8190 0.6960 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value.
154
Table 4.15(a): Threshold GARCH Results for Trading Volume during Pre- and Post-
General Election (1994 - 2005) – Non-Government Link Companies (Non-GLCs)
Variables PBK RHBC GENM GENT MISC
Sector Finance Finance Trade and Services
Trade and Services
Trade and Services
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 3.6402 (0.0000)***
3.7746 (0.0000)***
6.2348 (0.0000)***
6.8774 (0.0000)***
4.4942 (0.0000)***
PGE -0.0676 (0.6098)
0.0925 (0.4105)
0.0287 (0.8200)
-0.0794 (0.4593)
0.1876 (0.4132)
PtGE 0.0059 (0.9285)
0.0307 (0.6421)
0.1167 (0.0969)*
0.0878 (0.2466)
0.2721 (0.0405)**
1tVol 0.7391 (0.0000)***
0.7318 (0.0000)***
0.5970 (0.0000)***
0.5384 (0.0000)***
0.6292 (0.0000)***
Variance Equation
0 0.0191 (0.0012)***
0.0640 (0.0000)***
0.1229 (0.0000)***
0.1441 (0.0032)***
0.0361 (0.0000)***
1 0.0029 (0.6698)
0.0009 (0.9331)
0.0103 (0.1311)
0.0274 (0.1197)
0.0251 (0.0000)***
i 0.0355 (0.0010)***
0.1054 (0.0000)***
0.0909 (0.0000)***
0.0353 (0.1466)
0.0317 (0.0001)***
1 0.9315 (0.0000)***
0.7989 (0.0000)***
0.6651 (0.0000)***
0.6374 (0.0000)***
0.9358 (0.0000)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.0129 0.3903 0.0906 0.2709 0.0325 10 lags 0.1007 0.5424 0.0099 0.3554 0.0518
Ljung-Box Q2 Statistic (p-value) 5 lags 0.0150 0.3930 0.0880 0.2930 0.0320
10 lags 0.1150 0.5450 0.0180 0.3290 0.0610 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value.
155
Table 4.15(b): Threshold GARCH Results for Trading Volume during Pre- and Post-
General Election (1994 - 2005) – Non-Government Link Companies (Non-GLCs)
Variables YTL NESZ PEP ROTH IOI KLK
Sector Trade and Services
Consumer Products
Consumer Products
Consumer Products
Plantation Plantation
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 6.3771 (0.0000)***
5.9684 (0.0000)***
5.7836 (0.0000)***
6.4712 (0.0000)***
6.4558 (0.0000)***
5.8326 (0.0000)***
PGE -0.1114 (0.3895)
-0.2762 (0.0913)*
0.0686 (0.6870)
-0.1096 (0.5239)
-0.0762 (0.5906)
-0.2019 (0.1406)
PtGE 0.1029 (0.1670)
-0.0564 (0.6778)
0.2829 (0.0161)**
-0.0873 (0.4935)
0.1034 (0.0989)*
0.0715 (0.4886)
1tVol 0.5814 (0.0000)***
0.4451 (0.0000)***
0.5356 (0.0000)***
0.4552 (0.0000)***
0.5857 (0.0000)***
0.5542 (0.0000)***
Variance Equation
0 0.0256 (0.0000)***
-0.0015 (0.1115)
0.0573 (0.0000)***
0.0839 (0.0000)***
0.0131 (0.0000)***
0.0197 (0.0000)***
1 0.0047 (0.4721)
0.0078 (0.0000)***
0.0055 (0.4904)
-0.0209 (0.0074)***
0.0072 (0.1900)
0.0092 (0.0740)*
i 0.0603 (0.0000)***
-0.0022 (0.1582)
0.0559 (0.0000)***
0.0786 (0.0000)***
0.0285 (0.0000)***
0.0404 (0.0000)***
1 0.9177 (0.0000)***
0.9948 (0.0000)***
0.9105 (0.0000)***
0.8816 (0.0000)***
0.9522 (0.0000)***
0.9478 (0.0000)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.4638 0.0006 0.4580 0.6550 0.2650 0.0097 10 lags 0.0300 0.0000 0.5065 0.7386 0.4904 0.0175
Ljung-Box Q2 Statistic (p-value) 5 lags 0.4640 0.0000 0.4560 0.6480 0.2580 0.0110
10 lags 0.0330 0.0000 0.4870 0.7460 0.4320 0.0200 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value.
