Global Finance Journal 19 (2009) 203–218

Contents lists available at ScienceDirect

Global Finance Journal

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / g f j

Return, volatility spillovers and dynamic correlation in the

BRIC equity markets: An analysis using a bivariate
EGARCH framework
Ramaprasad Bhar a,⁎, Biljana Nikolova b
a
School of Banking and Finance, The University of New South Wales Sydney 2052, Australia
b
nabCapital, A division of National Australia Bank Limited, Level 25, 255 George St, Sydney NSW 2000, Australia

a r t i c l e i n f o a b s t r a c t

Article history: This paper examines the level of integration and the dynamic
Received 14 July 2008 relationship between the BRIC countries, their respective regions
Accepted 6 September 2008 and the world. We find that India shows the highest level of regional
Available online 15 October 2008
and global integration among the BRIC countries, followed by Brazil
and Russia and lastly by China. There is a negative relationship
JEL classification:
between the location conditional volatility of India with that of the
E37
G15
Asia-Pacific region and of China with the world, which indicates a
presence of diversification opportunities for portfolio investors.
Keywords:
Portfolio investors can continue to receive sound returns from
Volatility spillover
Dynamic correlation taking positions in the index of these countries, however for an
BRIC outstanding investment performance, they should consider investing
Market integration in specific areas of growth within the economy rather than the
country index.
© 2008 Elsevier Inc. All rights reserved.

1. Introduction

The topics of financial liberalization and market integration have received great attention in the
financial literature, especially after the financial market crisis in 1997–1998. Experiences to date confirm
that while openness of financial markets contributes to economic development it also makes developing
countries more vulnerable to financial disruptions (Kaminsky and Schmukler, 2001, in press; Levine and
Schmukler, 2003). The increased awareness of the risks associated with financial market openness has
however not discouraged the process of financial liberalization of developing countries. On the contrary,
the experiences are used to broaden the knowledge and understanding of the process and have resulted in
great learnings from past mistakes. Amongst others, Ito (2006) has studied the process of financial

⁎ Corresponding author.
E-mail addresses: R.Bhar@unsw.edu.au (R. Bhar), Biljana.Nikolova@nab.com.au (B. Nikolova).

1044-0283/$ – see front matter © 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.gfj.2008.09.005
204 R. Bhar, B. Nikolova / Global Finance Journal 19 (2009) 203–218

liberalization of 87 less developed Asian countries and concluded that to benefit from more open cross-
border financial transactions, financial systems need to be equipped with reasonable legal and institutional
infrastructure. It is only under these circumstances that financial liberalization can be translated into
financial growth. To add further to the argument of the growing initiative towards greater liberalization and
integration of developing economies, there is a record of more than 50 regional and world, both trade and
non-trade, agreements concluded after 2000 or currently being negotiated by non-OECD countries
(Antkiewitcz and Whalley, 2006).
Wilson and Purushothaman of Goldman and Sachs Investment Bank produced a research paper in 2003
which has attracted the attention of many academic and industry professionals. In their paper called “Dreaming
with the BRICs: the Path to 2050” they identify Brazil, Russia, India and China (“BRIC”) as the new economic
growth force in the world. This forecast is subject to the BRICs maintaining policies and developing institutions
that are supportive of growth. They can develop to their potential only in the presence of a regulatory
environment that supports the openness to foreign investments and enables and supports free flow of capital.
Now that the potential of these countries has been unveiled and BRIC has become a common terminology
amongst the finance professionals, the weight of the BRICs in investment portfolios is expected to rise sharply.
Therefore it is very important to understand the nature of these markets and the way their equity market
returns and volatility of returns relate and respond to regional and global financial events, and vice versa.
All four of the BRIC countries have gone through the process of financial liberalisation in the early 1990s.
Bekaert, Harvey and Lundblad (2003) define the official equity liberalization of countries as “the date of
formal regulatory change after which foreign investors officially have the opportunity to invest in domestic
equity securities and domestic investors have the right to transact in foreign equity securities abroad”.
Based on this definition, Bekaert, Harvey and Lumsdaine (2002) identify the official liberalization dates for
the BRIC countries as May 1991 for Brazil, January 1994 for Russia, February 1992 for India and July 1993 for
China. If the liberalization is effective it is presumed to lead to market integration, which would in turn have
an effect on both the financial sector and the real sectors of developing countries.
Theory suggests that expected returns of emerging markets should reduce following integration of the
emerging market with the world economy (Bekaert et al., 2002). De Jong and De Roon (2005) found in their
study of time varying market integration and expected returns in 30 emerging markets from Latin America,
Asia, the Far East, Europe, the Mid-East and Africa, that the average annual decrease in local and regional
segmentation of 0.055, on a [0, 1] scale, induce a decrease in returns of about 4.5% per year for these
countries. The ultimate implication of the process of integration therefore is that portfolio managers would
no longer be able to add significant value by pursuing indexing strategies in emerging markets. Stock
selection strategies are imperative, if benchmarks are to be outperformed.
The purpose of this paper is to explore the level of integration of the BRICs with their respective regions and
the world by using the bivariate EGARCH structure, which allows for time varying conditional correlation of
index equity returns from these markets. The significance and behaviour of return and volatility spillovers
from these countries to their respective regions and the world, and vice versa, are used as proxies for the level
of integration of these markets regionally and globally. The dynamic conditional correlation aspect of the
model also allows us to observe the impact of a number of significant events in the BRIC markets on the
correlation of equity index returns in these markets with their respective regions and the world.
Using our weekly data set for the BRICs, their respective regions and the world over the period January
1995 to October 2006, we find that India shows the highest level of integration on a regional and world
level amongst the BRIC countries, followed by Brazil and Russia and lastly by China. There is a negative
relationship between the conditional volatility of India with that of the Asia-Pacific region, which, can be
attributed to the low level of impact of the Asian crisis on India. Given the relatively closed nature of the
financial system in China, there is no evidence of regional integration for China, only a negative relationship
with the conditional volatility of the world market returns, which indicates a presence of diversification
opportunities for portfolio investors. None of the BRIC countries impacts the equity price creation process
in their respective regions, none of them have a significant impact over the conditional volatility of world
market returns and only Russia has effect over the price creation process of world equity index prices.
The equity market index returns conditional correlation for Brazil with the region and the world projects
very similar trends for the observed time period, with an evidence of a slightly higher level of world conditional
correlation. Both Russia and India have relatively stable level of conditional correlation with the world, while
the regional conditional correlation is very volatile. Unlike the other BRIC countries, the results for the Chinese
 R. Bhar, B. Nikolova / Global Finance Journal 19 (2009) 203–218 205

