Economic Modelling 37 (2014) 408416

Contents lists available at ScienceDirect

Economic Modelling
journal homepage: www.elsevier.com/locate/ecmod

Stock market integration and risk premium: Empirical evidence for

emerging economies of South Asia
Ilyes Abid a,b,, Olfa Kaabia b, Khaled Guesmi c,b
a
b
c

Department of Finance, Mtis Lab, EM-Normandie, France

EconomiX-CNRS (UMR 7235), University of Paris Ouest Nanterre La Defense, France
Department of Finance, IPAG Lab, IPAG Business School, France

a r t i c l e

i n f o

Article history:
Accepted 12 November 2013
JEL classication:
G12
F31
C32
Keywords:
Time-varying integration
Asian markets
Risk premium
ICAPM
GDC-GARCH

a b s t r a c t
This article investigates the dynamics of regional nancial integration and its determinants in an international
setting. We test a conditional version of the International Capital Asset Pricing Model (ICAPM) accounting for
the deviations from Purchasing Power Parity (PPP) as well as temporal variations in both regional and local
sources of risk. Using data from ve major South Asian markets (Malaysia, Thailand, Singapore, Indonesia, and
Sri Lanka), our results support the validity of an ICAPM and indicate that the risk is regionally priced. Furthermore, we show that changes in the degree of regional stock market integration are explained principally by
the U.S. term premium, and the level of market openness, whatever the measure of currency risk. Finally,
and as expected, the degree of stock market integration varies considerably over time and from one market to
another. As intense market integration induces both benets and risks, our ndings should have signicant
implications for economic policies and market regulations in emerging, frontier-emerging and transition countries,
particularly for countries from the same region.
2013 Elsevier B.V. All rights reserved.

1. Introduction
While the empirical literature has shown the potential benets of
international diversication into stock markets, global investors often
face both direct and indirect barriers (Bekaert and Harvey, 1995).
Geographical distance between domestic and foreign markets is often
an important barrier, limiting most cross-border investment opportunities. The heterogeneous characteristics (e.g., level of nancial market
development and trade openness) among the different economic regions also matter greatly. Financial integration is, rst of all, the gradual
elimination of direct and indirect barriers that impede free movement
of goods, services and capital. These stylized facts have given rise to
the establishment of several large geographical centers that offer very
different risk-return proles.
Grouping by major geographical clusters should lead to nancial
integration as well as to the validity of the law of one price under the
impetus of trade and investment between countries in the same region.
We would expect adjustments in the foreign exchange markets for this
law to be applied. However, as far as international portfolio diversication in emerging countries is concerned, the hypothesis of unique price
of risk across markets is usually violated insofar as exchange rate
regimes are likely to be subject to more or less stringent regulations im Corresponding author.
E-mail addresses: ilyes.abid@em-normandie.fr (I. Abid), kaabia.olfa@yahoo.fr
(O. Kaabia), Khaled.guesmi@ipag.fr (K. Guesmi).
0264-9993/$ see front matter 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.econmod.2013.11.015

posed by local authorities. Several studies have examined the dynamics

of regional integration in emerging markets. Errunza and Losq (1985)
introduce a pricing structure, called mild segmentation, where access
to the various asset classes is not the same for two types of investors:
investors not subject to legal restrictions on holding assets have access
to all securities while investors subject to reference restrictions are
only able to conduct transactions on a subset of assets. Their empirical
results show that emerging markets are neither strictly segmented
nor perfectly integrated. In a different way, Claessens and Rhee (1994)
apply Stehle's (1977) procedure to study the risk-return linkages in 16
emerging markets. Their empirical nding contradicts the hypothesis
of integration in most of the markets, whereas the segmentation
hypothesis cannot be rejected in any of the markets.
Phylaktis and Ravazzolo (2002) derive the covariances of excess
returns on the stock markets for 1980 and 1998 using Asset Pricing
Models. They establish expressions for the excess returns of the local
and foreign stock markets as a function of the real interest rate, dividends paid, and other variables such as lagged returns and the exchange
rates so as to nd the determinants of returns in each country, and also
to derive the variances and covariances of the excess returns; the idea is
to nd variables that help to explain movements in the stock markets of
Hong Kong, Indonesia, Korea, Malaysia, Philippines, Singapore, Taiwan
and Thailand. They nd that variations in dividends paid are a signicant source of variance in stock returns. An interesting result that arises
is that co-movements in output growth are directly related to stock
prices. The paper unearths a close connection between Thailand and

I. Abid et al. / Economic Modelling 37 (2014) 408416

the U.S., and a high degree of integration between Korea, Taiwan and
Japan.
Adler and Qi (2003) extend the model of Bekaert and Harvey (1995)
which basically combines the domestic and international versions of an
Asset Pricing Model (ICAPM) to test the power of domestic factors,
relative to that of common factors, to explain expected returns, and
empirically infers segmentation when the weight of the domestic
factors is high. So, Adler and Qi (2003) investigate the evolution of the
process of integration between the Mexican and North American equity
markets between 1991 and 2002. They show that the degree of market
integration is higher at the end of the period than at the beginning, and
that Mexico's currency risk is priced. Furthermore, there is signicant
asymmetric volatility, which is strongly related to the asymmetric
volatility of the Mexican equity market return process.
Carrieri et al. (2007) extend the model of Errunza and Losq (1985).
They study the integration levels of eight emerging markets over the
period 19772000. They show that the local pricing factor continues
to be relevant in the valuation of emerging-market assets, but none of
the markets considered is completely segmented from the world
market. Furthermore, Chambet and Gibson (2008) estimate a multifactor asset pricing model of partial integration, an extension of that of
Errunza and Losq (1985) for 25 emerging markets, and show that
some markets still remain segmented.
Guesmi and Nguyen (2011) inspired by the model of Bekaert and
Harvey (1995) use a conditional version of an ICAPM to evaluate the dynamics of the global integration process of four emerging market regions (Latin America, Asia, Southeastern Europe, and the Middle East)
into the world market. They show that the integration degree in the
four emerging market regions varies widely through time over the period
19962008, and that this can be explained by the regional factors. Although the general trend is toward increasing nancial integration,
emerging market areas seem to be still signicantly segmented from
the global market.
Guesmi (2012) investigates the evolution of the South-East Asian
stock market integration with the regional one, and deduces that with
the exception of Singapore's market, emerging markets are not strongly
integrated in the study area. These results were conrmed by those of
Petri (1993), Frankel and Wei (1995), and Frankel and Romer (1999).
They show that the geographical proximity effects are not signicant
in the Southeast Asian region.
More recently, Berger and Pozzi (2013) suggest a measure of nancial integration based on the conditional variances of the countryspecic and common international risk premiums in equity excess
returns. The authors show that Germany, France, the U.K., the U.S., and
Japan exhibit several shorter periods of disintegration over the period
19702011. They conclude that stock market integration is measured
as a dynamic process that is uctuating in the short run while gradually
increasing in the long run.
In our work, we investigate the issue through a longitudinal study of
the South Asian region using monthly data from 1996:01 to 2007:12.
Our study differs from previous ones by considering intra-regional
integration instead of global integration, and by taking into account
the currency risk in addition to the sources of global and domestic
risks. The international asset-pricing model we use is built so as to
characterize the changes in market integration through time due to
the impacts of the gradual removal of barriers to emerging market
investments. We also examine the portions of the returns explained
by regional and domestic risk factors, respectively, by carrying out a
decomposition of the total risk premium.
The present study contributes to the literature by developing a
regime-switching ICAPM with a slip condition. Specically, expected
return can slip from a perfectly segmented regime to a perfectly integrated one or vice versa depending on the number of national and regional
factors that may inuence the process of regional nancial integration.
It is true that this model was inspired by that of Bekaert and Harvey
(1995), but it has been extended using a multivariate GDC-GARCH