156
Table 4.16(a): Threshold GARCH Results for Trading Volume during Pre- and Post-
General Election (2006 - 2015) – Government Link Companies (GLCs)
Variables AHB BIMB CIMB MAY MBS Sector Finance Finance Finance Finance Finance (p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1)
Mean Equation
0 3.1199 (0.0000)***
2.6901 (0.0000)***
6.7374 (0.0000)***
6.2437 (0.0000)***
2.8216 (0.0000)***
PGE -0.0195 (0.8790)
-0.0035 (0.9796)
0.1784 (0.0123)**
0.2192 (0.0066)***
-0.0399 (0.7007)
PtGE -0.0269 (0.7576)
-0.0135 (0.9020)
0.1235 (0.0384)**
0.1973 (0.0004)***
0.1019 (0.2588)
1tVol 0.7647 (0.0000)***
0.7870 (0.0000)***
0.5807 (0.0000)***
0.6034 (0.0000)***
0.7963 (0.0000)***
Variance Equation
0 0.0352 (0.0002)***
0.0444 (0.0000)***
0.0565 (0.0000)***
0.0194 (0.0000)***
0.0236 (0.0000)***
1 0.0259 (0.0068)***
0.0029 (0.6957)
0.0285 (0.1137)
-0.0087 (0.3723)
-0.0147 (0.0007)***
i 0.0543 (0.0021)***
0.0927 (0.0000)***
0.0866 (0.0009)***
0.0621 (0.0000)***
0.0492 (0.0000)***
1 0.8901
(0.0000)*** 0.9152
(0.0000)*** 0.6945
(0.0000)*** 0.9069
(0.0000)*** 0.9575
(0.0000)*** (Diagnostic Checking)
ARCH – LM Statistic (p-value) 5 lags 0.1688 0.0000 0.4437 0.8418 0.0022
10 lags 0.3466 0.0001 0.3850 0.9930 0.0033 Ljung-Box Q2 Statistic (p-value)
5 lags 0.1780 0.0000 0.4430 0.8400 0.0030 10 lags 0.3550 0.0000 0.3620 0.9920 0.0030
Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value.
157
Table 4.16(b): Threshold GARCH Results for Trading Volume during Pre- and Post-
General Election (2006 - 2015) – Government Link Companies (GLCs)
Variables BOUS TMK TNB UMWH CCM MRC
Sector Trade & Services
Trade & Services
Trade & Services
Consumer Products
Industrial Products
Property
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 4.5936 (0.0000)***
5.8369 (0.0000)***
6.1201 (0.0000)***
6.9978 (0.0000)***
2.8813 (0.0000)***
3.0884 (0.0000)***
PGE 0.0429 (0.7397)
0.0210 (0.7701)
0.1780 (0.0082)***
0.2291 (0.0354)**
0.3025 (0.1615)
0.0675 (0.5304)
PtGE 0.0141 (0.8872)
0.0129 (0.8112)
0.0602 (0.3605)
0.1751 (0.0508)*
0.5449 (0.0016)***
0.1299 (0.1024)
1tVol 0.6442 (0.0000)***
0.6252 (0.0000)***
0.6107 (0.0000)***
0.5002 (0.0000)***
0.7227 (0.0000)***
0.7986 (0.0000)***
Variance Equation
0 0.0309 (0.0000)***
0.0163 (0.0000)***
0.0196 (0.0000)***
0.0263 (0.0000)***
0.0935 (0.0000)***
0.0302 (0.0001)***
1 0.0244 (0.0001)***
0.0233 (0.0102)**
-0.0130 (0.0651)*
0.0424 (0.0000)***
0.0574 (0.0000)***
-0.0170 (0.0136)**
i 0.0432 (0.0000)***
0.0786 (0.0000)***
0.0932 (0.0000)***
0.0335 (0.0253)**
0.0303 (0.0388)**
0.0690 (0.0000)***
1 0.9127 (0.0000)***
0.8819 (0.0000)***
0.8934 (0.0000)***
0.8943 (0.0000)***
0.8708 (0.0000)***
0.9134 (0.0000)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.0033 0.5892 0.1901 0.2809 0.0027 0.0031 10 lags 0.0061 0.8449 0.1585 0.1177 0.0104 0.0119
Ljung-Box Q2 Statistic (p-value) 5 lags 0.0030 0.5680 0.1700 0.3030 0.0020 0.0040
10 lags 0.0030 0.8300 0.1220 0.0940 0.0110 0.0170 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value.