equity market index returns show that there is a very weak, positive, relatively constant conditional correlation
with the Asia-Pacific region, and also weak but very volatile and mainly negative conditional correlation with
the rest of the world.
Based on the results and findings in this paper it can be concluded that the process of integration of the
BRIC countries with their respective regions and globally is relatively low at this stage, however it is
expected to continue to progress further given the initiatives undertaken by the countries' policy makers
stimulating the level of cross-country openness.
At this stage, portfolio investors can certainly continue to receive sound returns from taking positions in
the index of these countries and benefit from the existing diversification opportunities, however for an
outstanding investment performance, they should consider investing in specific areas of growth within the
economy rather than the country index.
The rest of the paper is organised as follows: section two provides information on the data used in the
paper and details results of the preliminary statistics; section three defines the methodology used to
measure the level of spillover effects and time varying conditional correlation; section four contains the
empirical results and discussion and section five concludes.

2. Data and preliminary statistics

Data used in this paper are weekly closing equity market price indices for four emerging markets: Brazil
(Bovespa), Russia (AK&M Composite), India (Sensex) and China (Shanghai Composite), weekly equity
market price indices for their respective regions: Americas (Financial Times World Index All Countries
Americas), Europe (Financial Times World Index All Countries Europe) and Asia-Pacific (Financial Times
World Index All Countries Asia Pacific), and weekly equity market price index for the World (Morgan
Stanley's All Countries World Index). Regional Financial Times indices were used, as Morgan Stanley indices
for Europe and Asia-Pacific were not available for the period prior to Jan 1999. The data are sampled weekly
(Wednesdays) over the period January 1995 to October 2006. Weekly (Wednesday) price series data have
been used to avoid non-synchronous trading and day-of-the week effects, as discussed in Ramchand and
Susmel (1998), Aggarwal, Inclan and Leal (1999), and Ng (2000). The data were sourced from Bloomberg.
Weekly equity index returns were calculated as natural logarithm of the price index relative, measured in
terms of U.S. dollars. Summary statistics for the weekly local, regional and world returns are presented in Table 1.
The average weekly returns for Brazil, Russia, India and China are 0.2058, 0.7849, 0.1274 and 0.1694
respectively, and the standard deviations are 6.0003, 6.4953, 4.2118 and 3.8871. The skewness and excess
kurtosis indicate that negative shocks are more common than positive for Brazil, Russia and India and positive
are more common for China. In general, the distribution properties of all return series appear to be non-normal.
The first order autocorrelation for the local, regional and world indices ranges from −0.1438 to 0.0143. The
Portmanteau tests for serial correlation for the returns and the squared value of the returns confirm that there
is persistence of non-linear dependence, that is, there is a presence of conditional heteroscedasticity in the
returns of all indices in the sample.

Table 1
Summary statistics for weekly equity index returns for the local, regional and world indices

Index Mean Std. Dev. ρ1 Q1 (20) ρ2 Q2 (20) Skewness Kurtosis

Local
Brazil 0.2058 6.0003 −0.0833 36.6270 0.2218 155.9304 −0.8069 7.2714
Russia 0.7849 6.4953 0.0143 33.7728 0.4036 326.6919 −0.4547 12.8307
India 0.1274 4.2118 0.0138 35.2749 0.0249 105.7159 −0.3648 11.4782
China 0.1694 3.8871 −0.0598 27.1266 0.1184 60.8442 0.2492 8.3383
Regional
Americas 0.2107 2.3696 −0.0736 31.0635 0.2489 238.1970 −0.1866 3.8413
Europe 0.1697 2.3304 −0.1438 48.3560 0.3613 249.4070 −0.5305 6.4598
Asia-Pacific 0.0445 2.8165 0.0087 21.1901 0.0086 67.2645 0.1097 3.3386
World 0.1252 2.0065 −0.0655 30.0535 0.2489 238.1970 −0.2536 4.7276

Data used are weekly equity index returns for the period January 1995 to October 2006. Q1 (20) refers to the Portmanteau statistic
with the null hypothesis of no data series serial correlations measured with a lag of 20. Similarly, Q2 (20) Sq refers to the same test
with squared data series. Large p-value entries would indicate that there are no serial correlations in the data series.
206 R. Bhar, B. Nikolova / Global Finance Journal 19 (2009) 203–218

Table 2
First order unconditional correlations of the returns and squared returns

Brazil Russia India China

Regional 0.4446 0.2333 0.2529 0.0878
World 0.5137 0.2611 0.2324 −0.0095

Data used are weekly equity index returns for the period January 1995 to October 2006.

Table 2 contains the results of the unconditional correlation between the local markets and the
regional and world market. It is evident that there is a relatively strong and positive correlation among the
countries with their respective regions and the world, except for China, in which case there is a
comparatively weak positive regional correlation of 0.0878 and a weak negative correlation with the
world of − 0.0095.
The augmented Dicky Fuller and Phillip Perron unit root tests were conducted for the local, regional and
world data, and all of them rejected the null for presence of unit root.

