Cycles, Contagion and Volatility in the Stock Markets in Middle East and North Africa NFA Ver Version sion Abstract. Using weekly data for ten stock markets in the Middle East and North Africa (MENA) area grouped into three regions, we decompose the market indices in each region into short term cycles and long term trends. We then construct GARCH models for the volatile index cycles in each region to examine their behaviors in response to cycle contagion from other markets of the common region, changes in trends, political events, and changes in oil prices and global financial markets. We make recommendations on spillover versus diversification strategies based on the average cycles’ behaviors. We also measure the persistence of shocks to cycle volatility in each market and discuss the implications to traders, investors, and policy makers. JEL: F 3 1
70
Embed
Cycles, Contagion and Volatility in the Stock Markets in ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Cycles, Contagion and Volatility in the Stock Markets in Middle East
and North Africa NFA VerVersion sion
Abstract. Using weekly data for ten stock markets in the Middle East and North Africa
(MENA) area grouped into three regions, we decompose the market indices in each
region into short term cycles and long term trends. We then construct GARCH models
for the volatile index cycles in each region to examine their behaviors in response to
cycle contagion from other markets of the common region, changes in trends, political
events, and changes in oil prices and global financial markets. We make
recommendations on spillover versus diversification strategies based on the average
cycles’ behaviors. We also measure the persistence of shocks to cycle volatility in each
market and discuss the implications to traders, investors, and policy makers.
Acknowledgement: The authors wish to thank Professor Nadir Habibi of Brandies University and Aksel Kibar of the National Bank of Abu Dhabi for helpful comments, and Dr. Sulayman Al-Qudsi of the Arab Bank for helpful comments, encouragement and for providing us with data. The usual disclaimer applies.
1
Cycles, Contagion and Volatility in the Stock Markets in Middle East
and North Africa
1. Introduction
Speculative attacks such as the 1982 Kuwait market crash, the 1987 US market
crash, the 1994 Mexican crisis are short-lived phenomena that inflect damage the world’s
stock markets. Several stock markets in the Middle East and North Africa (MENA) area,
particularly those in the Gulf region, experienced a devastating collapse in 2006 and
damage in January 2008 despite their strong fundamentals. In these crashes or crises the
markets experience a large drop in stock prices and a dramatic jump in volatility in the
short run. The speculative attacks can be caused by noisy traders who under price
misperceptions bid stock or oil prices away from the fundamentals, leading to
underpricing (De Long et al, 1990).
The stock price or index has two components: transitory known as the short-term
cycle, and permanent known as the long-term trend. The cycle may account for a larger
size of the aggregate index (or price), making harder to understand the behavior of this
index. Moreover, the cycle volatility usually reverts to normal levels quickly, but this
transitory volatility may sometimes linger on longer for some markets more than others
and can cause tremendous damage to wealth and social wellbeing, as happened in Kuwait
in 1982 and the Gulf countries in 2006.
It is important to examine the transitory component of an aggregate index in order
to understand its cycle’s central return and volatility behaviors. This becomes more
important when the research that uses the aggregate market indices does not detect causal
2
relations or information flows between markets. Moreover, the behaviors of the cycle and
its volatility are relevant to all traders, investors and options traders, particularly to those
in the MENA area. Methodologically, because the cycles are stationary they are used in
forecasting to achieve more predictive accuracy. However, the long-run trends are
random walks or nonstationary and thus are not predictable.
Our research on the regions of this area show limited information flows between
the GCC aggregate market indices but hardly any relations between the aggregate indices
of the markets in North African.1 This has motivated us to study the short-run cycle
components of the MENA aggregate indices. As the MENA markets become more
internationalized, they are more subject to market herding, fads, and speculative shocks
as capital starts to flow quickly into and out of these markets. The internationalization of
stock markets in developed countries and many emerging economies is a well-established
fact. However, in many of the stock markets in the Middle East and North Africa
(MENA) area, internationalization is a new phenomenon that is still taking shape.
With strong increases in oil prices and consequently strong economic growth, the
MENA markets have concurrently witnessed varying degrees of liberalization in the
forms of privatized state-owned enterprises and relaxed restrictions on cross-border firm
listing, capital movements and foreign stock ownership. The confluence of the
liberalization reforms with the sharp increases in oil revenues, excess liquidity in some of
these countries and petrodollar spillovers to others have brought about impressive
increases in their market capitalizations, turnovers, IPOs and returns. It is certain that the
movements in the stock market prices have crossed national borders to other countries in
1 The results of VAR/VEC models for the market indices of the regions in the MENA area using the same sample data used in this study are available upon request.
3
the MENA area. The collapse of the stock market in Dubai in 2006 had spread to Saudi
Arabia and the rest of the GCC like a hurricane. The spillover affected other neighboring
countries in varying degrees, hitting for example Egypt more then Jordan. It will also be
interesting to discern whether the short term cyclical components of the index
movements, which are influenced by psychological factors, spill over from one market to
another in the regions.
The MENA countries under study for which we have uninterrupted weekly data
include four Gulf Cooperation Council (GCC) countries (Kuwait, Oman, Saudi Arabia
and UAE),2 three Levant or Mediterranean countries (Jordan, Lebanon and Turkey) and
three North African countries (Egypt, Morocco and Tunisia).
Several studies have examined the dynamic long-run movements and lead/lag
linkages among stock price indexes of MENA stock markets but without decomposing
the indices into cycles and trends to separate the impacts of the psychological factors
from the fundamental. To understand how the market indices behave, we should also
understand how the indices’ components move in addition to understanding their
aggregate movement. Darrat at el, (2000) examine a subset (Egypt, Jordan and Morocco)
of the MENA stock markets, indices and find these countries to be globally segmented
but highly integrated within the region. They also find Egypt to be a dominant force
driving other markets in this region. Omran and Gunduz (20001) examine the dynamic
behavior of the MENA subset (Turkey, Israel, Egypt, Morocco and Jordan) over the
weekly period August 1997 through July 2000 and find no evidence of cointegration
among those five MENA markets. In contrast, Bley (2007) investigates the dynamic
2 Bahrain changed its index in 2004 and Qatar’s uninterrupted data is available starting from 1999 which is shorter than our data. Palestine, Syria and Algeria do not have lengthy and adequate market index data.