158
Table 4.17(a): Threshold GARCH Results for Trading Volume during Pre- and Post-
General Election (2006 - 2015) – Non-Government Link Companies (Non-GLCs)
Variables PBK RHBC GENM GENT MISC
Sector Finance Finance Trade and Services
Trade and Services
Trade and Services
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 5.8027 (0.0000)***
6.2867 (0.0000)***
5.1987 (0.0000)***
5.5374 (0.0000)***
4.5732 (0.0000)***
PGE 0.0557 (0.6303)
0.1175 (0.4428)
0.0202 (0.8561)
0.0944 (0.2518)
0.3235 (0.0041)***
PtGE 0.0147 (0.8368)
0.0257 (0.7700)
0.0192 (0.8017)
-0.1163 (0.1090)
0.1562 (0.2001)
1tVol 0.6061 (0.0000)***
0.5472 (0.0000)***
0.6626 (0.0000)***
0.6330 (0.0000)***
0.6711 (0.0000)***
Variance Equation
0 0.1109 (0.0003)***
0.3328 (0.0000)***
0.3341 (0.0000)***
0.1600 (0.0000)***
0.0193 (0.0000)***
1 0.0037 (0.7888)
0.1938 (0.0000)***
0.0815 (0.0014)***
0.0115 (0.5086)
0.0025 (0.6486)
i 0.1129 (0.0007)***
-0.1168 (0.0002)***
0.0302 (0.4283)
0.2145 (0.0000)***
0.0587 (0.0000)***
1 0.5985
(0.0000)*** 0.3595
(0.0000)*** 0.0578
(0.6368) 0.4452
(0.0000)*** 0.9413
(0.0000)*** (Diagnostic Checking)
ARCH – LM Statistic (p-value) 5 lags 0.5251 0.9660 0.9884 0.9229 0.2841
10 lags 0.7912 0.9983 0.9956 0.9602 0.5958 Ljung-Box Q2 Statistic (p-value)
5 lags 0.5430 0.9660 0.9880 0.9200 0.2970 10 lags 0.7850 0.9980 0.9950 0.9570 0.5910
Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value.
159
Table 4.17(b): Threshold GARCH Results for Trading Volume during Pre- and Post-
General Election (2006 - 2015) – Non-Government Link Companies (Non-GLCs)
Variables YTL NESZ PEP ROTH IOI KLK
Sector Trade and Services
Consumer Products
Consumer Products
Consumer Products
Plantation Plantation
(p, q) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) (1, 1) Mean Equation
0 6.6284 (0.0000)***
5.6604 (0.0000)***
5.9638 (0.0000)***
7.1552 (0.0000)***
5.2645 (0.0000)***
6.3827 (0.0000)***
PGE 0.1188 (0.2160)
0.6086 (0.0224)**
0.0359 (0.7393)
-0.1692 (0.0831)*
0.0685 (0.4666)
0.0043 (0.9731)
PtGE 0.0389 (0.5925)
0.3465 (0.0311)**
0.0986 (0.3116)
0.1728 (0.0892)*
0.1738 (0.0128)**
0.0971 (0.3070)
1tVol 0.5740 (0.0000)***
0.3709 (0.0000)***
0.5435 (0.0000)***
0.3913 (0.0000)***
0.6636 (0.0000)***
0.5335 (0.0000)***
Variance Equation
0 0.0022 (0.0098)***
0.1716 (0.0000)***
0.2169 (0.0000)***
0.0020 (0.0085)***
0.0117 (0.0000)***
0.2147 (0.0000)***
1 0.0181 (0.0000)***
0.1075 (0.0000)***
0.0436 (0.0771)*
0.0187 (0.0000)***
0.0184 (0.0033)***
0.0177 (0.3757)
i 0.0109 (0.0218)**
-0.0408 (0.0303)**
0.0769 (0.0342)**
0.0049 (0.0845)*
0.0161 (0.0123)**
0.1072 (0.0002)***
1 0.9714 (0.0000)***
0.8466 (0.0000)***
0.5236 (0.0000)***
0.9757 (0.0000)***
0.9341 (0.0000)***
0.5122 (0.0000)***
(Diagnostic Checking) ARCH – LM Statistic (p-value)
5 lags 0.1922 0.1230 0.6653 0.2048 0.0004 0.4774 10 lags 0.3505 0.2328 0.6199 0.4296 0.0060 0.8200
Ljung-Box Q2 Statistic (p-value) 5 lags 0.1910 0.1400 0.6480 0.1950 0.0010 0.4800
10 lags 0.3480 0.2580 0.6240 0.4090 0.0090 0.8290 Note: ***, ** and * denote significance at 1, 5 and 10% levels respectively. Numbers in parentheses depict p-value.