3. Bivariate EGARCH model

A well documented empirical finding in the finance literature is the asymmetric impact of news on
the volatility transmission (see Bae and Karolyi, 1994; Koutmos and Booth, 1995 and Booth; Martikainen,
and Tse, 1997). The asymmetric phenomenon in combination with the observed volatility clustering in
equity market returns validates the use of a bivariate EGARCH framework. The bivariate EGARCH
model, as developed by Nelson (1991), captures the potential asymmetric behaviour of equity market
returns and avoids imposing non-negativity constraints in GARCH modelling — by specifying the
logarithm of the variance ln(σt2), it is no longer necessary to restrict parameters in order to avoid
negative variances.
The purpose of this paper is to analyse the sources of volatility for equity markets in the BRIC countries.
The volatility spillover mechanism in the BRIC equity markets is modeled by assuming three sources of
shocks: local, regional and world. Regional shocks are represented by the returns of the respective
regional equity market indices, while world shocks are represented by the returns in the world equity
market index.
Here is a brief description of the bivariate EGARCH model with time varying correlations relating the
equity return from the BRIC countries and the regional equity index and also between the BRIC countries
and the world index return. We denote the return from one of the BRIC countries by rj,t where the subscript
j represents one of the BRIC index return, and by rI,t one of the regional indices. For Brazil the regional index
is that of Americas, for China and India it is the Asia-Pacific and for Russia it is the European regional index.
We will use the symbol rw,t to represent the world index return. The bivariate model we are analyzing
would, therefore, consist of either (rj,t, rI,t) or (rj,t, rw,t).
The mean spillover effect is captured by the following relationship:
        
rj;t βj;0 βj;1 βj;2 rj;t−1 ε
= + + j;t ; ð1Þ
rI;t βI;0 βI;1 βI;2 rI;t−1 εI;t

where,

j
 
εj;t
Xt fNð0; ∑t Þ ð2Þ
εI;t

Here, Ωt indicate all relevant information known at time t, and ∑t is the time varying covariance matrix
defined below. The diagonal elements of the (2 × 2) covariance matrix are given by:
 

 
ln σ 2j;t = α j;0 + α j;1 f1 zj;t−1 + α j;2 f2 zI;t−1 + γ j ln σ 2j;t−1 ; and ð3Þ

 

 
ln σ 2I;t = α I;0 + α I;1 f1 zj;t−1 + α I;2 f2 zI;t−1 + γI ln σ 2I;t−1 ð4Þ
 R. Bhar, B. Nikolova / Global Finance Journal 19 (2009) 203–218 207

In Eqs. (3) and (4), f1 and f2 are functions of standardized innovations. These innovations are defined as
zj,t = εj,t/σj,t and zI,t = εI,t/σI,t. The functions f1 and f2 capture the effect of sign and the size of the lagged
innovations as:



f1 zj;t−1 = jzj;t−1 j−E jzj;t−1 j + δj zj;t−1 ; ð5Þ

f2 zI;t−1 = jzI;t−1 j−E jzI;t−1 j + δI zI;t−1 ð6Þ

The first two terms in Eqs. (5) and (6) capture the size effect and the third term measures the sign effect.
If δ is negative, a negative realisation of zt will increase the volatility by more than a positive realisation of
equal magnitude. Similarly, if the past absolute value of zt is greater than its expected value, the current
volatility will rise. This is called the leverage effect and is documented by Black (1976) and Nelson (1991)
among others.
The asymmetric effect of standardised innovations on volatility may be measured as derivatives from
Eqs. (5) and (6):
 
1 + δi zi N 0
Afi ðzit Þ=Azit = ð7Þ
−1 + δi zi b 0

Relative asymmetry is defined as |−1 + δi| / (1 + δi). This quantity is greater than, equal to, or less than 1 for
negative asymmetry, symmetry and positive asymmetry respectively.
The persistence of volatility may also be quantified by an examination of the half life (HL), which
indicates the time period required for the shocks to reduce to one half of their original size:

lnð0:5Þ
HL = ð8Þ
ln jγi j

The off diagonal elements of the covariance matrix ∑t are defined in a manner similar to that in Darbar
and Deb (2002). The key is to define a time varying conditional correlation which when combined with the
conditional variances given the Eqs. (3) and (4) generate the required conditional covariance. The
conditional correlation is allowed to depend on the lagged standardized innovations and transformed using
a suitable function so that it lies between (−1,1). This is given by the following equation:
 
1
σ j;I;t = ρj;I;t σ j;t σ I;t ; ρj;I;t = 2 −1; t = c0 + c1 zj;t−1 zI;t−1 + c2 t−1 ð9Þ
1 + expð−t Þ

Although the function ξt may be unbounded, the sin function transformation will restrict it to the
desired range for correlation.
For a given pair of return series the 19 parameters to be estimated is conveniently labelled as:
 
Θu βj;0 ; βj;1 ; βj;2 ; βI;0 ; βI;1 ; βI;2 ; α j;0 ; α j;1 ; α j;2 ; γj ; δj ; α I;0 ; α I;1 ; α I;2 ; γ I ; δI ; c0 ; c1 ; c2 ð10Þ

The estimation of these parameters is achieved by numerical maximisation of the joint likelihood
function under the distributional assumption of this model. If the sample size is T then the log likelihood
function to be maximised with respect to the parameter set Θ is:
 
T T   ε j;t
LðΘÞ = −Tlnð0:5π Þ−0:5 ∑ lnj∑t j−0:5 ∑ εj;t εI;t ∑−1 ð11Þ
t=1 t=1
t εI;t

4. Empirical results

The bivariate EGARCH model applied in this analysis allows for both price and volatility spillovers as
well as for time varying correlation structure. The parameters of the model are estimated by the numerical
maximisation of the above discussed joint likelihood function with the algorithm developed by Berndt,
208 R. Bhar, B. Nikolova / Global Finance Journal 19 (2009) 203–218

Hall, Hall and Hausman (1974; BHHH in GAUSS™) without any parameter restrictions imposed. The results
for each of the BRIC countries are presented in Tables 3, 4, 5 and 6.

4.1. Mean and volatility spillover effects

Prices in the Brazilian and Indian markets are predominantly determined by price spillovers from the
regional and world market, rather than their own past values, as indicated by the βj,1 and βj,2 coefficients. It
should be noted that in both cases, the world influence is higher than the regional influence. Neither the
regional (Americas and Asia-Pacific markets respectively) nor the world market prices are influenced by
price spillovers from these local markets. Prices in the Russian market are determined by a combination of
price spillovers from the European market and its own past values to a lesser degree. While the regional
(European) market prices are not influenced by past values from the Russian market, the world market
prices are likely to be affected by Russian market spillovers, which might be related to the significant
impact of the Russian financial markets crisis in 1997–1998 on the returns of a large number of foreign
investors, particularly investors from the U.S. Prices in the Chinese market are predominantly determined

Table 3
Parameter estimates for the bivariate EGARCH model with dynamic correlation Brazil, regional and world

Brazil Regional
Mean equation
βj,0 0.0025 (1.41) βI,0 0.0023 (2.90)
βj,1 −0.0114 (−0.26) βI,1 −0.0031 (−0.19)
βj,2 0.1985 (2.26) βI,2 −0.0405 (−0.95)