4
behavior for fourteen MENA stock markets and finds that these markets to have three
cointegrating vectors over the daily period January 2000-June 2002 and eight vectors
over the period July 2002-December 2004. Using also daily data, Hammoudeh and Aleisa
(2004) analyze the dynamic linkages among five GCC countries (excluding Qatar) over
the period January 1994-December 2001and find two long-run relationships and short-
term lead/lag feedback among the market indices. Bley and Chen (2004) analyze the
dynamic relationships among the six GCC countries for the daily period January 2000 to
September 2004 and also find evidence of increasing market integration. Saadi-Sediq and
Petri (2006) find that the Jordan stock market to be cointegrated with other Arab markets
but not with emerging or developed markets over the weekly period March 17, 1998-
December 25, 2006. They contend that two of the main GCC markets, namely Kuwait
and Saudi Arabia, Granger cause the Jordanian stock market.
More recent literature on MENA market volatility uses the GARCH models to
study volatility behavior and volatility transmission. Hammoudeh and Choi (2007)
employ the univariate GARH approach with Markov-switching to study the volatility
behavior for the transitory and permanent components of the GCC market indices, while
allowing for two volatility regimes to exist. Hammoudeh and Li (2008) examines sudden
changes in volatility for five GCC stock markets using the iterated cumulative sums of
squares (ICSS) algorithm and analyzes their impacts on the estimated persistence of
volatility. They find that most of these stock markets are more sensitive to major global
events than to local and regional factors. Malik and Hammoudeh (2007) use multivariate
GARCH models to analyze return volatility transmission among GCC markets, the oil
price and S&P 500 index.
5
With the exception of Hammoudeh and Choi (2007), all those studies indicated
above do not distinguish between the short run and long run components that make up the
stock market index. The dynamic structures for non-stationary stock price and other
economic time series respond to forces that operate in both the short run and long run.
Most of these series are thought to be driven by at least two factors of consequence. One
of these factors is non-stationary and its dynamic is represented in some form of a
random walk structure. In the financial literature, this has been referred to as long-term
component, permanent component, and trend. We are interested in the decomposed
movements of the cycles and the trends that make up those indexes. In particular, we are
keen on knowing whether the short-run cycles of MENA stock markets are linked and
subject to cross border spillovers. We are also interested in knowing if oil prices affect
these short cycles in any of the three regions that make up the MENA area. The findings
should have implications for type of investment and hedging strategies that portfolio
mangers should follow, particularly whether they should follow a diversification strategy
or a spillover-based leapfrogging strategy.
The decomposition of stationary economic or financial time series into a random
walk, representing the permanent component, and a cycle, representing the transitory
component, is originated in Beveridge and Nelson (1981). Stock and Watson (1988) and
Gonzalo and Granger (1995) extended the B-N decomposition to a multivariate system
where each variable in the system can be expressed as a linear combination of several
random walks and cycles. Later work by Engle and Kozicki (1993) and Vahid and Engle
(1997) has shown that it is possible that the variables in the multivariate system can have
common stochastic trends and common stochastic cycles, with the total number of
6
common trends and common cycles to always be equal to or greater than the number of
the variables. While the V-E decomposition requires that the total number of common
cycles and trends be equal to the number of the variables to have efficient estimates of the
permanent and transitory components, Proetti (1997) and Hecq, Palm and Urbain (2000)
extended the efficient estimation to the general case when the total number is greater than
the number of the variables.
Shirvani and Wilbratte (2007) adopted a component-based approach to
decompose stock prices into long-term trends and short-term cyclical components. In
turn, they are able to explain the long-run trend component with the help of fundamentals
for stock valuations. However, they find the cyclical components to be unrelated to such
fundamentals. The cyclical components for the different countries appear to be related to
contagion spillovers. They then discuss the implications of these findings in term of
international diversification strategies.
As in Shirvani and Wilbratte (2007), our next step in this research would be to
determine the relevant explanatory variables in these components. At this stage, although
it is just a conjecture, we believe the long-term component of a stock price index would
be determined by the supply and demand of stocks. However, the short-term component
would most likely be governed by various transient events (e.g., geo-political events,
psychological factors, etc).
The paper makes the following contributions to the literature on emerging stock
markets. It decomposes the market indices for the ten MENA markets into short term
cycles and long term trends. The decomposition demonstrates that the proportion of the
decomposed cycle component in the market index increases for all the MENA regions
7
during crises, particularly during the 2006 collapse. The GARCH results show that for
the GCC and Levant region, cycles feed on themselves suggestion there is cycle
contagion in those regions. Strengthening in the decomposed trends can also reduce the
size of the cycles, focusing the markets on the fundamentals. It is also interesting to note
that the findings pins UAE as the source of cycle contagion in the GCC region, Turkey
for the Levant region and Egypt for the North Africa region. The change in the oil price
does not show a much direct impact on the cycles including those of the GCC. Cycles
are usually sensitive to psychological factors. The impact of past cycle volatilities
strongly dominates that of past cycle shocks for all the regions, implying that past cycle
volatilities can be used to predict future ones. Oil and S&P 500 dispersions have mixed
impacts on the cycles’ volatilities within the regions.
The paper is organized as follows. After this introduction, section 2 discusses the
relevance and merits of studying the MENA stock markets. Section 3 provides the
descriptive statistics and the correlations for the MENA markets, two global stock market
indices (S&P 500 and FSTE 100), US interest rate and oil price for comparison purposes.
Section 4presents the decomposition methodology and the GARCH model for the MENA
index cycles. Section 5 examines the results. Section 6 concludes and section 7 highlights
the future research on the trend components of the market indices for the MENA area.
2. The Relevance and Merits of the Study
We contribute to the literature on emerging markets by investigating some of the
features initially observed in the developing markets (Bekaert et al. (2000; 2003); Achour
et al (1999)). We deem that the merit of this study lays on the representation of a more
8
thorough understanding of markets in the Middle East which denotes a part of the flow of
studies in emerging market (e.g., Geert Bekaert et al (2007), G. Bekaert and H. Campbell
(2003))
It is genuine that the study of MENA market presents some challenges for
researchers due to the lack of database and because these markets are in transition and
hence not stable, yet they offer an opportunity to investors aiming to diversify by adding
risk and increasing reward. Moreover, some known models in finance literature may not
capture the nature of these markets but the recent strong interest in emerging markets in
general and frontier markets has provided momentum for future investigation in that
direction and called for further study of current models to new conditions in these
markets. Here in this study, we explore the financial nature of the MENA markets, and
some specific circumstances arising in these markets. We also consider a host of other
issues such as cycle contagion, changes in trends, political events and changes in oil
prices and diversification strategies in some selected MENA markets.