160
Figure 4.1: Relationship between Trading Volume and Stock Price Volatility for GLCs and Non-GLCs from 1994 – 2005
Pre-General Election Post-General Election
GLC
s - V
olat
ility
Non
-GLC
s - V
olat
ility
Trading Volume Trading Volume
161
Figure 4.2: Relationship between Trading Volume and Stock Price Volatility for GLCs
and Non-GLCs from 2006 – 2015
Pre-General Election Post-General Election
GLC
s - V
olat
ility
Non
-GLC
s - V
olat
ility
Trading Volume Trading Volume
162
Figure 4.3: Trading Volume and Stock Prices for the Highly Traded GLCs before the General Elections
Pre-General Election 2008 Pre-General Election 2013
CIM
B
MA
Y
TNB
UM
WH
163
Figure 4.4: Trading Volume and Stock Prices for the Highly Traded Non-GLCs before the General Elections
Pre-General Election 2008 Pre-General Election 2013
MIS
C
164
CHAPTER 5
GENERAL CONCLUSION
This study thoroughly examines the effect of the Malaysian general elections on its stock
market volatility from the year 1994 to 2015. Overview of Malaysian general elections and
the impact of political events on the stock market are presented in Chapter 1. Next, in Chapter
2, the analysis on election effect is firstly conducted on the benchmark composite indices,
including the Shariah-compliant indices, which represent the large, medium and small market
capitalization in the stock market. Then, in Chapter 3, the analysis is extended to the industry
by investigating the ten sectoral indices in the stock market. Lastly, at the firm level, eleven
Government-Link-Companies (GLCs) and eleven Non-Government-Link-Companies (Non-
GLCs) are selected as the sample of analysis in Chapter 4. By performing the analysis level
by level, this study is able to provide a complete understanding of election effects on the
Malaysian stock market.
In Chapter 2, evidence of election effect is found in the sample period of 2007 to
2015, which covers the 12th and 13th Malaysian general elections. These are the most two
recent general elections with a turbulent political condition. Generally, the result is consistent
with Wang and Lin (2009), Smales (2016), and Lean and Yeap (2017), who found higher
volatility in the pre-election periods. Specifically, this chapter shows the relevance of market
capitalization to stock market volatility when there is political uncertainty surrounding
elections. Companies with small capital experienced higher stock volatility prior to general
election. Conversely, the stock volatility is lower for larger companies' stock. Furthermore,
lower stock volatility is observed in Shariah-compliant stock indices which suggest that
Shariah-compliant companies have a lower risk during the pre-general election periods. The
165
finding from Chapter 2 implies that risk-averse investors could mitigate the political risk by
diversifying their portfolio in large companies’ stock and Shariah-compliant companies
stock. On the other hand, investors should be vigilant during the pre-general election periods
as their profits are underlying high volatility and compensation for abnormal high returns is
negligible.