Variance equation
αj,0 −0.3405 (−4.43) αI,0 −1.1404 (−3.45)
αj,1 0.2330 (6.12) αI,1 0.1093 (2.44)
αj,2 −0.0287 (−1.00) αI,2 0.1456 (2.43)
γj 0.9416 (73.42) γI 0.8494 (19.46)
δj −0.4976 (−3.17) δI −0.7875 (− 2.28)

Correlation function
c0 0.0251 (1.31)
c1 0.0287 (1.19)
c2 0.9644 (42.28)
Half Life 11.52 4.25
Relative Asymmetry 2.98 8.41

Brazil World
Mean equation
βj,0 0.0022 (1.27) βw,0 0.0013 (2.06)
βj,1 −0.0524 (−1.11) βw,1 −0.0009 (− 0.06)
βj,2 0.3216 (2.52) βw,2 −0.0288 (− 0.64)

Variance equation
αj,0 −0.3264 (−4.24) αw,0 −0.3437 (−4.18)
αj,1 0.2162 (5.32) αw,1 0.0257 (0.78)
αj,2 0.0603 (1.84) αw,2 0.1614 (6.38)
γj 0.9437 (71.91) γw 0.9568 (93.38)
δj −0.3258 (−2.16) δw −0.7865 (− 5.66)

Correlation function
c0 0.0517 (2.00)
c1 0.0457 (2.12)
c2 0.9441 (44.01)
Half Life 11.96 15.70
Relative Asymmetry 2.64 8.40

The numbers in parentheses indicate t-statistics. Half life represents the time it takes for the shocks to reduce its impact by one-half.
Relative asymmetry may be greater than, equal to or less than 1 indicating negative asymmetry, symmetry and positive asymmetry
respectively.
 R. Bhar, B. Nikolova / Global Finance Journal 19 (2009) 203–218 209

Table 4
Parameter estimates for the bivariate EGARCH model with dynamic correlation Russia, regional and world

Russia Regional
Mean equation
βj,0 0.0068 (3.33) βI,0 0.0021 (2.44)
βj,1 0.1123 (7.45) βI,1 0.0115 (0.58)
βj,2 0.1371 (10.16) βI,2 − 0.0546 (− 3.01)

Variance equation
αj,0 −0.3479 (− 16.85) αI,0 − 0.3397 (−16.52)
αj,1 0.2708 (25.05) αI,1 − 0.0400 (−2.14)
αj,2 −0.0183 (− 0.94) αI,2 0.2568 (22.93)
γj 0.9368 (11587.15) γI 0.9570 (25483.92)
δj −0.0011 (− 0.06) δI − 0.3297 (−36.13)
Correlation function
c0 0.9584 (27317.81)
c1 −0.0675 (−3.83)
c2 −0.7953 (−937.11)
Half life 10.62 15.77
Relative asymmetry 1.00 1.98

Russia World
Mean equation
βj,0 0.0061 (3.17) βw,0 0.0011 (3.33)
βj,1 0.1122 (2.75) βw,1 0.0173 (1.81)
βj,2 0.0679 (0.73) βw,2 − 0.0505 (− 2.39)

Variance equation
αj,0 −0.2976 (−3.86) αw,0 − 0.3599 (−4.49)
αj,1 0.2653 (5.45) αw,1 0.0239 (0.66)
αj,2 −0.0308 (−1.28) αw,2 0.1218 (4.98)
γj 0.9457 (71.24) γw 0.9549 (97.03)
δj −0.0126 (−0.15) δw − 0.9339 (−67.22)

Correlation function
c0 0.8140 (23.79)
c1 −0.0236 (−0.39)
c2 −0.3888 (−2.42)
Half Life 12.42 15.02
Relative Asymmetry 1.01 2.89

The numbers in parentheses indicate t-statistics. Half life represents the time it takes for the shocks to reduce its impact by one-half.
Relative asymmetry may be greater than, equal to or less than 1 indicating negative asymmetry, symmetry and positive asymmetry
respectively.

by price spillovers from the world market, rather than its own past values. Neither the regional (Asia-
Pacific) nor the world market prices are influenced by price spillovers from the Chinese market. Brazil, India
and China all report significant price spillover effects from the world market. Given that U.S. stocks
represent approximately fifty percent of the world market index, it can be argued that U.S. plays a
significant role in the price creation process in the Brazilian, Indian and Chinese equity markets.
Parameters αj1 and αI2/αw2 capture the impact of the market's own lagged standardised innovations on
the conditional volatility for the local and the regional/world markets respectively. The behaviour of the
market returns is summarised by the quantity of relative asymmetry detailed in the respective markets'
tables. Both αj1 and αI2/αw2 parameters are highly statistically significant for all BRIC countries, their
respective regions and the world, which indicates that the volatility in these markets depends highly on
their respective lagged standardised innovations.
There is a support for the presence of asymmetric volatility in the local, regional and world markets for
Brazil and India. The relative asymmetry is greater than one for these markets, which indicates that
negative innovation in the previous period will result in a higher conditional volatility in the current period
for all markets. An interesting finding for the Russian equity market, which is quite different to the other
210 R. Bhar, B. Nikolova / Global Finance Journal 19 (2009) 203–218

Table 5
Parameter estimates for the bivariate EGARCH model with dynamic correlation India, regional and world

India Regional
Mean equation
βj,0 0.0043 (3.11) βI,0 0.0012 (51.54)
βj,1 −0.0180 (−0.73) βI,1 −0.0019 (−0.20)
βj,2 0.1288 (4.65) βI,2 0.0481 (33.02)

Variance equation
αj,0 −1.2817 (−53.65) αI,0 −0.5628 (−84.06)
αj,1 0.4118 (19.62) αI,1 0.1500 (6.18)
αj,2 −0.2088 (−7.69) αI,2 0.0909 (3.13)
γj 0.7984 (514.90) γI 0.9212 (5608.09)
δj −0.2071 (−3.10) δI −0.1698 (−9.88)

Correlation function
c0 0.0127 (1.64)
c1 0.0994 (5.86)
c2 0.9280 (8153.81)
Half Life 3.08 8.44
Relative Asymmetry 1.52 1.41

