2013 GhulamGhouse

Tracing Dynamic Linkages and Volatility Spillover
Effect between Pakistani and Foreign Stock Markets
By
Ghulam Ghouse
Registration No. 03/M.Phil-ETS/PIDE/2013
Supervised By
Dr. SAUD AHMED KHAN

Assistant Professor
IESE (SCEE)
NUST
Department of Econometrics and Statistics

Pakistan Institute of Development Economics
Islamabad, Pakistan
25 May, 2015
i
Tracing Dynamic Linkages and Volatility Spillover Effect
between Pakistani and Foreign Stock Markets
By
Ghulam Ghouse
Registration No. 03/M.Phil-ETS/PIDE/2013
MPhil Econometrics
A Dissertation Submitted to the Pakistan Institute of Development Economics,
Islamabad, in partial fulfillment of the requirements of the Degree of Master of
Philosophy in Econometrics
Supervised By
Dr. SAUD AHMED KHAN

Assistant Professor
IESE (SCEE)
NUST
Department of Econometrics and Statistics

Pakistan Institute of Development Economics
Islamabad, Pakistan
25 May, 2015
i
IN THE NAME OF
ALLAH
The Most Beneficent
The Most Merciful
“To Allah belongs whatever is in the heavens and whatever is in the
earth. Whether you show what is within yourselves or conceal it,
Allah will bring you to account for it. Then He will forgive whom
He wills and punish whom He wills, and Allah is over all things
competent.”
(Al-Baqarah, 2:284)
ii
GOLDEN SAYING OF
THE HOLY PROPHET
(Peace and Blessings of Allah be Upon Him)
“The Prophet Muhammad (peace be upon him) said: "If anyone travels on
a road in search of knowledge, Allah will cause him to travel on one of the
roads of Paradise. The angels will lower their wings in their great pleasure
with one who seeks knowledge. The inhabitants of the heavens and the
Earth and (even) the fish in the deep waters will ask forgiveness for the
learned man. The superiority of the learned over the devout is like that of
the moon, on the night when it is full, over the rest of the stars. The learned
are the heirs of the Prophets, and the Prophets leave (no monetary
inheritance), they leave only knowledge, and he who takes it takes an
abundant portion”.
(Sunan of Abu-Dawood, Hadith 1631)
iii
Abstract
This study traces the degree of integration and volatility spillover effect between the Pakistani
and foreign stock markets by analyzing the Meteor shower hypothesis. Daily data are used
from nine worldly equity markets (KSE 100, NIKKEI 225, HIS, S&P 500, NASDAQ 100,
DOW JONES, GADXI, FTSE 350 and DFMGI) for the period of 3rd Jan, 2005 to 28th Nov,
2014. At first we used the whole data set to explore the mean and volatility spillover effect
between stock markets. Then we split data set into two subsets. First part of data contains the
era of global financial crisis of 2008 from 3rd January 2005 to 31st December 2009. Second
subset is after global financial crisis time period from 4th January 2010 to 28th November 2014
(The global crisis prevailed till end of 2009). Univariate GARCH type models i.e. GARCH and
GJR are employed to estimate volatility of Pakistani and leading foreign stock markets. Then
following the technique of Hamao et al (1990) the same GARCH type models are utilized to
explore the dynamic linkages between Pakistani and foreign stock markets. This study
emphasis on exploring the direct linkages between Pakistani (KSE 100) and US stock markets
(S&P 500, NASDAQ 100 and DJI). We also analyze the indirect linkages between Pakistani
and US markets through Dubai financial market (DFMGI).
The results of whole data set from 3rd Jan, 2005 to 28th Nov, 2014 illustrate that there is mixed
co-movements between leading foreign stock markets and Pakistani stock market. The
unidirectional mean and volatility spillover effect from S&P 500, NASDAQ 100 and DJI to
KSE 100 is found. The bidirectional mean spillover effect between DFMGI and S&P 500,
NASDAQ 100 and DJI is found. The bidirectional mean and volatility spillover effect between
Pakistan and Dubai equity market is also traced.
The results from first data subset from 3rd Jan, 2005 to 31st Dec, 2009 provide evidences of
unidirectional mean and volatility spillover effect from S&P 500, NASDAQ 100, DJI and
DFMGI to KSE 100. The results also describe that there is no co-movement between DFMGI
iv
and S&P 500 and DJI but unidirectional mean and volatility spillover effect from DFMGI to
NASDAQ 100 is found. There is unidirectional mean and volatility spillover effect from
DFMGI to KSE 100 but there is no sign of co-movement from KSE 100 to DFMGI.
The second subset of data includes period after the global financial crisis from 4th Jan, 2010 to
28th Nov, 2014. The results from second Period show that there exists only unidirectional
volatility spillover effect from S&P 500, NASDAQ 100 and DJI to KSE 100. The mean and
volatility spillover effect from KSE 100 to DFMGI and only mean spillover effect from
DFMGI to KSE 100 is traced. There is sign of unidirectional volatility spillover effect from
S&P 500, NASDAQ 100 and DJI to DFMGI. The mean and volatility spillover effect from
DFMGI to S&P 500, NASDAQ 100 and DJI is also observed. This reveals that these markets
are directly interlinked with each other during the period from 4th Jan, 2010 to 28th Nov, 2014.
After this discussion we came to know there is only one indirect linkage through which may
the information transmitted to KSE 100. This is linkage is developed due to the co-movement
among KSE 100, DFMGI and NASDAQ 100 in crisis period. Like, the mean and volatility
spillover effect from DFMGI to KSE 100 and NASDAQ 100. There is also mean and volatility
spillover effect from NASDAQ 100 to KSE 100. This integration between these markets may
provide a sign of indirect linkage. It also exhibits the volatility in Pakistan stock market returns
is instigated through direct effects as well as indirect effects. Our study brings important
conclusions for financial institutions, portfolio managers, market players and academician to
diagnose the nature and level of linkages between the financial markets.
Key words: Volatility, Equity market, Spillover effect, ARCH model, Asymmetric GARCH.
v
DEDICATED
TO
HAZRAT FATIMA (R.A)
AND HER
DEVOTEE
SHAMIM BIBI (MOTHER, LATE)
vi
Acknowledgements
First of all a special gratitude and special appreciation goes to ALLAH almighty; without His
blessings I would not be able to think of completing this work. After that, I offer my admirations
and respect from the core of my heart to the Holy Prophet Muhammad (Peace be upon him) who urges
his followers to “Seek knowledge from cradle to grave”. I would also like to pay my humble thank
to my beloved MOTHER, who prayed for me a lot and always encouraged me and guide me
in a proper way and advise me not to lose heart.
Being a fresh researcher and a student in the field of econometrics, this dissertation would not
have been possible without the help, provision and patience of my dedicated supervisor, Dr.
Saud Ahmed Khan, Assistant Professor, NUST, Islamabad, who supported me and guided me
right from the first day of my research. He made valuable and fruitful comments that helped
me to improve my research. I deeply appreciate and recognize all that I have received from
him.
I would never have been able to finish my dissertation without the kind leadership of my
teachers, Dr. Saqlain Raza, Dr. Hafza Hina, Prof. Muhammad Arshad, Prof. Muhammad
Jamil assistance from friends Adil Hussian Shahzada, Ali Umar, Ehtisham Noor, Nayyar
Rafiq, Saqib Aslam and moral support from my Sister and Brothers who helped me a lot,
give me his precious time and valuable suggestions in completing this piece or work.
I would like to thank all of my Friends, my Class Fellows who helped me to grasp some key
concepts regarding subject knowledge during course work and morally supported me on
regular basis throughout the session.
Ghulam Ghouse
vii
TABLE OF CONTENTS
Abstract………………………………………………………………………………………IV
Dedication ...............................................................................................................................VI
Acknowledgements…..….………………………………………………………………….VII
Tables of Contents………………………………………………………………………….VIII
List of Tables…………………………………………………………………………………X
List of Figures ........................................................................................................................XII
List of Acronyms…………………………………………………………………………..XIII
CHAPTER 1
INTRODUCTION....................................................................................................01
1.1 Objectives of the study.........................................................................................05
1.2 Research Question ...............................................................................................05
1.3 Significance of the study......................................................................................06
CHAPTER 2
REVIEW OF LITERATURE .............................................................................................07
2.1 International prospective ......................................................................................08
2.2 Global financial crisis impact on Dubai’s economy ............................................09
2.3 The global financial crisis 2007-08 impact on Pakistan economy ......................11
2.4 Possible channels of information transmission ....................................................13
2.5 Previous empirical studies ..................................................................................16
CHAPTER 3
MODELLING VOLATILITY AND EXPLORING SPILLOVER EFFECT ................18
3.1Econometric Methodology and Model Specification ...........................................19
3.1.1ARCH Model ................................................................................................21
3.1.2 GARCH Models................................................................................................22
3.1.3 GARCH (1, 1) ...................................................................................................23
viii
3.1.4 Asymmetric GARCH Models ...........................................................................24
3.1.4.1 GJR-GARCH Models ............................................................................24
3.1.4.2 GARCH (p, q) Model for Exploring Spillover Effect ………………...25
3.1.5 Residual Analysis..............................................................................................26
3.2 Methodological Structure.....................................................................................27
3.3 Description of Data and sources ..........................................................................27
CHAPTER 4
ESTIMATIONS AND ANALYSIS ....................................................................................29
4.1 Graphical Analysis ..............................................................................................30
4.2 Summary Statistics...............................................................................................37
4.3 Volatility Model specifications of Return Series .................................................38
4.4 Tracing Spillover Effect .......................................................................................44
4.5 Tracing direct and indirect linkages between Pakistan and US stock markets ....86
4.6 Tracing Spillover effect (3rd Jan, 2005 to 31st Dec, 2009) ...................................95
4.7 Tracing Spillover effect (4th Jan, 2010 to 28th Nov, 2014) ...................................104
CHAPTER 5
CONCLUSION AND POLICY IMPLICATIONS ...........................................................116

5.1 Conclusion ..........................................................................................................116
5.2 Policy Implication ...............................................................................................118
REFERENCES .....................................................................................................................119
ix
List of Tables
Table No. Description Page No.
Table 1.1 Exports, Imports and Relevant growth Rates between PAK-UAE ................. 37
Table 4.2 Summary of statistics ....................................................................................... 37
Table 4.3.1 Volatility models of Asian Stock markets Return series ................................. 40
Table 4.3.2 Volatility models of American stock markets Return series ........................... 42
Table 4.3.3 Volatility models of European and Gulf Stock markets Return series ............ 43
Table 4.4.1 Volatility Spillover effect between KSE 100 and NIKKEI 225 ...................... 46
Table 4.4.2 Volatility Spillover effect between KSE 100 and HSI .................................... 48
Table 4.4.3 Volatility Spillover effect between KSE 100 and FTSE 350 .......................... 49
Table 4.4.4 Volatility Spillover effect between KSE 100 and GDAXI .............................. 50
Table 4.4.5 Volatility Spillover effect between S&P 500 and NASDAQ100 .................... 52
Table 4.4.6 Volatility Spillover effect between S&P 500 and NIKKEI 225 ...................... 54
Table 4.4.7 Volatility Spillover effect between S&P 500 and HSI .................................... 55
Table 4.4.8 Volatility Spillover effect between S&P 500 and FTSE 350 .......................... 56
Table 4.4.9 Volatility Spillover effect between S&P 500 and GDAXI .............................. 57
Table 4.4.10 Volatility Spillover effect between NASDAQ100 and DJI ............................. 59
Table 4.4.11 Volatility Spillover effect between NASDAQ100 and NIKKEI225 ............... 61
Table 4.4.12 Volatility Spillover effect between NASDAQ100 and HSI ............................ 62
Table 4.4.13 Volatility Spillover effect between NASDAQ100 and FTSE350 ................... 63
Table 4.4.14 Volatility Spillover effect between NASDAQ100 and GDAXI ...................... 64
Table 4.4.15 Volatility Spillover effect between DJI and NIKKEI 225 ............................... 65
Table 4.4.16 Volatility Spillover effect between DJI and HIS ............................................. 67
Table 4.4.17 Volatility Spillover effect between DJI and FTSE 350 ................................... 68
Table 4.4.18 Volatility Spillover effect between DJI and GDAXI ....................................... 69
Table 4.4.19 Volatility Spillover effect between NIKKEI 225and HSI ............................... 71
Table 4.4.20 Volatility Spillover effect between NIKKEI225 and FTSE ............................ 73
Table 4.4.21 Volatility Spillover effect between NIKKEI225 and GDAXI ......................... 74
Table 4.4.22 Volatility Spillover effect between NIKKEI225 and DFMGI......................... 75
Table 4.4.23 Volatility Spillover effect between HSI and FTSE 350................................... 77
x
Table 4.4.24 Volatility Spillover effect between HSI and GDAXI ...................................... 79
Table 4.4.25 Volatility Spillover effect between HSI and DFMGI ...................................... 80
Table 4.4.26 Volatility Spillover effect between FTSE 350 and GDAXI ............................ 82
Table 4.4.27 Volatility Spillover effect between FTSE 350 and DFMGI ............................ 84
Table 4.4.28 Volatility Spillover effect between GDAXI and DFMGI................................ 85
Table 4.5.1 Volatility Spillover effect between KSE 100 and S&P 500 ............................ 87
Table 4.5.2 Volatility Spillover effect between KSE100 and NASDAQ100 ..................... 89
Table 4.5.3 Volatility Spillover effect between KSE 100 and DJI ..................................... 90
Table 4.5.4 Volatility Spillover effect between KSE 100 and DFMGI .............................. 91
Table 4.5.5 Volatility Spillover effect between S&P 500 and DFMGI .............................. 92
Table 4.5.6 Volatility Spillover effect between NASDAQ 100 and DFMGI ..................... 93
Table 4.5.7 Volatility Spillover effect between DJI and DFMGI ....................................... 94
Table 4.6.1 Volatility Spillover effect between KSE 100 and S&P 500 ............................ 96
Table 4.6.2 Volatility Spillover effect between KSE100 and NASDAQ100 ..................... 98
Table 4.6.3 Volatility Spillover effect between KSE 100 and DJI ..................................... 99
Table 4.6.4 Volatility Spillover effect between KSE 100 and DFMGI ............................ 100
Table 4.6.5 Volatility Spillover effect between S&P 500 and DFMGI ............................ 101
Table 4.6.6 Volatility Spillover effect between NASDAQ 100 and DFMGI ................... 102
Table 4.6.7 Volatility Spillover effect between DJI and DFMGI ..................................... 103
Table 4.7.1 Volatility Spillover effect between KSE 100 and S&P 500 .......................... 106
Table 4.7.2 Volatility Spillover effect between KSE100 and NASDAQ100 ................... 107
Table 4.7.3 Volatility Spillover effect between KSE 100 and DJI ................................... 108
Table 4.7.4 Volatility Spillover effect between KSE 100 and DFMGI ............................ 109
Table 4.7.5 Volatility Spillover effect between S&P 500 and DFMGI ............................ 110
Table 4.7.6 Volatility Spillover effect between NASDAQ 100 and DFMGI ................... 111
Table 4.7.7 Volatility Spillover effect between DJI and DFMGI ..................................... 112
xi
List of Figures
Figure No. Description Page No.
Figure 4.1.1.a Graph of series at level of stock indices .......................................................... 30
Figure 4.1.1.b Graph of series at level of stock indices ....................................................... 31
Figure 4.1.2 Graph of return series....................................................................................... 32
Figure 4.1.3 Graph of Squared return series ........................................................................ 33
Figure 4.1.4 Graph of distribution of the return series ......................................................... 34
Figure 4.1.5 Graph of ACF and PACF of return series ........................................................ 35
Figure 4.1.6 Graph of ACF and PACF of squared return series .......................................... 36
xii
List of Acronyms
KSE 100 Karachi Stock Exchange Index (Pakistan)
S&P 500 Standards & Poor's Stock Market Index (USA)
NASDAQ 100 NASDAQ Stock Market Index (USA)
DJI DOW JONES Industrial Average’s Index (USA)
NIKKEI 225 NIKKEI Stock Market Index (Japan)
HIS HANG SENG Stock Exchange Index (Hong Kong)
FTSE 350 London Stock Exchange Index (UK)
GDAXI German Stock Market Index (Germany)
DFMGI Dubai Financial Market Index (Dubai)
ARCH Autoregressive Conditional Hetroscedastic
GARCH Generalized Autoregressive Conditional Hetroscedastic
GJR Glosten, Jagannathan and Runkle
EGARCH Exponential Generalized Autoregressive Conditional Hetroscedastic
IMF International Monetary Fund
LM test Lagrange Multiplier Test
WB World Bank
xiii
Chapter 1
INTRODUCTION
Modern econometric tools are used for investigating volatility co-movement between
the financial markets. If the return or volatility of one financial market affects the other
market’s returns or volatility, it is known as spillover effect. Global financial integration started
in the mid-1980s, consequently risk and return Co-movements between the financial markets
were observed at that time. The growing economic integration of intercontinental financial
markets has gotten significance since last three decades. The major factors behind this observed
globalization are extensive growth of technology, easy capital flow and financial links between
the economies. That is why analysis of the nature and level of linkages between different
financial markets is significant for financial institutes, portfolio managers and market players.
Engle et al., (1990) proposed two hypotheses First, “Heat wave hypothesis analysis the intra-
sector / intra-firm co-movements”. Second, “Meteor Shower hypothesis traces intra-market co-
movements”.
The global financial crisis of 2008 was one of the worst financial crises of US history.
It not only triggered imbalances in US economy but also impacted a major part of overall global
economy. Most of the global financial crises initiated from US economy and due to the strong
interdependence of US economy with other economies these crises impacted all integrated
economies at some extend. This crisis emerged primarily due to creation and expansion of
bubbles in housing and subprime markets of US in 2007.The key reasons identified by the
academic researchers behind this crisis were excessively relaxed monetary policy, regulatory
failures in macroprudential and microprudential levels, the accumulation of global balance of
payment inequalities and flaws in the international financial planning (Kawai et al., 2012).
1
This financial crisis offers the chance to reconsider the degree of interdependencies,
between the financial markets of different economies. Occurrence of some past and present
events explain the effects of crises on a particular market. Our evaluation is constructed on an
examination of such events to describe the degree of spillover effect between the Pakistani and
foreign financial markets and their direct and indirect linkages, daily data is used from 3rdJan,
2005 to 28th Nov, 2014.
US global financial crisis triggered disparities in most of the developed and developing
economies having direct or indirect links with US economy. Dubai’s economy due to its
investment linkages with US was also impacted by the US mortgage crisis, Ibrahim Onour
(2010). The influence of global financial crisis impacted on different sectors of the economy,
some had prominent position in Dubai economy. Real estate sector was most effected sector of
Dubai economy. Real estate sector contributed more than 29% of GDP of Dubai in 2007. Due
to the shortage of the loaning from the banks, the investment in this sector was reduced.
Likewise, Dubai financial market had also impressively impacted Dubai’s economic
growth. Portfolio investment in Dubai financial market reduced 24% in 2009. Dubai banking
sector also stuck in this vicious circle of financial crisis, limited capital reserves and imposing
the regulation on bank lending. Dubai exports were decreased 15% during 2008-09. At
international level the oil prices were reduced, especially it was not good news for oil producing
countries.
The global financial crisis effects transmitted to Pakistan economy through direct
channels (i.e. Exports, capital flow, remittances, equity values), Amjad and Din (2010). At that
time Pakistan economy was also facing some internal issues like political instability, bad
governance, deficit in current account, rising unemployment, energy crisis and failure of
macroeconomic policies. Owing to all these circumstances Pakistan economy met the deficit
2
in balance of payment. The exports were reduced and high stagflation emerged in the economy.
The capital reserves curtailed approximately RS0.10 trillion but the banking sector was less
affected from this spillover effect. Pakistan government got assistance from international
monetary fund (IMF) through Stand-By Arrangement of $7.6 billion in 24 Nov, 2008 and
released first installment of $3.1 billion to meet budget deficit.
Pakistan and Dubai both countries have significant relationship in different sectors of
economy. Dubai is one of the emerging markets of UAE. Over 1.2 million emigrants of
Pakistan are providing their services in UAE. Their remittances significantly contribute to
Pakistan’s foreign reserve. UAE is the second prominent source of remittances from Pakistani
emigrants. Pakistan expatriates provided $2.52 billion remittances in 2013-14 with share in
total remittances of 19.57%. Similarly, UAE has major share in Pakistan exports and imports.
In 2013-14 UAE has 8% share of total Pakistan’s exports and 17% share in imports.
Following table explains the export, import and relevant growth rates between Pakistan and
UAE.
Table 1.1 Exports, Imports and Relevant growth Rates between PAK-UAE
DIRECTION OF TRADE RS(Million)

yaer Exports change% Imports change%
2003-04 54,304 -10 98,392 11.5
2004-05 65,054 20 101,054 2.7
2005-06 785,873 1,108 203,923 101.8
2006-07 83,990 -89 167,907 -17.7
2007-08 130,549 55 214,559 27.8
2008-09 114,074 -13 253,294 18.1
Source: (Ministry of finance Pakistan)
One of the major reasons behind the reduction of exports was that economy of Pakistan
remained defenseless to exterior shocks (i.e. oil market and global financial crises) with the
3
growing of current global financial crisis all the main Pakistan’s export markets were likely to
observe reduction. As per prediction of IMF exports of Pakistan declined around 3%.
Second reason was that in 2006-07 a lot of money moved from many other countries to
Dubai, when property and capital markets were flourishing in Pakistan. At that time credit was
available at low cost and traders were operating the market to gain large dividends. These
people had no knowledge what would be the result of this easy money. A huge amount of new
money arrived in Dubai. A large ratio of it wiped out when Dubai capital and the property
market crashed. The flow of finance lost balance as Dubai’s economy gone into depression in
2008.
At the occasion of UAE cityscape “7th annual property and real estate exhibition” 2008
in Dubai more than 100 Pakistani investors invested over $100 million for the booking of
construction projects. There were a large number of Pakistani investors out of 40,000 visitors
all over the world took part in this exhibition and booked their investment in the construction
companies’ offices. Within preceding 10 to 11 months after November 3, 2007 when the
emergency statements varies, outflow of capital from Pakistan economy, frequently to Dubai,
estimated $30 billion to around $45 billion. A top constructor was convinced of huge outflow
of $30 billion which was nearly 20% of Pakistan’s aggregate economy (Kashif Aziz, 2008).
In 2014 Pakistan exports to Dubai touched 21 billion Dirham, these exports were 4%
of total exports volume and the average growth rate is 29% per annum. In preceding year a
huge amount of investment shifted to UAE (Dubai) due to energy crises and security situation.
In last 6 months of preceding year investors invested $1 billion in Dubai in different sectors.
Pakistani investors are at third number regarding investment after India and UK. Pakistani
investors invest Dh4.5 billion in “Dubai real estate market” in the first half of the current year
4
2014 with 3,064 transactions, Dubai land department (DLD) report. It is a ranked top third
foreign investment in real estate market of Dubai.
In year 2008 the Dubai’s real estate industry collapsed down. Pakistani investors faced
losses more than they were facing in their local investment or industries in Pakistan. They had
stalled property projects in Dubai of worth approximately US$72.35 billion. The key factors
behind this distortion were the rumors circulating in Dubai’s real estate sector, that all these
projects, carried out by renowned construction company for unspecified time period. Owing to
these rumors most of the investors stopped their construction projects. Even there was no
official declaration for developers regarding the work stoppage. Those investors who visited
the Dubai, they had seen that work on 50 projects worth of $3.8 million to about $70 million
were stopped. After that no one ready to invest in these projects.
The above discussion elaborates the linkages between Pakistan, US and Dubai economies.
Most of the researchers explored the direct linkages between Pakistani and foreign financial
markets before this study. In this study we analyze direct linkages between Pakistani and
foreign stock markets, also trace out indirect linkages between Pakistani and US stock markets
through Dubai financial market.
1.1 Objective of this research
The objective of this study is to investigate the direct and indirect linkages between
Pakistani and foreign stock markets in general and particularly during the global financial crisis
of 2007-08. Keeping in view the objective of this research, we will address the following
research questions.
1.2 Research questions
 Are there any dynamic linkages between Pakistani and leading foreign financial (stock)
markets?
5
 Are these linkages between Pak-US markets direct or indirect (Via Dubai Financial
market)?
1.3 Significance of the study