409

model to take into account the asymmetric responses of expected

returns to different shocks.
One of the advantages of our approach is to authorize the prices of
domestic and world market risks, betas and correlations to vary asymmetrically through time. It is clear that this will help us to understand
the dynamics of interdependencies and correlations between South
Asian stock markets in order to facilitate decision-making. In fact, investors are normally risk-averse, they are concerned about market downturns more than upturns. Consequently, this risk-aversion behavior
will be reected in market prices, resulting in greater market responses
to downturns than upturns.
Asymmetry can be justied differently by the presence of information barriers, the behavioral standpoint of investor psychology (Verma
and Verma, 2010), other sources of market segmentation (Bekaert and
Harvey, 1995), heterogeneous transaction costs (Anderson, 1997), the
coexistence of different shareholders, and noise trading (De Grauwe
and Grimaldi, 2006). Also another source of asymmetry effects is industry concentration and imperfectly competitive behavior. It implies that
wholesalers, or middlemen with power over price, may exercise pricing
strategies that result in a slow and incomplete pass-through of increases
in the international price and a fast and complete transmission of
decreases in the international price to prices upstream.
Although asymmetric adjustment may also be the outcome of
market imperfections, it is plausible that price support policies result
in positive and negative changes in the international price affecting
the domestic market in different ways. Moreover, the effect of positive
shock and negative shock is different.
Our ndings clearly show that the degree of market integration of
the ve emerging markets varies over the period 19962007. Moreover,
the U.S. term premium and the level of market openness mainly explain
the degree of integration in emerging markets. Even though this degree
reaches high values during periods of turmoil, and exhibits an upward
trend toward the end of the estimation period. Hence, emerging markets still remain substantially segmented from the regional market.
Also, the total risk premium decomposition shows that the variance
risk related to the local market index (the local risk factor), explains
more than 70% of the total risk premium on average for the ve emerging
markets.
Tracking the integration level is a critical task. It is important to
know if the emerging countries in the Asia region are globally or regionally integrated for most of the sample period. In fact, if there are regional
integration, and if suddenly during an Asian crisis the intraregional
correlations between the countries rise dramatically, this may lead to
contagion effects.
Our analysis is relevant for both policymakers and investors that pay
a particular attention to stock markets and their degree of integration.
Also, analyzing the links between stock markets is of particular interest
for nancial players. Portfolio managers look at stock market uctuations to infer the trend of each market, and make diversication decisions. Moreover, studying the degree of integration becomes a central
issue for the world economy during turmoil periods. In fact, comparing
the impact of the nancial crisis on the degree of integration provides
useful information about possible substitution strategies between
stock classes. In particular, the integration level plays a key role regarding hedging possibilities, and impacts asset allocation, and their riskreturn trade-off.
The remainder of the article is organized as follows: Section 2
presents the empirical methodology; Section 3 describes the data;
Section 4 presents and discusses the results; and Section 5 draws the
appropriate conclusions.
2. Empirical approach
Our empirical asset pricing model takes as its point of departure
that of Bekaert and Harvey (1995), and is inspired by the theoretical
models of partial integration of Black (1974), Stulz (1981), Cooper

410

I. Abid et al. / Economic Modelling 37 (2014) 408416

and Kaplanis (2000), Hardouvelis et al. (2006), De Santis et al.

(2003), Carrieri et al. (2007), and Tai (2007). All these authors
conrm the partial integration hypothesis and time-varying world
market integration for most individual markets. Exchange rate risk
is also found to be priced in the context of both developed and
emerging markets.
In our study, we adopt a partially integrated conditional ICAPM
with three sources of systematic risk that globally reect uctuations in regional stock market, national stock market, and exchange
rate. Generally, the conditional mean excess return can be written
as:
"
#
l

 X



c 
c
c
c
c
Et1 Ri ; t pi;t1 reg;t1 Cov Ri;t ; Rreg:t =t1
k;t1 Cov Ri;t ; Rk;t =t1 1



 k1
c
1pi;t1 i;t1 Var Ri;t =t1

where Et 1(Rci,t) is the excess return issued in country i, conditionally on a set of information t 1 that is available to investors up to
time t 1. Exponent c indicates that returns are expressed in the
currency of the reference country. Rcreg,t is the return on the regional
market portfolio. Rck;l is the return on the exchange rate of the
currency of country k against the currency of the reference country
c. Cov is the conditional covariance between the security returns
and the region market returns. reg,t 1 refers to the conditionally
expected regional price of covariance risk. l is the number of markets included in the sample. i,t 1 is the conditionally expected
local price of variance risk. k,t 1 expresses the expected price of
the exchange risk for currency k. pi,t 1 is the conditional probability of transition between segmentation and integration states
which falls within the interval [0,1], and can be thus interpreted
as a conditional measure of integration of market i into the regional
market. If p i,t 1 = 1, only the covariance risk is priced, and the
strict segmentation hypothesis is rejected. If p i,t 1 = 0, the
unique source of systematic risk is the variance, and the pricing
relationship in a strictly segmented market applies.
Furthermore, Eq. (1) can be written as a risk premium decomposition. More specically, the total risk premium (TPRM) can be broken
down into three components.
TPRMi;t RPRM i;t EPRM i;t LPRMi;t
where the rst component is called the regional risk premium (RPRM),
c
and is given by: TPRMi,t = reg,t 1Covt 1(Ri,t
, Rcreg,t/t 1)pi,t 1. The
second one is the exchange rate risk premium (EPRM) expressed as


l
c
c
follows: EPRMi;t pi;t1 k;t1 Cov Ri;t ; Rk;t =t1 and the third
k1

one refers to the local risk premium (LPRM), written as: LPRMit =
(1 pi,t 1)i,t 1Vart 1(Rci,t/t 1).
The following Eqs. (2), (3), and (4) describe the expected return
on the regional market portfolio, and the expected returns for Asia,
country, and currency:






c
c
c
c
Et1 Rreg;t reg;t1 Vart1 Rreg;t =t1 M;t1 Covt1 Rreg;t ; RM;t =t1




c
c
c
c
T;t1 Covt1 Rreg;t ; RT;t =t1 S;t1 Covt1 Rreg;t ; RS;t =t1




c
c
c
c
I;t1 Covt1 Rreg;t ; RI;t =t1 N;t1 Covt1 Rreg;t ; RN;t =t1

 