India World
Mean equation
βj,0 0.0045 (3.80) βw,0 0.0016 (4.66)
βj,1 −0.0293 (−1.09) βw,1 0.0059 (0.48)
βj,2 0.3830 (21.66) βw,2 −0.0146 (−0.58)

Variance equation
αj,0 −2.0395 (−4.56) αw,0 −0.3410 (−4.21)
αj,1 0.4160 (7.44) αw,1 −0.0191 (−0.53)
αj,2 0.0574 (1.51) αw,2 0.1331 (6.30)
γj 0.6824 (9.87) γw 0.9578 (92.72)
δj −0.1821 (−2.04) δw −0.9440 (−98.11)

Correlation function
c0 0.8926 (25.55)
c1 0.0386 (0.87)
c2 −0.8635 (−54.78)
Half Life 1.81 16.08
Relative Asymmetry 1.49 2.34

The numbers in parentheses indicate t-statistics. Half life represents the time it takes for the shocks to reduce its impact by one-half.
Relative asymmetry may be greater than, equal to or less than 1 indicating negative asymmetry, symmetry and positive asymmetry
respectively.

BRIC countries, is that the market is likely to respond equally to positive and negative innovations from the
previous period, that is, the level of conditional volatility is likely to be the same regardless of the type
(positive or negative) of the previous period innovation. This is indicated by the relative asymmetry
coefficient of 1 for the local market as stated in Table 4. Also, unlike the other local markets in the group, the
Chinese equity market's relative asymmetry is less than one, which indicates that negative innovation in
the previous period will result in a lower conditional volatility in the current period for the market, and vice
versa.
The persistence in volatility is measured by the parameter γj/I/w. The values of γ are less than one for all
local, regional and world markets, which is a necessary condition for the volatility process to be stable. The
magnitude of the γj/I/w parameters suggests the tendency for the volatility shocks to persist. Using the HL
parameter, the volatility persistence can be expressed in day terms. Based on the HL results for the BRICs,
the Brazilian market takes the longest to reduce the impact from its shocks by half (11.52 weeks) and the
Indian market takes the least time (3.08 weeks), which suggests that India has the lowest level of volatility
persistence out of all BRIC countries.
 R. Bhar, B. Nikolova / Global Finance Journal 19 (2009) 203–218 211

Table 6
Parameter estimates for the bivariate EGARCH model with dynamic correlation China, regional and world

China Regional
Mean equation
βj,0 0.0011 (0.84) βI,0 0.0006 (0.57)
βj,1 −0.0437 (−1.06) βI,1 0.0329 (1.34)
βj,2 0.0392 (0.86) βI,2 0.0303 (0.83)

Variance equation
αj,0 −0.4382 (−1.89) αI,0 −0.4374 (−7.43)
αj,1 0.2609 (4.07) αI,1 −0.0299 (−0.88)
αj,2 0.0278 (0.59) αI,2 0.1196 (2.92)
γj 0.9333 (26.42) γI 0.9392 (115.63)
δj 0.1734 (1.48) δI −0.5220 (−2.09)

Correlation function
c0 0.4432 (3.06)
c1 0.0010 (0.05)
c2 −0.9754 (−34.52)
Half Life 10.04 11.05
Relative Asymmetry 0.70 3.18

China World
Mean equation
βj,0 0.0007 (0.51) βw,0 0.0015 (2.54)
βj,1 −0.0480 (−1.19) βw,1 −0.0101 (−0.71)
βj,2 0.1014 (1.82) βw,2 −0.0300 (−0.77)

Variance equation
αj,0 −0.3012 (− 2.27) αw,0 −0.3832 (−3.49)
αj,1 0.2152 (4.45) αw,1 −0.0374 (−1.04)
αj,2 −0.0457 (−1.63) αw,2 0.1633 (4.32)
γj 0.9542 (47.69) γw 0.9525 (70.23)
δj 0.1376 (1.10) δw −0.7236 (−3.37)

Correlation function
c0 −0.0018 (− 0.02)
c1 0.0792 (1.10)
c2 −0.3027 (−0.71)
Half Life 14.78 14.24
Relative Asymmetry 0.73 3.61

The numbers in parentheses indicate t-statistics. Half life represents the time it takes for the shocks to reduce its impact by one-half.
Relative asymmetry may be greater than, equal to or less than 1 indicating negative asymmetry, symmetry and positive asymmetry
respectively.

Parameters αj2, αI1 and αw1 capture the impact of cross-market standardised innovations for the local,
regional and the world market. Based on the results, the conditional volatility of the Russian market is not
affected by past innovations in the regional and world markets, that is, there is no evidence of volatility
spillover effects from the European region and the world to Russia. The conditional volatility of the Brazilian
market is affected by past innovations in the world market, but not the region. This can be linked to the
impact of the Asian and Russian financial market crisis of 1997–1998, which caused capital flight from
Brazil in 1998–1999 (Baig and Goldfajn, 2000). In case of India, there is an evidence of volatility spillover
effects from both the Asia-Pacific region and the world to the local Indian market, with an indication of a
negative relationship between past period innovations from the region and the conditional volatility of
India. This can be explained by the relatively unique position of India during the Asian financial crisis, as it
was largely unaffected by the impact of the crash of the majority South Asian countries. There is also a very
low negative volatility spillover from the world market to China and no evidence of spillovers from the
Asia-Pacific market to China.
There is no evidence of volatility spillovers from any of the BRIC countries to the world, there are
volatility spillovers on a regional level from Brazil, Russia and India, such that there is a positive relationship
212 R. Bhar, B. Nikolova / Global Finance Journal 19 (2009) 203–218

Table 7
Diagnostics tests (Brazil, regional and world)

Brazil Regional
p-values for Ljung–Box Q(20) statistics
z 0.369 0.416
z2 0.995 0.659
z1·z2 0.540
p-values for Engle and Ng (1993) diagnostic tests
Sign bias test 0.900 0.748
Negative size bias test 0.628 0.323
Positive size bias test 0.970 0.895
Joint test 0.298 0.583

Brazil World

p-values for Ljung–Box Q(20) statistics

z 0.359 0.483
z2 0.989 0.770
z1·z2 0.960
p-values for Engle and Ng (1993) diagnostic tests
Sign bias test 0.887 0.753
Negative size bias test 0.592 0.557
Positive size bias test 0.948 0.890
Joint test 0.419 0.674

z represents the standardised residual for the corresponding equation i.e. either country index return or regional or world index
return. z1·z2 indicate product of the two standardised residuals.

between past period innovations from Brazil and India and the respective regions' conditional volatility,
and a negative relationship between the past period innovations from Russia and the conditional volatility
of the European region. These past innovations for all markets exert an asymmetric influence on the
conditional volatility of the regional market in a sense that a price decline in the local market will result in
greater reaction in the regional market than a corresponding rise in prices. There is no evidence of spillover
effects from China to the Asia-Pacific region and the world.