Substantial literature is available on the impact of US financial crisis on different
countries which are directly or indirectly integrated with international financial system and
trade, some of studies like Kawai et al. (2012), Cheng et al. (2011), Angkinand et al. (2009)
and Patricia et al. (2001) traced the spillover effect from US to other developed economies.
Ashraf (2010) Ibrahim (2010) and Thomson et al. (2009) analyzed the spillover effect from US
to Dubai economy. From Pakistan Afzal et al. (2012), Nazir et al. (2012) and Amjad et al.
(2010) investigated direct effect of global financial crisis on Pakistan economy. In Pakistan all
these researches investigate how the US financial crisis directly impacted Pakistan economy.
No study reported channels of return and volatility spillover. We use one of major channel
through which the influence of global financial crises came to Pakistan economy, i.e. Karachi
stock market. This research is first of its kind in Pakistan to examine the direct and indirect
spillover effect of global financial crisis on Pakistani and leading foreign stock markets.
We explore that the global financial crisis negatively impacted returns and enhanced
the volatility in Pakistani and foreign financial markets. All these findings help us making
more effective short run and long-run policies to tackle the effect of such global crises in favor
of sustainable economic growth of Pakistan.
6
Chapter 2
LITERATURER REVIEW
This chapter briefly discusses previous studies. There’s a long debate on volatility spillover
effect between the intra and cross financial markets in financial Econometrics literature. In
financial Econometrics many researchers have presented an enormous empirical and
theoretical work to validate their particular selected models. This review emphasizes on tracing
dynamic linkages in Pakistan. Particularly, we try to point out the possible direct and indirect
linkages between Pakistani and foreign stock markets.
7
2.1 International perspective
The strong integration of global economies through different financial or real links,
crisis in one part of the world is much likely to transmit to other parts. In2007 when global
markets experienced a huge wave of financial crisis due to United State sub-prime mortgage
crisis. It not only impacted domestic economy of USA but also other economies of the world
which are integrated directly or indirectly with US economy. “This crisis emerged mainly due
to creation and expansion of bubbles in housing and subprime markets of US” Kawai et al.
(2012). At that time US economy was facing current account deficit whereas East Asian and
oil producing economies were with a persistent current account surplus. In September 2008,
the devastation in the financial sector shifted towards the real sector of US economy. In2007
this crisis started from small segment of the financial system. In 2008-2009, it became a reason
of global financial crisis Mishkin (2011).
In addition to other major or minor factors, Insolvency of the Lehman Brothers (Bank)
was the main cause of failure of most of the European and United State banks and this
bankruptcy at last became a major reason of financial crisis in other integrated countries. Due
to global financial crisis Taiwan stock market index unexpectedly drop down almost 60% from
9858 points to 3955 points (Cheng, 2012). Angkinand et al. (2009) investigated how financial
crisis in US markets impacted 17 developed economies from 1973-2009, and they found that
spillover effects from the US to other industrial countries were highest after the appearance of
the U.S subprime mortgage market collapse in the summer of 2007. Patricia et al. (2001)
explained the links of United State stock market with UK and European stock markets. The
results showed strong bilateral relationship between US and UK stock markets while
relationship of US market with other European economies also exists but not with the same
potential as with UK. The business cycle movements in the United Kingdom economy are more
sensitive to disturbances in US relative to other European economies. Also there exist
8
increasingly associated long run relationship between the stock markets of both economies.
The excitement in the other markets exists due to the change in rules and treaties and specially
the exchange rate by the United Kingdom market. Ashamu and James (2012) inspected that the
global financial crisis impacted Nigerian economy specially, the banking system.
Ashraf (2010) explored that US mortgage crisis 2008 impacted Dubai financial market,
banking system, GDP and Dubai’s companies credit ratings. Initially Dubai’s economy
tolerated the effect of global financial crisis but on 25 Nov, 2009 Dubai demanding suspension
on debt repayment from world. The foreign banks already pumped huge quantity of money as
loans and as investment in different sectors of Dubai, knowing that if UAE defaulted on debt,
they would face massive losses. US financial crisis also provided base for Dubai financial
crisis. As a result of Dubai financial crisis the investment in real estate sector was reduced.
Sleiman et al. (2009) examined that the neighbor country of Dubai, Abu Dhabi presenting a
loan of $10 billion for the management of its debt. At that time Dubai’s amount of debt was
roughly $59 billion and the overall global debt was 10 times more than that of Dubai’s debt.
Ibrahim (2010) investigated that spillover effect of US Mortgage crisis 2008, badly impacted
oil producing countries including Dubai. The portfolio investment in Dubai financial market
decreased 42%, due to this spillover effect Dubai also faced internal debt crisis in 2009.
2.2 Global financial crisis impact on Dubai’s economy
The global financial crisis 2008 smashed Dubai’s economy. Hasan (2010) examined
following financial crisis crushed the major sector of the economy. Global financial crisis
barely affected Dubai among all the other oil producing countries. It provided a reason for
reducing the oil prices in the developed countries. Ellaboudy (2010) explored that from mid-
2008 to March 2009, oil prices drop down from $140 to $50 at international market.
9
Global financial crisis triggered down the growth of banking sector. Mehta (2012)
studied the ongoing financial crisis constrained the substantial growth of UAE banking system.
Hamdan (2012) investigated that due to the heavy amount of world loans, the Dubai’s banks
hesitated to take further loans until they were totally recovered. Financial crisis badly impacted
Dubai banking system as compare to UAE banking system. Shafique et al. (2012) found that
the Islamic banking performance is much better than the conventional banking system.
Dubai financial market was also crashed by the global financial crisis. In Dubai
financial market the portfolio investment was declined from 42% post to 18% pre crisis period,
resulted as debt crisis in Dubai which seemed on apparent in 2009.In June 2008 the Dubai stock
market fell down. 57 companies were listed on Dubai financial market (DFM) with an
accumulated market capitalization of about $360 billion at 2007 (Oxford 2008). On 13
November 2007 the Dubai financial market index (DFMGI) closing price was 5489.37 with
change -2.22%. At31 January 2008Dubai financial market index (DFMGI) closing price was
5615.95 with change 1.49% but at 21 December 2008 (DFMGI) closing price was 1793.26
with change -5.58%.
The real estate sector is one of the emerging sectors in Dubai’s economy. More than
29% of total GDP of UEA contributed by Dubai, the most important construction and real
estate sectors is contributing almost 23% of GDP of Dubai. Financial sector of Dubai was
contributing 11% of GDP while exports and other trade contributed 31% of total GDP. In Dubai
property sharply slump due to global recession. Home prices were declined 50% from their
peak in 2008. Most of the speculator acquired loans from the banks to put down 10% of the
property that had not until now been built, and offhand it for massive profit to any other buyer.
This act was more favorable when property prices surging quickly. Dubai government
authorities were putting restrictions to reduce this speculation. According to the oxford, 2008
in real estate government of Dubai had hampered on liabilities $59 billion and total debt was
10
$80 billion in few weeks. Banks were restricting on loaning, business finance and construction
plans to solve the problems of credit Emirates central bank had US$13.6 billion on 22
September, 2008. The shortage of the credit reduced demand for mortgage. Real estate broker
was waiting for favorable span. Dubai had a massive budget and surpluses into the current
account. The Emirates alliance government was willing to use these surpluses to tackle the
credit problems.
2.3 The global financial crisis 2007-08 impact on Pakistan economy
When the global financial crisis 2008 effect came into Pakistan, Pakistan economy was
also facing an entire financial crisis at that time since 2006 due to some economic reasons.
Pakistan economy was suffering budget deficit of about 7% of GDP. The trade deficit of
Pakistan economy was more than $15 billion a year, current account deficit widened from 4.7%
of GDP in 2007 to 8.4% in 2008 (Amjad and Din 2010). The deficit in the balance of payment
(BOP) 2007 to 2008 existed in the economy due to increment in the oil prices and prices of
other commodities. The Oil prices streamed in this fiscal year from $55 per barrel in Jan, 2007
to those above $130 per barrel in May, 2008. This increment was more than 136 %. Owing to
there was a dreadful decline in the foreign exchange reserve. Sumra (2014) explored that
Instead of financial crisis the internal factors such as bad governance, energy crisis and absence
of effective market strategies were the main causes of down fall in textile industry in Pakistan.
McCartney (2011) pointed out that Pakistan economy was affected by some external factors
but the economic growth determined domestic dynamic factors. Umar (2011) examined that
Pakistan economy faced five financial crises (1958, 1974, 1979, 1997 and 2008) out of four
significantly affected Pakistan economy besides that internal factors also impacted the
Pakistan economy badly. Also explored that Pakistan economy was less effected from global
financial crisis 2008 due to the lack of global integration and small structure.
11
Pakistan had experienced food inflation, about 17.5% in during 2007-08 and 26.6% in
2008-09. The overall inflation in 2008-09 was 22.4%. Pakistan economy was trapped in vicious
circle of stagflation (Rukhsana et al. 2012). International monetary fund (IMF) announced a
Stand-By Arrangement of $7.6 billion in 24 Nov 2008 and released first installment of $3.1
billion.
Banking system of Pakistan remained unaffected since 2007, though economy was
suffering with weaken macroeconomic environment. The overall bank deposits dropdown from
RS3.77 trillion in Sep, 2008 to RS3.67 trillion in Oct 2008 approximately decrease of RS0.10
trillion. State bank of Pakistan increased the rate of interest on three month Treasury bill In
January 2008 from 9.09% to 14% in January, 2009 furthermore, discount rate also increased
20%. The shortage of liquidity resulted in crowding out of private sector of the economy.
Government of Pakistan increased rate of taxation from 10% to 10.5% of the GDP that
augmentation in the taxation rate slowdown the economic growth. The crowding out from the
private sectors and increment in taxation were resulting as the rise in unemployment.
In 2008 the law and order situation was deteriorating, due to this foreign direct
investment (FDI) declined $8 billion to $3.5 billion. In War against terror Pakistan got almost
$12 billion through Coalition Support Fund (CSF) in which the part of economic aid was only
$3.1 billion and approximately $9 billion for the maintenance of law and order. The foreign
exchange reserve of Pakistan were reduced significantly from $8 billion at June, 2008 to $6.4
billion in November, 2008 (Jamal 2009). The decline in the foreign exchange reserve in
between these months was more than that of whole year 2008. Finally, the inflow of $3.1 billion
from (IMF) through macroeconomic stabilization program increased foreign reserve capacity
(IMF 2008). Imports recovery ratio of Pakistan was decreased at scratchy level of 16.8% in
June 2008 to 9.1% in October 2008 but it upgraded to 12.4% after (IMF) assistance.
12
In 2008, Pakistani rupee depreciated between March and December against US dollar
approximately 21%. There were many other factors like administrative uncertainty, reduction
in foreign direct investment, trade deficit and speculative activities which enhanced the
depreciation. This depreciation improved with (IMF) aid, but rupee again depreciated 13.5%
from July, 2008 to February, 2009.
2.4 Possible channels of information transmission
Previous studies found following possible channels through which global financial crises
transmitted into Pakistan economy. Global financial crisis impacted on developing countries
including Pakistan through four major channels. These are stock market, trade in commodities,
remittances and capital flows (Amjad and Din 2010). The global crisis also caused dipping
remittances, exports, stock markets decay, flight of capital and local currencies depreciation.
This global crisis impacted critically on foreign direct investment, exports and portfolio
investment (Iqbal 2010).
Direct channels
1. Failure of the banking system
When the financial crises affected the banking system of the developed countries as a result
the wave of crises also reached in the banking sector of developing countries including Pakistan
and due to that spillover, financial institutions of developing countries were negatively
exaggerated. Nazir et al. (2012) explored that financial institutions of the Pakistan especially
the commercial banks were directly affected by the financial crises and also significantly
compressed their market policies, working strategies and financial arrangements. Phulpoto et
al. (2012) examined that in Pakistan during the global recession 2008 the Islamic banking
system less affected than the conventional baking system.
13
2. Reduced earnings from exports
When the recession came into the advance economies, it badly affected the exports of the
developing countries like Pakistan. The total exports reduced from $20,427 million in 2008 to
$19,121 million in 2009 (SBP 2009).USA and UEA had a major part of Pakistan’s exports. In
2007-08 Pakistan exports to US around $3719.4 million, UEA $2071.0 million and UK
$1030.0 million. In 2008-09 exports to these countries were US $3339.5 million, UEA $1470.0
million and UK $874.6 million. This reduction into exports owed to fall into the prices of
commodities at international level, and reduce demand of commodities. While Pakistan was
tackling shortage of energy and high inflation which were causing prices of commodities high.
Abdul Latif et al. (2011) investigated that the export of Pakistan reduced mainly due to down
fall in world trade prices rather than the volume of trade.
3. Fall in financial inflows to Pakistan
In period of financial crises Pakistan needed financial inflows in different shape from other
countries to improve economic growth. Financial inflow towards the developing countries
reduced approximately US$300 billion. Investors shifted Portfolio investment and foreign
direct investment to those countries which were not affected by global financial crises (Cali et
al. 2008). Global trade mostly depends upon the trade credit and in traditional trade 90% short
term credit policies were adopted. Trade credit was going to reduce due to the shortage of
capital reserve in banks and they tried to short their risk exposure. Reduction in the trade credit
was also a problem of international trade.
4. Remittances reduction
Remittances are also a big source of capital inflow for developing countries. Pakistan
receive $7.8bn form expatriates were working in developed countries in fiscal year 2009. About
70% of the total remittances were coming from four countries as US, UAE, UK and Sudia
14
Arabia (SBP 2008). In 2007 the remittances were increased US$240 billion towards developing
countries, it was more than of double of the aid which developing countries receive (Ratha et
al. 2007).
5. Stock market prices
At 26 December 2007 Karachi Stock Exchange index KSE-100 closed at its highest points
14,814.85 with accumulated market assets of RS4.57 trillion. Accumulated stock market
capitalization was only RS1.85 trillion with points 5,865.01 at 31 December 2008 in term of
dollar $23 billion. More critical situation arose at 23 January 2009 when stock market index
had only points 4,929.54 with total capitalization of RS1.58 trillion in term of dollar $20 billion
(Din et al. 2010). It means the loss of approximately 65% from highest point of 26 December
2007. According to the report of state bank of Pakistan in 2006-07 the foreign direct investment
in KSE listed companies was $500 million. Moody’s investor service reviewed the credit rating
policies at 5 November 2007. All this happened due to global financial crises and internal issues
like political instability, insecurities in the market, bad governance, doubting on
macroeconomic policies, budget deficit and reservations on the coming general election of
Pakistan, stagflation and reduction in the foreign reserves.
 Indirect channel
1. Decline in the internal loaning
The major fact was that with the stock market the housing prices also decreased. The
demand for loans increased. This reduced capital reserves of banks. It was a problem for large
scale firms who had not enough amounts in their hand and banks were not ready to provide
huge amount of capital. Banks had only one option i.e. to reduce the loaning otherwise faced
the problem of bankruptcy. This situation triggered other problems like reduction in the loans,
15
declined investment, worse growth and expend the unemployment. All this resulted in less
government revenue.
2.5 Previous empirical studies
Global financial crisis is one of the major factors which have shifted concentration on
the dynamic linkages between the financial markets. Owing to the dynamic linkages the
information transmission is also existed between financial markets. The information
transmission mechanisms were quantified through returns and volatilities Padhi and Lagesh
(2012). Mishra and Mukherjee (2008) exposed that volatility spillover effect and linkages
between India and other twelve emerging Asian and developed countries markets. Jeyanthi
(2010); Pankaj and Gyanesh (2010) investigated the relationships and volatility spillover effect
between India, UK, Japan and USA, incorporating the structural change; Concluded that the
Japan and USA stock market’s volatility impacted Indian stock markets. Abd Karim et al.
(2010) examined that volatility spillover between five countries of ASEAN, Japan and USA.
They found that USA stock market has more mean and volatility spillover effect on ASEAN
stock markets as compare to Japan stock market. Li and Giles (2013) inspected the spillover
effect from US stock markets to Japan and other six Asian developing countries stock markets.
They found uni-directional return and volatility spillover from US to other markets. Gahlot
(2013) found the bidirectional volatility spillover effect and causality from USA stock market
to Indian stock market.
Alikhanov (2013), Grobys (2010) examined the volatility spillover effect between the
eight European stock markets and oil price market, found a strong spillover effect US to
European stock and oil markets. Owing to economic liberalization and interconnected
synchronization equity and financial markets have impacted co-moments between the financial
markets. Sugimoto et al. (2013) investigated the volatility spillover effect from US to African
16
countries equity markets. They concluded that the African markets are badly impacted by the
global financial crisis and foreign exchange markets. Wongswa (2006) studied that there was
strong indications of transmission of information from US and Japan to Thailand and Korea.
Due to the information transmission there was co-moment between the markets and also
revealed the transmission from developed to emerging equity markets. Yang and Doong (2004)
explored the relationship between the stock market prices and the foreign exchange market
prices and concluded that stock market prices impacted the future foreign exchange market
prices but the reverse may not be possible in case of G-7countries. Choi et al. (2009) examined
the integration between the exchange market and stock market in case of New Zealand. They
found that there was volatility spillover effect New Zealand stock market return to exchange
market.
All the studies which we have reviewed in this chapter conclude that the global financial
crisis 2008 primarily originated due to the creation and expansion of bubbles in housing and
subprime markets of US. It triggered imbalances in US economy and also in those economies
which were directly and indirectly integrated with US economy. Pakistan and Dubai economies
also impacted by global financial crisis 2008 at some extend. The previous empirical studies
also explored that the global financial markets affected due to global financial crisis 2008. At
that time co-movements were also observed between the financial markets (particularly stock
markets) due to direct and indirect linkages with global financial system. Studies on Pakistan
we reviewed it is clear that the effect of global financial crisis 2008 transmitted into Pakistan
economy through four main linkages, one of them is stock market. In this study we trace out
direct and indirect linkages between Pakistani and leading foreign stock markets.
17
Chapter 3
MODELING VOLATILITY AND EXPLORING SPILLOVER