c
c
c
c
c
Et1 Rk;t M;t1 Covt1 Rk;t ; RM;t =t1 T;t1 Covt1 Rk;t ; RT;t =t1




c
c
c
c
S;t1 Covt1 Rk;t ; RS;t =t1 I;t1 Covt1 Rk;t ; RI;t =t1 4


c
c
N;t1 Covt1 Rk;t ; RN;t =t1


 



pi;t1 Exp  0 v1 F i;t1 

with i = M (Malaysia), T (Thailand), S (Sri Lanka), I (Indonesia) and

N (Singapore). RcM;t ; RcT;t ; RcS;t ; RcI;t and RcN;t are respectively the real
exchange rate returns of the ve markets under study. reg,t 1 ,
M,t 1 , T,t 1 , S,t 1 , I,t 1 and N,t 1 refer to the expected
prices of the exchange rate risk. Exp (.) denotes an exponential function.
|| is the absolute value. 0 and 1 are respectively a constant and a vector of region-specic parameters. Fi,t 1 is a vector of region-specic
predetermined information variables related to convergence toward a
regional market at time t 1.
The risk prices are modeled as a function of information variables as
follows:



reg;t1 Exp reg F reg;t1




i;t1 Exp i F i;t1




k;t1 k F reg;t1

er reg;t reg;t1 hregreg;t M;t1 hregM;t T;t1 hregT;t S;t1 hregS;t

I;t1 hregI;t N;t1 hregN;t reg;t

eri;t pi;t1 reg;t1 hireg;t M;t1 hiM;t T;t1 hiT;t S;t1 hiS;t
 

I;t1 hiI;t N;t1 hiN;t 1pi;t1 i;t1 hii;t i;t
rk;t M;t1 hkM;t T;t1 hkT;t S;t1 hkS;t I;t1 hkI;t N;t1 hkN;t k;t

7




with er t r M;t ; rT;t ; r S;t ; r I;t ; r N;t ; r t r M;t ; r T;t ; r S;t ; rI;t ; r N;t : So, r t


er reg;t ; er t ; r t refers to the (11 1) vector of excess returns which are assumed to be normally distributed. Also, t


reg;t ; M;t ; T;t ; S;t ; I;t ; N;t ; M;t ; T;t ; S;t ; I;t ; N;t =t1  N0; Ht is a
vector of unexpected excess returns given the set of information,
t 1, and Ht is a conditional variancecovariance matrix of excess
returns following a multivariate GDC-GARCH process1 given by:

H t Dt Rt Dt t

where:



3
c
c
c
c
reg;t1 Covt1 Ri;t ; Rreg;t =t1 M;t1 Covt1 Ri;t ; RM;t =t1
6



 7
c
c
c
c
6
7
pi;t1 6 T;t1 Covt1 Ri;t ; RT;t =t1 S;t1 Covt1 Ri;t ; RS;t =t1
7 3
4



 5
c
c
c
c
I;t1 Covt1 Ri;t ; RI;t =t1 N;t1 Covt1 Ri;t ; RN;t =t1




c
1pi;t1 i;t1 Vart1 Ri;t =t1
2

Et1 Ri;t

where Freg,t 1 and Fi,t 1 are respectively a set of regional and local
variables.
The estimated model consists of a system of eleven equations (ve
equations of excess returns for each country i, one equation of excess
returns for the region, and ve equations of real exchange rate indices).
More precisely, the econometric specication of the model to be
estimated, i.e., Eqs. (2), (3), and (4), is characterized by the following
system of equations:

 
p
Dt dijt ; dijt iit i; dijt 0ij
 
t ijt
ijt ij ai t1 t1 a j g i H t1 g i i; j
ai ; g i ; i 1; 11 are 11  1 vectors of parameters
ij ; ii 0i; ij ji

1
This multivariate framework is more suitable than the bivariate one for taking into account the dynamic interactions between all the variables included in the system.

I. Abid et al. / Economic Modelling 37 (2014) 408416

The dynamic correlation structure Rt is specied by Tse and Tsui

(2002) as follows: Rt = (1 1 2)R + 1t 1 + 2Rt 1 with
0 1 + 2 b 1,where R = (ij) is a symmetric (11 11) positive
denite matrix with ii = 1, and t 1 is the (11 11) correlation
matrix2 of for = t M,t M + 1,,t 1.3 Its ijth element is
given by:
M
X

ui;tm ; u j;tm

m1
i; j;t1 v
!
!
u M
M
X
X
u
2
2
t
u i;tm
u j;tm
m1

m1

i;t
. The matrix t 1 can be expressed as t 1 =
where ui;t p

hii;t

1
1
B
B
t 1Lt 1Lt 1
t 1 in which Lt 1 = (ut 1,,ut M) is a (11 M)
matrix and Bt 1 is a (11 11) diagonal matrix with ith diagonal ele"
#1=2

ment given by ui;th

h1

where ut = (u1,t, u2,t, u11,t) .

The estimation of the vector of unknown parameters is carried out

by the Quasi-Maximum Likelihood Estimation (QMLE) method as
proposed by Bollerslev and Wooldridge (1992) to avoid the problem
of non-normality in excess returns. Given the highly non-linear structure of the model and the large unknown parameter number, the simultaneous estimation of the model is not feasible. So, we follow the
common literature, i.e., Hardouvelis et al. (2006) and Guesmi and
Nguyen (2011) to estimate the system (Eq. (7)) in two steps; and
thus study the regional integration process of the ve emerging markets
(M, T, S, I and N). In the rst stage, we estimate a subsystem of six equations for excess returns on regional and individual markets, and ve real
exchange rates plus the relevant variancecovariance elements of
Eq. (8). This step allows us to obtain the conditional variances of regional market and real exchange rate, their conditional covariances as well
as the prices of regional market and exchange rate risks. In the second
stage, we estimate the price of local market risk and the time-varying
level of integration for each emerging market in the system (Eq. (7)).
We maintain the same prices of regional market and exchange rate
risks across different emerging markets by imposing the estimators
obtained from the rst stage.
3. Data
3.1. Stock market and exchange rate returns
The market indices for Malaysia, Thailand, Singapore, Indonesia, and
Sri Lanka are obtained from Thomson Datastream International from
January 1996 through December 2007.4 We use monthly stock returns
in excess of the one-month Eurodollar interest rate, which is considered
as a risk-free rate. Monthly stock returns are calculated from stock
market indices with dividends reinvested.
Real exchange rates represent the value of the local currency against
the U.S. dollar and are extracted from the IMF's International Financial
Statistics (IFS) and the U.S. Federal Reserve databases. The real effective
exchange rate index is the geometric average of bilateral real exchange
rates among the countries under consideration.
3.2. Regional and local informational variables
As regional instrumental variables are used to explain changes in the
prices of regional markets and foreign exchange risks, we use: the
2

A necessary condition to ensure the positivity of both t 1 and Rt is that M N = 1.