Surprisingly, the growing body of research on contagion and political events
primarily focuses on the developing markets; arguably the most organized and
established markets in the world. In contrast, our study focuses on markets where
political events and contagion may be particularly strong, namely MENA markets. In
fact, country-specific risks and various barriers to international investment caused by the
lack of information, discriminatory taxes, and restrictions on funds flows or simply fear
of expropriation were always mentioned as some of the reasons that prevented some
foreign institutional investors from investing in MENA markets and thus, prompted the
observed bias of investors toward their domestic stocks in the developed countries.
9
However, we deem that the aforementioned risk premiums are important features of these
markets; thus, the focus on MENA markets should help and maybe yield exceptionally
powerful tests and valuable evidence.
In line with the contention made by Bekaert and Harvey (2002), we believe that
MENA market integration is in progress and it is surely a gradual process which will
depend on the circumstances in each country and it is most likely that all barriers will
disappear either within the MENA region or eventually with the rest of global market.
Notably, that MENA equity returns are highly volatile and significantly less correlated
with equity returns in the developed markets, making the construction of low-risk
portfolios optimal and achievable .
Furthermore, the globalization of equity markets including the MENA countries
has been increased dramatically with the surge of access to various international
products. For instance, as of June 2006 the Morgan Stanley Emerging Market index
includes some of MENA countries used in our paper such as Egypt, Jordan, Morocco,
Turkey3. Also, in 1996 the Standard and Poor’s started to track frontier market and since
2007 the S&P launched the first investable index called Select Frontier index (30 of the
largest companies from 11 countries) and the Extended Frontier Index (150 companies
from 27 countries). Four out of the six MENA countries included in the initial frontier
index are used in the present paper such as Kuwait, Lebanon, Oman, Tunisia, and United
Arab Emirates.
3. Descriptive Statistics
3 For instance Turkey and Morocco have begun to open up their markets and "emerge" onto the global scene.
10
In this study, we use weekly data obtained from Global Financial Data for the closing
prices of the market price indexes of ten stock markets in the MENA area. We used
weekly data because the countries have different weekends. Saudi Arabia, for example,
considers Thursday and Friday as its weekend. Jordan observes Friday and Saturday as
the weekend, while Turkey’s weekend covers Saturday and Sunday. All the ten indexes
are expressed in US dollars to facilitate the comparison of descriptive statistics. We also
include the statistics for S&P 500 index and FTSE 100 index, US interest rate and the oil
price for meaningful comparisons. The weekly sample covers the period January 6, 1998
to December 4, 2007 for the markets in the Levant and North Africa regions and the
period from August 25, 1998 to December 4, 2007 for the GCC markets.
The price for oil (WTI, hence after) is the spot price quoted for immediate delivery of
WTI crude at Cushing, Oklahoma, and is expressed in U.S. dollars per barrel. Data for
the WTI price is accessed from the EIA website.
Table 1 confirms that Egypt is the one with the highest number of listed companies in the
sample followed by Turkey and Jordan. As a result if the total market capitalization of all
companies in the world was estimated to be around US$51 trillion in 2007, the result in
Table 1 implies that the MENA market capitalizations represents (3.20%) of the total
market capitalization which ranges from $518 billion for the Saudi Arabia followed by
Turkey ($286 billion) and then Egypt with $138 billion. Saudi Arabia’s stock market
remains the leading market in the MENA with about $147 billion of trading value in the
last quarter of 2007 followed by Turkey and Dubai, and it is considered somewhat
competitve, with the average price-earnings ratio hovering at about 23.2.
11
It is clearly illustrated that the Dubai followed by Abu Dhabi, and Kuwait markets
stand out in the MENA region for the high relative liquidity of their stocks. Dubai is
ranked first for turnover velocity, an indicator which is defined as total trading volume
divided by exchange market capitalization such that a high turnover rate may indicate
excessive trading and commissions. In a context of growing international investment
activities, another key liquidity indicator is the trading turnover which is the percentage
of outstanding traded shares during the fourth quarter 2007. Table (1) indicates that
Turkish stock market (129.7) followed by Dubai (39.1) and Saudi Arabia (28.4) stock
markets are among the top three most liquid stock markets in the region. Nevertheless,
the data confirm that the figures of trading turnover in the region are still significantly
below those of the most liquid markets. With the exception of Turkey, all MENA
countries reveal trading turnover rates which range from 39.1 (Dubai) to 3.7 (Lebanon),
while Tokyo and NYSE have 138 and 167, respectively. With the exception of Turkey.
the liquidity ratios have remarkably improved in between 1989-1999 in Morocco, GCC
countries and Egypt but they have remained relatively in the two-digit range.
Yet, by looking at some market characteristics between the first and the fourth
quarter of the year 2007, we could suggest that these countries showed a noticeable
growth in market capitalization, value of traded shares, turnover ratio, and number of
listed companies. This growth is associated with the massive privatization plans
introduced in the region; the sale of government assets to private foreign companies; and
considerable efforts towards improving the efficiency, suggestiing that liquidity is
improving in the MENA stock markets4.
4 For instance, it is widely noticeable the massive privatization plans introduced in Morocco, Egypt and Turkey and further significant efforts devoted towards enhancing the efficiency, depth, and liquidity of the three stock markets.
12
In terms of earnings multiple ratio, Morocco stands out in the MENA region with
a high rate followed by Jordan and Saudi Arabia stock markets which implies that their
listed companies are considered expensive, and investors in these three countries are
paying more for each unit of earnings. Whereas other markets such as Turkey, Tunisia,
Dubai and Bahrain have low earnings multiple ratios suggesting that companies in those
markets could be considered cheap and attractive investments; in fact those ratios are in
the range of those of developed economies (e.g., P/E of S&P 500 was 15.32 in 2007 Q.4 )
The descriptive statistics for the log of the weekly ten MENA and the two world’s
major market indices, US federal funds rate (FFR) and oil price are reported in Table 2.
They MENA statistics suggest that Turkey and Kuwait have the highest weekly index
mean, while Morocco has the lowest over the sample period. In terms of historical
volatility as represented by the standard deviation, Kuwait and Egypt have the highest
while Tunisia has the lowest. In terms of the risk-to-mean ratio as defined by the
coefficient of variation (c.v.), Oman has the lowest while Egypt has the highest. What is
most informative about these statistics is that both historical volatility and coefficient of
variation are much higher in the MENA markets than for S&P 500 and FTSE 100 indices
but not for FFR and oil price WTI.