EFFECT
Volatility modeling is one of the issues addressed by financial econometrics.
Predictability of time varying volatility is the elementary purpose of financial econometric
modeling. In financial markets risk is a synonym of volatility. In econometrics the main
purpose of modeling of time series is to estimate the conditional mean, some theoretical models
are used to estimate conditional variance, it is also known as volatility. These models are
employed to analyze the historical behavior of volatility, future prediction of volatility,
examining series of asset return by considering volatility clustering, leverage effect and
persistence.
The spillover effect also termed as co-movement or information transmission.
Exploring spillover effect means to investigate whether return or volatility of one financial
market affects the other market’s return or volatility. It is quite important to analyze
information transmission between the financial markets, due to information transmission the
dynamic linkages developed between the financial markets. Padhi and Lagesh (2012) found
that Information transmission mechanisms persists through return and volatility, it plays a
significant role in determining the distribution and financial integration across the global
financial markets. The understanding and predictability of spillover effect and volatility
modeling are significantly important for asset allocation, strategies of global hedging and
pricing of internal securities.
18
The volatility modeling and tracing its spillover effect have been studied in the financial
econometric literature in case of Pakistan. Most of the studies considered the direct linkages
through which the return and volatility spillover effect transmitted from global financial
markets to Pakistani financial markets. These studies empirically explored direct linkages
through which global financial crisis impacted Pakistani stock markets [Ali 2012; Zia-Ur-
Rehman et al. 201; Khalid and Rajaguru 2010; Attari and Safdar, 2013; Gulzar et al. 2014;
Tahir et al. 2013; Khan et al. 2013]. [Qayyum and Khan 2014; Qayyum and Kemal 2006;
Khalil et al. 2013; Zia and Zahid 2011; Hassan et al. 2014; Rahman and Jasmin 2009]
investigated the volatility spillover effect from foreign exchange market to Pakistan stock
markets by using different econometric tools. These studies also analyzed historical behavior
of foreign exchange market. These studies only investigated direct dynamic linkages between
the financial markets (stock) no one find out indirect information transmission. In this study
we find out direct linkages between the leading global stock markets and also indirect dynamic
linkages between Pak-US stock markets through Dubai financial market.
3.1 Econometric Methodology and Model Specification
To describe the variation of conditional variance with respect to time, Engle (1982) proposed
Autoregressive conditional hetroscedastic (ARCH) model. Although ARCH model is a
substantial contribution in econometric tools, it has some problems like long lag length and
non-negativity restriction on parameters. Bollerslev (1986) introduced generalized
autoregressive conditional heteroskedastic (GARCH) model, which improves the unique
specification with the addition of lag value of conditional variance, which acts like smoothing
term. GARCH model cannot analyze leverage effect. For this Glosten, Jagannathan & Runkle
(1993) proposed GJR-GARCH model. GJR-GARCH model is a significant extension of
standard GARCH model, it contains asymmetric term in conditional variance equation.
19
There are dozens of univariate and multivariate (ARCH) type model. The DCC-
MGARCH model captures the dynamics of the conditional correlations and volatility spillover
effect Hwang et al. (2009). The multivariate GARCH and Asymmetric-BEKK models are
better to identify the spillover effect Yanan et al. (2013). Sayani et al. (2014) used multivariate
GRACH model to explore spillover effect. They used Islamic and conventional indices data
and investigated that Islamic indices are less risky than conventional indices. Parameter
convergence is a serious problem in MGARCH models, as a single specification can increase
many parameters.
To avoid any non-convergence problem in this study we employ appropriate univariate
GARCH type model such as GARCH (p, q) and GJR-GARCH (p, q) to estimate volatility
models and to explore mean and volatility spillover effect. Following the technique of Hamao
et al. (1990) we explore spillover effect between Pakistani and foreign stock markets. The
GARCH (p, q) and GJR-GARCH (p, q) Univariate models are capable of exploring better
volatility dynamics.
The financial series at level are trendy in nature. It is impossible to estimate a robust
model if the series is trendy. To deal with trend we used the log difference return.
𝑅𝑡 = 𝑙𝑜𝑔𝑒 (𝑙𝑡 /𝑙𝑡−1 )
𝑙𝑡 = Financial time series at level i.e. stock indices and exchange rates at the end of time t.
𝑙𝑡−1= First lag of financial time series.
Granger and Andersen in (1978) anticipated that the conditional variance depends upon the
predicted past value of return series.
𝛾𝑡 = 𝜀𝑡 𝑟𝑡−1 …………………………………………. (3.1)
The conditional variance is
20
𝛾𝑡 2
𝑉 (𝑟 ) = 𝜎 2 𝑟𝑡−1 ………………………………………… (3.2)
𝑡−1
There is no restriction for unconditional variance, either it is unspecified or zero. Then another
famous approach came at front to find the ARCH effect in return series.
3.1.1 ARCH (q) Model
Robert F. Engle in (1982) introduced the Autoregressive conditional hetroscedastic
(ARCH) model. This model overcomes all short comings which exists in previous models. In
this model Engle, introduced conditional mean and conditional variance equations. Empirically
the conditional mean equation follows ARMA (p, q) process and the conditional variance
depends upon the square of past values of error process 𝜀𝑡 .
The general description of ARCH model is
Conditional mean equation
𝑅𝑡 = 𝛼0 + 𝛽𝑋𝑡 + 𝜀𝑡 …………………………………………. (3.3)
Where 𝜀𝑡 ~𝑁(0, 𝜎𝑡2 )
Conditional variance equation
𝑞
𝜎𝑡2 = 𝜃0 + ∑𝑖=1 𝜃𝑖 𝜀𝑡−1
2
…………………………………………. (3.4)
Where 𝜃0 > 0, 𝜃𝑖 ≥ 0 𝑖= 1,2,…….., q
In conditional mean equation Rt represents the return which is linear function of Xt.
𝑤ℎ𝑒𝑟𝑒 𝛽 shows the vector of parameters. Empirically 𝛽𝑋𝑡 illustrates ARMA (m, n) process
with different specifications. In some cases it may be ARMA (0, 0). According to the “Efficient
Market Hypothesis (EMH)” Rt represents mean reversion behavior and it is unpredictable. In
conditional variance equation the restriction on coefficients is that they must be non-negative.
21
𝜎𝑡2 Represents conditional variance which depends upon lags of squared past value of 𝜀𝑡
process.
3.1.2 GARCH (p, q) Model
Linear ARCH (q) model has some problems first, sometime takes long lag length ‘q’
due to this number of parameters are going to increase as result loss of degree of freedom.
Second, non- negativity condition of parameters. Bollerslev (1986) proposed generalized
extension of ARCH (q) model Generalized autoregressive conditional hetroscedastic
(GARCH) model.
The general description of GARCH model is
𝑅𝑡 = 𝛼0 + 𝛽𝑋𝑡 + 𝜀𝑡 …………………………………………. (3.5)
𝑞 𝑝
𝜎𝑡2 = 𝜃0 + ∑𝑖=1 𝜃𝑖 𝜀𝑡−1
2 2
+ ∑𝑖=1 𝜑𝑗 𝜎𝑡−1 ………………………………………. (3.6)
Where 𝜃0 > 0,𝜃𝑖 ≥ 0, 𝜑𝑗 ≥ 0
In GARCH (p, q) model the conditional variance depends upon square of past values of
2
process𝜀𝑡 and lag of conditional variance𝜎𝑡−1 . The condition of non-negativity of parameter
also applied in this model.
22
3.1.3 GARCH (1, 1)
GARCH (1, 1) is frequently used in financial econometric literature for volatility
modeling. Sajid et al. (2012) employed ARMA-GARCH for measurement of inflation and
inflation uncertainty. Jabeen and Saud (2014) employed GARCH model to find out “Exchange
rate volatility by macroeconomic fundamentals in Pakistan”. The GARCH (1, 1) is the modest
form in dispersion models family. The GARCH (1, 1) provide most robust estimations than
other volatility models. GARCH (p, q) mostly use when data is very large and require higher
lags.
The general representation of GARCH (1, 1) is
𝑅𝑡 = 𝛼0 + 𝛽𝑋𝑡 + 𝜀𝑡 ………………………………. (3.7)
𝜎𝑡2 = 𝜃0 + 𝜃1 𝜀𝑡−1
2 2
+ 𝜑1 𝜎𝑡−1 ………………………………. (3.8)
Where 𝜃0 > 0, 𝜃1 ≥ 0, 𝜑1 ≥ 0
These are the restrictions𝜃0 > 0, 𝜃1 ≥ 0, 𝜑1 ≥ 0of non-negativity on coefficients
of conditional variance equation. Here 𝑅𝑡 represents the return series of stock market index.
Equation (3.7) the conditional mean equation follow ARMA (p, q) process. 𝜀𝑡 is Error series
with normal distribution of zero mean and 𝜎𝑡2 conditional variance. GARCH (1, 1) model
(Bollerslev, 1986) established statistical properties for unconditional moment of residual (𝜀𝑡 ).
(𝜃𝑖 + 𝜑𝑗 < 1) is sufficient and necessary condition represents the persistence of shock to
volatility, it satisfies the wide sense stationary condition. The unconditional variance is Var
23
(𝜀𝑡 ) = E (𝜎𝑡2 ) = (𝜃0 / 1-𝜃1 -𝜑1 ). The fourth moment of error (𝜀𝑡 ) is ( 3𝜃12 + 2𝜃1 𝜑1 + 𝜑12 ). It is
sufficient and necessary condition and the kurtosis is more than 3.
3.1.4 Asymmetric GARCH models
Simple GARCH type models deal with the symmetric effect of bad and good news on
volatility. These models do not take into account the asymmetries which are associated with
the distribution. In financial econometrics literature Asymmetric GARCH type models
consider the asymmetries of response to bad or good news. Asymmetric GARCH models
account for leverage effect. The leverage effect indicates the negative correlation between the
assets returns and the volatility of the assets return (Black 1976), means the magnitude of bad
and good news are different.
Engle and Victor (1993) conducted a brief discussion on how univariate GARCH type
model capture the impact of bad news. They have used Japan stock market data. They argued
that the GJR model is the best model to capture the asymmetries. According to them EGARCH
model capture the Asymmetries but when we employ the EGARCH model the standard
deviation is going too high, as compare to GJR model. They also concluded that GJR model
is best for capturing the asymmetries.
3.1.4.1GJR-GARCH (p, q) Model
Glosten, Jagannathan and Runkle introduced (GJR) model in 1993. GJR model is a
significant extension in simple GARCH model. This model also capture the asymmetries in
ARCH process. GJR model also account for the leverage effect in a financial series.
The general representation of the GJR model is:
24
𝑅𝑡 = 𝛼0 + 𝛽𝑋𝑡 + 𝜀𝑡 …………………………. (3.9)
𝑞 2 𝑞
2 2𝑝
𝜎𝑡2 = 𝜃0 + ∑𝑖=1 𝜃𝑖 𝜀𝑡−𝑖 + ∑𝑖=1 𝛿𝑖 𝜀𝑡−𝑖 𝐺𝑡 + ∑𝑖=1 𝜑𝑗 𝜎𝑡−𝑗 ………………. (3.10)
Where 𝜃0 > 0, 𝜃𝑖 ≥ 0, 𝜑𝑖 ≥ 0
0 ≤ 𝛿𝑖 < 1 Range of leverage effect parameter.
Gt = 1 when 𝜀𝑡−1 < 0 and Gt = 0 when 𝜀𝑡−1 ≥ 0
Gt = 1 when 𝜀𝑡−1 < 0 illustrates bad news or the negative shock and Gt = 0 when
𝜀𝑡−1 ≥ 0 indicates good news or positive shock. GJR model also shows that bad news has more
impact (𝜃𝑖 + 𝛿𝑖 ). The good news has less impact (𝜃𝑖 ). If the 𝛿𝑖 > 0 means that there is leverage
effect and shows that response to shock is distinct. If the 𝛿𝑖 = 0 means symmetric response to
𝛿
distinct shock (In other words both news have same impact). Condition (𝜃𝑖 +𝜑𝑖 + 2𝑖 <1) shows
the persistence of shock.
3.1.4.2 GARCH (p, q) Model (For Exploring Spillover Effect)
Conditional Mean equation
R t,k = α0 + βXt + π1 R t,s + εt ……………………………………. (3.11)
Conditional Variance equation
q p
σ2t,k = θ0 + ∑i=1 θi ε2t−1 + ∑i=1 φj σ2t−1 + π2 R2t,s …………………………………. (3.12)
25
R t,k Shows the return series of K market. R t,s Describes the return series of S market
which is used as a regressor in conditional mean equation of K markets return series. π1 ,
Represents the parameter of S market returns series. σ2t,k , Denotes the conditional variance of
K market. R2t,s Indicates the squared return series of S markets which is used as a regressor in
conditional variance equation of K markets return series. π2 , Demonstrates the parameter of
squared return series of S market.
We trace out the co-movements among these markets by following the technique of
Hamao et al. (1990). According to Hamao et al. (1990), the residuals of one return series
introduce as a regressor in conditional mean equation of other return series for mean spillover
effect. For volatility spillover effect the squared residuals of one return series introduce as a
regressor in conditional variance equation of other return series.
In this instead of using residuals and squared residuals we use return series and squared
return series. According to “The Efficient Market Hypothesis (EMH) return are unpredictable
and show mean reversion behavior”. To check the mean spillover effect between two series,
the return series of one market is introduced as regressor in other market return series. For
volatility spillover affect the square of return series of one market is introduced as regressor in
the conditional variance equation of other market.
3.1.5 Residual Analysis
To identify the good fitness of employed model we use post estimation results (Residual
analysis). The Jarque Bera test (Normality test) employs to check the null hypothesis that
distribution of return series is normal. Q-stat (return series) employs to validate the null
hypothesis, there is no serial autocorrelation in standardized residuals. Q2-stat (return series)
checks the null hypothesis, there is no serial autocorrelation in squared standardized residuals.
26
LM-ARCH with the Null hypothesis, there is no ARCH effect in return series. Due to
convergence problem we check Q-stat and Q2-stat up to 10th lag. LM-ARCH test up to 5th lag.
3.2 Methodological Structure
First the graphical analysis is used to understand the behavior of financial time series.
The descriptive statistics are used to investigate the characteristic of the return series. After that
models are employed to Model volatility and exploring spillover effect. The residuals analysis
is carried out post estimation for the validity of employed model.
3.3 Description of Data and sources
The daily data of stock market indices are used form 2005 to 2014. These stock markets
are taken from ASIA, Europe, America and Gulf countries. From US S&P 500, DOW JONES
(DJI), and NASDAQ 100 are used. From EU London (FTSE 350) and German (GDAXI) stock
exchange data are taken. From Asia Pakistan (KSE 100), Japan (NIKKEI 225) and Hong Kong
(HIS) stock market indices are used. Dubai financial market index (DFMGI) is taken from Gulf
countries.
Variables Sources
Pakistan, Official website of Karachi Stock Exchange
KSE 100 (www.kse.com.pk)
US, Official website of S&P 500 Stock Market
S&P 500 (www.us.spindices.com)
US, Official website of NASDAQ 100 Stock Market
NASDAQ 100 (www.nasdaq.com)
US, Official website of DOW JONES Industrial Averages
DJI (www.djaverages.com)
Japan, Official website of NIKKEI Stock Market
NIKKEI 225 (www.indexes.nikkei.co.jp)
Hong Kong, Official website of HANG SENG Stock
HSI Exchange(www.hsi.com.hk)
UK, Official website of London Stock Exchange
FTSE 350 (www.londonstockexchange.com)
Germany, Official website of German stock market
GDAXI (www.dax-indices.com)
27
Dubai, Official website of Dubai Financial Market
DFMGI (www.dfm.ae)
Official website (www.finance.yahoo.com)
Official website of Pak finance
(www.pkfinance.info.com)
Official website of Market today
(www.marketstoday.net)
State bank of Pakistan (SBP) bulletins
OTHERS Survey reports of ministry of finance of Pakistan
Survey reports of ministry of finance of UAE
IMF annual reports
28
Chapter 4
ESTIMATIONS AND ANALYSIS
In this chapter the estimations and analysis are presented. First, the graphical analysis
is presented to describe the performance of financial time series in section (4.1, page 29). The
summary statistics are used to investigate the characteristic of the return series (4.2, page 36).
In section (4.3, page 38) univariate GARCH type models are employed for volatility modeling
of stock market return series [KSE 100 (Pakistan), NASDAQ 100 (US), DOW JONES (US),
S&P 500 (US), GADXI (Germany), NIKKEI 225 (Hong Kong), HIS (Japan), DFMGI (Dubai)
and FTSE 350 (UK)]. To validate the good fitness of employed model used residual diagnostic
test. After that in section (4.4, page 43) we explore volatility spillover effect to find out direct
and indirect linkages between Pakistani stock market and leading foreign stock markets by
using whole data set 3rd Jan, 2005 to 28th Nov, 2014. In this section we do not explore the co-
movement between Pakistani, US and Dubai financial markets. We trace out the co-movements
among these markets by following the technique of Hamao et al. (1990). We explore the
information transmission between individual markets to check the ‘Meteor Shower’ hypothesis
of Engle. Accordingly, to check the mean spillover effect between two series, the return series
of one market is introduced as regressor in other market return series. For volatility spillover
affect the square of return series of one market is introduced as regressor in the conditional
variance equation of other market. The residuals analysis is carried out post estimation for the
validity of employed model. In section (4.5, page 83), (4.6, page 92) and (4.7, page 100) we
explore the information transmission between Pakistani (KSE 100), US (NASDAQ 100, S&P
500 and DJI) and Dubai (DFMGI) stock markets.
In the same way to meet our second research question, In section (4.5, page 83 ) we
explore volatility spillover effect between Pakistani, US and Dubai stock markets and find out
29
direct and indirect linkages between them. In this section we investigate spillover effect among
these markets by using whole data set i.e. daily data from 3rd Jan, 2005 to 28th Nov, 2014. To
establish the sharp cut spillover effect we divide the data in two sub groups before and after
the 2008 crisis. In next step we separate data set into 2 parts before and after the 2008 crisis. In
this step we particularly trace out the direct linkages between Pakistani and US stock markets
and indirect linkages between both markets through Dubai financial market. In section (4.6,
page 92) we use first sub group of data includes the era of global financial crisis 2008 from 3rd
Jan, 2005 to 31st Dec, 2009 because the global financial crisis started from mid-2007, 7th Sep,
2008 at its boom and eliminates its effect until end of 2009. Second sub group after global
financial crisis period from 4th Jan, 2010 to 28th Nov, 2014 used in section (4.7, page 100).
4.1 Graphical Analysis
Figure 4.1.1.a Graphs of series at level of stock indices
KSE100 NIKKI_225
30000
15000
20000
10000 10000
0 400 800 1200 1600 2000 2400 0 400 800 1200 1600 2000 2400
30000 HSI S&P500

2000
1500
20000
1000
0 400 800 1200 1600 2000 2400 0 400 800 1200 1600 2000 2400
DJI
15000
10000
0 400 800 1200 1600 2000 2400
30
Figure 4.1.1.b Graphs of series at level
NASDAQ_100 FTSE_350
4000
3000 3000
2000
2000
0 400 800 1200 1600 2000 2400 0 400 800 1200 1600 2000 2400
10000 GDAXI DFMGI

7500
7500
5000
5000 2500
0 400 800 1200 1600 2000 2400 0 400 800 1200 1600 2000 2400
Forex
100
75
0 400 800 1200 1600 2000 2400
Figure 4.1.1.a and 4.1.1.b given above show in the beginning all series have upward
trend than sharp decline and then again there is an upward trend continuously. This means that
series are trendy at level. In figure 4.1.1.a the series are Karachi stock market (KSE 100),
Nikkei 225, Hang Seng (HIS), Standard and Poor (S&P 500) and Dow Jones. In figure 4.1.1.b
the series are Nasdaq 100, FTSE 350, GDAXI, Dubai financial market (DFMGI). Daily data
is used from 3rd Jan, 2005 to 28th Nov, 2014.
Figure 4.1.2 given below represents return series of Karachi stock market indices. It is
impossible to find out robust model if the series is trendy, we use log difference return series
to deal with trend. In financial econometrics, spread characterized as volatility. In return series spread
31
does not remain constant, it is known as Hetroscedasticity. The circles in figure 4.1.2 are indicating
the low and high volatility which denote the spread autocorrelation. According to “The Efficient
Market Hypothesis (EMH) return are unpredictable and show mean reversion behavior”. That’s why
all return series have mean reversion behavior. If we combine all effects it indicate ARCH (Auto-
Regressive Conditional Hetroscedasticity) effect. We can easily distinguish between low volatility
clustering and high volatility clustering period. The greater depreciation from constant level
(mean of return series) indicates high volatility clustering and less depreciation illustrate low
volatility clustering. In the same way we can plot and analyze return series of other stock
markets.
Figure 4.1.2 Graph of given return series
High volatility
0.08 DLKSE100 clustering
0.06 High volatility

clustering
0.04 Low volatility

clustering
0.02
0.00
-0.02
-0.04
-0.06
0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400
In figure 4.1.3 given below shows squared returns series of KSE 100. The graph of
square return series have “spiky” look signifying variation in square return. Circles indicate
32
high volatility and low volatility. It also shows that extreme values (outliers) of return series
contribute more to the high volatility. Square of the return series is also known as variance of
the return series means these graphs illustrate the dispersion. In the same way we can analyze
square return series of other stock markets’ indices.
Figure 4.1.3 Graph of Squared return series
High volatility
0.007
DLKSE100^2
clustering
0.006
0.005 High volatility

clustering
0.004
Low volatility
0.003 clustering
0.002
0.001
0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400
The figure 4.1.4 given below illustrates the distribution of the return series. The
distribution of return series is non-normal. In this graph green line shows the normal reference
distribution of return series. The red line indicates the actual distribution of the return series.
Histograms describe the outliers (extreme values) in return series. The distribution of return
series have heavy tails and is leptokurtic. This all is due to different response of market players
by having same information from the same market.
33
Figure 4.1.4 Graphs distribution of the return series
Density
70
DLKSE100 N(s=0.0133)
60
50
40
30
20
10
-0.06 -0.05 -0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09
In figure 4.1.5 given below presents ACF (Auto-correlation function) and PACF
(Partial Auto-correlation function) of return series. The green straight lines in this graph show
95 percent confidence interval, if any bar of ACF and PACF outside these lines means at that
lag the values are auto correlated in other words significantly vary from zero. The ARMA (p,
q) process specify through the significant lags of ACF and PACF. The ACF specify the MA
(q) process PACF specify the AR (p) process. In this graph 1st, 2nd, 3rd, 4th, 10th, 17th and 18th
lags of ACF are significant and 1st, 3rd, 4th, 10th, 11th, 12th, 17th and 18th lags of PACF are
significant, these lags format ARMA (p, q) process in conditional mean equation. It means
34
auto correlation and partial autocorrelation exist in the return series. We can also analyze
cyclical behavior in return series through ACF and PACF graphs.
Figure 4.1.5 Graphs of ACF and PACF of return series
1.00
ACF-DLKSE100 PACF-DLKSE100
0.75
0.50
0.25
0.00
-0.25
-0.50
-0.75
0 5 10 15 20
Figure 4.1.6 given below show the graph of ACF and PACF of square return series. 1st
to 20th lags of ACF are significantly differ from zero and 1st……...8th, 10th, 13th, 14th, 19th and
20th lags of PACF are statistically significant. In the same manner square return series ACF
and PACF may provide an indication about the critical lags in conditional variance equation
structure of GARCH (p, q) model. Means there is autocorrelation and partial autocorrelation in
the square return series.
35
Figure 4.1.6 Graphs of ACF and PACF of square of return series
1.00
ACF-DLKSE100^2 PACF-DLKSE100^2
0.75
0.50
0.25
0.00
-0.25
-0.50
-0.75
0 5 10 15 20
The initial statistics of return series of stock markets indices are given below unveil
some indications about the behavior of stock markets. The distributions of return are non-
normal, heavy tails and leptokurtic. The mean of all return series are about zero which implies
that return series show mean reversion behavior. Standard deviation of return series describe
the dispersion from mean value which return series have greater standard deviation it means
more deviation from mean value. The skewness deals with the asymmetry of the distribution.
The distributions of KSE 100, S&P 500, NASDAQ 100, DJI, NIKKEI 225, FTSE 350 and
DFMGI return series are negatively skewed which means that the return of these stock markets
are less than average return. The distributions of HIS and GDAXI are positively skewed which
36
imply the returns of these markets are more than average return. The Jarque-Bera test with null
hypothesis of normal distribution is employed. Jarque-Bera statistics of all return series are
significant means the distribution of all return series are non-normal.
Table 4.2 Summary statistics
Summary statistics
Variables Mean Standard Skewness Jarque Excess Q-stat Q2-stat ARCH KPSS
deviation Bera Kurtosis (5) (5) 1-2
KSE 0.0006 0.0132 -0.3854 1098.1 3.1075 76.120 1167.51 266.88 0.2073
100 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
S&P 0.0002 0.0127 -0.3409 14088 11.448 45.484 1131.31 266.72 0.1965
500 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
NASDAQ 0.0003 0.0136 -0.1587 7985.9 8.6282 24.928 765.777 156.96 0.2005
100 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
DJI 0.0001 0.0116 -0.0851 14168 11.499 45.037 1123.85 283.89 0.1548
(0.077) (0.000) (0.000) (0.000) (0.000) (0.000)
NIKKEI 0.0001 0.0153 -0.5737 8597.9 8.8850 10.564 1396.71 489.45 0.1994
225 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
HIS 0.0002 0.0156 0.0459 10971 10.120 8.3870 1361.38 361.66 0.0525
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
FTSE 0.0001 0.0118 -0.1879 7288.8 8.2401 39.367 1130.0 147.29 0.0569
350 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
GDAXI 0.0003 0.0137 0.0297 5510.5 7.1719 16.783 686.71 111.39 0.07411
(0.537) (0.000) (0.000) (0.000) (0.000) (0.000)
DFMGI 0.0001 0.0183 -0.8778 13612 11.135 32.381 166.23 44.647 0.4874
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Null Hypotheses (All Null Hypotheses are for nth order)
KPSS H0: Return series is level stationary, Asymptotic significant values 1% (0.739), 5% (0.463),
10% (0.347). Q-stat (return series) there is no serial autocorrelation. Q2-stat (square return series) H0:
there is no serial autocorrelation. Jarque-Bera H0: distribution of series is normal. LM-ARCH H0:
there is no ARCH effect. Use these Asymptotic Significance values of t-stat 1% (0.01), 5% (0.05), 10% (0.1)
and compare these critical values with P-values (Probability values). P-values are in the parenthesis.
The Excess kurtosis of all returns series are significant which means that return series
distributions are leptokurtic and also indicates that probability of large values is more than
normal return series. Q-stat of return series are significant, rejecting the null hypothesis of no
37
autocorrelation return series. This shows that there is serial autocorrelation in return series. Q-
stat of squared return series are significant, rejecting the null hypothesis of no autocorrelation
in squared return series. This shows that there is serial autocorrelation in square return series.
LM-ARCH test validates that there is ARCH effect in return series. KPSS is a unit root test
with null hypothesis of stationary series. KPSS test results of all variable show that the
estimated values lies in acceptance region [less than given three significance values 1% (0.739),
5% (0.463), 10% (0.347)] means the null hypothesis is accepted, return series are level stationary.
4.3 Volatility Model specifications of Return Series

In this section volatility models of stock markets (area under study) are presented to
understand the Data Generating Process of all financial return series (area under study). It will
be helpful to understand the mean and volatility structure of financial return series (area under
study). Volatility modeling is a striking issue for market players, portfolio managers,
academicians and policy makers. A lot of empirical work on volatility modeling exists in
financial econometrics, the predictability and modeling of volatility is still a challenge for
researchers. Many researcher in their studies employed GARCH type model for volatility
modeling. In these studies researchers employed different ARCH-GARCH family models to
describe volatility modeling and volatility forecasting [Vijayalakshmi and Gaur (2013); Pasha
et al. (2007); Kamal et al. (2011); Khan and Parvez (2013); Chand et al. (2013); Jabeen and
Saud (2014); Sajid et al. (2012) and Faisal et al. (2012)]. In this study we employ GARCH (p,
q) and GJR-GARCH (p, q) model for volatility modeling and exploring spillover effect. The
GARCH and GJR models mostly employed by the researchers due to unique characteristics of
these models. These univariate models are best to give a better explanation of asset volatility
modeling.
38
The GARCH model is employed for HIS volatility modeling. The estimated conditional
mean equation (4.1) is from equation (3.5) and the estimated conditional dispersion equation
(4.2) is from equation (3.6). The P-values are in parenthesis.
𝑅𝑡 = 0.0005 …………………………………………………………… (4.1)

(0.0000)
σ2t = 0.0077 + 0.0523ε2t−2 + 2.0077σ2t−1 + 1.7204σ2t−2 + 0.6600σ2t−3 ... (4.2)

(0.0830) (0.0000) (0.0000) (0.0000) (0.0000)
The GJR-GARCH model is employed for KSE 100 volatility modeling. The estimated
conditional mean equation (4.3) is from equation (3.9) and the estimated conditional dispersion
equation (4.4) is from equation (3.10). The P-values are in parenthesis.
𝑅𝑡 = 0.0008 + 1.0000𝑅𝑡−1 − 0.9000𝜀𝑡−1 ……………………………… (4.3)

(0.0000) (0.0000) (0.0000)
σ2t = 0.0000 + 0.1460ε2t−2 + 0.3234ε2t−1 𝐺𝑡 + 0.8031σ2t−1 ……...….…… (4.4)

(1.0000) (0.0000) (0.0000) (0.0000)
The employed models in table 4.3.1 given below describe the data generating process
of the return series. The estimated parameters of the employed models are statistically
significant. In KSE 100 model AR (1) term is statistically significant which means that current
return of KSE 100 depends upon 1st lag. MA (1) term in this model is also differ from zero,
shows relationship between past and current variations. The leverage effect term 𝛿1 in KSE
100 and NIKKEI 225 models are significant, indicates that the current return negatively
correlated with future volatility, no leverage effect is found in HIS stock return series. Most of
the parameters are statistically significant at 5% level of significance. ARCH and GARCH
terms are also significant in three models means the return series are subject to ARCH effect.
The persistence of shock of the return series are KSE 100, NIKKEI 225 and HSI all are close
to 1 which means that the persistence of ARCH and GARCH effect take long time to decay.
39
Table 4.3.1 Volatility models of Asian Stock markets Return series
Return series KSE-100 NIKKEI-225 HIS

ARMA(1,1) GJR (1,1) ARMA(0,0) GJR (1,1) ARMA(0,0) GARCH (3,2)
Parameters
Conditional Mean Equation
Constant 0.0009 0.0006 0.0006
𝛼0 (0.7538) (0.0093) (0.0013)
AR(1) 1.0000
𝜗1 (0.0000) ----------- -----------
MA(1) -0.9000
∅1 (0.0000) ----------- -----------
Conditional Variance Equation
Constant 0.0000 0.0475 0.0070
θ0 (1.0000) (0.0007) (0.0830)
ARCH(1) 0.1460 0.0273
𝜃1 (0.0003) (0.0240) -----------
ARCH(2) 0.0523
𝜃2 ----------- ----------- (0.0000)
GARCH(1) 0.8032 0.8839 2.0078

𝜑1 (0.0000) (0.0000) (0.0000)
GARCH(2) 1.7205
𝜑2 ----------- ----------- (0.0000)
GARCH(3) 0.6600
𝜑3 ----------- ----------- (0.0000)
GJR(1) 0.3234 0.1287
𝛿1 (0.0000) (0.0000) -----------
Persistence of shock 1.1109 0.9755 0.9996