For a complete review of the choice of the parameter M, see Duchesne and Lalancette
(2003).
4
Our sample excludes the episodes of the last Global Financial Crisis that could generate
biased estimates.
3

411

dividend yield of the region in excess of the 30-day Eurodollar interest

rate (RIDY), the regional market index return (RRENT), and the region
term spread (RPRM).
As local instrumental variables, we consider the dividend yield of a
market portfolio (DDIV), the return on the stock market index in excess
of the 30-day Eurodollar interest rate (RSRI), and the variation in the ination rate (DINF). Data are extracted from MSCI and Datastream
International.
3.3. Financial integration instrumental variables
Fluctuations in the regional stock market constitute a source of
systematic risk within the context of an ICAPM model with partial integration. The theory suggests that this risk is relevant and priced; so we
hint at a number of instrumental variables that may help to describe
the prices of risk. The commonly used variables are summarized below:
List of integration instrumental variables
Determinant variables

Measurements

Market openness (MO)

Total trade with the world/

nominal GDP

References

Bekaert and Harvey (1997,

2000), Rajan and Zingales
(2001), Bhattacharya and
Daouk (2002), Carrieri
et al. (2007).
Stock Market
Market value/nominal GDP
Levine and Zervos (1998)
Development (SMD)
Bekaert and Harvey (1995,
1997), Bekaert et al.
(2002), and Carrieri et al.
(2007).
Industrial Production (IP) log (Industrial Production)
King and Levine (1992,
1993), Savides (1995) and
Odedokun (1996).
Ination Rate (IR)
(CPIt CPIt 1) / CPIt 1
Boyd et al. (2001).
Harvey (1995) and
U.S. Term Spread (UTS)
Ln (U.S. Treasury 10 year
bond U.S. risk free 30 day Hardouvelis et al. (2006).
rate)
Dividend Yield
DY of country i-DY world;
Bekaert and Harvey (1995,
Differential (DYD)
with DY = dividend/price
2000) and Hardouvelis
et al. (2006).
Jorion (1991), De Santis
Exchange Rate Volatility
Conditional volatility
and Gerard (1998), and
(ERV)
generated from an AR(1)
Bollerslev and Wooldridge
with GARCH(1,1) errors on
log exchange rate expressed (1992).
in USD.
Economic Growth Rate
Ln (Gross Domestic Product) King and Levine (1992,
(EGR)
1993), Savides (1995) and
Odedokun (1996).
Current Account Decit
Ln (export import)
Guesmi (2011).
(CAD)
Market Returns (MR)
Ln (Pt / Pt 1)
Bekaert and Harvey (1997,
2000).
Interest Rate (IR)
Ln (short term interest rate, Arouri (2006) and Carrieri
TB rate or interbank rate)
et al. (2007).
Difference in Industrial
IP country i-IP G7
Gurley and Shaw (1967),
Production (DIP)
King and Levine (1993),
and Arouri (2006).

3.4. Descriptive analysis of data

Table 1 presents the descriptive statistics for stock market and real
exchange rate returns. The average stock returns are negative for the
considered countries, and range from 0.017 (Sri Lanka) to 0.006
(Indonesia). Thailand is the least volatile market with a standard deviation of 0.071. While the highest market is that of Singapore (0.113), for
which the skewness coefcients are negative, denoting that the return
distributions are skewed toward the left and that the probability of
observing extreme negative returns is higher than that of a normal
distribution. The kurtosis coefcients are signicant, and greater than
three in all cases, and thus reveal the leptokurtic behavior of return distributions. Altogether, the non-normality of all the return series is clearly conrmed by the JarqueBera test. Besides, the Engle (1982) test

412

I. Abid et al. / Economic Modelling 37 (2014) 408416

Table 1
Descriptive statistics of return series.
Std. dev.

Skewness

Kurtosis

JB

Q(12)

ARCH(6)

Panel A: Excess returns on stock market indices

Malaysia
0.014
Singapore
0.009
Sri Lanka
0.017
Thailand
0.008
Indonesia
0.006

Mean

0.072
0.113
0.144
0.071
0.121

0.941+
0.075
0.961
0.442
0.312

5.395++
6.660
5.614
4.312
5.322

55.332+++
79.957+++
62.749+++
14.926+++
17.116+++

68.86+++
101.97+++
52.018+++
58.43+++
78.53+++

0.403+++
0.016+++
0.072+++
0.472+++
0.972+++

Panel B: Real exchange rate returns

Malaysia
0.034
Singapore
0.269
Sri Lanka
0.291
Thailand
0.655
Indonesia
0.036

0.007
0.038
0.045
0.075
0.008

0.703
1.109
1.905
1.362
0.903

2.587
3.594
4.965
3.575
3.587

12.914+++
31.689+++
11.342+++
46.543+++
14.114+++

41.79+++
55.55+++
30.02+++
27.72+++
21.59+++

0.093+++
0.169+++
0.260+++
0.322+++
0.223+++

Notes: This table shows the basic statistics and the stochastic properties for stock returns in excess of the Eurodollar rates at 1 month and the exchange rate. +, ++, and +++ indicate that
the null hypothesis of normality, of no autocorrelation, and of no ARCH effect is rejected at the 10%, 5% and 1% rate, respectively.

highlights the existence of ARCH effects in all the returns series, which
obviously supports our decision to model the conditional volatility of
returns by a GARCH-type process.
Also, all the exchange rate returns are positive, and range from an
average of 0.034 (Malaysia) to 0.655 (Thailand). Their return distributions deviate signicantly from normality. The JarqueBera test statistic
strongly rejects the hypothesis of normally distributed returns. Moreover, we nd the presence of ARCH effects for all the series. Similar to
stock returns, the LjungBox test of order 12 reveals that exchange
rate returns are subject to serial correlation.

4.1. Regional market prices and foreign exchange risks

We report in Table 2 the regional market prices and real exchange
rate risks respectively in panels A and B.
It appears, from Panel A, that the price of currency risk for Malaysia
and Thailand is explained by three variables (RIDY), (RRENT) and
(RPRM). For Singapore's market, the price of currency risk is mainly
determined by (RRENT) and (RPRM). Also, the price of exchange rate
risk for Indonesia is, mainly and positively, determined by (RIDY) and
(RRENT).
Also, it appears that the price of regional market risk (in Panel B) is
also signicantly and positively explained by all the regional variables.
Moreover, we investigate the economic signicance of the risk
factors considered by testing the null hypotheses that the prices of
risk are equal to zero or constant, respectively. The Wald test results, reported in Table 3, indicate the rejection of these null hypotheses at 1%
level for all the markets considered. Also, the hypothesis that the price
Table 2
Regional market prices and real exchange rate risks.
Constant

Sri Lanka
Thailand
Indonesia

(0.044)
0.546
(0.129)
0.122
(0.111)
0.111
(0.134)

Null hypothesis

p-Value

The price of market risk of the South Asian

region is equal to zero? H0:reg = 0
The price of market risk of the South East
Asian region is constant? H0:reg = 1
The price of exchange rate risk of the South
Asian market is jointly zero? H0:k = 0
The price of exchange rate risk of the South
Asian market is jointly constant? H0:k = 1

111.23

0.0000

224.111

0.0000

114.152

0.000

111.455

0.000

Note: indicates that the coefcients are signicant at the 1% levels.

