Accepted Manuscript

Title: An Investigation of Return and Volatility Linkages

among Equity Markets: A Study of Selected European and
Emerging Countries

Author: Burhan F. Yavas Lidija Dedi

PII: S0275-5319(16)30025-3
DOI: http://dx.doi.org/doi:10.1016/j.ribaf.2016.01.025
Reference: RIBAF 479

To appear in: Research in International Business and Finance

Received date: 24-6-2015

Revised date: 19-12-2015
Accepted date: 27-1-2016

Please cite this article as: Yavas, B.F., Dedi, L.,An Investigation of Return and
Volatility Linkages among Equity Markets: A Study of Selected European and
Emerging Countries , Research in International Business and Finance (2016),
http://dx.doi.org/10.1016/j.ribaf.2016.01.025

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
An Investigation of Return and Volatility Linkages among Equity Markets:
A Study of Selected European and Emerging Countries*
By

Burhan F. Yavas

t
Department of Accounting, Finance & Economics, California State University-Dominguez Hills,

ip
Carson, CA 90747, USA. Tel: 310-243-3501 E-mail: byavas@csudh.edu

cr
Lidija Dedi
Department of Managerial Economics, University of Zagreb, Faculty of Economics & Business,

us
Croatia, Tel: +385-1-238-3109 E-mail: lidija.dedi@efzg.hr

ABSTRACT

an
This paper investigates the linkages among equity returns (based on exchange traded funds,

ETF) and transmission of volatilities in the following countries: Germany, Austria, Poland,
M
Russia and Turkey. Multivariate Autoregressive Moving Averages (MARMA) and the

Generalized Autoregressive Conditional Heteroskedasticity (GARCH) methodologies are

d
te

utilized. The findings include the existence of significant co-movement of returns among

countries in the sample. Also, Turkish and Russian markets were found to be more volatile
p

than Austria, Germany and Poland. However, volatilities in Russia and Turkey do not persist
ce

very long. Finally, there is strong evidence of volatility spillovers. All of the countries in the

sample, with the exception of Turkey, experience volatility spillovers from other markets. The
Ac

presence of spillovers among return series and persistence of volatilities are useful to

investors interested in diversifying their portfolios and to traders/fund managers who are

interested in maximizing returns.

*
A shorter version of this paper was submitted for presentation in the EIBA conference on
December 1-3, 2015
Key words: Volatility transmission, Exchange Traded Funds, MARMA, GARCH

JEL codes: G01, G11, G15, G17, C58

Page 1 of 32
 t
ip
cr
us
ABSTRACT

This paper investigates the linkages among equity returns (based on exchange traded funds,

an
ETFs) and transmission of volatilities in the following countries: Germany, Austria, Poland,
Russia and Turkey. The study covers the period from November 9, 2010 to January 30, 2015.
Multivariate Autoregressive Moving Averages (MARMA) and the Generalized
M
Autoregressive Conditional Heteroskedasticity (GARCH) methodologies are utilized. The
findings include existence of significant co-movement of returns among countries in the
sample. Also, Turkish and Russian markets were found to be more volatile than Austria,
d

Germany and Poland. However, volatilities in Russia and Turkey do not persist very long.
te

There is strong evidence of volatility spillovers. All of the countries in the sample, with the
exception of Turkey, experience volatility spillovers from other markets. Finally, we tested
p

the hypothesis of positive correlation between the expected risk and the expected return, and
ce

found that increases in volatility lead to a rise in future returns in Polish market but not in the
other country markets. The findings including presence of spillovers among return series and
persistence of volatilities are useful to investors interested in diversifying their portfolios and
Ac

to traders/fund managers who are interested in maximizing returns.

*
A shorter version of this paper was submitted for presentation in the EIBA conference on
December 1-3, 2015

Key words: Volatility transmission, Exchange Traded Funds, MARMA, GARCH

JEL codes: G01, G11, G15, G17, C58

1. Introduction
2
Page 2 of 32
This paper investigates the linkages among country equity returns (based on representative
broad market ETFs) and transmission of volatilities. The study covers the period from
November 9, 2010 to January 30, 2015. The sample consists of five countries: Austria,
Germany, Poland, Russia and Turkey. The list includes three of the fastest growing emerging
countries of the last decadePoland, Russia and Turkey. The selection was made mainly on

t
the basis of close economic relationships between Germany and the rest of the sample

ip
countries. The study continues our previous research on the same topic but extends it both in
terms of the countries studied and the time period included.