Null Hypotheses(All Null Hypotheses are for nth order)
AR (p) H0: 𝜗𝑖 = 0 No AR Process, MA (q) H0: ∅𝑖 = 0 No MA Process, ARCH H0: 𝜃𝑖 = 0 No ARCH effect,
GARCH H0: 𝜑𝑖 = 0 No GARCH effect, Leverage effect H0: 𝛿𝑖 = 0 No leverage effect. P-values are in the
parenthesis.
Residual Analysis
Parameter Jarque Q-Stat Q-Stat Q2-Stat Q2-Stat LM -ARCH LM-ARCH
Series Bera (5) (10) (5) (10) (1-2) (1-5)
128.04 1.7793 5.1972 5.2051 13.983 0.3964 0.5536
HIS (0.0000) (0.8787) (0.8776) (0.3913) (0.5267) (0.6727) (0.7356)
509.46 2.0605 6.0220 0.6678 5.1253 0.2047 0.1380
NIKKEI 225 (0.0000) (0.8407) (0.8134) (0.8807) (0.7440) (0.8149) (0.9835)
7.0272 0.0019 0.0024 0.0020 0.0039 0.0004 0.0004
KSE 100 (0.0000) (0.9999) (1.0000) (0.9999) (1.0000) (0.9999) (1.0000)
Q-stat (return series) there is no serial autocorrelation. Q2-stat (square return series) H0: there is no
serial autocorrelation. Jarque-Bera H0: distribution of series is normal. LM-ARCH H0: there is no
ARCH effect. P-values are in the parenthesis.
40
Tables 4.3.1 also illustrate the post estimation results (Residual analysis). The Jarque
Bera test (Normality test) results show non normal residuals. The Q-stat are insignificant up to
10th lags accept null hypothesis means no serial autocorrelation in the standardized residuals.
The Q-stat on squared standardized residuals are insignificant up to 10th lags accept null
hypothesis means no serial autocorrelation in squared standardized residuals. LM-ARCH test
is also insignificant up to 5thlags accept null hypothesis means no ARCH effect remain in series
residuals.
The employed models given below in table 4.3.2 describe the data generating process
of the return series. The estimated parameters of the fitted models are statistically significant.
In S&P 500 and DOW JONES models AR term is statistically significant which means that
current return of these market are only depends upon lag values. In NASDAQ model AR term
is insignificant which means current return of this market not depends upon lag values. MA (1)
term in S&P 500 and DOW JONES model is differ from zero, shows relationship between past
and current variations of return series. Most of the parameters are statistically significant at 5%
level of significance. ARCH and GARCH terms are also significant in three models means
these return series encompass ARCH and GARCH effect. The persistence of shock of the return
series are S&P 500, NASDAQ 100 and DOW JONES all are close to 1 which means that the
persistence of ARCH and GARCH effect take long time for decay.
Table 4.3.2 also illustrate the post estimation results (Residual analysis). The Jarque
hypothesis means no serial autocorrelation in squared standardized residuals.
41
Table 4.3.2 Volatility models of American Stock markets Return series
Return series S&P 500 NASDAQ 100 DOW JONES

Parameters ARMA(1,1) GARCH ARMA(0,0) GARCH (1,1) ARMA(2,1) GARCH (1,1)
(1,2)
Constant 0.0008 0.0011 0.0008
𝛼0 (0.0000) (0.0000) (0.0000)
AR(1) 0.7395 -0.9511
𝜗1 (0.0000) ----------- (0.0000)
AR(2) -0.0560
𝜗2 ----------- ----------- (0.0039)
MA(1) -0.7992 0.8958
∅1 (0.0000) ----------- (0.0000)
Constant 0.0219 0.0232 0.0130
θ0 (0.0000) (0.0000) (0.0005)
ARCH(1) 0.0890 0.1112
𝜃1 ----------- (0.0000) (0.0000)
ARCH(2) 0.1433
𝜃2 (0.0000) ----------- -----------
GARCH(1) 0.8481 0.8991 0.8841

𝜑1 (0.0000) (0.0000) (0.0000)

Null Hypotheses(All Null Hypotheses are for nth
order)
GARCH H0: 𝜑𝑖 = 0 No GARCH effect. P-values are in the parenthesis.
Residual Diagnostic Test
Series Bera (5) (10) (5) (10) (1-2) (1-5)
524.96 4.5842 7.4715 1.2465 9.7895 0.2319 0.2597
S&P 500 (0.0000) (0.2049) (0.4867) (0.5361) (0.2008) (0.7930) (0.9350)
213.74 2.8643 5.1957 5.8315 14.408 2.7272 1.1573
NASDAQ 100 (0.0000) (0.2387) (0.6360) (0.1201) (0.0717) (0.0656) (0.3279)
438.61 5.2781 8.9697 8.5410 17.475 3.8500 1.6483
DJI (0.0000) (0.0714) (0.2548) (0.0360)* (0.0255)* (0.0214)* (0.1438)
LM-ARCH test is also insignificant up to 5thlags accept null hypothesis means no ARCH effect
remain in series residuals.
42
Table 4.3.3 Volatility models of European and Gulf Stock markets Return series
Return series FTSE 350 GDAXI DFMGI

Parameters ARMA(0,0) GARCH ARMA(1,1) GARCH (1,1) ARMA(1,1) GARCH (1,1)
(1,1)
Constant 0.0006 0.0010 0.0005
𝛼0 (0.0000) (0.0000) (0.2032)
AR(1) 0.9396 0.8848
𝜗1 ----------- (0.0000) (0.0000)
MA(1) -0.9551 -0.8330
∅1 ----------- (0.0000) (0.0000)
Constant 0.0139 0.0218 0.0313
θ0 (0.0025) (0.0025) (0.0633)
ARCH(1) 0.1145 0.0988 0.0630
𝜃1 (0.0000) (0.0000) (0.0000)
GARCH(1) 0.8790 0.8943 0.9311
𝜑1 (0.0000) (0.0000) (0.0000)

Null Hypotheses(All Null Hypotheses are for nth
order)
GARCH H0: 𝜑𝑖 = 0 No GARCH effect, Mean spillover. P-values are in the parenthesis.
Residual Analysis
Parameter Jarque Q-Stat Q-Stat Q2-Stat Q2-Stat LM -ARCH LM-
Series Bera (5) (10) (5) (10) (1-2) ARCH
(1-5)
134.01 3.5698 5.2425 3.7790 5.0251 0.6032 0.7742
FTSE 350 (0.0000) (0.6128) (0.8743) (0.2863) (0.7548) (0.5471) (0.5682)
309.27 4.5445 7.4357 6.9227 8.9998 0.8606 1.4246
GDAXI (0.0000) (0.2083) (0.4904) (0.0744) (0.3423) (0.4230) (0.2121)
12835 6.7335 12.502 3.4343 8.3277 0.2255 0.6888
DFMGI (0.0000) (0.0808) (0.1301) (0.3293) (0.4021) (0.7981) (0.6319)
In table 4.3.3 given above describes the data generating process of the return series. The
estimated parameters of the fitted models are statistically significant. In GDAXI and DFMGI
models AR terms are statistically significant which means that current return of markets are
only depends upon 1st lag. In FTSE 350 model AR term is insignificant which means current
43
return of this market not depends upon lag values. MA (1) term in GADXI and DFMGI models
are differ from zero, shows relationship between past and current variations of return series.
Most of the parameters are statistically significant at 5% level of significance. ARCH and
GARCH terms are also significant in three models means these return series encompass ARCH
and GARCH effect. The persistence of shock of the return series are FTSE 350 (0.99359),
GDAXI (0.99321) and DFMGI (0.99417) all are close to 1 which means that the persistence
of ARCH and GARCH effect take long time for decay.
residuals.
4.4 Tracing Spillover Effect
In section 4.4 we explore the direct and indirect linkages between Pakistani (KSE 100)
and leading foreign stock markets (S&P 500, NASDAQ 100, DOWJONES, FTSE 350, HIS,
GDAXI, NIKKEI 225 and DFMGI) except Pakistani (KSE 100), US [S&P 500, NASDAQ
100, DOW JONES (DJI)] and Dubai financial market (DFMGI). We investigate the co-
movements among these markets by using the technique of Hamao et al. (1990). We trace out
the information transmission between individual markets to check the ‘Meteor Shower’
hypothesis of Engle. To explore mean spillover effect between two markets, the return series
of one market is introduced as regressor in conditional mean equation of other market.
44
For volatility spillover effect the square return series of one market is introduced as regressor
in the conditional variance equation of other market.
The GJR-GARCH model is employed for exploring mean and volatility spillover effect
between KSE 100 and NIKKEI 225. These estimated equations analyze mean and volatility
spillover effect from KSE 100 to NIKKEI as given below in table 4.4.1.a. To trace out the
mean spillover effect we introduced the return series of KSE 100 in conditional mean equation
of NIKKEI 225 return series. For exploring volatility spillover effect from KSE 100 to NIKKEI
225 we introduced squared return series of KSE 100 into the conditional variance equation of
NIKKEI 225. The P-values are in parenthesis.
R t,NIK = 0.0004 + 0.0739R t,KSE ……………………… (4.5)

(0.0449) (0.0000)
σ2t,NIK = 0.0502 − 0.0010𝑅𝑡,𝐾𝑆𝐸

2
+ 0.0251𝜀𝑡2 + 0.1361ε2t−1 𝐺𝑡 0.8817σ2t−1 .(4.6)
(0.0003) (0.6263) (0.0391) (0.0000) (0.0000)
Table 4.4.1 given below explains that the parameter of return series 𝜋1 and parameter
of squared return series 𝜋2 of NIKKEI 225 are statistically significant in conditional mean and
variance equations of KSE 100. It means that there exist mean and volatility spillover effect
from NIKKEI 225 to KSE 100. Results also describe that the return series parameter of KSE
100 is statistically significant in conditional mean equation of NIKKEI 225 which means there
also found mean spillover effect from KSE 100 to NIKKEI 225. The parameter values explore
that there is 7.3% mean spillover form KSE 100 to NIKKEI 225 and 0.83% from NIKKEI 225
to KSE 100. It means KSE 100 mean spillover effect is more than NIKKEI 225. The parameter
of squared return series of KSE 100 in conditional dispersion equation of NIKKEI is
insignificant mean there is no volatility spillover effect from KSE 100 to NIKKEI 225.
45
Table 4.4.1 Volatility Spillover effect between KSE 100 and NIKKEI 225 (Bidirectional
analyses for period 3rd Jun, 2005 to 28th Nov, 2014)
Return series KSE 100 to NIKKEI 225 NIKKEI 225 to KSE 100
Parameters ARMA(0,0) GJR (1,1) ARMA(0,2) GJR (1,1)
Constant 0.0004 0.0011
𝛼0 (0.0449) (0.0000)
𝑅𝑡 0.0739 0.0080
𝜋1 (0.0001) (0.0000)
MA(1) 0.1305
∅1 ----------- (0.0000)
MA(2) 0.0291
∅2 ----------- (0.0497)
Constant 0.0502 0.0603
θ0 (0.0003) (0.0000)
𝑅𝑡2 -0.0010 -0.0009
𝜋2 (0.6263) (0.0000)
ARCH(1) 0.0251 0.0652
𝜃1 (0.0391) (0.0000)
GARCH(1) 0.8817 0.8113
𝜑1 (0.0000) (0.0000)
GJR(1) 0.1361 0.1619
𝛿1 (0.0000) (0.0000)
Persistence of shock 0.9701 0.9552

GARCH H0: 𝜑𝑖 = 0 No GARCH effect, Mean spillover H0:𝜋1 = 0 No mean spillover, volatility spillover H0: 𝜋2 =
0 No volatility spillover. Leverage effect H0: 𝛿𝑖 = 0 No Leverage effect. P-values are in the parenthesis.
Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
KSE 100 to 550.13 2.5642 6.8771 0.6754 5.1295 0.2381 0.1403
NIKKEI 225 (0.0000) (0.7667) (0.7369) (0.8789) (0.7436) (0.7881) (0.9829)
NIKKEI 225 to 1886.3 9.6411 25.498 0.7510 2.1603 0.1100 0.1515
KSE 100 (0.0000) (0.0218)* (0.0127)* (0.8611) (0.9756) (0.8958) (0.9796)
The parameter of squared return series of NIKKEI 2525 in conditional variance
equation of KSE 100 is highly significant means the volatility spillover effect from NIKKEI
225 to KSE 100 is found. This co-movement between equity markets specify that the equity
46
markets are directly interlinked with each other. MA term is significant in conditional mean
equation of both models at second lag this show relationship between past and current
variations of return series at second lag value. The AR term is insignificant in both models
current return of markets are not depends upon lag values. ARCH and GARCH terms in
conditional dispersion equations are significant means ARCH and GARCH effect also exist in
return series. 𝛿1 Term describes leverage effect is significant in GJR-GARCH models which
indicates that negative correlation between assets returns and assets volatility. The persistence
of shock of the most of them are close to 1 which means that ARCH and GARCH effects are
existed and take long time for decay. Those values which are not close to 1 show ARCH and
GARCH effects are existed and take short time for decay.
residuals.
In the same fashion table 4.4.2, 4.4.3 and 4.4.4 given below Results explain that the
parameter of return series 𝜋1 and parameter of squared return series 𝜋2 of other series are
significant in conditional mean and variance equations of KSE 100. It means that the mean and
volatility spillover effect from other equity markets to KSE 100. This co-movement between
equity markets specify that the equity markets are directly interlinked with each other. Results
also represent that the mean spillover effect from KSE 100 to HIS, FTSE, GDAXI and volatility
spillover effect from KSE 100 to DFMGI.
47
Table 4.4.2 Volatility Spillover effect between KSE 100 and HIS (Bidirectional analyses
for period 3rd Jun, 2005 to 28th Nov, 2014)
Return series KSE 100 to HIS HSI to KSE 100

Parameters ARMA(0,0) GARCH (1,2) ARMA(1,2) GJR (1,1)
Constant 0.0005 0.0010
𝛼0 (0.0026) (0.0000)
𝑅𝑡 0.0368 0.0012
𝜋1 (0.0141) (0.5187)
AR(1) 0.9268
𝜗1 ----------- (0.0000)
MA(1) -0.8055
∅1 ----------- (0.0000)
MA(2) -0.0441
∅2 (0.0516)
Constant 0.0129 0.0302
θ0 (0.0566) (0.0000)
𝑅𝑡2 -0.0006 -0.0003
𝜋2 (0.6088) (0.0009)
ARCH(1) 0.1492
𝜃1 ----------- (0.0046)
ARCH(2) 0.0849
𝜃2 (0.0000) -----------
GARCH(1) 0.9139 0.7578
𝜑1 (0.0000) (0.0000)
GJR(1) 0.2770
𝛿1 ----------- (0.0000)
Persistence of shock 1.0000*

0.9989
Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) ARCH
(1-5)
KSE 100 to 138.76 1.5570 5.0265 5.5846 12.487 0.3758 1.1184
HSI (0.0000) (0.9063) (0.8894) (0.0612) (0.0856) (0.6867) (0.3483)
HSI to 17291 1.3596 9.5849 1.7442 3.7525 0.3774 0.3612
KSE 100 (0.0000) (0.5067) (0.2133) (0.6271) (0.8787) (0.6857) (0.8752)
48
Table 4.4.3 Volatility Spillover effect between KSE 100 and FTSE 350 (Bidirectional
Return series KSE 100 to FTSE 350 FTSE 350 to KSE 100
Constant 0.0006 0.0009
𝛼0 (0.0001) (0.0000)
𝑅𝑡 0.0238 0.0111
𝜋1 (0.0493) (0.0000)
AR(1) 0.668268
𝜗1 ----------- (0.0000)
AR(2) -0.0193
𝜗2 ----------- (0.7210)
MA(1) -0.5484
∅1 ----------- (0.0215)
Constant 0.0139 0.0452
θ0 (0.0054) (0.0000)
𝑅𝑡2 0.0001 -0.0016
𝜋2 (0.9398) (0.0000)
ARCH(1) 0.1146 0.0605
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.8787 0.8338
𝜑1 (0.0000) (0.0000)
GJR(1) 0.159515
𝛿1 ----------- (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) ARCH
(1-5)
KSE 100 to 137.63 3.7332 5.7011 4.1279 5.4643 0.6816 0.8478
FTSE 350 (0.0000) (0.5884) (0.8397) (0.2479) (0.7069) (0.5059) (0.5156)
FTSE 350 to 2468.8 2.4525 15.930 0.88392 2.5439 0.1512 0.1760
KSE 100 (0.0000) (0.2933) (0.0257)* (0.8293) (0.9596) (0.8596) (0.9716)
49
Table 4.4.4 Volatility Spillover effect between KSE 100 and GDAXI (Bidirectional
Return series KSE 100 to GDAXI GDAXI to KSE 100
Parameters ARMA(0,0) GJR (1,2) ARMA(1,1) GJR (1,1)
Constant 0.0006 0.0009
𝛼0 (0.0009) (0.0000)
𝑅𝑡 0.0279 -0.0030
𝜋1 (0.0651) (0.0000)
AR(1) 0.747531
𝜗1 ----------- (0.0000)
MA(1) -0.626688
∅1 ----------- (0.0000)
Constant 0.0316 0.0552
θ0 (0.0000) (0.0000)
𝑅𝑡2 -0.0007 -0.0016
𝜋2 (0.6734) (0.0000)
ARCH(1) -0.0799 0.0470
𝜃1 (0.0000) (0.0000)
ARCH(2) 0.0649
𝜃2 (0.0000) -----------
GARCH(1) 0.8941 0.8271
𝜑1 (0.0000) (0.0000)
GJR(1) 0.1994 0.1653

𝛿1 (0.0000) (0.0000)
Persistence of shock 1.0000* 0.9470

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) ARCH
(1-5)
KSE 100 to 238.10 5.2023 6.8032 2.3595 6.1688 0.9100 0.4765
GDAXI (0.0000) (0.3916) (0.7438) (0.3073) (0.5201) (0.4027) (0.7940)
GDAXI to 1651.3 4.4115 15.677 1.1346 2.3629 0.2401 0.2266
KSE 100 (0.0000) (0.2203) (0.0472)* (0.7687) (0.9678) (0.7865) (0.9511)
50
The 𝛿1 Term describes leverage effect is significant in GJR-GARCH models which
indicates that negative correlation between assets returns and assets volatility. The significant
AR terms mean the current return of markets are only depends upon lag values. The significant
MA terms show relationship between past and current variations of return series. ARCH and
GARCH terms are also significant in above models means these return series encompass
ARCH and GARCH effect. The persistence of shock of the most of them are close to 1 which
means that ARCH and GARCH effects are existed and take long time for decay. Those values
which are not close to 1 show ARCH and GARCH effects are existed and take short time for
decay.
The validity of results is also approved by the residual analysis. Table 4.4.2, 4.4.3, 4.4.4
illustrate the post estimation results (Residual analysis). The Jarque Bera test (Normality test)
results show non normal residuals. The Q-stat are insignificant up to 10th lags accept null
hypothesis means no serial autocorrelation in the standardized residuals. The Q-stat on squared
standardized residuals are insignificant up to 10th lags accept null hypothesis means no serial
autocorrelation in squared standardized residuals. LM-ARCH test is also insignificant up to
5thlags accept null hypothesis means no ARCH effect remain in series residuals. the persistence
of shock of the models are close to 1 which means that the ARCH and GARCH effects are
existed and take a long time for decay.
Table 4.4.5 given below explains that the explains that the parameter of return series
𝜋1 and parameter of squared return series 𝜋2 of S&P 500 are statistically significant in
conditional mean and variance equations of NASDAQ 100. This show the mean and volatility
spillover effect from S&P 500 to NASDAQ 100 is found. Results also describe that the
parameter of return series 𝜋1 and parameter of squared return series 𝜋2 of NASDAQ 100 is
statistically significant in conditional mean and variance equation of S&P 500 which means
there also found mean and volatility spillover effect from NASDAQ 100 to S&P 500. The
51
parameter values in conditional mean equation explore that there is 3.4% mean spillover from
S&P 500 to NASDAQ 100 and 77.5% from NASDAQ 100 to S&P 500.
Table 4.4.5 Volatility Spillover effect between S&P 500 and NASDAQ100 (Bidirectional
Return series S&P 500 to NASDAQ 100 NASDAQ 100 to S&P 500
Parameters ARMA(0,0) GARCH (1,1) ARMA(1,1) GARCH (1,1)
Constant 0.0002 -0.0001
𝛼0 (0.0017) (0.3469)
𝑅𝑡 1.0341 0.7757
𝜋1 (0.0000) (0.0000)
AR(1) -0.8368
𝜗1 ----------- (0.0000)
MA(1) 0.826348
∅1 ----------- (0.0000)
Constant 0.0054 0.0085
θ0 (0.0106) (0.0163)
𝑅𝑡2 0.0036 0.0224
𝜋2 (0.0733) (0.0007)
ARCH(1) 0.0719 0.0957
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.8868 0.6906
𝜑1 (0.0000) (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) ARCH
(1-5)
S&P 500 to 165.34 0.9930 1.3501 2.3659 12.180 1.0636 0.4819
NASDAQ 100 (0.0000) (0.9631) (0.9993) (0.5000) (0.1433) (0.3454) (0.7900)
NASDAQ 100 to 30.667 1.5821 5.3171 4.1925 8.1283 0.1667 0.8510
S&P 500 (0.0000) (0.6634) (0.7232) (0.2414) (0.4210) (0.8464) (0.5134)
52
The parameter values in conditional mean equation explore that there is 3.4% mean
spillover from S&P 500 to NASDAQ 100 and 77.5% from NASDAQ 100 to S&P 500. It
means NASDAQ 100 mean spillover effect is more than S&P 500. The parameter values of
squared return series of S&P 500 in conditional dispersion equation of NASDAQ 100 explain
there is 7.3% volatility spillover effect from S&P 500 to NASDAQ 100 and 0.07% from
NASDAQ 100 to S&P 500 which is minor as compare S&P 500. This co-movement between
equity markets specify that the equity markets are directly interlinked with each other. MA
term is significant in conditional mean equation of NASDAQ 100 to S&P 500 model at first
lag this show relationship between past and current variations of return series at first lag value.
The AR term is significant in NASDAQ 100 to S&P 500 model which show current return of
markets are depends upon lag values. ARCH and GARCH terms in conditional dispersion
equations are significant means ARCH and GARCH effect also exist in return series. The
persistence of shock of the most of them are close to 1 which means that ARCH and GARCH
effects are existed and take long time for decay. Those values which are not close to 1 show
ARCH and GARCH effects are existed and take short time for decay.
residuals.
In the same way Table 4.4.6, 4.4.7, 4.4.8 and 4.4.9 given below results describe that the
parameter of return series 𝜋1 and parameter of squared return series 𝜋2 of S&P 500 are
significant in conditional mean and variance equations of other markets return series. It means
53
that the mean and volatility spillover effect from S&P 500 to NIKKEI 225, HIS and GDAXI
equity markets.
Table 4.4.6 Volatility Spillover effect between S&P 500 and NIKKEI 225 (Bidirectional
Analyses for period 3rd Jun, 2005 to 28th Nov, 2014)
Return series S&P 500 to NIKKEI 225 NIKKEI 225 to S&P 500
Parameters ARMA(1,0) GJR (1,1) ARMA(1,0) GARCH (1,2)
Constant 0.0006 0.0007
𝛼0 (0.0008) (0.0000)
𝑅𝑡 0.1961 0.0892
𝜋1 (0.0000) (0.0000)
AR(1) -0.114845 -0.127305
𝜗1 (0.0000) (0.0000)
Constant 0.0709 0.0212
θ0 (0.0087) (0.0005)
𝑅𝑡2 0.1039 0.0014
𝜋2 (0.0035) (0.5293)
ARCH(1) 0.0348
𝜃1 (0.0386) -----------
ARCH(2) 0.1408
𝜃2 ----------- (0.0000)
GARCH(1) 0.8085 0.8468
𝜑1 (0.0000) (0.0000)
GJR(1) 0.1121
𝛿1 (0.0039) -----------

nth
Null Hypotheses(All Null Hypotheses are for order)
Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) ARCH
(1-5)
S&P 500 to 487.60 1.2799 0.8647 1.3321 7.4262 0.0240 0.2646
NIKKEI 225 (0.0000) (0.8647) (0.9383) (0.7215) (0.4914) (0.9763) (0.9325)
NIKKEI 225 to 389.83 6.6272 11.986 1.7992 11.589 0.3438 0.3865
S&P 500 (0.0000) (0.1569) (0.2140) (0.4067) (0.1148) (0.7091) (0.8583)
54
Table 4.4.7 Volatility Spillover effect between S&P 500 and HIS (Bidirectional Analyses
Return series S&P 500 to HIS HSI to S&P 500
Constant 0.0005 0.0006
𝛼0 (0.0031) (0.0000)
𝑅𝑡 0.2346 0.1483
𝜋1 (0.0000) (0.0000)
AR(1) -0.0726 -0.1487
𝜗1 (0.0008) (0.0000)
Constant 0.0466 0.0199
θ0 (0.0316) (0.0009)
𝑅𝑡2 0.2068 0.0189
𝜋2 (0.0002) (0.0084)
ARCH(2) 0.1252
𝜃2 ----------- (0.0000)
ARCH(3) 0.1135
𝜃3 (0.0000) -----------
GARCH(1) 0.7373 0.8320
𝜑1 (0.0000) (0.0000)