4. Empirical results

Panel A: Price of exchange rate risk

Malaysia
0.311
(0.146)
Singapore
0.113

Table 3
Specication test for prices of regional and exchange rate risks.

RIDY

RRENT

RPRM

0.024
(0.005)
0.0022
(0.0054)
0.012
(0.014)
0.014
(0.001)
0.015

0.050
(0.020)
0.022

0.033
(0.007)
0.012

(0.005)
0.056
(0.002)
0.05
(0.001)
0.06
(0.004)

(0.001)
0.018
(0.017)
0.013
(0.026)
0.017
(0.025)

0.007
(0.0005)

0.004
(0.001)

(0.003)

Panel B: Price of regional market risk

Asia
0.06
0.061
(0.011)
(0.072)

Note: and indicate that the coefcients are signicant at the 5% and 1% levels.

of currency and local risk are equal to zero or constant can also be
rejected at the 1% signicance level. These ndings effectively concur
with those of previous studies, including for example Adler and
Dumas (1983), Hardouvelis et al. (2006), and Carrieri et al. (2007).
4.2. Financial integration factors
To identify the determinants of the nancial integration, we
estimate the model (Eq. (7)) jointly for all studied markets and for
each factor at a time, using the Multivariate Nonlinear Least Squares
Method. Following Bhattacharya and Daouk (2002), we impose the
same coefcients on the system (Eq. (7)) to estimate the determinant
factors' coefcients (0 and 1) of stock market integration in emerging
market returns. This assumption allows us to capture the impact of each
candidate factor on the integration of individual markets. Referring to
previous studies (Bekaert and Harvey, 1997, Grifn, 2001; Karolyi and
Stulz, 2002), we use the U.S. dollar as the reference currency (column
(I) of Table 4). However, when taking into account the regional integration, the benchmark portfolio is that of the regional market, this suggests that the estimation results may be sensitive to a benchmark
currency at a regional level if the member countries have different currencies. In the considered countries, Thailand has the largest share of
GDP, and its currency (Baht) is most commonly used in international
and regional trade. Therefore, we consider the Baht as the new reference
currency instead of the U.S. dollar to study the impact of changing the
reference currency on the estimation of nancial integration determinants. So, we re-estimate the system (Eq. (7)) for each integration
factor. The results are presented in column (II). In addition, we use a
real effective exchange rate (REER) index as a proxy of the bilateral exchange rates presented in column (III). For each emerging market, the
REER index is computed as the geometric weighted average of countries
regional members' exchange rates against the U.S. dollar, where the
weights are the share of each country in the foreign trade with the
rest of the world. By construction, the REER index also allows for
cross-country comparisons of changes in trade competitiveness.

I. Abid et al. / Economic Modelling 37 (2014) 408416

413

Table 4
Robustness tests of the choice of currency reference.
Bilateral exchange rates against the
dollar (I)

Bilateral exchange rates against region

currency (II)

Real effective exchange rate index (III)

v0

v0

v0

1.944 (0.08)
7.764 (2.339)

v1
4.486 (2.073)
13.057 (3.614)

Trade Openness
Stock Market Development
National Industrial Production
0.27 (0.739)
0.0115 (0.373)
World Industrial Production
1.080 (0.114)
5.589 (4.761)
Differences in Industrial Production Growth Rates
0.283 (0.425)
0.255 (0.549)
Ination Rate
0.230 (0.607)
0.048 (0.103)
Exchange Rate Volatility
4.960 (5.620)
1.250 (5.545)
Economic Growth Rate
0.704 (0.073) 1.464 (0.081)
Dividend Yield on the Local Market Index
0.495 (1.043) 4.597 (0.893)
Dividend Yield on the Regional Market Index
0.288 (0.474)
0.001 (0.030)
Dividend Yield on the World Market Index
0.080 (0.180)
0.140 (0.760)
Differences in Dividend Yield
0.043 (0.213)
0.075 (0.078)
U.S. risk free 30 day rate
0.201 (0.540)
0.822 (0.423)
U.S. Treasury 10 year bond
0.143 (0.432)
0.020 (0.01)
U.S. term spread
0.263 (0.093)
0.100 (0.021)
Current Account Decit
0.290 (0.771)
0.023 (0.364)
Local Market Returns
0.498 (0.475)
4.596 (5.147)
Regional Market Returns
11.706 (1.643)
6.180 (0.951)
World Market Returns
0.021 (0.054)
0.041 (0.543)
World Interest Rate
0.383 (0.524)
0.155 (0.748)

v1

v1

7.480 (2.431)
8.914 (2.825)

5.654 (1.654)
5.530 (1.637) 11.127 (3.142)
0. 789 (0.028)
2.342 (1.499)
3.603 (3.469)
0.285 (0.762) 1.214 (0.384)
0.286 (0.641)
2.399 (0.315)
0.243 (0.067) 0.005 (0.002)
0.073 (0.167)
0.811 (0.664)
0.383 (0.525)
0.155 (0.749)
0.129 (0.169)
0.045 (0.775)
0.063 (0.353)
2.010 (0.073)
0.064 (0.803)
0.702 (0.526)
0.143 (0.432)
0.020 (0.001)
2.384 (0.889)
0.001 (0.875)
1.230 (2.920)
1.563 (7.345)
1.519 (1.659)
0.201 (1.654)
0.807 (0.953) 1.732 (0.619)
0.646 (0.644) 7.198 (0.732)
0.213 (0.343)
0.023 (0.364)
0.161 (0.132)
0.025 (0.415)
0.569 (0.730) 4.050 (0.987)
1.569 (1.320) 3.750 (1.450)
1.060 (1.230)
0.030 (0.155)
0.437 (0.664) 2.849 (0.862)
0.507 (1.053) 4.597 (0.892)
0.339 (0.140)
0.153 (0.192)
0.158 (0.471)
0.254 (0.162) 5.031 (0.744)
5.346 (0.767)
0.383 (0.024)
0.165 (0.017) 0.090 (0.008)
0.016 (0.005)
0.042 (0.032)
0.254 (0.943)
0.490 (0.766)
0.040 (0.449)
0.078 (0.184)
0.137 (0.755)
0.035 (0.008)
0.008 (0.047)
7.480 (2.431)
6.045 (1.546)
4.530 (0.637)
8.273 (1.102)
8.179 (1.258)
0.892 (0.008)
3.042 (2.049)
3.036 (3.496)
0.285 (0.762) 1.214 (0.384)
0.286 (0.641)
2.399 (0.315)

Notes: We estimate the system (Eq. (7)) for all countries and consider one candidate factor for nancial integration at a time. Columns (I), (II) and (III) report the estimation results
respectively for the bilateral exchange rates against the U.S. dollar, the bilateral exchange rates against the Baht, and the REER. The numbers in parentheses are the associated standard
deviations. , , and indicate signicance at the 10%, 5% and 1% levels, respectively.