The objective of Chapter 3 is to identify the influence of general elections on the
movement of the ten selected sectoral indices in Malaysian stock market for the period of
1994 to 2015. Beside the full sample period, the five general elections period is divided into
two stages. The first sub-sample covers the 9th, 10th and 11th Malaysia general election where
the ruling party continued to win 2/3 majority seats in all the three elections. The second sub-
sample period represents drastic shock periods during the 12th and 13th Malaysia general
election, from the year 2006 to 2015. Interestingly, the finding of the first sub-sample period
is obviously different from the second sub-sample period. While volatility on the stock return
is lower during the pre-general election periods of 1994-2005, it did show its negative and
significant influence in the 2008 and 2013 general election years. The result is quite in accord
with the political condition. For the first sub-sample period of 1994 to 2005, lower volatility
in the market is a good indicator showing that there is no uncertainty before the general
elections. Nevertheless, during the pre-general election period in the second sub-sample of
2006 to 2015, higher volatility in the market was induced by the uncertainties associated with
the general elections. Hence, the break-down of the full sample period into two sub-sample is
able to illustrate the impact of general elections more precisely. The result sheds light on the
importance of addressing the difference of political condition when testing for asymmetry
effect during election periods. Besides, the finding also indicates that the sectors of
Construction, Finance, Mining, and Property are more sensitive to the market condition with
166
significant result found in stock volatility. This is in line with Tuyon and Ahmad (2016)
where they also classified these few sectors as the cyclical sector. While Consumer Product is
a defensive sector where the estimated results are mostly insignificant.
Provided the evidence of election effect on the Malaysian main stock indices in
Chapter 2 and sectoral indices in Chapter 3, Chapter 4 further explores the reaction of stock
returns and volatility in the firm level. Eleven GLCs and eleven Non-GLCs are selected as
the sample and the sample period covers from January 4, 1994, through December 31, 2015.
Similar to Chapter 3, the full sample period is divided into two sub-samples. The finding in
this chapter also shows that the pattern of the stock volatility in GLCs and Non-GLCs is
clearly different in the two sub-samples, and thus, lends support to the observation in Chapter
3. As well, lower volatility of returns is found before the general elections in years 1994-
2005, for both the GLCs and Non-GLCs stock indices. In the general election years of 2006-
2015, the finding shows that most of the GLCs and Non-GLCs stock prices were highly
volatile before the general elections. Interestingly, further analysis on trading volume shows
that those GLCs and Non-GLCs with higher market capitalization that encountered higher
volatility are also associated with a significant higher trading volume. This pattern of trading
was probably due to the willingness of liquidity traders to trade in the periods where the
prices are more volatile (Admati and Pfleiderer, 1988). Another interesting point found in the
finance sector is that investors are still willing to actively trade the GLCs stock despite
market uncertainties. This indicates that investors are very careful during the time of market
uncertainties as it is always safer to trade GLCs stock which is more liquid than others.
Nonetheless, this trading pattern appears only in the finance sector, but not in other sectors.
167
Overall, the findings of this study indicate that the Malaysian stock market volatility
is associated with investors' behaviour during the period of the general elections. The
presence of a political shock in the 12th and 13th Malaysia general election changed the
trading pattern in the market that reflects how investors react to the market uncertainty.
Therefore, this study is of great importance to risk managers, portfolio managers,
policymakers, and market participants to understand the pattern of volatility in the Malaysian
stock market during general election years. Thus, the results of this study perhaps provide an
insight for investors in adjusting their portfolio around the next general election. Future work
in this area can proceed in several directions. First, microdata on investors' personal
investment choices can be used to study their influence on stock market performance during
general election. Second, future study can be conducted to compare the market performance
of different stocks characteristics to evaluate the volatility during general election. This study
provides an exemplar for further studies to explore further details by employing a
comprehensive disaggregated data for different sectors and firm characteristics and sectors.
168
References
Admati, A. and Pfleiderer, P. (1988). “A theory of intraday patterns: Volume and price variability”, Review of Financial Studies, Vol. 1, pp. 3 – 40.
Lean, H. H. and Yeap, G. P. (2017). “Asymmetric Effect of Political Elections on Stock Returns and Volatility in Malaysia”, in Munir, Q. and Kok, S. C. (Ed.), Information Efficiency and Anomalies in Asian Equity Markets, Routledge, Taylor and Francis Group, pp. 228 - 245.
Smales, L. A. (2016). “The role of political uncertainty in Australian financial markets”, Accounting and Finance, Vol. 56, No. 2, pp. 545–575.
Tuyon, J. and Ahamd, Z., (2016). “Behavioural finance perspectives on Malaysian stock market efficiency”, Borsa Istanbul Review "Vol.16, No.1, pp. 43 - 61. Wang, Y. H. and Lin, C. T. (2009). “The Political Uncertainty and Stock Market Behavior in Emerging Democracy: The Case of Taiwan”, Quality and Quantity, Vol. 43, No. 2, pp. 237 - 248.