0 No volatility spillover. P-values are in the parenthesis.
Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) ARCH
(1-5)
S&P 500 to 63.752 3.5190 6.9497 3.0945 6.7552 0.4248 0.6085
HIS (0.0000) (0.4749) (0.6423) (0.0785) (0.3440) (0.6539) (0.6934)
HSI to 219.29 6.4978 13.163 1.3563 11.529 0.2039 0.2903
S&P 500 (0.0000) (0.1649) (0.1553) (0.5075) (0.1171) (0.8155) (0.9186)
55
Table 4.4.8 Volatility Spillover effect between S&P 500 and FTSE 350 (Bidirectional
Return series S&P 500 to FTSE 350 FTSE 350 to S&P 500
Constant 0.0001 0.0003
𝛼0 (0.0486) (0.0002)
𝑅𝑡 0.6661 0.652
𝜋1 (0.0000) (0.0000)
MA(1) -0.3104 -0.3442
∅1 (0.0000) (0.0000)
Constant 0.0324 0.0169
θ0 (0.0000) (0.0002)
𝑅𝑡2 0.0816 0.0590
𝜋2 (0.0000) (0.0000)
ARCH(1) 0.0783 0.0784
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.7294 0.7963
𝜑1 (0.0000) (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) ARCH
(1-5)
S&P 500 to 177.24 3.6704 5.3683 4.0957 13.736 0.2129 0.8241
FTSE 350 (0.0000) (0.4524) (0.8010) (0.2513) (0.0889) (0.8082) (0.5323)
FTSE 350 to 61.622 2.3450 4.6627 6.1567 9.0712 2.2960 1.1979
S&P 500 (0.0000) (0.6725) (0.8626) (0.10422) (0.33632) (0.1009) (0.3076)
56
Table 4.4.9 Volatility Spillover effect between S&P 500 and GDAXI (Bidirectional
Return series S&P 500 to GDAXI GDAXI to S&P 500
Constant 0.0004 0.0002
𝛼0 (0.0000) (0.0064)
𝑅𝑡 0.8152 0.5227
𝜋1 (0.0000) (0.0000)
MA(1) -0.2711 -0.3093
∅1 (0.0000) (0.0000)
Constant 0.0667 0.0144
θ0 (0.0000) (0.0084)
𝑅𝑡2 0.1129 0.0261
𝜋2 (0.0000) (0.0041)
ARCH(1) 0.0790 0.1277
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.7096 0.7899
𝜑1 (0.0000) (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) ARCH
(1-5)
S&P 500 to 302.55 0.709672 3.1866 6.3130 9.1005 2.7801 1.2152
GDAXI (0.0000) (0.9501) (0.9564) (0.0973) (0.3338) (0.0622) (0.2992)
GDAXI to 155.05 1.6001 2.7980 3.8698 4.5361 1.8031 0.8014
S&P 500 (0.0000) (0.8087) (0.9717) (0.2758) (0.8058) (0.1650) (0.5485)
This co-movement between equity markets specify that these equity markets are
directly interlinked with S&P 500. The results also explore that mean and volatility spillover
effect from HIS, FTSE 350 and GDAXI to S&P 500 and only mean spillover effect from
NIKKEI 225 to S&P 500. 𝛿1 Term describes leverage effect is significant in GJR-GARCH
57
models which indicates that negative correlation between assets returns and assets volatility.
The significant AR terms mean the current return of markets are only depends upon lag values.
The significant MA terms show relationship between past and current variations of return
series. ARCH and GARCH terms are also significant in above models means these return series
encompass ARCH and GARCH effect. The persistence of shock of the most of them are close
to 1 which means that ARCH and GARCH effects are existed and take long time for decay.
Those values which are not close to 1 show ARCH and GARCH effects are existed and take
short time for decay.
The validity of results is also approved by the residual analysis. Table 4.4.6, 4.4.7, 4.4.8
and 4.4.9 also illustrate the post estimation results (Residual analysis). The Jarque Bera test
(Normality test) results show non normal residuals. The Q-stat are insignificant up to 10th lags
accept null hypothesis means no serial autocorrelation in the standardized residuals. The Q-stat
on squared standardized residuals are insignificant up to 10th lags accept null hypothesis means
no serial autocorrelation in squared standardized residuals. LM-ARCH test is also insignificant
up to 5th lags accept null hypothesis means no ARCH effect remain in series residuals.
Table 4.4.10 given below explain that the parameter of return series 𝜋1 and parameter
of squared return series 𝜋2 of DJI are statistically significant in conditional mean and variance
equations of NASDAQ 100. This show the mean and volatility spillover effect from DJI to
NASDAQ 100 is found. Results also describe that the parameter of return series 𝜋1 and
parameter of squared return series 𝜋2 of NASDAQ 100 is statistically significant in conditional
mean and variance equation of DJI which means there also found mean and volatility spillover
effect from NASDAQ 100 to DJI. The parameter values in conditional mean equation explore
that there is 107% mean spillover from DJI to NASDAQ 100 and 6.8% from NASDAQ 100 to
DJI. It means DJI mean spillover effect is more than NASDAQ 100.
58
Table 4.4.10 Volatility Spillover effect between NASDAQ100 and DJI (Bidirectional
Return series NASDAQ 100 to DJI DJI to NASDAQ 100

Constant 0.0001 0.0003
𝛼0 (0.3662) (0.0022)
𝑅𝑡 0.685135 1.071909
𝜋1 (0.0000) (0.0000)
Constant 0.0117 0.0087
θ0 (0.1910) (0.0511)
𝑅𝑡2 0.0191 0.0058
𝜋2 (0.1011) (0.0991)
ARCH(1) 0.0768 0.0548
𝜃1 (0.0087) (0.0002)
GARCH(1) 0.7545 0.9023
𝜑1 (0.0000) (0.0000)
Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) ARCH
(1-5)
NASDAQ 100 to 94.866 5.2324 7.1708 1.3959 3.4951 0.6412 0.2764
DJI (0.0000) (0.3881) (0.7092) (0.7064) (0.8995) (0.5267) (0.9262)
DJI to 180.66 1.0454 2.3475 5.3200 13.494 1.6702 1.0769
NASDAQ 100 (0.0000) (0.9588) (0.9929) (0.1498) (0.0959) (0.1884) (0.3710)
The parameter values of squared return series of DJI in conditional dispersion equation
of NASDAQ 100 explain there is 1.91% volatility spillover effect from DJI to NASDAQ 100
and 0.5% from NASDAQ 100 to DJI which is less as compare DJI. This co-movement between
equity markets specify that the equity markets are directly interlinked with each other. MA
term is insignificant in both models conditional mean equation this show no relationship
between past and current variations of return series at first lag value. The AR term is
59
insignificant in both models conditional mean equations which shows current return of markets
are not depends upon lag values. ARCH and GARCH terms in conditional dispersion equations
are significant means ARCH and GARCH effect also exist in return series. The persistence of
shock of the most of them are close to 1 which means that ARCH and GARCH effects are
Table 4.4.10 illustrate the post estimation results (Residual analysis). The Jarque Bera
test (Normality test) results show non normal residuals. The Q-stat are insignificant up to 10th
lags accept null hypothesis means no serial autocorrelation in the standardized residuals. The
Q-stat on squared standardized residuals are insignificant up to 10th lags accept null hypothesis
means no serial autocorrelation in squared standardized residuals. LM-ARCH test is also
insignificant up to 5thlags accept null hypothesis means no ARCH effect remain in series
residuals.
In the same fashion Table 4.4.11, 4.4.12, 4.4.13 and 4.4.14 given below results explain
that the parameter of return series 𝜋1 and parameter of squared return series 𝜋2 of NASDAQ
100 are significant in conditional mean and variance equations of NIKKEI 225, HIS, FTSE
350 and GADXI return series. It means that the mean and volatility spillover effect from
NASDAQ 100 to NIKKEI 225, HIS, FTSE 350 and GDAXI equity markets. This co-
movement between equity markets specify that these equity markets are interlinked with
NASDAQ 100. The results also explore that mean and volatility spillover effect from HIS,
FTSE 350 and GDAXI to NASDAQ 100 and only mean spillover effect from NIKKEI 225 to
NASDAQ 100. The significant AR terms mean the current return of markets are only depends
upon lag values. The significant MA terms show relationship between past and current
variations of return series. ARCH and GARCH terms are also significant in above models
means these return series encompass ARCH and GARCH effect. The persistence of shock of
60
the most of them are close to 1 which means that ARCH and GARCH effects are existed and
take long time for decay. Those values which are not close to 1 show ARCH and GARCH
effects are existed and take short time for decay.
Table 4.4.11 Volatility Spillover effect between NASDAQ100 and NIKKEI225
(Bidirectional Analyses for period 3rd Jun, 2005 to 28th Nov, 2014)
Return series NASDAQ 100 to NIKKEI 225 NIKKEI 225 to NASDAQ 100
Constant 0.0006 0.0009
𝛼0 (0.0008) (0.0000)
𝑅𝑡 0.1702 0.0967
𝜋1 (0.0000) (0.0000)
AR(1) -0.1110 -0.0883
𝜗1 (0.0000) (0.0000)
Constant 0.0290 0.0213
θ0 (0.0667) (0.0010)
𝑅𝑡2 0.0717 0.0014
𝜋2 (0.0016) (0.5458)
ARCH(1) 0.0859 0.0879
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.8478 0.8990
𝜑1 (0.0000) (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
NASDAQ 100 to 277.95 1.6230 3.5850 6.8752 7.5595 2.866 1.3340
NIKKEI 225 (0.0000) (0.8046) (0.9365) (0.07598) (0.4776) (0.0571) (0.2467)
NIKKEI 225 to 184.81 3.2599 6.9860 5.6753 15.538 2.7165 1.1314
NASDAQ 100 (0.0000) (0.5153) (0.6385) (0.1285) (0.0494)* (0.0663) (0.3414)
61
Table 4.4.12 Volatility Spillover effect between NASDAQ100 and HIS (Bidirectional
Return series NASDAQ 100 to HSI HSI to NASDAQ 100
Constant 0.0004 0.0009
𝛼0 (0.0139) (0.0000)
𝑅𝑡 0.1827 0.1838
𝜋1 (0.0000) (0.0000)
AR(1) -0.0631 -0.1185
𝜗1 (0.0023) (0.0000)
Constant 0.0105 0.0258
θ0 (0.5071) (0.0011)
𝑅𝑡2 0.0886 0.0266
𝜋2 (0.0237) (0.0012)
ARCH(1) 0.080562
𝜃1 ----------- (0.0000)
ARCH(3) 0.1142
𝜃3 (0.0000) -----------
GARCH(1) 0.8160 0.8699
𝜑1 (0.0000) (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
NASDAQ 100 to 93.116 3.2325 7.1566 2.6453 8.1493 0.5918 0.5309
HIS (0.0000) (0.5196) (0.6208) (0.1038) (0.2273) (0.5534) (0.7530)
HSI to 139.42 3.2357 7.5141 5.4243 10.842 2.6095 1.0842
NASDAQ 100 (0.0000) (0.5191) (0.5837) (0.1432) (0.2107) (0.0738) (0.3670)
62
Table 4.4.13 Volatility Spillover effect between NASDAQ100 and FTSE350
(Bidirectional Analyses for period 3rd Jun, 2005 to 28th Nov, 2014)
Return series NASDAQ 100 to FTSE 350 FTSE 350 to NASDAQ 100
Constant 0.0001 0.0005
𝛼0 (0.1636) (0.0000)
𝑅𝑡 0.4486 0.6435
𝜋1 (0.0000) (0.0000)
MA(1) -0.193129 -0.220695
∅1 (0.0000) (0.0000)
Constant 0.0289 0.0290
θ0 (0.0010) (0.0088)
𝑅𝑡2 0.0689 0.0510
𝜋2 (0.0000) (0.0041)
ARCH(1) 0.0520 0.0643
𝜃1 (0.0063) (0.0000)
ARCH(3) 0.0492
𝜃3 (0.0222) -----------
GARCH(1) 0.7317 0.8512
𝜑1 (0.0000) (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
NASDAQ 100 to 125.01 4.0924 5.4616 3.1853 6.8453 0.0464 0.6353
FTSE 350 (0.0000) (0.3936) (0.7923) (0.0743) (0.3353) (0.9546) (0.6727)
FTSE 350 67.278 2.0826 4.3335 6.0212 9.1625 1.2473 1.1912
NASDAQ 100 (0.0000) (0.7205) (0.8881) (0.1105) (0.3287) (0.2875) (0.3109)
63
Table 4.4.14 Volatility Spillover effect between NASDAQ100 and GDAXI (Bidirectional
Return series NASDAQ 100 to GDAXI GDAXI to NASDAQ 100
Constant 0.0004 0.0005
𝛼0 (0.0041) (0.0000)
𝑅𝑡 0.5780 0.5445
𝜋1 (0.0000) (0.0000)
MA(1) -0.1833 -0.2194
∅1 (0.0000) (0.0000)
Constant 0.0415 0.0249
θ0 (0.0092) (0.0086)
𝑅𝑡2 0.0640 0.0199
𝜋2 (0.0018) (0.0461)
ARCH(1) 0.0850 0.0965
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.7882 0.8460
𝜑1 (0.0000) (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
NASDAQ 100 to 255.47 2.5963 4.4714 3.1317 3.6513 0.8767 0.6241
GDAXI (0.0000) (0.6274) (0.8777) (0.3717) (0.8871) (0.4163) (0.6814)
GDAXI to 119.71 1.8524 3.4458 1.2184 5.2169 0.5335 0.2451
NASDAQ 100 (0.0000) (0.7628) (0.9439) (0.7485) (0.7341) (0.5866) (0.9424)
The validity of results is also approved by the residual analysis. Table 4.4.11, 4.4.12,
4.4.13 and 4.4.14 also illustrate the post estimation results (Residual analysis). The Jarque Bera
64
insignificant up to 5th lags accept null hypothesis means no ARCH effect remain in series
residuals.
Table 4.4.15 Volatility Spillover effect between DJI and NIKKEI 225 (Bidirectional
Return series DJI to NIKKEI 225 NIKKEI 225 to DJI
Constant 0.0006 0.0006
𝛼0 (0.0010) (0.0000)
𝑅𝑡 0.2005 0.0774
𝜋1 (0.0000) (0.0000)
AR(1) -0.1077 -0.1153
𝜗1 (0.0001) (0.0000)
Constant 0.0665 0.0191
θ0 (0.0102) (0.0004)
𝑅𝑡2 0.1082 0.0019
𝜋2 (0.0038) (0.3648)
ARCH(1) 0.0351
𝜃1 (0.0314) -----------
ARCH(2) 0.1416
𝜃2 ----------- (0.0000)
GARCH(1) 0.8192 0.8439
𝜑1 (0.0000) (0.0000)
GJR(1) 0.1077
𝛿1 (0.0044) -----------

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) ARCH
(1-5)
DJI to 330.92 4.3422 8.3268 2.0637 9.6626 0.3423 0.4200
NIKKEI 225 (0.0000) (0.3616) (0.5015) (0.3563) (0.2085) (0.7101) (0.8350)
NIKKEI 225 to 572.25 1.3622 3.6471 1.2427 7.4384 0.0317 0.2460
DJI (0.0000) (0.8507) (0.9330) (0.7427) (0.4901) (0.9688) (0.9419)
65
Table 4.4.15 clarifies that the parameter of return series 𝜋1 and parameter of squared
return series 𝜋2 of DJI are statistically significant in conditional mean and variance equations
of NIKKEI 225. This show the mean and volatility spillover effect from DJI to NIKKEI 225 is
found. Results also describe that parameter of return series 𝜋1 of NIKKEI 225 is statistically
significant in conditional mean equation of DJI which means there also found mean spillover
effect from NIKKEI 225 to DJI. The parameter values in conditional mean equation explore
that there is 20% mean spillover from DJI to NIKKEI 225 and 7.7% from NIKKEI 225 to DJI.
It means DJI mean spillover effect is more than NIKKEI 225. The parameter values of squared
return series of DJI in conditional dispersion equation of NIKKEI 225 explain there is 10%
volatility spillover effect from DJI to NIKKEI 225 and there is no volatility spillover effect
from NIKKEI 225 to DJI. This co-movement between equity markets specify that the equity
markets are directly interlinked with each other. 𝛿1 Term describes leverage effect is significant
in GJR-GARCH models which indicates that negative correlation between assets returns and
assets volatility. MA term is insignificant in both models conditional mean equation this show
no relationship between past and current variations of return series at first lag value. The AR
term is significant in both models conditional mean equations which shows current return of
markets are depends upon lag values. ARCH and GARCH terms in conditional dispersion
66
residuals.
Table 4.4.16 Volatility Spillover effect between DJI and HIS (Bidirectional Analyses for
period 3rd Jun, 2005 to 28th Nov, 2014)
Return series DJI to HIS HSI to DJI
ARMA(0,1) GARCH (1,3) ARMA(1,0) GARCH (1,2)
Parameters
Constant 0.0005 0.0006
𝛼0 (0.0026) (0.0000)
𝑅𝑡 0.2261 0.1260
𝜋1 (0.0000) (0.0000)
AR(1) -0.1350
𝜗1 ----------- (0.0000)
MA(1) -0.0655
∅1 (0.0031) -----------
Constant 0.0374 0.0184
θ0 (0.0748) (0.0006)
𝑅𝑡2 0.1836 0.0131
𝜋2 (0.0024) (0.0208)
ARCH(2) 0.1288
𝜃2 ----------- (0.0000)
ARCH(3) 0.1115
𝜃3 (0.0000) -----------
GARCH(1) 0.7744 0.8329
𝜑1 (0.0000) (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) ARCH
(1-5)
DJI to 72.632 2.3151 5.4920 2.5126 6.7660 0.4021 0.4924
HSI (0.0000) (0.6780) (0.7894) (0.1129) (0.3430) (0.6690) (0.7821)
HSI to 213.16 3.8207 8.8865 2.4596 7.9560 0.5828 0.4963
DJI (0.0000) (0.4308) (0.4478) (0.2923) (0.3364) (0.5584) (0.7792)
67
Table 4.4.17 Volatility Spillover effect between DJI and FTSE 350 (Bidirectional
Return series DJI to FTSE 350 FTSE 350 to DJI
Constant 0.0002 0.0002
𝛼0 (0.0324) (0.0006)
𝑅𝑡 0.6953 0.5897
𝜋1 (0.0000) (0.0000)
MA(1) -0.2716 -0.2808
∅1 (0.0000) (0.0000)
Constant 0.0304 0.0132
θ0 (0.0000) (0.0007)
𝑅𝑡2 0.0965 0.0449
𝜋2 (0.0000) (0.0000)
ARCH(1) 0.0908 0.0882
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.7231 0.8038
𝜑1 (0.0000) (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) ARCH
(1-5)
DJI to 173.57 2.4236 5.7618 4.3258 13.180 0.0493 0.8636
FTSE 350 (0.0000) (0.6583) (0.7634) (0.2283) (0.1057) (0.9518) (0.5048)
FTSE 350 to 69.466 1.4047 2.2146 2.2230 5.0735 1.0141 0.4495
DJI (0.0000) (0.8433) (0.9876) (0.5274) (0.7496) (0.3629) (0.8139)
68
Table 4.4.18 Volatility Spillover effect between DJI and GDAXI (Bidirectional Analyses
Return series DJI to GDAXI GDAXI to DJI
Constant 0.0004 0.0002
𝛼0 (0.0002) (0.0098)
𝑅𝑡 0.8589 0.4770
𝜋1 (0.0000) (0.0000)
MA(1) -0.2473 -0.2656
∅1 (0.0000) (0.0000)
Constant 0.0581 0.0105
θ0 (0.0004) (0.0095)
𝑅𝑡2 0.1205 0.0165
𝜋2 (0.0001) (0.0087)
ARCH(1) 0.0739 0.1159
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.7369 0.8212
𝜑1 (0.0000) (0.0000)

nth
Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) ARCH
(1-5)
DJI to 262.16 0.7696 3.3513 7.0556 8.0888 3.2141 1.353
GDAXI (0.0000) (0.9424) (0.9487) (0.0701) (0.4248) (0.0404)* (0.2387)
GDAXI to 118.97 3.9253 4.8567 3.4139 4.6976 0.4534 0.6958
DJI (0.0000) (0.4162) (0.8466) (0.3321) (0.7893) (0.6355) (0.6266)
In the same way Table 4.4.16, 4.4.17 and 4.4.18 Results clarifies that the parameter of
return series 𝜋1 and parameter of squared return series 𝜋2 of DJI are significant in conditional
mean and variance equations of HIS, FTSE 350 and GADXI return series. It means that the
mean and volatility spillover effect from DJI to HIS, FTSE 350 and GDAXI equity markets.
69
This co-movement between equity markets specify that these equity markets are interlinked
with DJI. The results also explore that mean and volatility spillover effect from HIS, FTSE 350
and GDAXI to DJI. 𝛿1 Term describes leverage effect is significant in GJR-GARCH models
which indicates that negative correlation between assets returns and assets volatility. The
significant AR terms mean the current return of markets are only depends upon lag values. The
significant MA terms show relationship between past and current variations of return series.
ARCH and GARCH terms are also significant in above models means these return series
The validity of results is also approved by the residual analysis. Table 4.4.16, 4.4.17
up to 5thlags accept null hypothesis means no ARCH effect remain in series residuals.
Table 4.4.19 given below clarifies that the parameter of return series 𝜋1 and parameter
of squared return series 𝜋2 of HIS are statistically significant in conditional mean and variance
equations of NIKKEI 225. This show the mean and volatility spillover effect from HIS to
NIKKEI 225 is found. Results also describe that the parameter of return series 𝜋1 of NIKKEI
225 is statistically significant in conditional mean equation of HIS which means there also
found mean spillover effect from NIKKEI 225 to HIS. The parameter values in conditional
70
mean equation explore that there is 52% mean spillover from HIS to NIKKEI 225 and 20%
from NIKKEI 225 to HIS.
Table 4.4.19 Volatility Spillover effect between NIKKEI 225and HIS (Bidirectional
Return series HSI to NIKKEI 225 NIKKEI 225 to HIS

Constant 0.0001 0.0002
𝛼0 (0.4475) (0.0971)
𝑅𝑡 0.5230 0.4458
𝜋1 (0.0000) (0.0000)
Constant 0.0276 0.0063
θ0 (0.0107) (0.0908)
𝑅𝑡2 0.0102 0.0052
𝜋2 (0.0352) (0.1016)
ARCH(1) 0.0563 0.0614
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.8773 0.9270
𝜑1 (0.0000) (0.0000)
GJR(1) 0.0663
𝛿1 (0.0270) -----------
Persistence of shock
0.9633 0.9884
Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) ARCH
(1-5)
NIKKEI 225 to 81.765 3.8168 9.7411 3.9138 13.712 0.7521 0.7671
HIS (0.0000) (0.5760) (0.4634) (0.2709) (0.0895) (0.4715) (0.5734)
HSI to 395.06 9.5432 12.916 9.6568 12.386 3.4036 1.9034
NIKKEI 225 (0.0000) (0.0892) (0.2283) (0.0217)* (0.1347) (0.0334)* (0.0905)
71
It means HIS mean spillover effect is more than NIKKEI 225. The parameter values of
squared return series of HIS in conditional dispersion equation of NIKKEI 225 explain there
is 3.5% volatility spillover effect from HIS to NIKKEI 225 and there is no volatility spillover
effect from NIKKEI 225 to HIS. This co-movement between equity markets specify that the
equity markets are directly interlinked with each other. 𝛿1 Term describes leverage effect is
significant in GJR-GARCH models which indicates that negative correlation between assets
returns and assets volatility. MA term is insignificant in both models conditional mean equation
this show no relationship between past and current variations of return series at first lag value.
The AR term is insignificant in both models conditional mean equations which shows current
return of markets are not depends upon lag values. ARCH and GARCH terms in conditional
dispersion equations are significant means ARCH and GARCH effect also exist in return series.
The persistence of shock of the most of them are close to 1 which means that ARCH and
GARCH effects are existed and take long time for decay. Those values which are not close to
1 show ARCH and GARCH effects are existed and take short time for decay.
residuals.
72
Table 4.4.20.a Volatility Spillover effect between NIKKEI225 and FTSE (Bidirectional
Return series FTSE to NIKKEI 225 NIKKEI 225 to FTSE
Constant 0.0004 0.0003
𝛼0 (0.0074) (0.0018)
𝑅𝑡 0.4406 0.2091
𝜋1 (0.0000) (0.0000)
MA(1) -0.1654 -0.1485
∅1 (0.0000) (0.0000)
Constant 0.0553 0.0105
θ0 (0.0064) (0.0147)
𝑅𝑡2 0.0865 0.0053
𝜋2 (0.0009) (0.0302)
ARCH(1) 0.0581 0.1110
𝜃1 (0.0004) (0.0000)
GARCH(1) 0.8129 0.8729
𝜑1 (0.0000) (0.0000)
GJR(1) 0.0824
𝛿1 (0.0193) -----------
Persistence of shock
0.9840 0.9106
Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
NIKKEI 225 to 107.68 3.7223 5.4932 4.8830 7.9069 0.5806 1.0100
FTSE 350 (0.0000) (0.4448) (0.7893) (0.1805) (0.4426) (0.5596) (0.4101)
FTSE 350 to 155.19 1.9531 3.8672 4.1730 6.4061 0.7651 0.8310
NIKKEI 225 (0.0000) (0.7443) (0.9199) (0.2433) (0.6018) (0.4654) (0.5274)
73
Table 4.4.21 Volatility Spillover effect between NIKKEI225 and GDAXI (Bidirectional
Return series GDAXI to NIKKEI 225 NIKKEI 225 to GDAXI
Constant 0.0004 0.0007
𝛼0 (0.0204) (0.0000)
𝑅𝑡 0.3316 0.2507
𝜋1 (0.0000) (0.0000)
MA(1) -0.1643 -0.1418
∅1 (0.0000) (0.0000)
Constant 0.0432 0.0249
θ0 (0.0212) (0.0112)
𝑅𝑡2 0.0481 0.0133
𝜋2 (0.0112) (0.0088)
ARCH(1) 0.0588
𝜃1 (0.0002) -----------
ARCH(2) 0.1228
𝜃2 ----------- (0.0000)
GARCH(1) 0.8296 0.8499
𝜑1 (0.0000) (0.0000)
GJR(1) 0.0909
𝛿1 (0.0063) -----------