The results show that a higher degree of market openness leads to an

increase in the exposure of national markets to global risk factors.
Besides, this factor affects positively the evolution of regional nancial
integration in the case of the different currency specications (columns
I, II and III). Bekaert et al. (2002, 2005), Rajan and Zingales (2001), and
Bhattacharya and Daouk (2002) document that higher degree of market
openness. Thus, as the markets become more open to foreign trade and
capital ows, their level of economic integration rises, and asset
exchanges become signicant. Consequently, the degree of market
openness can be a potential factor in promoting nancial integration.
Moreover, the U.S. Term Spread is found to have signicant impacts
on the evolution of nancial integration in the case of the three currency
specications. The U.S. Term Spread has substantial effects on valuation
and on nancial asset allocation in an international context. Adler and
Qi (2003) use the interest rate spread as a factor of nancial integration,
and nd that this variable affects the mobility of international capital
ows that, in turn, leads to changes in the level of market integration.
If we consider the regional market return factor, the estimated coefcients are signicant for the three specications and for the considered
countries. Moreover, they are positive for all the markets indicating a
positive correlation between the increase of regional stock returns and
intra-regional nancial integration. Levine et al. (2000) show that indicators of economic growth are positively related to the stock markets'
integration.
To conclude, we note that the main results remain the same despite
the change in base currency due to the dependence of these currencies
on the dollar.

4.3. Regional integration

We shall focus on the dynamics of stock market integration, reported
in Fig. 1, and estimated using two factors: the U.S. term premium (UTS),
and the level of market openness (MO). In fact, since there is a numerical
convergence problem at the estimation stage when we have more than
two unknown parameters, only two information variables are used to
capture the evolution of market integration. On the light of the previous
analysis, and in regard to the better statistical results of the Bayesian
Information Criterion (BIC), we choose two retain the U.S. term premium
(UTS), and the level of market openness (MO) as information variables.

At rst sight, we notice that Singapore, Malaysia and Thailand exhibit the same feature, displaying high integration degrees approaching
70% at the end of the sample. It appears clearly that from the beginning
of the 2000s, there was a general increase in the case of the precited
countries. This may be explained by the regional cooperation process.
Such cooperation pursues both market-sharing and resource-pooling
strategies and achieves greater economic integration. We also remark
that the increase in the degree of integration for Malaysia is higher
than that for Singapore and Thailand.
Moreover, the Malaysian market reached the highest integration
level, exceeding 70%. It is clearly the most integrated market in the
South Asian region. This result was expected since Malaysia is one of
the most important nancial markets in the South Asian region. The
Malaysian market tends to compensate for the shortcomings of local
markets, which are insufciently open and which liaise with less developed neighboring markets such as Thailand to transfer technologies and
services not available on the domestic market.
The Sri Lankan and Indonesian markets show a far lower regional integration level than the other countries in the region during 20002007.
The graphical inspection (Fig. 1) shows that the intra-regional nancial
integration does not register any particular trend upward or downward.
This nding may be related to the no signicant interdependence
between Sri Lankan and Indonesian stock markets and the other Asian
countries.
To complete our analysis, we report in Table 5 the dynamics of stock
market integration levels.
With an average level of about 0.512, Thailand is the least integrated
country within the regional market even if its process of nancial integration has begun with structural reforms aimed at stimulating the
private sector, and the opening of markets to foreign investors in the
late 1980s.
The Singapore market has an average of 60.1%, followed by the
Malaysian one with an average of 55.3%, and the Sri Lankan market
with an average of 53.1%. We can deduce that, with the exception of
the Indonesian and Sri Lankan markets, the degree of integration has become very important in the study area from the 2000s. Petri (1993)
nds that the effects of geographical proximity are not signicant in
the Asian region, indicating that the strategy of developing Asian countries turned to the conquest of foreign markets. These results are veried by Frankel and Romer (1999) and Guesmi (2012). In fact, they

414

I. Abid et al. / Economic Modelling 37 (2014) 408416

1.1Malaysia

1.2 Singapore
.8

.8

.7

.7

.6

.6

.5

.5

.4

.4

.3

96

97

98

99

00

01

02

Integration

03

04

05

06

.3

96

07

97

98

99

HP-Filtered

00

01

02

Integration

1.3 Sri Lanka

03

04

05

06

07

HP-Filtered

1.4 Thailand
1.0

.8
.7

0.8

.6

0.6

.5
.4

0.4

.3

0.2

96

97

98

99

00

01

Integration

02

03

04

05

06

.2

07

96

97

98

99

HP-Filtered

00

01

02

Integration

1.5 Indonesia

03

04

05

06

07

HP-Filtered

.9
.8
.7
.6
.5
.4
.3

96

97

98

99

00

01

Integration

02

03

04

05

06

07

HP-Filtered

Fig. 1. Dynamic integration of emerging markets into the South Asian regional market.

show that intra-regional trade integration in Asia is more inuenced by

the rapid growth of the country than by a genuine commitment to economic integration. Moreover, there is no obvious indication of intensied regional nancial market integration. Nonetheless, this seems to
reveal a close correspondence between measures of nancial integration, and the extent of the development of nancial markets in general.
The high-income economies of Singapore are fairly highly integrated
with regional capital markets. The recent pace of liberalization in South
Asia post-crisis is also intensifying the extent of the country's regional
and international nancial integration. The lower-middle-income
Southeast Asian countries, Thailand and Indonesia as well as Sri Lanka,
are relatively less nancially integrated, though evidence suggests a
gradual movement toward enhanced integration. The evidence on
Malaysia is mixed (a low integration level until 2000 and an upward

trend throughout the rest of the period); also, there is no evidence on

Sri Lanka. The fact of not having a common trend for the markets
under consideration is due to the short period of the study. These
ndings may be due to the non-inclusion of smaller economies like
Cambodia and Vietnam that are relatively integrated with the Asian
regional market thanks to their liberalization politics, and nancial
market deregulation.
In order to examine the relevance of the local risk price in the valuation of nancial assets issued by Asian markets, we use the robust
Wald test (Table 6) to check the nullity of the coefcients associated
with the information variables. The results from the Wald test clearly
reject the hypotheses according to which the local risk prices are individually equal to zero. In parallel, the assumptions of constant local
risk price are rejected for the considered markets. These ndings are

I. Abid et al. / Economic Modelling 37 (2014) 408416

415

Table 5
Dynamics of stock market integration.
Panel A: Parameters of the market integration measure

Sri Lanka
Malaysia
Singapore
Thailand
Indonesia

Constant

MO

UTS

0.196 (0.035)
0.277 (0.01)
0.561 (0.059)
0.181 (0.222)
0.221 (0.342)

0.132 (0.031)
0.151 (0.066)
0.061 (0.002)
0.307 (0.013)
0.207 (0.011)

0.156 (0.003)
0.155 (0.053)
0.117 (0.007)
0.052 (0.002)
0.032 (0.001)

p mean

p max

p min

0.531 (0.092)
0.553 (0.130)
0.601 (0.115)
0.512 (0.114)
0.525 (0.08)

0.846
0.788
0.790
0.767
0.844

0.214
0.314
0.312
0.266
0.361

Panel B: Statistics of market integration

Sri Lanka
Malaysia
Singapore
Thailand
Indonesia

Notes: The numbers in parentheses are the associated standard deviations. , , and indicate that the coefcients are signicant at the 10%, 5%, and 1% levels, respectively.

consistent with those of previous studies, including that of Carrieri et al.