It is also interesting that all the MENA indices are skewed to the right, while both
S&P 500 and FTSE 100 indices are skewed to the left. This means that there is a higher
probability for investors to earn positive returns from the MENA area than to get negative
returns. The highest skewed index in the MENA area is for Morocco and the lowest is for
Tunisia. The Jarque-Bera statistic points to a distribution that is different from the normal
distribution for all the MENA markets except Tunisia.
13
The weekly contemporaneous correlation coefficients among the market indices, US
interest rate and oil price over the sample period are displayed in Table 3. The highest
index correlation MENA area is between UAE and Jordan (0.99), followed by the
correlation between Kuwait and Jordan (0.96). This is not surprising because Jordan
depends on remittances and direct and portfolio investments flowing from the GCC
countries. These correlations are much higher than the correlations between the S&P 500
and FTSE 100 indices (0.89) and between these two world’s major indices and oil price
(0.35 with S&P and 0.04 with FTSE). They are also much higher than the correlations
between these world’s indices and the US federal funds rate.
The MENA country that has the highest correlation with oil price is Saudi Arabia
followed by Kuwait. Saudi Arabia is the world’s oil exporter and Kuwait is a major
exporter as well. The highest positive correlation between MENA markets and US
interest rate is with Lebanon. However, Kuwait and Saudi Arabia have a negative
correction with the US federal fund rate, probably caused by their actual or effective peg
to the US dollar. It is also noticeable that GCC countries have lower correlations with the
two world’s major indices than the markets in the other regions of MENA. This is also
not surprising because the GCC markets were segmented from the world markets but
recently the correlation is rising after the recent liberalization and reforms in the GCC
markets. It is also worth nothing that the markets in North Africa have higher correlations
with S&P 500 and FSTE 100 than with any other countries in the MENA area. This could
be due to their geographical proximity and business links with the European countries.
Turkey has its highest correlation with Egypt. Those two countries are among the most
populous in the area and similar in other economic aspects.
14
It is worth mentioning that the index correlations among markets within the same
region have changed significantly since the collapse of the MENA markets in early 2006.
This is particularly true for correlations in the GCC countries where correlations are not
only declined but also become negative in some cases. For example, the index correlation
between Kuwait and Saudi Arabia changed from 0.95 over the sample period to -0.17
during the period since February 2006.5 The correlation between Kuwait and UAE also
changed from 0.94 to -0.02 over the corresponding periods. The correlation also turned
negative between Oman and each of Kuwait and Saudi Arabia. Interestingly, the index
correlations between Saudi Arabia and UAE remained high. These drastic changes in
correlations among the GCC stock market reflect unusual lack of policy coordination
among policy makers in this region during crises, particularly the 2006 markets’ collapse.
This is something those decision makers should stoke concerns about as they plan to form
a monetary union which is scheduled for 2010 and a common stock market after that.6
Drastic changes in index correlations over the corresponding periods also occurred in
the Levant region. These changes are pertinent to correlations between Jordan and
Turkey, and Lebanon and Turkey. However, the correlation increased between Jordan
and Lebanon since the 2006 collapse. In contrast, the correlations for the North Africa
region increased since the 2006 collapse for most of the region.
4. Methodology
We first present a state-space approach that cast the Beveridge -Nelson
decomposition into a multivariate system representation for each region in the MENA
5 The correlations among the MENA markets since the early 2006 collapse are available upon request.6 http://archive.gulfnews.com/business/Markets/10194636.html
15
area. This approach is provided by Morley (2002), Proetti (1997) and Hecq et al (2000).
Next, the methodology presents the univariate GARCH model for the generated cycle
component of each stock market index in the MENA regions, where in its mean equation
it depends on the lagged cycles and changes in the trends of market indices in its
respective group, and on a dummy variable.
State space representation
Let xt be a vector of n I(1) variables which can be expressed as a VAR system in
levels of order p by:
(1)
If Eq. (1) can be most accurately described by a VAR system of its first difference with
order p-1, then this system can be expressed as:
(2)
The AR(1) representation of the VAR in Eq. (1) using state-space is straightforward and
can be given by state equation and vector:
(3)
(4)
The state vector and transition matrix are represented by:
16
where Z=[In, 0nxn, …, 0nxn] is an n× n(p-1) matrix, c’=[ ’, 01xn, …, 01xn] a 1×n(p-1)
matrix, ft an n(p-1)×1 vector, and H an n(p-1× n(p-1) transition matrix.
If xt is best described by a VECM system, then this system can be expressed as:
(5)
where and ’ are two n×r matrices, is the matrix of r cointegrating vectors and is
the matrix of adjustment coefficients. Following Hecq (2000), the state-space
representation is the same as equations (3) and (4), but the definition of state vector and
transition matrix are expanded:
ft then is an (n(p-1)+r)×1 vector, H an (n(p-1)+r)×(n(p-1)+r) matrix, Z=[In, 0nxn, …, 0-
nxn ,0nxr] an n×(n(p-1)+r) matrix, and c’=[ ’, 01xn, …, 01xn,01xr] a 1×(n(p-1)+r) vector.
Multivariate Beveridge-Nelson decomposition
The multivariate Beveridge-Nelson decomposition is expressed as the sum of a
random walk trend (Tt) and a stationary cycle (Ct):
17
(6)
The trend is defined as:
(7)
where is the ith-step best unbiased linear predictor given all information available
at time t. the cyclical component is then defined by:
(8)
Both VAR representation in Eq. (2) and VECM representation in Eq. (5) have drifts,
therefore the unconditional mean . To simplify Eq. (8), we modify Eqs. (3) and
(4) by distracting the mean of xt from the state vector. Following Proetti (1997) and
Hecq (2000), we assume the drift has a constant mean over time. Using Eq. (4) the
expected mean of drift is . Under stability condition that all eigenvalues of
H stay inside the unit circle, As t , so that while t ,
, where m=n(p-1) for VAR system in Eq. (2) without cointegration and
m=n(p-1)+r for VECM system in Eq. (5). Plugging the expected value of drift into
equations (3) and (4), we get:
(9)
(10)
where . Therefore the best unbiased linear predictor of is
(11)
18
where is the updated estimate of the state vector based on Kalman filter. Since all
elements of the state vector are observed at time t, we get . On the other
hand, by taking the advantage of the stability conditions again, we plug Eq. (11) into Eq.