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
NIKKEI 225 to 227.90 9.1178 10.290 1.4453 5.8019 0.1807 0.2803
GDAXI (0.0000) (0.0582) (0.3274) (0.4854) (0.5630) (0.8346) (0.9241)
GDAXI to 207.66 1.5820 3.8762 3.0676 4.8036 0.0745 0.6067
NIKKEI 225 (0.0000) (0.8120) (0.9193) (0.3813) (0.7783) (0.9281) (0.6948)
74
Table 4.4.22 Volatility Spillover effect between NIKKEI225 and DFMGI (Bidirectional
Return series NIKKEI 225 to DFMGI DFMGI to NIKKEI 225
Constant 0.0003 0.0004
𝛼0 (0.3265) (0.0502)
𝑅𝑡 0.1927 0.0747
𝜋1 (0.0000) (0.0000)
AR(1) 0.8473
𝜗1 (0.0000) -----------
AR(2) 0.0458
𝜗2 (0.0674) -----------
MA(1) -0.8398
∅1 (0.0000) -----------
Constant 0.0155 0.0514
θ0 (0.3331) (0.0002)
𝑅𝑡2 0.0123 -0.0006
𝜋2 (0.1737) (0.6274)
ARCH(1) 0.0684 0.0277
𝜃1 (0.0000) (0.0190)
GARCH(1) 0.9238 0.8820
𝜑1 (0.0000) (0.0000)
GJR(1) 0.1290
𝛿1 ----------- (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
NIKKEI 225 to 15008 5.2588 12.207 2.8701 6.6643 0.0863 0.5704
DFMGI (0.0000) (0.0721) (0.0939) (0.4120) (0.5732) (0.9173) (0.7227)
DFMGI to 439.42 3.0023 0.6999 0.84060 5.5456 0.1593 0.1718
NIKKEI 225 (0.0000) (0.6996) (0.7481) (0.8397) (0.6979) (0.8527) (0.9731)
In the same way Table 4.4.20, 4.4.21 and 4.4.22 Results simplifies that the parameter
of return series 𝜋1 and parameter of squared return series 𝜋2 of NIKKEI 225 are significant in
75
conditional mean and variance equations of FTSE 350 and GADXI return series. It means that
the mean and volatility spillover effect from NIKKEI 225 to FTSE 350 and GDAXI equity
markets. This co-movement between equity markets specify that these equity markets are
interlinked with NIKKEI 225. The results also explore that mean and volatility spillover effect
from FTSE 350 and GDAXI to NIKKEI 225. There is only mean spillover effect between
NIKKEI 225 and DFMGI. 𝛿1 Term describes leverage effect is significant in GJR-GARCH
models which indicates that negative correlation between assets returns and assets volatility.
The significant AR terms mean the current return of markets are only depends upon lag values.
The significant MA terms show relationship between past and current variations of return
series. ARCH and GARCH terms are also significant in above models means these return series
The validity of results is also approved by the residual analysis. Table 4.4.20, 4.4.21
up to 5th lags accept null hypothesis means no ARCH effect remain in series residuals.
Table 4.4.23 given below explains that the parameter of return series 𝜋1 and parameter
of squared return series 𝜋2 of HIS are statistically significant in conditional mean and variance
equations of FTSE 350. This show the mean and volatility spillover effect from HIS to FTSE
350 is found.
76
Table 4.4.23 Volatility Spillover effect between HSI and FTSE 350 (Bidirectional
Return series FTSE 350 to HIS HSI to FTSE 350

Constant 0.0003 0.0003
𝛼0 (0.0257) (0.0000)
𝑅𝑡 0.5244 0.3154
𝜋1 (0.0000) (0.0000)
MA(1) -0.1283 -0.1664
∅1 (0.0000) (0.0000)
Constant 0.0169 0.0160
θ0 (0.0888) (0.0026)
𝑅𝑡2 0.0807 0.0404
𝜋2 (0.0339) (0.0000)
ARCH(1) 0.0457 0.0894
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.8844 0.8157
𝜑1 (0.0000) (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
HSI to 36.727 2.0681 2.8867 5.5206 10.051 1.5137 1.1366
FTSE 350 (0.0000) (0.7232) (0.9686) (0.1374) (0.2614) (0.2203) (0.3386)
FTSE 350 to 76.596 4.9466 6.9803 4.9889 11.416 1.6487 0.9939
HIS (0.0000) (0.2928) (0.6391) (0.1726) (0.1792) (0.1925) (0.4198)
Results also describe that the explains that the parameter of return series 𝜋1 and
parameter of squared return series 𝜋2 of FTSE 350 is statistically significant in conditional
mean equation of HIS which means there also found mean and volatility spillover effect from
FTSE 350 to HIS. The parameter values in conditional mean equation explore that there is 31%
77
mean spillover from HIS to FTSE 350 and 52% from FTSE 350 to HIS. It means HIS mean
spillover effect is less than FTSE 350. The parameter values of squared return series of HIS in
conditional dispersion equation of FTSE 350 explain there is 8.0% volatility spillover effect
from HIS to FTSE 350 and 4% volatility spillover effect from FTSE 350 to HIS. This co-
movement between equity markets specify that the equity markets are directly interlinked with
each other. MA term is significant in both models conditional mean equation this show
relationship between past and current variations of return series at first lag value. The AR term
is insignificant in both models conditional mean equations which shows current return of
markets are not depends upon lag values. ARCH and GARCH terms in conditional dispersion
Table 4.4.23 show the post estimation results (Residual analysis). The Jarque Bera test
In the same fashion Table 4.4.24 and 4.4.25 given below results explain that the
parameter of return series 𝜋1 and parameter of squared return series 𝜋2 of HSI are significant
in conditional mean and variance equation of GADXI return series. It means that the mean and
volatility spillover effect from HSI to GDAXI equity markets. This co-movement between
equity markets specify that these equity markets are interlinked with HSI. The results also
78
explore that mean and volatility spillover effect from GDAXI to HSI and only mean spillover
effect between HIS and DFMGI is found.
Table 4.4.24 Volatility Spillover effect between HSI and GDAXI (Bidirectional Analyses
Return series GDAXI to HSI HSI to GDAXI

Constant 0.0003 0.0007
𝛼0 (0.0229) (0.0000)
𝑅𝑡 0.3520 0.3262
𝜋1 (0.0000) (0.0000)
MA(1) -0.0955 -0.1294
∅1 (0.0000) (0.0000)
Constant 0.0169 0.0275
θ0 (0.1976) (0.0040)
𝑅𝑡2 0.0775 0.0251
𝜋2 (0.0057) (0.0071)
ARCH(1) 0.0849
𝜃1 ----------- (0.0000)
ARCH(3) 0.1202
𝜃3 (0.0000) -----------
GARCH(1) 0.7995 0.8637
𝜑1 (0.0000) (0.0000)

nth
Residual Analysis
Parameter Jarque Q-Stat Q-Stat Q2-Stat Q2-Stat LM - LM-ARCH
Series Bera (5) (10) (5) (10) ARCH (1-5)
(1-2)
HSI to 71.734 6.6523 7.8033 7.6461 9.3427 1.4372 1.5356
GDAXI (0.0000) (0.1554) (0.5540) (0.0539) (0.3142) (0.2378) (0.1753)
GDAXI to 68.913 4.5527 5.8500 2.8994 7.9257 1.0234 0.5853
HIS (0.0000) (0.3363) (0.7548) (0.0886) (0.2435) (0.3595) (0.7113)
79
Table 4.4.25 Volatility Spillover effect between HSI and DFMGI (Bidirectional Analyses
Return series HSI to DFMGI DFMGI to HIS
Constant 0.0003 0.0005
𝛼0 (0.1385) (0.0000)
𝑅𝑡 0.1374 0.0540
𝜋1 (0.0000) (0.0000)
AR(1) 0.8786 0.7766
𝜗1 (0.0000) (0.0000)
MA(1) -0.8436 -0.7907
∅1 (0.0000) (0.0000)
Constant 0.0835 0.0123
θ0 (0.0613) (0.0971)
𝑅𝑡2 0.0035 0.0001
𝜋2 (0.6912) (0.9329)
ARCH(1) 0.1829
𝜃1 (0.0038) -----------
ARCH(2) 0.0828
𝜃2 ----------- (0.0000)
GARCH(1) 0.8888 0.9149
𝜑1 (0.0000) (0.0000)

GARCH H0: 𝜑𝑖 = 0 No GARCH effect, Mean spillover H0: 𝜋1 = 0 No mean spillover, volatility spillover H0: 𝜋2 =
Residual Analysis
Parameter Jarque Q-Stat Q-Stat Q2-Stat Q2-Stat LM-ARCH LM-ARCH
Series Bera (5) (10) (5) (10) (1-2) (1-5)
HSI to 27400 11.123 17.612 1.8807 4.1401 0.2039 0.3718
DFMGI (0.0000) (0.0110)* (0.0243)* (0.5974) (0.8442) (0.8155) (0.8682)
DFMGI to 127.89 2.2084 4.9181 6.0725 12.644 0.2475 1.1828
HIS (0.0000) (0.5302) (0.7662) (0.0137)* (0.0490) (0.7807) (0.3150)
The significant AR terms mean the current return of markets are only depends upon lag
values. The significant MA terms show relationship between past and current variations of
return series. ARCH and GARCH terms are also significant in above models means these return
80
series encompass ARCH and GARCH effect. The persistence of shock of the most of them are
close to 1 which means that ARCH and GARCH effects are existed and take long time for
decay. Those values which are not close to 1 show ARCH and GARCH effects are existed and
take short time for decay.
The validity of results is also approved by the residual analysis. Table 4.4.24 and 4.4.25
also illustrate the post estimation results (Residual analysis). The Jarque Bera test (Normality
test) results show non normal residuals. The Q-stat are insignificant up to 10th lags accept null
hypothesis means no serial autocorrelation in the standardized residuals. The Q-stat on squared
standardized residuals are insignificant up to 10th lags accept null hypothesis means no serial
autocorrelation in squared standardized residuals. LM-ARCH test is also insignificant up to
5thlags accept null hypothesis means no ARCH effect remain in series residuals.
Table 4.4.26 explains that the parameter of return series 𝜋1 and parameter of squared
return series 𝜋2 of GDAXI are statistically significant in conditional mean and variance
equations of FTSE 350. This show the mean and volatility spillover effect from GDAXI to
FTSE 350 is found. Results also describe that explains that the parameter of return series 𝜋1
and parameter of squared return series 𝜋2 of FTSE 350 is statistically significant in conditional
mean equation of GDAXI which means there also found mean and volatility spillover effect
from FTSE 350 to GDAXI. The parameter values in conditional mean equation explore that
there is 69% mean spillover from GDAXI to FTSE 350 and 1.0% from FTSE 350 to GDAXI.
It means GDAXI mean spillover effect is more than FTSE 350. The parameter values of
squared return series of GDAXI in conditional dispersion equation of FTSE 350 explain there
is 12% volatility spillover effect from GDAXI to FTSE 350 and 6% volatility spillover effect
from FTSE 350 to GDAXI. This co-movement between equity markets specify that the equity
markets are directly interlinked with each other.
81
Table 4.4.26 Volatility Spillover effect between FTSE 350 and GDAXI (Bidirectional
Return series FTSE 350 to GDAXI GDAXI to FTSE 350

Constant 0.0003 -0.0001
𝛼0 (0.0030) (0.8484)
𝑅𝑡 1.0194 0.6939
𝜋1 (0.0000) (0.0000)
Constant 0.0587 0.0393
θ0 (0.0004) (0.0007)
𝑅𝑡2 0.0615 0.0384
𝜋2 (0.0002) (0.0000)
ARCH(1) 0.1503 0.1235
𝜃1 (0.0000) (0.0002)
ARCH(4) 0.1000
𝜃4 ----------- (0.0013)
GARCH(1) 0.4450
𝜑1 ----------- (0.0000)
GARCH(2) 0.2398
𝜑2 (0.0154) -----------
GARCH(3) 0.2988
𝜑3 (0.0024) -----------

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
FTSE 350 to 107.17 1.0499 7.1911 3.9220 9.1864 1.0006 0.7806
GDAXI (0.0000) (0.9584) (0.7072) (0.0476)* (0.1633) (0.3678) (0.5635)
GDAXI to 202.53 3.2520 8.9544 8.5635 18.236 0.9688 0.9617
FTSE 350 (0.0000) (0.6611) (0.5364) (0.0729) (0.1962) (0.3796) (0.4399)
MA term is insignificant in both models conditional mean equation this show no
relationship between past and current variations of return series at first lag value. The AR term
82
is insignificant in both models conditional mean equations which shows current return of
markets are not depends upon lag values. ARCH and GARCH terms in conditional dispersion
residuals.
Table 4.4.27 and 4.4.28 given below results describe that the parameters of return series
𝜋1 of FTSE 350 and GDAXI are significant in conditional mean equations of DFMGI return
series. It means that the mean spillover effect from FTSE 350 and GDAXI to DFMGI equity
market. This co-movement between equity markets specify that these equity markets are
interlinked. The significant AR terms mean the current return of markets are only depends upon
lag values. The significant MA terms show relationship between past and current variations of
series encompass ARCH and GARCH effect. The persistence of shock of the most of them
are close to 1 which means that ARCH and GARCH effects are existed and take long time for
83
Table 4.4.27 Volatility Spillover effect between FTSE 350 and DFMGI (Bidirectional
Return series FTSE 350 to DFMGI DFMGI to FTSE 350

Constant 0.0004 0.0005
𝛼0 (0.2398) (0.0001)
𝑅𝑡 0.1693 0.0260
𝜋1 (0.0000) (0.0021)
AR(1) 0.8451
𝜗1 (0.0000) -----------
AR(2) 0.0488
𝜗2 (0.0439) -----------
MA(1) -0.8410
∅1 (0.0000) -----------
Constant 0.0232 0.0180
θ0 (0.1740) (0.0026)
𝑅𝑡2 0.0083 -0.0007
𝜋2 (0.3912) (0.2004)
ARCH(1) 0.0659 0.1172
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.9281 0.8745
𝜑1 (0.0000) (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
FTSE 350 to 13621 5.5221 11.183 2.8170 7.1601 0.0624 0.5619
DFMGI (0.0000) (0.0632) (0.1308) (0.4206) (0.5194) (0.9395) (0.7293)
DFMGI to 118.41 3.4716 4.7733 3.7189 5.3709 0.6161 0.7641
FTSE 350 (0.0000) (0.6276) (0.9057) (0.2934) (0.7172) (0.5401) (0.5756)
84
Table 4.4.28 Volatility Spillover effect between GDAXI and DFMGI (Bidirectional
Return series GDAXI to DFMGI DFMGI to GDAXI
Constant 0.0004 0.0009
𝛼0 (0.2555) (0.0000)
𝑅𝑡 0.1436 0.0317
𝜋1 (0.0000) (0.0039)
AR(1) 0.8520 0.9381
𝜗1 (0.0000) (0.0000)
MA(1) -0.8450 -0.9544
∅1 (0.0000) (0.0000)
MA(2) 0.0501
∅2 (0.0349) -----------
Constant 0.0306 0.0246
θ0 (0.0541) (0.0160)
𝑅𝑡2 -0.0020 -0.0004
𝜋2 (0.4851) (0.7247)
ARCH(1) 0.0637 0.0998
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.9322 0.8918
𝜑1 (0.0000) (0.0000)

nth
Residual Analysis
Parameter Jarque Q-Stat Q-Stat Q2-Stat Q2-Stat LM - LM-
Series Bera (5) (10) (5) (10) ARCH ARCH
(1-2) (1-5)
GDAXI to 13501 6.6508 13.082 3.1148 6.9068 0.2242 0.6248
DFMGI (0.0000) (0.0359)* (0.07013) (0.3742) (0.54671) (0.7991) (0.6809)
DFMGI to 259.39 4.0615 7.1350 7.6228 9.4610 1.0130 1.5710
GDAXI (0.0000) (0.25489) (0.52211) (0.05448) (0.3049) (0.3633) (0.1648)
The validity of results is also approved by the residual analysis. Table 4.4.27 and 4.4.28
also demonstrate the post estimation results (Residual analysis). The Jarque Bera test
85
4.5 Tracing direct and indirect linkages between Pakistan, US and Dubai
stock markets
Now we explore the spillover effect between Pakistani (KSE 100), US (S&P 500,
NASDAQ 100 and DJI) and Dubai financial market (DFMGI) by using daily data from 3rd Jan,
2005 to 28th Nov, 2014. After that we split the data set into two parts. First part of data
incorporate the era of global financial crisis 2008 from 3rd Jan, 2005 to 31st Dec, 2009. Second
part after global financial crisis period from 4th Jan, 2010 to 28th Nov, 2014. We again explore
the spillover effect by using these sub sets.
Table 4.5.1 given below clarifies that the parameter of return series 𝜋1 and parameter
of squared return series 𝜋2 of S&P 500 are statistically significant in conditional mean and
variance equations of KSE 100. This show the mean and volatility spillover effect from S&P
500 to KSE 100 is found.. The parameter values in conditional mean and variance equations
explore that there is no mean and volatility spillover effect from KSE 100 to S&P 500. The
parameter values of return series of S&P 500 in conditional mean equation of KSE 100 explain
there is 0.6% mean spillover effect from S&P 500 to KSE 100. The parameter values of squared
return series of S&P 500 in conditional variance equation of KSE 100 explain there is -0.1%
volatility spillover effect from S&P 500 to KSE 100.
86
Table 4.5.1 Volatility Spillover effect between KSE 100 and S&P 500 (Bidirectional
Return series KSE 100 to S&P 500 S&P 500 to KSE 100
Constant 0.0008 0.0011
𝛼0 (0.0000) (0.0000)
𝑅𝑡 0.0169 0.0067
𝜋1 (0.1293) (0.0000)
AR(1) 0.7428 0.9250
𝜗1 (0.0000) (0.0000)
MA(1) -0.8050 -0.7804
∅1 (0.0000) (0.0000)
MA(2) -0.0905
∅2 ----------- (0.0000)
Constant 0.0205 0.0365
θ0 (0.0008) (0.0000)
𝑅𝑡2 0.0015 -0.0010
𝜋2 (0.4331) (0.0000)
ARCH(2) 0.1427 0.0722
𝜃2 (0.0000) (0.0000)
GARCH(1) 0.8473 0.841
𝜑1 (0.0000) (0.0000)
GJR(1) 0.1380
𝛿1 ----------- (0.0000)

0 No volatility spillover. Leverage effect H0: 𝛿𝑖 = 0 No leverage effect. P-values are in the parenthesis.
Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
KSE 100 to 519.91 4.5809 7.5260 1.1042 9.8357 0.2195 0.2308
S&P 500 (0.0000) (0.2051) (0.4810) (0.5757) (0.1980) (0.8029) (0.9492)
S&P 500 to 4157.1 1.0655 9.6531 0.8109 2.9143 0.1368 0.1614
KSE 100 (0.0000) (0.5869) (0.2090) (0.8468) (0.9396) (0.8721) (0.9765)
This co-movement between equity markets specify that the equity markets are directly
interlinked with each other. MA term is significant in both models conditional mean equation
87
this show relationship between past and current variations of return series at first lag value. The
AR term is significant in both models conditional mean equations which shows current return
of markets are depends upon lag values. ARCH and GARCH terms in conditional dispersion
residuals.
In the same fashion Tables 4.5.2, 4.5.3, 4.5.4, 4.5.6 and 4.5.7 given below results
explain that the parameter of return series 𝜋1 and parameter of squared return series 𝜋2 of S&P
500, NASADQ 100, DJI and DFMGI in conditional mean and variance equations of KSE 100
are statistically significant. It means that the mean and volatility spillover effect from S&P 500,
NASADQ 100, DJI and DFMGI to KSE 100. This co-movement between equity markets
specify that the equity markets are interlinked with each other. The results also represent that
there is no mean and volatility spillover effect from KSE 100 to S&P 500, NASADQ 100, and
DJI. Table 4.5.4 results explain bidirectional mean and volatility spillover effect between KSE
100 to DFMGI. Table 4.5.5, 4.5.6 and 4.5.7 results represent that there is only bidirectional
mean spillover effect between S&P 500, NASADQ 100, DJI and DFMGI.𝛿1 Term describes
88
leverage effect is significant in GJR-GARCH models which indicates that negative correlation
between assets returns and assets volatility.
Table 4.5.2 Volatility Spillover effect between KSE100 and NASDAQ100 (Bidirectional
Return series KSE 100 to NASDAQ 100 NASDAQ 100 to KSE 100