Table 8
Diagnostics tests (Russia, regional and world)

Russia Regional
p-values for Ljung–Box Q(20) statistics
z 0.757 0.603
z2 0.632 0.900
z1·z2 0.904
p-values for Engle and Ng (1993) diagnostic tests
Sign bias test 0.810 0.800
Negative size bias test 0.884 0.976
Positive size bias test 0.165 0.552
Joint test 0.330 0.261

Russia World
p-values for Ljung–Box Q(20) statistics
z 0.755 0.568
z2 0.548 0.805
z1·z2 0.898
p-values for Engle and Ng (1993) diagnostic tests
Sign bias test 0.892 0.875
Negative size bias test 0.926 0.862
Positive size bias test 0.190 0.824
Joint test 0.211 0.557

z represents the standardised residual for the corresponding equation i.e. either country index return or regional or world index
return. z1·z2 indicate product of the two standardised residuals.
 R. Bhar, B. Nikolova / Global Finance Journal 19 (2009) 203–218 213

Table 9
Diagnostics tests (India, regional and world)

India Regional
p-values for Ljung–Box Q(20) statistics
z 0.036 0.544
z2 0.860 0.046
z1·z2 0.894
p-values for Engle and Ng (1993) diagnostic tests
Sign bias test 0.021 0.012
Negative size bias test 0.058 0.008
Positive size bias test 0.608 0.550
Joint test 0.095 0.036

India World
p-values for Ljung–Box Q(20) statistics
z 0.035 0.545
z2 0.709 0.775
z1·z2 0.001
p-values for Engle and Ng (1993) diagnostic tests
Sign bias test 0.046 0.799
Negative size bias test 0.095 0.751
Positive size bias test 0.537 0.848
Joint test 0.255 0.674

z represents the standardised residual for the corresponding equation i.e. either country index return or regional or world index
return. z1·z2 indicate product of the two standardised residuals.

In summary, if we use the level of spillovers as an indication of the level of integration of the BRIC
countries with their respective regions and the world, India shows the highest level of integration on a
regional and world level amongst the BRIC countries, followed by Brazil and Russia and lastly by China. It
should be noted however, that there is a negative relationship between the conditional volatility of India
with that of the Asia-Pacific region, which, as mentioned earlier, can be attributed to the low level of impact
of the South-Asian crisis on India. Given the relatively closed nature of the financial system in China, there is

Table 10
Diagnostics tests (China, regional and world)

China Regional
p-values for Ljung–Box Q(20) statistics
z 0.499 0.518
z2 0.981 0.064
z1·z2 0.721
p-values for Engle and Ng (1993) diagnostic tests
Sign bias test 0.667 0.004
Negative size bias test 0.368 0.010
Positive size bias test 0.321 0.400
Joint test 0.588 0.026

China World
p-values for Ljung–Box Q(20) statistics
z 0.521 0.543
z2 0.982 0.805
z1·z2 0.949
p-values for Engle and Ng (1993) diagnostic tests
Sign bias test 0.690 0.790
Negative size bias test 0.445 0.707
Positive size bias test 0.251 0.872
Joint test 0.520 0.658

z represents the standardised residual for the corresponding equation i.e. either country index return or regional or world index
return. z1·z2 indicate product of the two standardised residuals.
214 R. Bhar, B. Nikolova / Global Finance Journal 19 (2009) 203–218

no evidence of regional integration for China, only a negative relationship with the conditional volatility of
world market returns, which indicates a presence of diversification opportunities for portfolio investors.
None of the BRIC countries impacts the equity price creation process in their respective regions, none of
them have a significant impact over the conditional volatility of the world market returns and it is only
Russia that has effect over the price creation process of world equity index prices.

4.2. Diagnostics tests

The diagnostics statistics for the BRIC equity market indices, their respective regions and the world are
detailed in Tables 7, 8, 9 and 10. The test statistics include the 20th order serial correlation in the level and
squared standardised innovations as well as the asymmetry test statistics following Engle and Ng (1993).
The Ljung–Box statistics indicate the absence of linear and non-linear dependence in the standardised
innovations for all equity markets. The validity of the Ljung–Box test is confirmed by the Engle and Ng test,
which confirms that there are no sign biases, that is, there is no asymmetry effect, except for the possible
sign bias for the Asia-Pacific region. This bias could potentially affect the outcomes of the Ljung–Box test,
however we are still comfortable with the outcomes of the test, as the joint test bias can be rejected at 90%
confidence level. The Engle and Ng test also indicate a good fit of the bivariate EGARCH model to the
available data set.

4.3. Time-varying conditional correlation

The estimated dynamic conditional correlation between the BRIC countries and their respective regions,
and BRIC countries and the world are displayed in Figs. 1–4. The equity market index returns conditional
correlation for Brazil with the region and the world project very similar trends for the observed time period,
with an evidence of a slightly higher level of world conditional correlation.
There is a significant increase in the level of conditional correlation to mid 1998 on both regional and
world level, which can be associated mainly with four events. First the liberalisation of the Brazilian market
in May 1991 (Bekaert et al., 2003). Second, the introduction of the Real Plan stabilization program
introduced in June 1994 (Dornbusch and Cline, 1997) which linked the nominal value of Brazil's currency to
the dollar, restored price stability moving the economy rapidly from triple to single digit inflation, and
expedited the process of trade liberalization. Third, in 1995, constitutional amendments and legislative
changes (Amendments 5–8 of 1995 and Law 8987 of the same year) opened up to private capital, and by
extension to FDI, public-service concessions and certain other activities previously reserved for federally
owned and controlled entities. In addition, the principle of equal treatment between companies controlled
by Brazilian nationals and Brazilian entities that are subsidiaries of foreign entities or that are controlled by