(8) to get the cyclical component of Beveridge-Nelson decomposition for each market in
the MENA area:
(12)
The trend component for each market is:
(13)
GARCH models for cycles
Following the derivation of the index cyclical and trend components for all the
markets in the respective region, we can employ the GARCH model to examine the
volatility clustering behavior of cycles, their linkages across markets within each of the
three MENA regions, their responses to changes of trends within the same region, their
sensitivity to changes in the oil price and their reaction to structural dummies. In the
GARCH model, we construct the mean equation for the cycles of the ith country in a gives
MENA region as:
and the variance equation for that cycles as7:
7 We also replaced S&P 500 alternatively by FTSE 100 and FFR in both the mean and volatility equations but the results were not as significant, and thus are not reported.
19
(15)
where and
, i, j are country indexes, N is the number
of countries in a zone, is the cycle component of country i at week t as in Eq. (13) and
is that of country j at week t-1. is the first difference of trend component of
country j at week t. Both trends and cycles are derived with the multivariate Beveridge-
Nelson decomposition discussed above. Moreover, dlwtit is the log price change of
Nymex spot WTI price of week t, dlsp500t is the log change of S&P 500 index of week t,
and D is the dummy variable representing a break as a break in the data. The structural
dummies that have been identified are D00 representing the change in the OPEC oil
pricing mechanism to a price band in February 2000, the D003 denoting the break as a
result of the 2003 Iraq war and D911 for the 2001 attack on New York. In Eqs.14 and 15,
is the long term drift, is the error term for the ith country at week t, where | It – 1
~ N(0, ) and where N (.) represents the conditional normal density with mean 0 and
variance , and It – 1 is the information set available up to time t –1. We note that the
above mean equation has an AR(1) term because of the existence of significant serial
correlation among cycles.
In the variance equation, stands for the conditional variance, and are the
squared residuals from the mean equation, which is the ARCH term. The coefficient j
represents the GARCH or past volatility effect and j captures the ARCH or the past
shock effect. The sum of j +j measures the degree of convergence to long-run
equilibrium or volatility persistence for cycle j in this model. A high value for this sum
20
indicates slow convergence or high volatility persistence. The half-life of cycle volatility
shocks, which are the sum of the AARCH and GARCH coefficients in the variance
equation, is defined as
Half-life = ln(1/2)/ln(α + )
This equation gives the time period (or number of weeks in our case) for the volatility to
reach half of its life after a shock hits. The shorter the half life, the faster the volatility
shock will vanish. In this study we expect the volatility shock to be transitory because we
are dealing with short run cycles.
Three exogenous variables are included in the variance equation, which are one of
the three dummy variables: D00 for the OPEC price band, D01 for the 2001 New York
City attack and D03 for the 2003 Iraq war used in the mean equation, and the squared
deviations of oil price and S&P 500 index from their respective sample means.
5. Empirical results
In this section, we present various empirical results for a large set of Middle East
and North African countries combined into regions over the sample period. More than at
any time in its long history, the Middle East today represents a real patchwork of
individual regions and states, defying the attempts of outsiders to define it as a
homogeneous entity. Based on the results of the return correlations, geographic and
economic similarities, we will initially classify the MENA markets into three primary
regions: The GCC, Levant (Mediterranean) and North Africa. The GCC in this study
includes Kuwait, Oman, Saudi Arabia and UAE. These countries share geographic
proximity and high dependence on oil export and labor imports and their stock market
21
returns have relatively higher historical correlations. The Levant region includes Jordan,
Lebanon and Turkey.8 These countries are some what close in terms of geographic
proximity and are labor exporters. The return correlation coefficients show high
correlation between Jordan and each of Lebanon and Turkey. Finally, the North Africa
regions include Egypt, Morocco and Tunisia. They share geographic proximity and are
labor exporters. Morocco and Tunisia and Egypt and Tunisia have relative high return
correlations. The three primary regions will be augmented with stock markets from other
MENA regions and two major world’s stocks markets to determine the impact of
spillover across border whether a results of international trade, neighborhood or/and
financial flows.
5.1. Decomposition of MENA Stock indices: Cycles and Trends
Since we are interested in extracting the trend and cycle components and apply
these to one-step-ahead forecasts, we simply specify the component model in discrete
time directly. In this section, we will examine the behaviors of those decomposed cycles
in each region and their proportions of market indices during periods of crises. In the
following subsection, we will focus on finding explanatory variables for the extracted
components in addition to exploring any relationships between the components of the
stock indices.
The GCC region
All the decomposed index cycles in this region are highly volatile, but there is still
a common pattern that characterizes all of them (see Fig. 1). The cycles increased in size
8 We could not include Palestine and Syria in the Levant region and Algeria in the North Africa region because these countries do not adequate time series for their market indices.
22
in the late 1990s; a period that witnessed the 1997, 1998 and 1999 Asian, Russian and
Argentinean crises, and the collapse of oil prices in 1999.9 The cycle patterns show that
GCC cycles increased during the periods of global economic crises. These cycles also
increased during the period 2005-2006 which overlaps with the regional stock market
collapse. The proportions of the decomposed cycles in the actual stock indices in the
respective markets of this region also increased in those periods (see Fig. 2).
The long-run trends of the four GCC countries have also moved up since 2000
which marks the creation of the oil price band by OPEC, reflecting recent economic
growth. These trends were later reinforced by the 2003 Iraq war and the consequent
increase in commodity prices. Interestingly, the trend dropped for two of the GCC
countries, namely Saudi Arabia and UAE, after the 2006 stock market collapse, but at the
same time continued to grow for Oman and Kuwait, which were affected the least by the
collapse. In sum, this index trend pattern shows that the GCC stock markets do not
behave the same during crises.
The Levant region
Contrary to the VARs in GCC and North Africa regions, the index VAR for this
region is cointegrated. The cycle pattern similarity in this region reflects the
cointegrating trend that co-moves the markets (see Fig. 3). The trends generally move
upward in this region, moving up again in 2005 and then down in 2006 as in the case of
the GCC countries. But the cycles show general decline until 2005 when they reversed
themselves in a strong jump compared to the GCC trends during the same period. The
proportions of cycle to index in this region generally decline because of the upward
cointegrating trend. But during the 2006 regional market collapse, the proportion of
cycles in the indices increased similar to the case in the GCC region.
The North Africa region
The trends in this region started to move up in early 2003, long after they started
to move in the GCC region. The trend seems to be more aggressive in Egypt than in the
other two countries in this region and in the other regions. While the cycles in this region
share with those in the GCC the up and down patterns that developed since 2005 they did
not move erratically in late 1990s as the GCC did. This probably has to do with
differential impact on the regions of the 1999 collapse in the oil price. The cycles in this
region increased during the 2000-2002 period, which is not shared with the other two
regions.