Constant 0.0010 0.0014
𝛼0 (0.0000) (0.0000)
𝑅𝑡 0.0119 0.0116
𝜋1 (0.4178) (0.0000)
AR(1) -0.0372 0.5603
𝜗1 (0.0475) (0.0000)
MA(1) -0.4115
∅1 ----------- (0.0000)
Constant 0.0207 0.0320
θ0 (0.0016) (0.0000)
𝑅𝑡2 0.0045 -0.0009
𝜋2 (0.0973) (0.0000)
ARCH(1) 0.0886 0.0985
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.8961 0.8387
𝜑1 (0.0000) (0.0000)
GJR(1) 0.1166
𝛿1 ----------- (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
KSE 100 to 211.08 2.7090 5.7941 5.9844 15.662 2.7596 1.1853
NASDAQ 100 (0.0000) (0.6074) (0.7603) (0.1123) (0.0474) (0.0635) (0.3138)
NASDAQ 100 to 5847.7 3.5112 16.882 0.8004 2.6895 0.1104 0.1585
KSE 100 (0.0000) (0.3193) (0.0313)* (0.8493) (0.9523) (0.8954) (0.9775)
89
Table 4.5.3 Volatility Spillover effect between KSE 100 and DJI (Bidirectional Analyses
Return KSE 100 to DJI DJI to KSE 100
series ARMA(0,0) GARCH (1,1) ARMA(1,1) GJR (1,1)
Parameters
Constant 0.0007 0.0006
𝛼0 (0.0000) (0.0000)
𝑅𝑡 0.0096 -0.0019
𝜋1 (0.3784) (0.0000)
AR(1) 0.7816
𝜗1 ----------- (0.0000)
MA(1) -0.6808
∅1 ----------- (0.0000)
Constant 0.0114 0.0935
θ0 (0.0033) (0.0000)
𝑅𝑡2 0.0022 -0.0024
𝜋2 (0.1750) (0.0000)
ARCH(1) 0.1095 0.0622
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.8827 0.7671
𝜑1 (0.0000) (0.0000)
GJR(1) 0.2599
𝛿1 ----------- (0.0000)

nth
Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
KSE 100 to 441.53 11.755 14.776 8.3740 18.127 3.8385 1.6138
DJI (0.0000) (0.038)* (0.1404) (0.0388)* (0.0202)* (0.0216)* (0.1529)
DJI to 1198.1 4.8240 16.462 0.86921 2.1575 0.0613 0.1691
KSE 100 (0.0000) (0.1851) (0.036)* (0.8328) (0.9757) (0.9405) (0.9740)
90
Table 4.5.4 Volatility Spillover effect between KSE 100 and DFMGI (Bidirectional
Return series KSE 100 to DFMGI DFMGI to KSE 100
Constant 0.0005 0.0009
𝛼0 (0.1725) (0.0000)
𝑅𝑡 0.0650 0.0153
𝜋1 (0.0443) (0.0000)
AR(1) 0.8814 0.1268
𝜗1 (0.0000) (0.0000)
MA(1) -0.8325
∅1 (0.0000) -----------
Constant 0.0178 0.0450
θ0 (0.2500) (0.0000)
𝑅𝑡2 0.0185 -0.0011
𝜋2 (0.0788) (0.0000)
ARCH(1) 0.0597 0.1783
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.9277 0.8085
𝜑1 (0.0000) (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
KSE 100 to 13571 4.7337 10.608 3.0709 7.4353 0.1503 0.6138
DFMGI (0.0000) (0.1923) (0.2248) (0.3808) (0.4904) (0.8604) (0.6893)
DFMGI to 2453.1 10.055 26.834 1.1986 4.4112 0.1127 0.2381
KSE 100 (0.0000) (0.0395)* (0.0148)* (0.7533) (0.8182) (0.8934) (0.9457)
91
Table 4.5.5 Volatility Spillover effect between S&P 500 and DFMGI (Bidirectional
Return series S&P 500 to DFMGI DFMGI to S&P 500
Constant 0.0005 0.0008
𝛼0 (0.1927) (0.0000)
𝑅𝑡 0.0820 0.0161
𝜋1 (0.0028) (0.0328)
AR(1) 0.8883 0.7296
𝜗1 (0.0000) (0.0000)
MA(1) -0.8392 -0.7904
∅1 (0.0000) (0.0000)
Constant 0.0265 0.0093
θ0 (0.0998) (0.0001)
𝑅𝑡2 0.0036 -0.0002
𝜋2 (0.5462) (0.1585)
ARCH(1) 0.0647 0.0522
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.9298 1.5895
𝜑1 (0.0000) (0.0000)
GARCH(2) 0.6451
𝜑2 ----------- (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
S&P 500 to 13131 7.6998 13.595 3.1527 7.9966 0.1441 0.7641
DFMGI (0.0000) (0.0526) (0.0929 (0.3686) (0.4337) (0.8658) (0.6638)
DFMGI to 491.25 4.6889 6.6039 5.8411 12.523 2.5586 1.1151
S&P 500 (0.0000) (0.1960) (0.5799) (0.0539) (0.0845) (0.0776) (0.3500)
92
Table 4.5.6 Volatility Spillover effect between NASDAQ 100 and DFMGI (Bidirectional
Return series NASDAQ 100 to DFMGI DFMGI to NASDAQ 100
Constant 0.0004 0.0010
𝛼0 (0.2140) (0.0000)
𝑅𝑡 0.0671 0.0217
𝜋1 (0.0056) (0.0257)
AR(1) 0.8888
𝜗1 (0.0000) -----------
MA(1) -0.8401
∅1 (0.0000) -----------
Constant 0.0274 0.0247
θ0 (0.0761) (0.0007)
𝑅𝑡2 0.0022 -0.0003
𝜋2 (0.7160) (0.6331)
ARCH(1) 0.0643 0.0901
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.9302 0.8976
𝜑1 (0.0000) (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) ARCH
(1-5)
NASDAQ 100 to 13236 7.6817 13.448 3.1950 7.9185 0.1746 0.6385
DFMGI (0.0000) (0.0530) (0.0973) (0.3625) (0.4414) (0.8398) (0.6704)
DFMGI to 191.57 6.8647 9.7689 5.6420 14.328 2.6055 1.1185
NASDAQ 100 (0.0000) (0.2308) (0.4609) (0.1303) (0.0736) (0.0741) (0.3482)
93
Table 4.5.7 Volatility Spillover effect between DJI and DFMGI (Bidirectional Analyses
Return series DJI to DFMGI DFMGI to DJI
Constant 0.0005 0.0007
𝛼0 (0.1949) (0.0000)
𝑅𝑡 0.0827 0.0159
𝜋1 (0.0056) (0.0240)
AR(1) 0.8878 -0.0590
𝜗1 (0.0000) (0.0000)
MA(1) -0.8386
∅1 (0.0000) -----------
Constant 0.0258 0.0082
θ0 (0.1089) (0.0001)
𝑅𝑡2 0.0046 -0.0001
𝜋2 (0.5194) (0.3720)
ARCH(1) 0.0645 0.0527
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.9302 1.5924
𝜑1 (0.0000) (0.0000)
GARCH(2) 0.6489
𝜑2 ----------- (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) ARCH
(1-5)
DJI to 13059 7.5169 13.317 3.2428 8.1727 0.1524 0.6473
DFMGI (0.0000) (0.0571) (0.1013) (0.3556) (0.4167) (0.8586) (0.6636)
DFMGI to 377.36 3.8701 6.8827 3.3924 9.9342 1.1531 0.6552
DJI (0.0000) (0.4238) (0.6493) (0.1833) (0.1923) (0.3158) (0.6575)
The significant AR terms mean the current return of markets are only depends upon lag
values. The significant MA terms show relationship between past and current variations of
94
series encompass ARCH and GARCH effect. The persistence of shock of the most of them are
close to 1 which means that ARCH and GARCH effects are existed and take long time for
The validity of results is also approved by the residual analysis. Table 4.5.2, 4.5.3, 4.5.4,
4.5.5, 4.5.6 and 4.5.7 illustrate the post estimation results (Residual analysis). The Jarque Bera
test (Normality test) results show, its parameter is statistically significant reject hull hypothesis
means the distribution of standardized residuals is still non-normal. The Q-stat are insignificant
up to 10th lags accept null hypothesis means no serial autocorrelation in the standardized
residuals. The Q-stat square parameters are insignificant up to 10th lags accept null hypothesis
residuals.
4.6 Tracing Spillover effect (3rd Jan, 2005 to 31st Dec, 2009)
First part of data incorporate the era of global financial crisis 2008 from 3rd Jan, 2005
to 31st Dec, 2009, because the global financial crisis started from mid-2007, 7th Sep, 2008 at its
boom and eliminate its effect end 2009. We explore the spillover effect among Pakistani (KSE
100), US (S&P 500, NASDAQ 100 and DJI) and Dubai financial market (DFMGI). The results
of first part of data explore the direct and indirect linkages between these markets in crisis
period.
Table 4.6.1 given below describes that the parameter of return series 𝜋1 and parameter
of squared return series 𝜋2 of S&P 500 are statistically significant in conditional mean and
95
500 to KSE 100 is found.
Analyses for period 3rd Jan, 2005 to 31st Dec, 2009)
Constant 0.0006 0.0008
𝛼0 (0.0001) (0.0000)
𝑅𝑡 0.0229 0.0087
𝜋1 (0.0709) (0.0000)
AR(1) 0.6767 0.9141
𝜗1 (0.0000) (0.0000)
MA(1) -0.7625 -0.8581
∅1 (0.0000) (0.0000)
Constant 0.0047 0.0714
θ0 (0.2418) (0.0000)
𝑅𝑡2 0.0013 -0.0018
𝜋2 (0.3111) (0.0000)
ARCH(1) 0.0861 0.1322
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.9135 0.8364
𝜑1 (0.0000) (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
KSE 100 to 755.67 3.5942 7.4527 3.4164 12.079 1.6175 0.7003
S&P 500 (0.0000) (0.3087) (0.4886) (0.3317) (0.1476) (0.1988) (0.6233)
S&P 500 to 727.74 6.2326 9.9602 5.4904 7.0424 2.3411 1.0386
KSE 100 (0.0000) (0.1008) (0.2678) (0.1392) (0.5320) (0.0966) (0.3933)
Results also describe that the parameter of return series 𝜋1 of KSE 100 is statistically
significant in conditional mean equation of S&P 500 which means there also found mean
96
spillover effect from KSE 100 to S&P 500. The parameter values in conditional mean equation
explore that there is 2% mean spillover from KSE 100 to S&P 500 and 0.8% from S&P 500 to
KSE 100. The parameter values of squared return series of S&P 500 in conditional dispersion
equation of KSE 100 explain there is -1.8% volatility spillover effect from S&P 500 to KSE
100. This co-movement between equity markets specify that the equity markets are directly
interlinked with each other. MA term is significant in both models conditional mean equation
this show relationship between past and current variations of return series at first lag value. The
AR term is significant in both models conditional mean equations which shows current return
of markets are depends upon lag values. ARCH and GARCH terms in conditional dispersion
residuals.
Table 4.6.2, 4.6.3, 4.6.4, 4.6.5, 4.6.6 and 4.6.7 Results defines that the parameter of
return series 𝜋1 and parameter of squared return series 𝜋2 of NASADQ 100, DJI and DFMGI
in conditional mean and variance equations of KSE 100 are statistically significant. It means
that the mean and volatility spillover effect from S&P 500, NASADQ 100, DJI and DFMGI to
KSE 100.
97
Constant 0.0006 0.0008
𝛼0 (0.0421) (0.0000)
𝑅𝑡 0.0187 0.0083
𝜋1 (0.2924) (0.0000)
AR(1) 0.1564
𝜗1 ----------- (0.0000)
AR(2) 0.0412
𝜗2 ----------- (0.0000)
AR(3) 0.0007
𝜗3 ----------- (0.0000)
Constant 0.0189 0.0725
θ0 (0.0965) (0.0000)
𝑅𝑡2 0.0033 -0.0017
𝜋2 (0.2286) (0.0000)
ARCH(1) 0.1428
𝜃1 ----------- (0.0000)
ARCH(3) 0.0926
𝜃3 (0.0000) -----------
GARCH(1) 0.8947 0.8281
𝜑1 (0.0000) (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
KSE 100 to 102.98 4.4689 7.3220 2.9350 16.236 0.8027 0.5791
NASDAQ 100 (0.0000) (0.4840) (0.6947) (0.0866) (0.0125)* (0.4483) (0.7160)
NASDAQ 100 to 754.25 4.1406 14.419 3.5894 4.9040 1.3555 0.6920
KSE 100 (0.0000) (0.1261) (0.044)* (0.3093) (0.7677) (0.2582) (0.6295)
98
for period 3rd Jan, 2005 to 31st Dec, 2009)
Return series KSE 100 to DJI DJI to KSE 100
Constant 0.0005 0.0006
𝛼0 (0.0003) (0.0000)
𝑅𝑡 0.0257 0.0027
𝜋1 (0.0450) (0.0000)
AR(1) 0.7069
𝜗1 (0.0000) -----------
MA(1) -0.7650 0.1225
∅1 (0.0000) (0.0000)
Constant 0.0034 0.126
θ0 (0.3433) (0.0000)
𝑅𝑡2 0.0017 -0.0033
𝜋2 (0.2011) (0.0000)
ARCH(1) 0.0817 0.0842
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.9165 0.7809
𝜑1 (0.0000) (0.0000)
GJR(1) 0.1628
𝛿1 ----------- (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
KSE 100 to 603.05 5.0334 9.6675 3.4379 12.471 1.6910 0.7115
DJI (0.0000) (0.1693) (0.2891) (0.3289) (0.1313) (0.1847) (0.6148)
DJI to 375.32 11.680 26.188 0.3739 1.9294 0.1046 0.0737
KSE 100 (0.0000) (0.0198)* (0.0190)* (0.9455) (0.9830) (0.9007) (0.9961)
99
Constant 0.0004 0.0015
𝛼0 (0.5766) (0.0000)
𝑅𝑡 -0.0031 0.0292
𝜋1 (0.9356) (0.0000)
AR(1) 0.6865 0.1756
𝜗1 (0.0000) (0.0000)
AR(3) 0.1159
𝜗3 (0.0000) -----------
MA(1) -0.6829
∅1 (0.0000) -----------
Constant 0.0255 0.1225
θ0 (0.3046) (0.0000)
𝑅𝑡2 0.0046 -0.0029
𝜋2 (0.4866) (0.0000)
ARCH(1) 0.0689 0.1150
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.9322 0.8031
𝜑1 (0.0000) (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
KSE 100 to 10851 4.7721 10.146 0.6018 2.9579 0.0953 0.1199
DFMGI (0.0000) (0.0289)* (0.1186) (0.8960) (0.9369) (0.9090) (0.9880)
DFMGI to 170.47 6.7159 16.182 7.0552 10.817 2.7588 1.3146
KSE 100 (0.0000) (0.1516) (0.0631) (0.0701) (0.2122) (0.0637) (0.2551)
100
Constant 0.0003 0.0005
𝛼0 (0.6309) (0.0024)
𝑅𝑡 -0.0015 0.0025
𝜋1 (0.9693) (0.8135)
AR(1) 0.6903
𝜗1 (0.0000) -----------
AR(3) 0.1175
𝜗3 (0.0003) -----------
MA(1) -0.6868 -0.0866
∅1 (0.0000) (0.0011)
Constant 0.0313 0.0069
θ0 (0.2223) (0.1403)
𝑅𝑡2 0.0106 -0.0001
𝜋2 (0.3314) (0.8328)
ARCH(1) 0.0676 0.0842
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.9306 0.9169
𝜑1 (0.0000) (0.0000)
Persistence of shock 0.9983 1.0000*

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) ARCH
(1-5)
S&P 500 to 10971 5.1041 10.180 0.6686 3.3012 0.0775 0.1334
DFMGI (0.0000) (0.0238) (0.1172) (0.8805) (0.9140) (0.9254) (0.9847)
DFMGI to 491.20 2.7170 8.2260 4.0943 14.818 1.8780 0.8505
S&P 500 (0.0000) (0.6062) (0.5115) (0.2514) (0.0627) (0.1533) (0.5139)
101
Constant 0.0003 0.0009
𝛼0 (0.6321) (0.0000)
𝑅𝑡 -0.0162 0.0402
𝜋1 (0.6396) (0.0026)
AR(1) 0.6887 0.9756
𝜗1 (0.0000) (0.0000)
AR(3) 0.1161
𝜗3 (0.0000) -----------
MA(1) -0.6851 -0.9921
∅1 (0.0000) (0.0000)
Constant 0.0373 0.0531
θ0 (0.1097) (0.0002)
𝑅𝑡2 0.0024 -0.0022
𝜋2 (0.8169) (0.0257)
ARCH(1) 0.0690 0.1362
𝜃1 (0.0000) (0.0000)
ARCH(3) 0.092347
𝜃3 ----------- (0.0000)
GARCH(1) 0.9309 0.8316
𝜑1 (0.0000) (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
NASDAQ 100 to 10596 4.9854 10.190 0.6827 3.0835 0.0925 0.1363
DFMGI (0.0000) (0.0250)* (0.1168) (0.8772) (0.9290) (0.9116) (0.9839)
DFMGI to 90.799 3.8728 4.6946 5.7397 9.7055 2.4636 1.0966
NASDAQ 100 (0.0000) (0.2755) (0.7896) (0.1249) (0.2863) (0.0855) (0.3604)
102
for period 3rd Jan, 2005 to 31st Dec, 2009)
Constant 0.0003 0.0006
𝛼0 (0.6203) (0.0000)
𝑅𝑡 -0.0001 0.0008
𝜋1 (0.9992) (0.9273)
AR(1) 0.6895 -0.0604
𝜗1 (0.0000) (0.0140)
AR(3) 0.1174
𝜗3 (0.0000)
MA(1) -0.6862
∅1 (0.0000)
Constant 0.0292 0.0062
θ0 (0.2533) (0.1482)
𝑅𝑡2 0.0122 -0.0001
𝜋2 (0.3262) (0.7943)
ARCH(1) 0.0680 0.0819
𝜃1 (0.0000) (0.0000)
GARCH(1) 0.9310 0.9185
𝜑1 (0.0000) (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
DJI to 11009 5.1051 10.190 0.66797 3.2496 0.07629 0.1332
DFMGI (0.0000) (0.0238)* (0.1168) (0.8807) (0.9177) (0.9265) (0.9847)
DFMGI to 419.48 3.2679 9.6662 3.7201 13.603 1.8006 0.7850
DJI (0.0000) (0.5140) (0.3781) (0.2933) (0.0927) (0.1656) (0.5605)
This co-movement between equity markets specify that the equity markets are
interlinked with each other. The results also represent that there is only mean spillover effect
103
from KSE 100 to NASADQ 100, and DJI. The results explain unidirectional mean and
volatility spillover effect from DFMGI to KSE 100 is found. The results also expose that there
is no mean and volatility spillover effect from DFMGI to S&P 500 and DJI. Table 4.6.6 results
illustrate that the unidirectional mean and volatility spillover effect from DFMGI to NASDAQ
100. 𝛿1 Term describes leverage effect is significant in GJR-GARCH models which indicates
that negative correlation between assets returns and assets volatility. The significant AR terms
mean the current return of markets are only depends upon lag values. The significant MA terms
show relationship between past and current variations of return series. ARCH and GARCH
terms are also significant in above models means these return series encompass ARCH and
GARCH effect. The persistence of shock of the most of them are close to 1 which means that
ARCH and GARCH effects are existed and take long time for decay. Those values which are
not close to 1 show ARCH and GARCH effects are existed and take short time for decay.
residuals.
4.7 Tracing Spillover effect (4th Jan, 2010 to 28th Nov, 2014)
Second part of data incorporate the time period of after global financial crisis 2008 from
4th Jan, 2010 to 28th Nov, 2014, because the after global financial crisis market moved towards
boom. We trace out the spillover effect among Pakistani (KSE 100), US (S&P 500, NASDAQ
104
100 and DJI) and Dubai financial market (DFMGI). The results of second part of data explore
the direct and indirect linkages between these markets in after crisis period.
Table 4.7.1 given below states that the parameter of return series 𝜋1 and parameter of
squared return series 𝜋2 of S&P 500 are statistically significant in conditional mean and
500 to KSE 100 is found. There is no mean and volatility spillover effect KSE 100 to S&P 500
is found. The parameter values in conditional mean equation explore that there is 20% mean
spillover from S&P 500 to KSE 100. The parameter values of squared return series of S&P
500 in conditional dispersion equation of KSE 100 explain there is 6.1% volatility spillover
effect from S&P 500 to KSE 100. This co-movement between equity markets specify that the
equity markets are directly interlinked with each other. MA term is significant in both models
conditional mean equation this show relationship between past and current variations of return
series at first lag value. The AR term is significant in both models conditional mean equations
which shows current return of markets are depends upon lag values. ARCH and GARCH terms
in conditional dispersion equations are significant means ARCH and GARCH effect also exist
in return series. The persistence of shock of the most of them are close to 1 which means that
ARCH and GARCH effects are existed and take long time for decay. Those values which are
not close to 1 show ARCH and GARCH effects are existed and take short time for decay.
residuals.
105
Analyses for period 4th Jan, 2010 to 28th Nov, 2014)
Constant 0.0007 14.4714
𝛼0 (0.0001) (0.0000)
𝑅𝑡 0.0022 0.0614
𝜋1 (0.9195) (0.2094)
AR(1) 0.9616
𝜗1 (0.0000) -----------
MA(1) -0.9878 0.1221
∅1 (0.0000) (0.0000)
Constant 3.7150 14.471
θ0 (0.0031) (0.0000)
𝛷2 -0.0084 0.0614
𝜋2 (0.4739) (0.0024)
ARCH(1) 0.1583 0.0816
𝜃1 (0.0000) (0.0125)
ARCH(2) 0.1013
𝜃2 ----------- (0.0252)
ARCH(5) 0.0935
𝜃5 ----------- (0.0959)
ARCH(10) 0.0739
𝜃10 ----------- (0.0227)
GARCH(1) 0.8247 0.3981
𝜑1 (0.0000) (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
KSE 100 to 125.18 1.6992 4.8306 10.181 13.011 4.5706 1.9114
S&P 500 (0.0000) (0.6371) (0.7755) (0.0170)* (0.1114) (0.0105)* (0.0896)
S&P 500 to 132.85 4.7610 11.472 12.716 46.567 0.2405 0.7452
KSE 100 (0.0000) (0.3127) (0.2446) (0.17583) (0.1890) (0.7862) (0.5896)
106
Constant 0.0009 0.0014
𝛼0 (0.0421) (0.0000)
𝑅𝑡 -0.0239 -0.0283
𝜋1 (0.2924) (0.2447)
AR(2) 0.9779
𝜗2 (0.0000) -----------
AR(3) -0.9925
𝜗3 (0.0000) -----------
MA(1) 0.1267
∅1 ----------- (0.0000)
Constant 0.0461 15.988
θ0 (0.0965) (0.0007)
𝑅𝑡2 -0.0055 0.0411
𝜋2 (0.2286) (0.0194)
ARCH(1) 0.1369 0.0923
𝜃1 (0.0000) (0.0075)
ARCH(2) 0.1121
𝜃2 ----------- (0.0155)
ARCH(5) 0.0993
𝜃5 ----------- (0.0730)
ARCH(10) 0.0768
𝜃10 ----------- (0.0160)
GARCH(1) 0.8391 0.3630
𝜑1 (0.0000) (0.0000)

th
Null Hypotheses(All Null Hypotheses are for n order)
Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
KSE 100 to 110.90 3.3126 4.0554 5.1164 8.5917 2.1784 0.9916
NASDAQ 100 (0.0000) (0.3458) (0.8520) (0.1634) (0.3778) (0.1137) (0.4215)
NASDAQ 100 to 133.79 4.7878 11.601 13.348 48.772 0.2321 0.7127
KSE 100 (0.0000) (0.3097) (0.2367) (0.3093) (0.1357) (0.7929) (0.6139)
107
for period 4th Jan, 2010 to 28th Nov, 2014)
Return series KSE 100 to DJI DJI to KSE 100
Constant 0.0008 0.0014
𝛼0 (0.0003) (0.0000)
𝑅𝑡 -0.0167 -0.0442
𝜋1 (0.4574) (0.1478)
AR(1) 0.9355
𝜗1 (0.0000) -----------
MA(1) -0.9616 0.1220
∅1 (0.0000) (0.0000)
Constant 2.9780 14.095
θ0 (0.0016) (0.0012)
𝑅𝑡2 -0.0047 0.0664
𝜋2 (0.6145) (0.0034)
ARCH(1) 0.1632 0.0832
𝜃1 (0.6145) (0.0134)
ARCH(2) 0.0937
𝜃2 ----------- (0.0363)
ARCH(5) 0.0948
𝜃5 ----------- (0.0911)
ARCH(10) 0.0741
𝜃10 ----------- (0.0231)
GARCH(1) 0.8196 0.4145
𝜑1 (0.0000) (0.0000)

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
KSE 100 to 104.37 3.1643 5.9019 9.6225 12.303 4.1240 1.7742
DJI (0.0000) (0.3669) (0.6582) (0.0220)* (0.1381) (0.0164)* (0.1152)
DJI to 135.89 5.0008 11.379 12.612 45.535 0.1805 0.7726
KSE 100 (0.0000) (0.2872) (0.2506) (0.1809) (0.2186) (0.8349) (0.5695)
108
Constant -0.0002 0.0013
𝛼0 (0.0647) (0.0000)
𝑅𝑡 0.0174 0.0490
𝜋1 (0.0006) (0.0000)
AR(1) 0.1756
𝜗1 ----------- (0.0000)
MA(1) -0.0148 0.1172
∅1 (0.0000) (0.0000)
Constant 0.0000 0.1172
θ0 (1.0000) (0.0000)
𝑅𝑡2 -0.0003 0.0060
𝜋2 (0.0000) (0.5071)
ARCH(1) 0.0000 0.1015
𝜃1 (1.0000) (0.0062)
ARCH(2) 0.0000 0.1236
𝜃2 (1.0000) (0.0122)
ARCH(3) 0.0000
𝜃3 (1.0000) -----------
ARCH(4) 0.0000
𝜃4 (1.0000) -----------
ARCH(5) 0.2509 0.1031
𝜃5 (0.0000) (0.0532)
ARCH(6) 0.0000 0.0733
𝜃6 (1.0000) (0.0117)
ARCH(7) 0.0000
𝜃7 (1.0000) -----------
GARCH(1) 0.0000 0.3437
𝜑1 (1.0000) (0.0000)
GARCH(2) 0.0000
𝜑2 (1.0000) -----------
GARCH(3) 0.0000
𝜑3 (1.0000) -----------
GARCH(4) 0.0000
𝜑4 (1.0000) -----------
GARCH(5) 0.8337
𝜑5 (0.0000) -----------