Fig. 1. Equity market index returns dynamic conditional correlation for Brazil with the region and the world.
 R. Bhar, B. Nikolova / Global Finance Journal 19 (2009) 203–218 215

Fig. 2. Equity market index returns dynamic conditional correlation for Russia with the region and the world.

foreign investors was established. Finally, the Auto Pact between Brazil and Argentina, effective in January
1996, established conditions that would essentially compel foreign manufacturers to locate production in
the two countries if they wished to maintain or increase their share of the Mercosur1 market (Neves, 2002).
Higher level of volatility is evident in the period between late 1998 and 1999, corresponding with the
financial market crisis aftermath and the introduction of a floating exchange rate in Brazil. The level of
conditional correlation for Brazil on regional and world level has been relatively stable ever since, with an
upward sloping trend evident since the beginning of 2006. This is most likely due to the higher level of
visibility and investor attention the country has achieved lately (Wilson and Purushothaman, 2003).
Unlike Brazil, Russia has reported relatively stable level of conditional correlation with the world, while
the regional conditional correlation has been very volatile throughout the whole period of our analysis. The
level of conditional correlation between the Russian equity market index returns and the European regional
equity market index returns was on the increase following the liberalisation of the Russian financial market
in January 1994 (Bekaert et al., 2003). The large level of volatility in the conditional correlation began in 1997,
when the first wave of the Asian crisis contagion hit though the heavy selling in Hong Kong, which caused a
domino effect in other emerging markets (Kaminsky and Schmukler, 1999). This persisted throughout 1998,
reflective of the severe cash-flow problems Russia incurred as investors continued to withdraw from the
government debt market and as international reserves dropped precipitously. The volatility of conditional
correlation was further motivated by the float of the ruble in early September 1998 (Baig and Goldfajn,
2000). The level of volatility reduced considerably in the period between 1999 and 2002, with an evidence of
a sharp positive spike in late 2002, which corresponded with the sovereign risk-1 A common market among
Argentina, Brazil, Paraguay and Uruguay, known as the rating upgrade of Russia by the sovereign risk-rating
agency Standard & Poor's, based on a three years prudent economic policy in Russia.
Similar to the Russian market, the conditional correlation of the Indian equity market index returns with
the world is relatively constant, while the regional conditional correlation with the Asia-Pacific equity
index returns is quite volatile and at times negative. The negative spikes are mainly evident in 1997, late
1998 to late 1999 and mid 2002. The negative regional conditional correlation in 1997 was caused mainly
by the political crisis in India at the time, created after the Congress Party withdrew its support of the
coalition government. The political concerns translated into low trading volumes with noticeable lack of
foreign interest. The level of correlation began to increase as the political situation stabilised, and following
the increase in the aggregate investment ceiling for foreign institutional investors from 5% to 10% of the
paid-up capital of a company, and 24% of the paid-up capital of a listed company (Chandrasekhar and Pal,

1
“Common Market of the South” (“Mercado Comun del Sur”). It was created by the Treaty of Ascunción on March 26, 1991, and
added Chile and Bolivia as associate members in 1996 and 1997.
216 R. Bhar, B. Nikolova / Global Finance Journal 19 (2009) 203–218

Fig. 3. Equity market index returns dynamic conditional correlation for India with the region and the world.

2006). The negative conditional correlation between India and the region in the period 1998–1999 can
most likely be related to the fact that India's economy was relatively unaffected by the South-Asian crisis,
whilst this was not the case for the majority of the other countries comprising the Asian-Pacific index. In
addition, India was in a quite unique position as the Group of Seven (G7) imposed sanctions on the country
following their nuclear testings conducted in 1998, and the subsequent downgrade of India's sovereign
rating from investment grade to speculative (Kumar, 1999). The last downward spike in 2002 was mainly
caused by political turmoil, as a suicide squad attacked the parliament in New Delhi. The conditional
correlation with the region has increased ever since, with large increase evident in late 2003 as the Indian
pension system was liberated, and further increase in foreign direct investment limits in 2004 from 49% to
74% was introduced (Bekaert et al., 2003).
Unlike any of the above discussed BRIC markets, the results for the Chinese equity market index returns
show that there is a very weak, positive, relatively constant conditional correlation with the Asia-Pacific region,
and also weak but very volatile and mainly negative conditional correlation with the rest of the world. This can
be explained by the relatively closed nature of the Chinese financial system. Bekaert et al. (2002) recognise July
1993 as the financial liberalization date for China, however unlike the other emerging market economies
regionally and globally, the financial liberalization in China is characterised by a gradual decline in the state
sector and a steady growth of importance of collective, individual and foreign enterprises. In addition, the low
level of conditional correlation of China with the region and the rest of the world can be explained by the fact
that the Asian financial crisis of 1997–1998 did not exert a negative effect on China. China has in fact absorbed a
considerable amount of foreign direct investment that could have otherwise channelled to neighbouring Asian

Fig. 4. Equity market index returns dynamic conditional correlation for China with the region and the world.
 R. Bhar, B. Nikolova / Global Finance Journal 19 (2009) 203–218 217

economies. This and the Standard & Poor's currency rating downgrade from stable to negative in mid 1998 can
explain the significant negative spike on the conditional correlation diagram with the world that began to form
in early 1999 (Li and Liu, 2001). Another negative spike in the conditional correlation is evident in mid 2002,
which can be related to the U.S. announcement that China is modernizing its military for a possible forcible
reunification with Taiwan (Bekaert et al., 2003). The level of conditional correlation of the Chinese equity index
returns with the world equity index returns has increased in 2003, most likely due to the reform in late 2002
allowing foreign companies to buy shares in listed Chinese companies. The level of conditional correlation has
remained relatively stable ever since.