5.2. The Cycle GARCH Results
We analyze the cycle volatility behavior for the markets in the three regions as
specified in Eqs. (14) and (15) of the previous section. We discuss the results of both the
cycle return mean and volatility equations, which make up the GARCH model, and pay
particular attention to volatility persistence and the half-life.
The GCC GARCH results
In the GCC region, the estimates of the short-term market index cycles in the
mean equations of the GARCH models are provided in Table 4. Kuwait and Saudi
Arabia are the most affected by the past cycles, originating from GCC members,
implying that GCC cycles feed on themselves and giving credence to the so-called
friendship bias hypothesis which conjectures that many countries prefer to invest in the
24
same geographic region (Berkel, 2007). Specifically, Kuwait and Saudi Arabia as well as
Oman are positively affected by the past cycles coming from the UAE markets. The UAE
cycle is however not affected by any past GCC cycle except its own. It seems that the
UAE markets are a source of contagion in this region. Anecdotally, the bubble that burst
and spread to other GCC markets in early 2006 originated from the UAE markets. This is
not surprising because Dubai has experienced astronomical growth in its economy. These
results suggest that the GCC index cycles are positively affected by contagion spillovers.
Thus, based on the cycles’ behavior these markets are less suitable for diversification
gains potential in the short run. Investors and traders in this region can be anticipatory
and position themselves in a certain GCC market to benefit from the contagion spillover
from other GCC markets. If the transition costs are not prohibitive, traders can switch
from one market, particularly from Dubai’s to another or switch to other asset types or
regions based on the relative persistence of the markets of interest.
Interestingly, trends in the previous period in this region also affect the current
cycle of any given market index. The impact is negative which suggests that any
strengthening in past week’s trends of other GCC markets reduces the current week’s
cycle contribution to the index for any market in the region. This implies that traders,
investors and portfolio managers should give more weight in their decision-making to
trends over cycles particularly when the trends are strengthening in this region.
In terms of structural breaks in the data, the R–squared and AIC select the
structural dummy D00 which represents a lift in the oil price to the central price parity of
the oil price band over the 9/11 and 2003 Iraq war dummies, D01 and D03, respectively.
The results show that this dummy variable is not statistically significant in affecting the
25
cycles for the GCC countries, except Kuwait. Also, the change in the oil price does not
show a significant direct impact on the cycles, which are usually sensitive to
psychological factors. The global factor S&P 500 return reduces the cycle of Saudi
Arabia only.10
In the variance equation, the impact of past cycle volatilities strongly dominates
that of past cycle shocks for all four GCC markets, implying that past volatilities can be
used to predict future ones. Cycle volatility has the least persistence and the fastest
convergence is in the case of Oman followed by UAE. It seems that Oman restores
stability faster than any market in this region. In fact, after the 2006 GCC stock market
collapse, Oman restored stability and ended with a modest positive gain. Additionally,
cycle volatility is more persistent for Saudi Arabia than for any market in this region.
Saudi Arabia which is characterized by “too much money chasing too few stocks”
experienced the second worst market collapse in 2006 and is still reeling from its
aftermath in 2008. Correspondingly, the half-life for volatility shock in Saudi Arabia is
about eight weeks compared to less than three weeks for Oman and UAE. These findings
should be particularly relevant in options valuations which use current volatility as a
determinant in pricing options. They are also informative for policy makers of those
countries as they plan to have a monetary union and a common currency. In terms of
sensitivity to global stock market variations or risks, cycle volatility in each of Oman,
Saudi Arabia and UAE is negatively sensitive to the global risks as representative by
S&P 500 index. In January 24, 2008, the GCC markets were affected by varying degrees
to the one-week drop in the S&P 500 index, showing more interaction with global world
10 FTSE 100 and federal funds rate (FFR) have no impact on this region: the results are not reported but are available upon request.
26
market and less decoupling than in the past. Interestingly, the oil price dispersion or
variation affected the cycle volatility of only UAE and Oman, which is unusual for these
oil-exporting countries.
The Levant GARCH results
Similar to the GCC markets, past market index cycles in the mean GARCH
equations for the Levant region also expand current cycles (see Table 5). The major
source of cycle contagion in this region is Turkey, which have been facing conflicts
inside the country and around it as well as financial crises. Turkey has by far the largest
market capitalization in this region and is also characterized by highest market
capitalization density per company which narrows the investment opportunities.11 One
difference with the GCC region is that cycle interaction among the markets in the Levant
region is much less than in the GCC region. Thus, cycle contagion or cycles-feed-on-
cycles behavior is less consistent in this region than in the GCC region.
An increase in the permanent components of the other market indices in this
region as represented by the long-run trends seem to reduce the index cycle component
for almost all of the countries in this region as is the case in the GCC region. This implies
that when the permanent component of the market index gains momentum, the transitory
cycle component in the index return decreases as indicated in the previous region.
The structural break dummy D00 is also selected in this region over the 9/11 and
2003 Iraq war dummies. Interestingly, the results for D00 suggest that that there has been
a permanent increase in the oil price since 2000, giving rise to a statistically significant
decline in the short term cycles for all countries in this region, a result which is not as
11 Turkey’s market capitalization at the end of 2007 is $286 billion and the number of the companies listed is 245 firms.
27
pervasive in the GCC region. However, the oil price itself does not suggest that there is a
significant direct impact on the cycles for any of the countries as is the case in the
previous region. However, the S&P 500 return seems to increase the cycles in all the
countries, indicating that these cycles are more susceptible to the global stock market
than the GCC countries. FFR has no effect but FTSE 100 affected Lebanon and Jordan.
Those two results are not reported.
The past cycle volatilities in the Levant have stronger impacts than the past cycle
shocks for all three markets in this region as is also the case in the GCC region. The half-
life for a volatility shock for Jordan is about nineteen weeks, the highest in the ten
markets, compared to surprisingly about five weeks for Lebanon. D00, the oil dispersion
risk and S&P 500 risk do not have any significant impact on cycle variance for any of the
countries.
The North Africa GARCH results
What distinguishes this region from the two previous ones is that in this region
there is not much interaction between current cycle of a given and past cycle returns of
the other markets in this region (see Table 6). The only exception is the interaction
between Morocco and Tunisia. This finding qualifies this region for cycle portfolio
diversification in the short run.12 However, there are strong impacts on the cycle
component coming from long-term trends of the markets. It is interesting to note that the
impact coming from Tunisia on Egypt and Morocco and from Morocco on Tunisia is
positive, implying an expansion in the cycle component. This is different from what we
12 Hammoudeh et al (2008) also find no cointegration among the market indices of this region, suggesting the presence of diversification benefits potential in the long run.