109
Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
KSE 100 to 1.0657 1.5481 2.0050 0.0788 0.2140 0.0031 0.0033
DFMGI (0.0000) (0.8180) (0.9913) (0.9999) (1.0000) (0.9968) (1.0000)
DFMGI to 154.08 4.9840 12.291 9.7777 45.508 0.1125 0.4484
KSE 100 (0.0000) (0.2889) (0.1973) (0.36877) (0.2194) (0.8936) (0.8146)
Constant 0.0001 0.0006
𝛼0 (0.5766) (0.0000)
𝑅𝑡 0.0014 0.0405
𝜋1 (0.1548) (0.0017)
AR(1) 0.9625
𝜗1 ----------- (0.0000)
MA(1) -0.0141 -0.9897
∅1 (0.0000) (0.0000)
Constant 0.0000 3.7556
θ0 (0.1340) (0.0001)
𝑅𝑡2 -0.0030 -0.0014
𝜋2 (0.0000) (0.0322)
ARCH(1) 0.0050 0.1583
𝜃1 (0.0000) (0.0000)
ARCH(2) 0.0001 0.1236
𝜃2 (0.0000) (0.0122)
ARCH(3) 0.0003
𝜃3 (0.0000) -----------
ARCH(4) 0.0001
𝜃4 (0.0000) -----------
ARCH(5) 0.2756
𝜃5 (0.0000) -----------
ARCH(10) 0.0001
𝜃10 (0.9947) -----------
GARCH(1) 0.0000 0.8177
𝜑1 (0.8533) (0.0000)
110
GARCH(2) 0.0004
𝜑2 (0.0000) -----------
GARCH(3) 0.0000
𝜑3 (0.9952) -----------
GARCH(4) 0.0000
𝜑4 (0.9973) -----------
GARCH(5) 0.8261
𝜑5 (0.0000) -----------

nth
Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
S&P 500 to 1.6469 1.9973 4.1004 0.0499 0.1249 0.0012 0.0017
DFMGI (0.0000) (0.7362) (0.9046) (1.000) (1.000) (0.9988) (1.0000)
DFMGI to 118.63 2.3156 5.5290 9.4953 13.311 4.2169 1.7836
S&P 500 (0.0000) (0.5095) (0.6998) (0.0233)* (0.1015) (0.0150)* (0.1132)
Constant 0.0001 0.0009
𝛼0 (0.0000) (0.0000)
𝑅𝑡 -0.0062 0.0402
𝜋1 (0.0000) (0.0026)
AR(1) 0.9756
𝜗1 ----------- (0.0000)
MA(1) -0.0089 -0.9921
∅1 (0.0000) (0.0000)
Constant 0.0001 0.0531
θ0 (0.0000) (0.0001)
𝑅𝑡2 -0.0008 -0.0022
𝜋2 (0.0000) (0.0257)
ARCH(1) 0.0029 0.1362
𝜃1 (0.0000) (0.0000)
111
ARCH(2) 0.0000 0.1236
𝜃2 (0.5619) (0.0122)
ARCH(3) 0.0000
𝜃3 (0.0000) -----------
ARCH(4) 0.0000
𝜃4 (0.6643) -----------
ARCH(5) 0.2570
𝜃5 (0.0000) -----------
GARCH(1) 0.0000 0.8316
𝜑1 (0.0034) (0.0000)
GARCH(2) 0.0000
𝜑2 (0.9659) -----------
GARCH(3) 0.0000
𝜑3 (0.8673) -----------
GARCH(4) 0.0000
𝜑1 (0.7118) -----------
GARCH(5) 0.7565
𝜑5 (0.0000) -----------

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
NASDAQ 100 to 4.651 3.1361 5.9590 0.9772 3.3843 0.0265 0.0342
DFMGI (0.0000) (0.5353) (0.7440) (0.9998) (1.0000) (0.9738) (0.9994)
DFMGI to 90.799 3.8728 4.6946 5.7397 9.7055 2.4636 1.0966
NASDAQ 100 (0.0000) (0.2755) (0.7896) (0.1249) (0.2863) (0.0855) (0.3604)
for period 4th Jan, 2010 to 28th Nov, 2014)
Constant -0.0000 0.0008
𝛼0 (0.9931) (0.0000)
𝑅𝑡 0.0007 0.0398
𝜋1 (0.7697) (0.0006)
AR(1) 0.9224
𝜗1 ----------- (0.0000)
112
MA(1) -0.0013 -0.9527
∅1 (0.0496) (0.0000)
Constant 0.0001 3.1528
θ0 (0.9673) (0.0001)
𝑅𝑡2 -0.0019 -0.0011
𝜋2 (0.0000) (0.1172)
ARCH(1) 0.0056 0.1639
𝜃1 (0.0000) (0.0000)
ARCH(2) 0.0000 0.1236
𝜃2 (1.0000) (0.0122)
ARCH(3) 0.0009
𝜃3 (0.5983) -----------
ARCH(4) 0.0000
𝜃4 (0.9989) -----------
ARCH(5) 0.3239
𝜃5 (0.0000) -----------
GARCH(1) 0.0003 0.8126
𝜑1 (0.8562) (0.0000)
GARCH(2) 0.0000
𝜑2 (1.0000) -----------
GARCH(3) 0.0007
𝜑3 (0.7291) -----------
GARCH(4) 0.0000
𝜑1 (1.0000) -----------
GARCH(5) 0.6889
𝜑5 (0.0000) -----------
Persistence of shock 0.9766 1.0000*

Residual Analysis
Series Bera (5) (10) (5) (10) (1-2) (1-5)
DJI to 46861 3.2184 6.6630 7.8177 35.876 0.1743 0.4187
DFMGI (0.0000) (0.5219) (0.6721) (0.6466) (0.6563) (0.8400) (0.8359)
DFMGI to 100.44 3.5579 6.3453 9.4192 13.456 4.0144 1.722
DJI (0.0000) (0.3133) (0.6086) (0.0242)* (0.0970) (0.0183)* (0.1264)
113
Table 4.7.2, 4.7.3, 4.7.4, 4.7.5, 4.7.6 and 4.7.7 Results defines that the parameter of
return series 𝜋1 and parameter of squared return series 𝜋2 of NASADQ 100 and DJI in
conditional mean and variance equations of KSE 100 are significant means that the mean and
volatility spillover effect from NASADQ 100 and DJI to KSE 100. The results also represent
that there is no mean and volatility spillover effect from KSE 100 to S&P 500, NASADQ 100,
and DJI. Table 4.7.4 results explain unidirectional mean and volatility spillover effect from
KSE 100 to DFMGI but there is only mean spillover effect from DFMGI to KSE 100. Table
4.7.6 results expose that there is bidirectional mean and volatility spillover effect between
DFMGI and NASDAQ 100. Table 4.7.5 results illustrate that the mean and volatility spillover
effect from DFMGI to S&P 500 but there is only volatility spillover effect from S&P 500 to
KSE 100. Table 4.7.7 results describe that the volatility spillover effect from DJI to DFMGI
but there is only mean spillover effect from DFMGI to DJI. 𝛿1 Term describes leverage effect
is significant in GJR-GARCH models which indicates that negative correlation between assets
returns and assets volatility. The significant AR terms mean the current return of markets are
only depends upon lag values. The significant MA terms show relationship between past and
current variations of return series. ARCH and GARCH terms are also significant in above
models means these return series encompass ARCH and GARCH effect. The persistence of
shock of the most of them are close to 1 which means that ARCH and GARCH effects are
114
residuals.
115
Chapter 5
CONCLUSION AND POLICY IMPLICATION
5.1 Conclusion
This study investigates the direct and indirect dynamic linkages between Pakistani and
leading global stock markets. Daily data are used from 3rd Jan 2005 to 28th Nov 2014. The
whole data set split into two sub sets. First sub of data incorporated the era of global financial
crisis 2008 from 3rd Jan, 2005 to 31st Dec, 2009. Second sub set contain data after global
financial crisis time period from 4th Jan, 2010 to 28th Nov, 2014.
The appropriate univariate GARCH type models i.e. GARCH (p, q) and GJR-GARCH
(p, q) are employed to examine information transmission between stock markets and modeling
volatility. Particularly it examined the fluctuating nature and the magnitude of the mean and
volatility spillover from US, European and Gulf equity markets to Pakistan stock market KSE
100.
The summary statistics illustrates that all the returns series are characterized with
stylized properties of financial return series. The distributions of the return series of equity
markets are non-normal. The return series have high volatility clustering and the persistence of
shock to volatility in most of the models is close to 1 which indicated that the ARCH and
GARCH effect take long time period for decay. The mean and volatility spillover effect are
captured by employing the standard GARCH and GJR model.
We results from whole data set explored the co-movement from leading foreign stock
markets Pakistani stock market. The unidirectional mean and volatility spillover effect from
US stock markets and Dubai financial market to Pakistani stock market is found. The mean
and volatility spillover effect from Pakistani stock market to Dubai financial market is also
116
observed. The results also clarified that there is bidirectional mean and volatility spillover
effect between Dubai financial and US stock markets.
The results from first sub set of data from 3th Jan, 2005 to 31th Dec, 2009 described that
there is only unidirectional mean and volatility spillover effect from Dubai financial and US
stock markets to Pakistani stock market. There is no sign of co-movements between US stock
markets and Dubai financial except NASDAQ 100. The unidirectional mean and volatility
spillover effect from Dubai financial market to NASDAQ 100 is found.
The second sub set of data set from 4th Jan, 2010 to 28th Nov, 2014 results indicated
that only volatility spillover effect from US stock markets to Pakistani stock market. The mean
and volatility spillover effect from Pakistani stock market to Dubai financial market and only
mean spillover effect from Dubai financial market to Pakistani stock market is traced. The
results also explored that the only volatility spillover effect from US stock market to Dubai
financial market and mean and volatility spillover effect from Dubai financial to US stock
markets.
All this means NASDAQ100 is the only one US stock markets which indirectly effect
KSE 100 through DFMGI in crisis period. The all sets which we have used exploring spillover
effect we found these three markets are interlinked. That’s why it may be an indirect linkage.
We concluded that the investors are using these markets in their diversified portfolios.
Despite the war and terror foreign investors are interested in Pakistani stock markets.
Particularly, in Pakistan the investment in energy sector is more attractive for foreign investors.
The boom in KSE 100 is not a bubble created by local investors.
117
5.2 Policy Implications
This study is an important tool for financial institutions, portfolio managers, market players
and academician to diagnose the nature and level of linkages and information transmission
between the financial markets. The financial managers get more understanding about the
management of portfolio which is badly impacted by the stock prices. The market players may
use this information for portfolio diversification and hedging. The policy makers can minimize
the effects of spread of stock prices. The stability of stock prices is very important for portfolio
and foreign direct investments, which improves macroeconomic stability and positively
impacted the economic growth. Through these results the investor/market player of one market
can guess the performance of other markets.
Future Research
This study unseals ways for future research in financial markets by investigating the mean and
volatility spillover effect between Pakistani and foreign financial markets. It can be further
extended by incorporating simultaneous equation model. Furthermore, it is possible to find out
the long run and short run impact of information transmission among financial markets. One
can find out the impact of this co-movement on macro-economic variables of economy.
118
REFERENCES
Andersen, C. G (1978) On the Invertibility of Time Series Models. Stochastic Process and
their Applications, Vol. 8, No. 1, 87-92.
Abou-Zaid, A. S (2011) Volatility Spillover Effects In Emerging MENA Stock Markets.
Review of Applied Economics, Vol. 7, 107-127.
Afzal, R. A ( 2012) Impact of Global Financial Crisis on Stock Markets: Evidence from
Pakistan and India. Journal of Business Management and Economics, Vol. 3 No. 7,
275-282.
Alikhanov, A (2013) To What Extent are Stock Returns Driven by Mean and Volatility
Spillover Effects? – Evidence from Eight European Stock Markets. Review 0f
Economics Perspectives, Vol. 13, 3–29.
ALshbiel, H. A.Z ( 2012) Multivariate Volatility and Spillover Effects in Financial Markets
Case Stusy USA and Major European Stock Markets. Interdisciplinary Journal of
Contemporary Research in Business, Vol. 4, No. 6, 109-113.
Azam Shafique, M. A ( 2012) Impact of Global Financial Crises on the Islamic Banking
System. Arabian Journal of Business and Management Review (OMAN Chapter), Vol.
1, 124-134.
Aziz, K. (2008). Investors from Pakistan buying property worth millions of dollars at UAE
Cityscape 2008. Retrieved from www.chowrangi.pk.
Black, F (1976) Studies of Stock Price Volatility Changes. Journal of American Statistical
Association, 177-181.
119
Bank, W (2009) The Impact of the Financial Crisis on the Developing World. World Bank
Research Digest, Vol. 3, 7-8.
Bollerslev, T (1986) Generalized Autoregressive Conditional Heteroskedasticity. Journal of
Econometrics, Vol. 31, 307-327.
Chan Sok-Gee, M. Z (2010) Volatility Spillovers of the Major Stock Markets in ASEAN-5
with the U.S. and Japanese Stock Markets. International Research Journal of Finance
and Economics, Vol. 44, 156-168.
Chang, H.-F (2012) Comparison of the Volatility of Stock Market Returns in Taiwan, USA
and Europe. International Journal of e-Education, e-Business, e-Management and e-
Learning, Vol. 2, 179-183.
Chaudhary, S. L (2014) Implications of Global Financial Crisis on Textile Industry of Pakistan.
Journal of Law, Policy and Globalization, Vol. 26, 106-114.
Daniel FS Choi, V. F (2009) Volatility Spillovers between New Zealand Stock Market Returns
and Exchange Rate Changes before and after the 1997 Asian Financial Crisis. Asian
Journal of Finance & Accounting, Vol. 1, No. 2, 106-117.
Din, R. A (2010) Economic and Social Impact of Global Financial Crisis: Implications for
Macroeconomic and Development Policies in South Asia. PIDE Monograph Series 1.
Doong, S.Y. Y.C ( 2004) Price and Volatility Spillovers between Stock Prices and Exchange
Rates: Empirical Evidence from the G-7 Countries . International Journal of Business
and Economics, Vol. 3, No. 2, 139-153.
Draz, M. U (2011) Impact of Financial Crisis on Pakistan and China: A Comparative Study of
Six Decades. Journal of Global Business and Economics, Vol. 3, No. 1, 174-186.
120
Ellaboudy, S (2010) The Global Financial Crisis: Economic Impact on GCC Countries and
Policy Implications. International Research Journal of Finance and Economics, 41,
180-193.
Engle, R. F. (1982) Autoregressive Conditional Heteroscedasticity with Estimates of the
Variance of United Kingdom Inflation. Econometrica, Vol. 50, No. 4, 987-1007.
Engle, R. F., & Ng, V. K (1993) Measuring and Testing The Impact of News on Volatality.
The Journal of Finance, Vol. 48, No. 5, 1749-177.
Giles, Y. L ( 2013) Modelling Volatility Spillover Effects Between Developed Stock Markets
and Asian Emerging Stock Markets, Econometrics Working Paper EWP1301, Victoria,
B.C, Canada.
Gaur, S. V (2013) Modeling Volatility: Indian Stock and Foreign Exchange Markets. Journal
of Emerging Issues in Economics, Finance and Banking (JEIEFB), Vol. 2, No. 1, 583-
598.
HAIDER, M. A (2004) Exchange Rate Behaviour after Recent Float: The Experience of
Pakistan. The Pakistan Development Review, Vol. 43, 829–852.
Hamdan, Y (2012) The UAE Country Risk Analysis. European Journal of Business and
Management, Vol 4, No.18, 27-32.
Hasan, Z (2010) Dubai financial crisis: causes, bailout and after - a case study. Journal of
Islamic Banking and Finance, 47-55.
IMF (2009) World Economic Outlook. Washington, DC: World Economic and Financial
Surveys.
121
Jeyanthi, B (2010) Interdependence And Volatility Spillovers Under Market Reforms: The
Case Of National Stock Exchange. International Business & Economics Research
Journal, Vol. 9, No. 9, 77-86.
Kalim, M. A (2012) Empirical Analysis of Food Price Inflation in Pakistan. World Applied
Sciences Journal , Vol.16, No. 7, 933-939.
Khan, A. Q (2014) Dynamic Relationship and Volatility Spillover between the Stock Market
and the Foreign Exchange Market in Pakistan: Evidence from VAR-EGARCH
Modelling. PIDE Working Papers 103.
Khan, M. J (2014) Modelling Exchange Rate Volatility by Macroeconomic Fundamentals in
Pakistan. International Econometric Review (IER), vol. 6, No. 2, 58-76.
Lagesh, P. P (2012) Volatility Spillover and Time-varing Correlation Among the Indian, Asian
and US Stock markets . Journal of Quantitative Economics, Vol. 10, No. 2, 78-90.
Luc Bauwens, S. L (2006) Multivariate GARCH Models: A Survey. Journal of Applied
Econometrics, Vol. 21, 79–109.
Lutuf Ali Phulpoto, A. B ( 2012) Global Financial Crises and its Impact on Banking Sector in
Pakistan. Asian Economic and Social Society, Vol. 2, No. 6, 142-152.
McCartney, M (2011) Pakistan, Growth, Dependency, and Crisis. The Lahore Journal of
Economics, Vol. 16, 71-94.
Mehta, A (2012) Financial Performance of UAE Banking Sector- A Comparison of before and
during Crisis Ratios. International Journal ofTrade, Economics and Finance, Vol. 3,
No. 5, 381-387.
Mishkin, F. S (2011) Over the Cliff: From the Subprime to the Global Financial Crisis. Journal
of Economic Perspectives, Vol. 25, No. 1, 49–70.
122
Mishra, D. K (2008) Sto ck Market Integration and Volatility Spillover:India and its Major
Asian Counterparts. MPRA Paper No. 12788.
Mohammed Ashraf, J. K (2012) An Investigation into the Global Financial Crisis Factors and
a Snap Shot of Their Impact on Different Segments of Economy in a Developing
Country: A Case of Pakistan. International Journal of Academic Research in Business
and Social Sciences, Vol. 2, No. 9, 384-391.
Muhammad Zahid Khalil, M. U (2013) Volatility Spill-over between the Stock Market and the
Foreign Exchange Market in Pakistan. Management and Administrative Sciences
Review, Vol. 2, No. 6, 695-704.
Muhammad Zia-ur-Rehman, S. Z (2011) Sources of Return and Volatility Spillover for
Pakistan: An Analysis of Exogenous Factors by using EGARCH Model. International
Conference on Business and Economics Research, Vol. 16, 151-157.
Nikolova, R. B (2007) Analysis of Mean and Volatility Spillovers Using BRIC . Journal of
Economic Integration, Vol. 22, No. 2, 369-381.
Omar Farooq, M. B (2013) “Day Of The Week” And Its Effect On Stock Market Volatility:
Evidence From An Emerging Market . The Journal of Applied Business Research , Vol.
29, No. 6, 711-716.
Oyefeso, P. F (2001) The UK Stock Market’s Relationship with US and European Stock
Markets: Is the UK Stock Market Snuggling-Up to the US - or to Europe?. University
of Aberdeen.
Qayyum, A. R (2006) Volatility Spillover between the Stock Market and the Foreign Market
in Pakistan. PIDE Working Paper 2006:7.
123
Rahman, Q. Z (2011) The Causality between Stock Market and Foreign Exchange Market of
Pakistan. Interdisciplinary Journal of Contemporary Research in Business, Vol. 3, No.
5, 906-919.
Rizwana Bashir, R. S (2014) The Efficiency of Foreign Exchange Markets in Pakistan: An
Empirical Analysis. The Lahore Journal of Economics, Vol. 19, No. 1 ,133–149.
Robert F. Engle, T. I.L (1990) Meteor Showers or Heat Waves? Heteroskedastic Intra-Daily
Volatility in the Foreign Exchange Market. Econometrica, Vol. 58, No. 3, 525-542.
Sajid Amin Javed, S. A ( 2012) Inflation and Inflation Uncertainty Nexus: Empirical Evidence
from Pakistan. International Journal of Economics and Financial Issues, Vol. 2, No. 3,
348-356.
Safdar Hussain Tahir, H. M (2013) Interdependence of South Asian & Developed Stock
Markets and Their Impact on KSE (Pakistan). Asian Economic and Financial Review,
Vol. 3, No. 1, 16-27.
Safdar, M. I (2013) The Relationship between Macroeconomic Volatility and the Stock Market
Volatility: Empirical Evidence from Pakistan. Pakistan Journal of Commerce and
Social Sciences, Vol. 7, No. 2, 309-320.
Sinha, P. j (2010) Volatility Spillover in India, USA and Japan Investigation of Recession
Effects. MPRA Paper No. 21873.
Suleman Gomez, H. A (2014) Volatility Transmission between Pals, Foes, Minions and Titans.
IOSR Journal of Business and Management (IOSR-JBM), Vol. 16, 122-128.
Tim Bollerslev, J. L (2006) Leverage and Volatility Feedback Effects in High-Frequency Data.
Journal of Financial Econometrics, Vol. 4, No. 3, 353–384.
124
Wongswa, J (2006) Transmissionof Informationacross InternationalEquity Markets. Review
of Financial Studies, Vol. 19, No. 4, 1157-1189.
Yoshida, K. S ( 2013) The global financial crisis: An analysis of the spillover effects on African
stock markets. MPRA Paper No. 50473.
Yasushi Hamao, R. w (1990) Correlations in Price Changes and Volatility across International
Stock Markets. The Review of Financial Studies, Vol. 3, No. 2, 281-307.
Zahoor Hussain Javed, M. F (2009) Economic Growth and Exchange Rate Volatility in Case
of Pakistan. Pakistan Journal of Life and Social Sciences, Vol. 7, No. 2, 112-118.
125

2013 GhulamGhouse

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

2013 GhulamGhouse

Caricato da

Copyright:

Formati disponibili

Tracing Dynamic Linkages and Volatility Spillover

Effect between Pakistani and Foreign Stock Markets

Registration No. 03/M.Phil-ETS/PIDE/2013

Dr. SAUD AHMED KHAN

Department of Econometrics and Statistics

Registration No. 03/M.Phil-ETS/PIDE/2013

Dr. SAUD AHMED KHAN

Department of Econometrics and Statistics

The Most Beneficent

The Most Merciful

“To Allah belongs whatever is in the heavens and whatever is in the

earth. Whether you show what is within yourselves or conceal it,

THE HOLY PROPHET

(Peace and Blessings of Allah be Upon Him)

inheritance), they leave only knowledge, and he who takes it takes an

(Sunan of Abu-Dawood, Hadith 1631)

and US markets through Dubai financial market (DFMGI).

Pakistan and Dubai equity market is also traced.

HAZRAT FATIMA (R.A)

SHAMIM BIBI (MOTHER, LATE)

in a proper way and advise me not to lose heart.

regular basis throughout the session.

List of Figures ........................................................................................................................XII

1.1 Objectives of the study.........................................................................................05

1.2 Research Question ...............................................................................................05

1.3 Significance of the study......................................................................................06

REVIEW OF LITERATURE .............................................................................................07

2.1 International prospective ......................................................................................08

2.2 Global financial crisis impact on Dubai’s economy ............................................09

2.4 Possible channels of information transmission ....................................................13

2.5 Previous empirical studies ..................................................................................16

MODELLING VOLATILITY AND EXPLORING SPILLOVER EFFECT ................18

3.1Econometric Methodology and Model Specification ...........................................19

3.1.1ARCH Model ................................................................................................21

3.1.2 GARCH Models................................................................................................22

3.1.3 GARCH (1, 1) ...................................................................................................23

3.1.4.1 GJR-GARCH Models ............................................................................24

3.1.4.2 GARCH (p, q) Model for Exploring Spillover Effect ………………...25

3.1.5 Residual Analysis..............................................................................................26

3.2 Methodological Structure.....................................................................................27

3.3 Description of Data and sources ..........................................................................27

ESTIMATIONS AND ANALYSIS ....................................................................................29

4.1 Graphical Analysis ..............................................................................................30

4.2 Summary Statistics...............................................................................................37

4.3 Volatility Model specifications of Return Series .................................................38

4.4 Tracing Spillover Effect .......................................................................................44

CONCLUSION AND POLICY IMPLICATIONS ...........................................................116

5.2 Policy Implication ...............................................................................................118

failures in macroprudential and microprudential levels, the accumulation of global balance of

2005 to 28th Nov, 2014.

released first installment of $3.1 billion to meet budget deficit.

DIRECTION OF TRADE RS(Million)

foreign investment in real estate market of Dubai.

were stopped. After that no one ready to invest in these projects.

through Dubai financial market.

1.1 Objective of this research

1.2 Research questions

1.3 Significance of the study

of sustainable economic growth of Pakistan.

financial Econometrics many researchers have presented an enormous empirical and

linkages between Pakistani and foreign stock markets.

of global financial crisis Mishkin (2011).

sensitive to disturbances in US relative to other European economies. Also there exist