(2007), Tai (2007), in the sense that the local risk is a relevant source of
risk in the valuation of nancial assets issued by emerging markets in
the Asian region. Also, the exposure to these local markets changes
over time.

It appears that normality of the estimated residuals can be accepted

for Malaysia, Singapore, Sri Lanka, and the regional market. The 1982
Engle's test for conditional heteroscedasticity of the standardized
residuals indicates that ARCH effects no longer exist in all cases, thus
revealing the appropriateness of the GARCH modeling approach. Such
evidence against normality warrants the use of QML testing procedures.

4.4. Formation of total risk premium

Table 7 indicates that the regional and local risk premiums are
signicantly different from zero at the 1% level for all the emerging
markets studied. Malaysia has the highest total risk premium market
(11.909%), followed by Sri Lanka (11.5%), Indonesia (8.592%),
Singapore (6.847%) and Thailand (5.189%). The Exchange risk premiums
are, on average, greater than the regional ones for all the countries. The
contribution of currency risk premium (EPRM) is also higher for
Malaysia, Singapore and Indonesia; the exchange risk premium is the
main component of the total risk premium for these economies. Our results conrm those of Arouri (2006), Phylaktis and Ravazzolo (2002),
Carrieri et al. (2007) and Guesmi (2012) who show that currency risk
is the most important risk factor.
Finally, throughout the study period, the premium associated with
the exchange risk is statistically and economically signicant for the
ve economies studied. However, the contribution of the exchange premium to the total premium is more pronounced for Malaysia, Singapore
and Indonesia. The contribution of the local risk factor is also statistically
signicant but economically weak. For the rest of countries, the total risk
premium is mainly determined by the regional market risk factor
(Arouri, 2006; Guesmi, 2012).
Table 8 presents an analysis of the model's residuals in terms of
normality, autocorrelation and conditional heteroscedasticity.

5. Conclusion
We developed a conditional ICAPM in the presence of exchange rate
risk to identify factors that may inuence the degree of nancial integration for ve major markets in Southeast Europe. The ndings are then
used to study the dynamics of nancial integration. Our empirical analysis is conducted on the basis of a nonlinear framework, which relies on
the multivariate GDC-GARCH model.
By allowing the prices of risk and the level of market integration to
vary through time, we show that the degree of trade openness and variation in the U.S. term premium are the most important determinants of
regional nancial integration. Moreover, the degree of market integration admits frequent changes over the study period and its dynamic patterns differ greatly across the markets under consideration. The average
premium for global risk appears to be only a small fraction of the average of the total premium. These results thus suggest that diversication
into emerging market assets continues to produce substantial prots,
and that the asset pricing rules should reect a state of partial integration. Our investigation, which addresses the evolution and formation
of total risk premiums, conrms this empirically.

Table 7
Decomposition of the total risk premium.

Table 6
Specication test of price of local risk.

LPRM (%)

Null hypothesis

p-Value

Is the local risk price in Thailand zero? H0:T = 0

Is the local risk price in Thailand constant? H0:T = 1
Is the local risk price in Singapore zero? H0:N = 0
Is the local risk price in Singapore constant? H0:N = 1
Is the local risk price in Sri Lanka zero? H0:S = 0
Is the local risk price in Sri Lanka constant? H0:S = 1
Is the local risk price in Malaysia zero? H0:M = 0
Is the local risk price in Malaysia constant? H0:M = 1
Is the local risk price in Indonesia zero? H0:I = 0
Is the local risk price in Indonesia constant? H0:I = 1

18.113
84.234
67.211
99.488
22.555
21.600
18.711
22.110
387.182
70.393

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000

Note: indicates that the coefcients are signicant at the 1% level.

Malaysia
Singapore
Sri Lanka
Thailand
Indonesia

+++

1.120
(0.130)
1.389+++
(0.149)
1.111+++
(0.152)
1.000+++
(0.166)
1.022+++
(0.225)

RPRM (%)
+++

4.412
(0.120)
2.145+++
(0.812)
5.203+++
(0.028)
1.745+++
(0.150)
3.751+++
(0.143)

EPRM (%)
+++

6.377
(0.244)
2.953+++
(0.011)
5.186+++
(0.178)
2.444+++
(0.131)
3.819+++
(0.122)

TPRM (%)
11.909+++
(0.170)
6.487+++
(0.151)
11.500+++
(0.125)
5.189+++
(0.213)
8.592+++
(0.203)

Note: +++ indicates that the average risk premiums are signicantly different from zero at
the 1% level with respect to the two-sided Student-t test.

416

I. Abid et al. / Economic Modelling 37 (2014) 408416

Table 8
Residuals analysis.

Malaysia
Singapore
Sri Lanka
Thailand
Indonesia
Region

Skewness

Kurtosis

JB

Q(12)

ARCH(6)

1.172+
0.382
1.418
0.291
0.333
1.514

5.441++
5.843
15.368
3.247
7.666
16.244

67.786+++
51.282+++
952.563+++
2.356
22.356+++
10.131+++

13.392
16.801
9.739
5.873
7.765
13.456

0.196
0.190
0.285
0.062
0.333
0.115

Notes: Numbers in parentheses are the associated standard deviations. JB, Q(12), and
ARCH(6) are respectively, the empirical statistics of the JarqueBera test for normality,
the LjungBox test for serial correlation of order 12, and Engle's (1982) test for conditional
heteroscedasticity. +, ++, and +++ indicate that the null hypothesis of normality and zero
autocorrelation is rejected at the 10%, 5% and 1% levels, respectively.