5. Conclusion

The purpose of this paper is to explore the level of integration of the BRICs with their respective regions
and the world in the post-liberalization period. The bivariate EGARCH model with time varying correlations
relating the equity index returns from the BRIC countries and the regional equity index and also between
the BRIC countries and the world index returns was used for the purpose of this study. The return and
volatility spillovers from these countries to their respective regions and the world, and vice versa, for the
period from January 1995 to October 2006 are used as proxies for the level of integration of these markets
regionally and globally. In addition, the dynamic conditional correlation allows us to observe the impact of
a number of significant events in the BRIC markets on the correlation of the equity index returns of these
markets with their respective regions and the world.
The results indicate that India has the highest level of integration on a regional and world level amongst
the BRIC countries, followed by Brazil and Russia and lastly by China. There is a negative relationship
between the conditional volatility of India with that of the Asia-Pacific region, which, can be attributed to
the low level of impact of the South-Asian crisis on India. Given the relatively closed nature of the financial
system in China, there is no evidence of regional integration for China, only a negative relationship with the
conditional volatility of world market returns, which indicates a presence of diversification opportunities
for portfolio investors. None of the BRIC countries impacts the equity price creation process in their
respective regions, none of them have a significant impact over the conditional volatility of world market
returns and it is only Russia that has effect over the price creation process of world equity index prices,
which can be related to the significant impact of the Russian financial markets crisis in 1997–1998 on the
returns of a large number of foreign investors, particularly investors from the U.S.
The results from the dynamic conditional correlation analysis indicate that market liberalisation of the
BRICs and sovereign risk upgrades from independent credit risk agencies such as Standard & Poor's and/or
Moody's result in increased levels of regional and world conditional correlations. The Asian crisis had a
significant impact over the level of volatility of the conditional correlation of Russia and India with their
respective regions, and Brazil and China with the world. Political turmoils and military actions result in
negative regional and world conditional correlations.
The negative relationship between the conditional correlation of India with that of the Asia-Pacific
region, and of China with the world indicates a presence of diversification opportunities for portfolio
investors. Portfolio investors can continue to receive sound returns from taking positions in the index of
these countries, however in view of the progressive integration of the BRICs regionally and globally, for an
outstanding investment performance, they should consider investing in specific areas of growth within the
economies rather than the country index.

References

Antkiewitcz, A., & Whalley, J. (2006). Recent regional agreements in the large population, rapidly growing non-OECD countries. North
American Journal of Economics and Finance, 17, 343−348.
Aggarwal, R., Inclan, C., & Leal, R. (1999). Volatility in emerging stock markets. Journal of Financial and Quantitative Analysis, 34, 33−35.
Bae, K. H., & Karolyi, G. A. (1994). Good news, bad news and international spillovers of stock return volatility between Japan and the
US. Pacific Basin Finance Journal, 2, 405−438.
Baig, T., & Goldfajn, I. (2000). The Russian default and the contagion to Brazil. World Bank, Working Paper.
Bekaert, G., Harvey, C. R., & Lumsdaine, R. (2002). Dating the integration of world capital markets. Journal of Financial Economics, 65(2),
203−248.
Bekaert, G., Harvey, C. R., & Lundblad, C. T. (2003). Equity market liberalisation in emerging markets. Journal of Financial Research, 3,
275−299.
218 R. Bhar, B. Nikolova / Global Finance Journal 19 (2009) 203–218

Berndt, E. K., Hall, H. B., Hall, R. E., & Hausman, J. A. (1974). Estimation and inference in non-linear structural model. Analysis of
Economic and Social Measurement, 4, 653−666.
Black, F. (1976). Studies of stock market volatility changes. Proceedings of the American Statistical Association Business and Economic
Studies Section (pp. 177−181).
Booth, G. G., Martikainen, T., & Tse, Y. (1997). Price and volatility spillovers in Scandinavian stock markets. Journal of Banking and
Finance, 21, 811−823.
Chandrasekhar, C. P., & Pal, P. (19 May 2006). Financial liberalisation in India: an assessment of its nature and outcomes. International
Development Economics Associates.
Darbar, S. M., & Deb, P. (2002). Cross-market correlations and transmission of information. Journal of Futures Markets, 22, 1059−1082.
De Jong, F., & Roon, D. e. F. A. (2005). Time varying market integration and expected returns in emerging markets. Journal of Financial
Economics, 78, 583−613.
Dornbusch, R., & Cline, W. R. (1997). Brazil's incomplete stabilisation and reform. Brookings Papers on Economic Activity, Vol. 1997 (1).
(pp. 367−404).
Engle, R. F., & Ng, V. K. (1993). Measuring and testing the impact of news on volatility. Journal of Finance, 48, 1749−1778.
Ito, H. (2006). Financial development and financial liberalization in Asia: thresholds, institutions and sequence of liberalization. The
North American Journal of Economic and Finance, 17, 303−327.
Kaminsky, G. L., & Schmukler, S. L. (1999). What triggers market jitters: a chronicle of the Asian crisis. Journal of International Money
and Finance, 18, 537−560.
Kaminsky, G. L., & Schmukler, S. L. (2001). Short or long run integration: do capital controls matter, Brookings Trade Forum 2000
(pp. 128−178). Washington DC: Brookings Institution.
Kaminsky, G. L., & Schmukler, S. L. (in press). On booms and crashes: Financial liberalization and stock market cycles, Washington DC:
World Bank.
Koutmos, G., & Booth, G. G. (1995). Asymmetric volatility transmission in international stock markets. Journal of International Money
and Finance, 14, 747−762.
Kumar, R. (1999). India's economy is relatively unaffected by Asia's financial crisis.: Asian Development Bank News Release, No. 031/99,
http://www.adb.org
Levine, R., & Schmukler, S. L. (2003). Migration, spillovers and trade diversion: The impact of internationalisation on stock market
liquidity. World Bank Policy Research Paper No. 3046.
Li, K. W., & Liu, T. (2001). Financial liberalization and growth in China's economic reform. The World, 5, 673−687.
Nelson, D. B. (1991). Conditional heteroscedasticity in asset returns: a new approach. Econometrica, 59, 347−370.
Neves, M. (2002). Designing policies that stimulate foreign investment in Brazil. The George Washington University, School of Business
and Public Management Issues, XV — Minerva Program — Fall 2002.
Ng, A. (2000). Volatility spillover effects from Japan and the US to the Pacific Basin. Journal of International Money and Finance, 19,
207−333.
Ramchand, L., & Susmel, R. (1998). Volatility and cross-correlation across major stock markets. Journal of Empirical Finance, 5, 397−416.
Wilson, D., & Purushothaman, R. (2003). Dreaming with BRICs: the path to 2050. Goldman Sachs Global Economic Paper No: 99.