28
have in the previous regions. However, the impacts from the Egyptian and Moroccan past
cycles are negative.
The R–squared and AIC select the structural dummy D03 which represents the
2003 Iraq war. The impact of this geopolitical event is marginal. Oil risk has a significant
impact on Tunisia’s cycle volatility. S&P 500 risk has no effect on this region.
The impact of past cycle volatilities also strongly dominates that of past cycle
shocks for all three markets in this region as in the previous two regions. Cycle volatility
has the most persistence and the slowest convergence as in the case of the Egyptian
market. The half-life for volatility shock in this market is about twelve weeks. The half-
life for Tunisia is five weeks.
6. Conclusions
This study examines the cycle components of the indices for ten stock markets in
three regions of the MENA area using weekly data in the presence of global shocks. In
the GCC region, there is a cycle spillover from one country to another with the UAE
market being the contagion source, suggesting that diversification within the region based
on movements of cycles is not the best strategy in the short run. Therefore, investors
could use leapfrogging from one market to another or get out in case of negative cycle
contagion once a problem starts in one of the markets particularly Dubai, weighing this
strategy over diversification in the short run. The GCC policy makers should coordinate
policies to deal with the negative spillovers. The 2000 adjustment in the oil pricing
mechanism from a single price targeting to band targeting affected the mean cycle
component of the stock index negatively only in the case of Kuwait but it did increase
29
the cycle volatility for any country in the region. The oil price volatility does not have
much direct effects on index cycle volatility. Moreover, since the half-life of cycle
volatility in the GCC region is on the order of three to eight weeks, fluctuations in cycle
volatility could certainly affect the values of financial derivatives such as options on
futures contracts in these countries because those derivatives have typically a duration of
several months. But those short-lived fluctuations should have not any considerable
impact on real options or the related investment decisions.
In the Levant region, the cycle-on-cycle behavior is present but weaker than in the
case of the GCC region. What distinguishes this region from the other two is that the
reinforcing feedback, giving cycles relatively more weight in decision-making.
Moreover, volatility persistence and half-lifes are stronger in this region than in the other
ones, making financial hedging more pressing for this region.
The North Africa Region, there is no significant cycle-feed on cycle spillover
among most of the country's mean cycles. The exception is between Tunisia and
Morocco. There is however volatility persistence which linger much more for Egypt than
for Tunisia and Morocco. Fluctuations in cycle volatility could also affect the values of
financial derivatives in this region but with no significant impact on real options and
related investment decisions. The 2003 Iraq war dummy marginally affected the mean
cycle component of the stock indexes, showing this region’s weak sensitivity to political
events. The oil volatility impacted the cycle volatility for Tunisia only. Tunisia turned
from a net oil exporter to a net oil importer recently. As in the other regions, this region’s
cycles are not sensitive to changes or volatility or S&P 500 index and the US federal
funds rate.
30
7. Future Research
This paper examines the mean and volatility behaviors of the cycle components of
the MENA’s stock market indices. It should also be useful to study the same behavior for
the market indices’ trends which are usually influenced by the fundamentals such as
CPIs, interest rates …etc. The trends are however are nonstationary and thus are not
predictable. This task will also be more challenging to be conducted at the weekly level
because the data is not available. We plan to carry out a study at the monthly data to
study the behavior of the trends and their volatility persistence.
31
References
Abd Majid, M. S. , Yousof, R. M. and Rayal, M. N. 2007. Dynamic financial linkages among selected OIC countries. Journal of Economic Cooperation 28, 25-56.
Bekaert, G., Harvey, C. R. and Lundblad. C. 2007. Liquidity and expected returns: Lessons from emerging markets. Review of Financial Studies, forthcoming.
Bekaert, G., Harvey, C. R. and Lundblad. C. 2003. Equity market liberalization in emerging markets. Journal of Financial Research 26, 275-299
Bekaert, G. and Harvey, C. R. 2003. Emerging markets finance. Journal of Empirical Finance 10, 3-56.
Berkel, B. 2007. Institutional determinants of international equity portfolios - A country-level analysis. The B.E. Journal of Macroeconomics 7.
Beveridge, S and Nelson, C. 1981. A new approach to the decomposition of economic times series into permanent and transitory components with particular attention to the measurement of the business cycle. Journal of Monetary Economics 7, 161-174.
Bley, J. 2007. How homogeneous are the stock markets of the Middle East and North Africa? Quarterly Review of Business and Economics.
Bley, J. and Chen, K. H. 2004. Gulf Cooperation Council (GCC) stock markets: The dawn of new era. Working paper, American University of Sharjah, Sharjah, UAE.
Blucemnt .2005. An empirical analysis of Istanbul stock exchange sub-indexes. Studies in Linear Dynamics 9, Article 5, 1-14.
Achour, D. and Harvey, C. R. 1999. Greg Hopkins and Clive Lang “Stock Selection in Emerging Markets: Portfolio Strategies for Malaysia, Mexico and South Africa," Emerging Markets Quarterly Winter, 38-91.
Darrat, A., Elkhal, K. and Hakim, J. On the integration of emerging stock markets in the Middle East. Journal of Economic Development 25, 61-78.
De Long, J. Shleifer, A., Summes, L. H. and Waldmann, R. J., 1990. Noise trader risk in financial markets. Journal of Political Economy 98, 703-738
Engle, R. and Kozicki, S. 1993. Testing for common feathers. Journal of Business and Economics Statistics 11, 369-380.
Gonzalo, J. and Granger, C. W. J. 1995. Estimation of common long memory components in cointegrated systems. Journal of Business and Economics Statistics 33, 27-35.
32
Hammoudeh, S. and Alesia, E. 2004. Dynamic relationships among GCC stock markets and NYMEX oil futures. Contemporary Economic Policy 22, 250-269.
Hammoudeh, S. and Choi, K. 2007. Characteristics of permanent and transitory returns in oil-sensitive emerging stock markets: the case of the GCC countries. International Review of Economics and Finance 17, 231-245.
Hammoudeh, S. and Li, H. 2008. Sudden changes in volatility in emerging markets: The case of Gulf Arab stock markets. International Review of Financial Analysis 17, 47-63.