References
Adler, M., Dumas, B., 1983. International portfolio selection and corporation nance: a
synthesis. J. Financ. 38, 925984.
Adler, M., Qi, R., 2003. Mexico's integration into the North American Capital market.
Emerg. Econ. Rev. 4, 91120.
Anderson, N.H., 1997. Intuitive physics: understanding and learning of physical relations.
In: Ward, T.B., Smith, S.M., Vaid, J. (Eds.), Creative thought: an investigation of conceptual structures and processes. APA, Washington, pp. 231265.
Arouri, M.H., 2006. Are stock markets integrated? Evidence from a partially segmented
ICAPM with asymmetric effects. Front. Finance Econ. 2, 7094.
Bekaert, G., Harvey, C.R., 1995. Time-varying world market integration. J. Financ. 50 (2),
403444.
Bekaert, G., Harvey, C.R., 1997. Emerging equity market volatility. J. Financ. Econ. 43, 2977.
Bekaert, G., Harvey, C.R., 2000. Foreign speculators and emerging equity markets.
J. Financ. 55, 565613.
Bekaert, G., Harvey, C.R., Lumsdaine, R., 2002. The dynamics of emerging market equity
ows. J. Int. Money Financ. 21, 295350.
Bekaert, G., Harvey, C.R., Lumsdaine, R., 2005. Does nancial liberalization spur growth?
J. Financ. Econ. 77, 355.
Berger, T., Pozzi, L., 2013. Measuring time-varying nancial market integration: an unobserved components approach. J. Bank. Financ. 37, 463473.
Bhattacharya, U., Daouk, H., 2002. The world price of insider trading. J. Financ. 57, 75108.
Black, F., 1974. International capital market equilibrium with investment barriers.
J. Financ. Econ. 1, 337352.
Bollerslev, T., Wooldridge, J.M., 1992. Quasi-maximum likelihood estimation and inference in dynamic models with time-varying covariances. Econ. Rev. 11, 143172.
Boyd, R.D., Johnston, M.E., Usry, J.L., Fralick, C.E., Sosnicki, A.A., Fields, B., 2001. Lysine level
required to optimize the growth performance to Paylean in PIC pigs. J. Anim. Sci. 79
(Suppl.1), 66 (Abstr.).
Carrieri, F., Errunza, V., Hogan, K., 2007. Characterizing world market integration through
time. J. Financ. Quant. Anal. 42 (04), 915940.
Chambet, A., Gibson, R., 2008. Financial integration, economic instability and trade
structure in emerging markets. J. Int. Money Financ. 27, 654675.
Claessens, S., Rhee, M., 1994. The effect of barriers to equity investment in developing
countries. In: Frankel, Jeffrey A. (Ed.), The Internationalization of Equity Markets.
University of Chicago Press, Chicago and London, pp. 231275.
Cooper, I.A., Kaplanis, E., 2000. Partially segmented international capital markets & international capital budgeting. J. Int. Money Financ. 19, 309329.
De Grauwe, P., Grimaldi, M., 2006. Exchange rate puzzles: a tale of switching attractors.
Eur. Econ. Rev. 50, 133.

De Santis, G., Gerard, B., 1998. How big is the premium for currency risk? J. Financ. Econ.
49, 375412.
De Santis, G., Gerard, B., Hillion, P., 2003. The relevance of currency risk in the EMU.
J. Econ. Bus. 55, 427462.
Duchesne, P., Lalancette, S., 2003. On testing for multivariate ARCH effects in vector time
series models. La Rev. Can. Stat. 31, 275292.
Engle, R., 1982. Autoregressive conditional heteroskedasticity with estimates of the variance of U.K. ination. Econometrica 50, 9871008.
Errunza, V., Losq, E., 1985. International asset pricing under mild segmentation: theory
and test. J. Financ. 40, 105124.
Frankel, J., Romer, D., 1999. Does trade cause growth? Am. Econ. Rev. 89, 379399.
Frankel, J., Wei, S., 1995. Emerging currency blocs. In: Genberged, H. (Ed.), The International Monetary System: Its Institutions and Its Future. Springer Verlag, Berlin,
pp. 111143.
Grifn, M.W., 2001. Complex cases. CAMHS Staff Seminar presented at Flinders Medical
Centre. Adelaide, (February).
Guesmi, K., 2011. What drive the regional integration of emerging stock markets? Econ.
Bull. 31 (3), 26032619.
Guesmi, K., 2012. Characterizing South-east Asian stock market integration through time.
Int. J. Bus. 17 (1), 100112.
Guesmi, K., Nguyen, Duc Khuong, 2011. How strong is the global integration of emerging
market regions? An empirical assessment. Econ. Model. 28, 25172527.
Gurley, J., Shaw, E., 1967. Financial structure and economic development. Econ. Dev. Cult.
Chang. 34 (2), 333346.
Hardouvelis, G.A., Malliaropulos, D., Priestley, R., 2006. EMU and European stock market
integration. J. Bus. 79 (1), 365373.
Harvey, C., 1995. Predictable risk and returns in emerging markets. Rev. Financ. Stud. 8,
773816.
Jorion, P., 1991. The pricing of exchange rate risk in stock market. J. Financ. Quant. Anal.
363376.
Karolyi, A.G., Stulz, R.M., 2002. Are nancial assets priced locally or globally?. NBER
Working Papers 8994, National Bureau of Economic Research, Inc.
King, R., Levine, R., 1992. Financial indicators and growth in a cross section of countries.
Working Paper 819. Policy Research. World Bank.
King, R., Levine, R., 1993. Finance and growth: Schumpeter might be right? Q. J. Econ. 108,
717737.
Levine, R., Loayza, N., Beck, T., 2000. Financial intermediation and growth: causality and
causes. J. Monet. Econ. 46 (1), 3177.
Levine, R., Zervos, A., 1998. Stock markets, banks and economic growth. Am. Econ. Rev. 88
(3), 537558.
Odedokun, M., 1996. Alternative econometric approaches for analyzing the role of the
nancial sector in economic growth: time-series evidence from LDCs. J. Dev. Econ.
50, 119146.
Petri, Peter A., 1993. The East Asian trading bloc: an analytical history. In: Frankel,
Jeffrey A., Kahler, Miles (Eds.), Regionalism and Rivalry (A National Bureau of
Economic Research Conference Report). University of Chicago Press, Chicago,
pp. 2152.
Phylaktis, K., Ravazzolo, F., 2002. Measuring nancial and economic integration with
equity prices in emerging markets. J. Int. Money Financ. 21, 879904.
Rajan, R., Zingales, L., 2001. The rm as a dedicated hierarchy: a theory of the origins and
growth of rms. Q. J. Econ. CXVI, 805852.
Savides, A., 1995. Economic growth in Africa. World Dev. 23 (3), 449458.
Stehle, R., 1977. An empirical test of the alternative hypotheses of national and international pricing of risky asset. J. Financ. 33, 493502.
Stulz, R., 1981. A model of international asset pricing. J. Financ. Econ. 9, 383406.
Tai, C.-S., 2007. Market integration and contagion: evidence from Asian emerging stock
and foreign exchange markets. Emerg. Mark. Rev. 8 (4), 264283.
Tse, Y.K., Tsui, K.C., 2002. A multivariate GARCH model with time-varying correlations.
J. Bus. Econ. Stat. 20 (3), 351362.
Verma, P., Verma, R., 2010. Response asymmetry of Latin American stock markets to the
U.S. money market. Glob. Econ. Financ. J 3 (2), 133147.