Hammoudeh, S., Yuan, Y. and Smimou, K. 2008. Equity market diversification in the MENA regions and impacts of oil and global major markets. Arab Bank Review 19
Hecq, Alain & Palm, Franz C & Urbain, Jean-Pierre, 2000. Permanent-transitory decomposition in VAR models with cointegration and common cycles. Oxford Bulletin of Economics and Statistics 62, 511-532.
Malik, S. and Hammoudeh, S. 2007. Shock and volatility transmission in the oil, US and Gulf equity markets. International Review of Economics and Finance 17, 357-368.
Morley, J. C. 2002. A state-space approach to calculating the Beveridge-Nelson decomposition. Economic Letters 75, 123-127.
Omran, M. and Gunduz, L. 2001. Stochastic trends and stock prices in emerging markets: The case of Middle East and North Africa region. Istanbul Stock Exchange Review 5, 3-7.
Proetti, T. 1997. Short term dynamics in cointegrated systems. Oxford Bulletin of Economics and Statistics 59, 405-422.
Saadi-Sediq, T. and Petri, 2006. The Jordanian stock market-Should you invest in it for risk diversification or for performance? IMF Working paper # WP/06/187.
Shirvani, H. and Wilbratte, B. 2007. The permanent-transitory decomposition of the stock markets of the G7 countries: A multivariate approach. The Quarterly Review of Economics and Finance, 47, 352-365.
Stock, J. and Watson, M. 1988. Testing common trends. Journal of the American Statistical Association 83, 1097-1107.
Vahid, F. and Engle, R. 1997. Common trends and common cycles. Journal of Applied Econometrics 8, 341-360.
North AfricaEgypt 435.0 138828.0 23924.0 5824.0 17.2 . . .Morocco 73.0 75494.5 11869.7 140.5 15.7 33.8 . 2.1Sudan 52.0 4931.1 191.4 18.7 3.9 . . .Tunisia 51.0 5338.9 310.0 25.5 5.8 16.2 1.7 2.9Source: Arab Monetary FundNotes: Algeria is not included in this table because it has only two firms listed. Market capitalization, trading value and trading volume are in million USD.Trading Turnover, which is a measure of liquidity, is 167 for NYSE, 303 NASDAQ, 154 London, 394 Shenzhen, and 138 Tokyo
38
Table 2: Descriptive Statistics for Market Indices, Interest Rate and Oil prices
Observations 518 518 518 518 518 518 518Notes: The statistics are for log of levels. The weekly sample period for the GCC markets spans the period from Aug. 25th, 1998 to Dec. 4th, 2007, while covers the period Jan. 6th, 1998 to Dec. 4th, 2007 for the markets in the Levant and North Africa regions. The GCC in this study includes Kuwait, Oman, and Saudi Arabia. The Levant covers Jordan, Lebanon and Turkey, while North Africa includes Egypt, Morocco and Tunisia. J-B statistic is significant for all courtiers except Tunisia
39
Table 3: Contemporaneous Correlations among Market Indices, US Interest Rate and Oil Price over the Sample Period
Notes: The correlations are for log of levels. The weekly sample period for the GCC markets spans the period from Aug. 25th, 1998 to Dec. 4th, 2007, while covers the period Jan. 6th, 1998 to Dec. 4th, 2007 for the markets in the Levant and North Africa regions. The GCC in this study includes Kuwait, UAE, Oman, and Saudi Arabia. The Levant covers Jordan, Lebanon and Turkey, while North Africa includes Egypt, Morocco and Tunisia. The symbols a, b and c represent statistical significance at the 1%, 5% and 10% levels, respectively
40
Table 4: GARCH Models for GCC Region’s Stock Market Index Cycles Kuwait UAE Oman S. Arabia
Mean EquationC 0.001927 a 0.002362 0.001274 -0.000179CLKUWA(-1) 0.139534 a 0.100718 0.509005 a 0.384683 a
CLNBAD(-1) 0.164492 a 0.448341 b -0.104149 -0.144692 a
CLOMAN(-1) 0.03111 0.050543 0.291779 b -0.091602CLSAUD(-1) -0.119773 a 0.208013 0.18427 c 0.358489 a
DTRLKUWA -0.159184 a -0.149054 a -0.042058 a
DTRLNBAD -0.091678 a -0.104869 a 0.062588 a
DTRLOMAN -0.11961 a -0.19591 a -0.042978 a
DTRLSAUD -0.093106 a -0.183211 a -0.13638 a
D00 -0.001363 a -0.000508 0.000397 0.000164DLWTI 0.002993 0.000542 -0.010198 -0.001434DLSP500 -0.009253 c -0.007845 0.011991 -0.010121 b
Variance EquationC 2.15E-06 b 5.11E-05 c 2.24E-05 b 3.65E-06 a
RESID(-1)^2 0.226198 a 0.148847 0.148019 0.226855 a
GARCH(-1) 0.604085 a 0.596768 b 0.596019 a 0.688517 a
D00 -6.22E-07 -1.48E-06 -5.08E-06 -1.17E-06DLWTIV -9.82E-05 -0.001751 a -0.000558 a -4.86E-05DLSP500V 0.00038 -0.004033 b -0.001681 a -0.000592 a
0.83 0.75 0.74 0.92 Half-life(weeks) 3.73 2.36 2.34 7.84 Adj. R^2 0.91 0.84 0.87 0.29 Log-Likelihood 2155.81 1630.68 1838.76 1939.90 F-Stats 319.23 163.84 199.05 13.19 DW 2.06 1.84 2.05 2.06 Notes: a, b an c represent significance at 1%, 5% and 10%, respectively. DLWTI in the mean equations is the first log difference of oil price, while DLWTIV is the squared deviation between the oil price and its mean as a measure of oil volatility in the variance equations. D00 stands for the change in OPEC pricing mechanism in 2000
41
Table 5: GARCH Models for Levant Region’s Stock Market Index Cycles
Jordan Lebanon Turkey Mean Equation
C 0.007276 a 0.012047 a 0.011565 a
CLJord(-1) 0.068116 0.026254 -0.010304CLLeba(-1) 0.300653 0.395231 a 0.371849CLTurk(-1) 0.303563 b 0.729863 a 0.519869 a
DTRLJord -0.966445 a -0.365837 a
DTRLLeba -0.950912 a -0.718531 a
DTRLTurk 0.032234 a -0.133711 a
D00 -0.005075 a -0.008895 b -0.008901 b
DLWTI -0.004885 -0.009115 -0.008696DLSP500 0.055236 b 0.112312 b 0.097001 b
Variance EquationC 3.76E-06 3.17E-05 1.12E-05RESID(-1)^2 0.080432 a 0.150188 a 0.078536 a