cr
The selection of Russia and Turkey is in line with the European Neighborhood Policy (ENP)
which has been established to create a ring of friendly, stable and prosperous countries around

us
the European Union (EU) in order to guarantee stability along the outer borders of the EU.
This goal is consistent with the broader aims of the European Security Strategy which would

an
promote close political cooperation and economic integration. The impact of the ENP on the
integration between neighboring countries and the EU in various areas such as trade flows,
factor mobility, human capital and technological activities are currently being studied. The
M
present paper references this literature by including the financial flows and financial
integration dimension to the ENP.
d

The main idea behind the study is simple: If, as recent research indicates, correlation among
te

equity markets has increased (that the markets move together) then, an unexpected event in
one market may affect not only returns, but also volatilities in other markets. The increasing
p

pace of integration of global financial markets has provided motivation for many studies to
investigate the mechanism(s) through which equity market movements are transmitted around
ce

the world. These studies make it clear that while real economic conditions and equity market
performances are linked, the performance of equity markets also vary based on international
Ac

factors. The implication is that market performances are not perfectly correlated across
countries. In fact, in the short run, equity performance may have less to do with expected
fundamentals of individual countries than financial inflows (outflows). For example, rounds
of quantitative easing (QE) by the Federal Reserve (FED) in the US in 2008 and the European
Central bank (ECB) in the European Union (EU) in 2014 have resulted in near zero short term
and very low long term interest rates. The lower rates encouraged capital to look for higher
returns elsewhere. Many emerging countries like Brazil, Indonesia, India, Turkey and Russia
became the recipients of capital flows from the US and Western Europe. The incoming
financial flows, it was pointed out, have been mostly responsible for many emerging equity

3
Page 3 of 32
markets spectacular performance after the 2008 up until the second half of 2013
(Morningstar, 2014). Starting with 2014 however, we witnessed a reversal of financial flows
primarily because the FEDs announcement that the long term bond purchases would be eased
and then stopped sometime in 2015. Since this news was interpreted by the bond market that
the long term rates would increase investors from the US and Europe started to bring their

t
funds back home, causing many emerging markets currencies to lose value. Turkey and

ip
Russia, among other emerging markets experienced sizable currency depreciations in late
2013 and 2014. The reaction of the equity markets was similar in that, many of the indices

cr
declined both in terms of local currencies as well as in dollar terms. By March of 2014, the
equity markets were down (Brazil 11%, China 6%, Russia 16%, Turkey 7%, and Mexico

us
10%) (Yardeni and Quintana, 2014). The global financial cycle which started with FED
policies of low interest rates resulted initially in capital flows into risky assets (high volatility)

an
may have already run its course due to changes in monetary policy by the FED and hence
expected higher interest rates. Recently, the European Central Bank (ECB) started its bond
buying program, driving down interest rates to the negative territory in much of the EU
M
countries. For example, current yields on two and five- year German government bonds are -
0.25 and -0.12 respectively (www.bloomberg.com). These rates are down from 1.2 to 1.3 one
year earlier. The table below indicates the trajectory of the rates in the last 1-year period in the
d

sample countries:
te

Table 1: Harmonized long-term interest rates for convergence assessment purposes (Feb 2014
Feb 2015) (% per annum; secondary market yields of government bonds with maturities of close to 10 years)
p

Country Feb. Mar. Apr. May June July Aug. Sep. Oct. Nov. Dec. Jan. Feb.
14 14 14 14 14 14 14 14 14 14 14 15 15
ce

Germany 1.56 1.51 1.46 1.33 1.26 1.11 0.95 0.92 0.79 0.72 0.59 0.39 0.30
Austria 1.95 1.87 1.77 1.62 1.65 1.47 1.28 1.22 1.10 0.98 0.81 0.54 0.44
Poland 4.47 4.25 4.10 3.80 3.54 3.34 3.36 3.10 2.72 2.54 2.55 2.21 2.20
Ac

Turkey 10.18 10.60 9.80 9.05 8.79 8.75 9.17 9.17 8.96 8.30 n/a 7.75 7.5
Russia n/a 8.17 8.18 8.14 8.12 8.22 8.57 8.73 8.88 8.63 10.22 10.37 n/a
Sources: ECB and European Commission (https://www.ecb.europa.eu/stats/money/long) and OECD Stat
Extracts http://stats.oecd.org/ (March, 2015)

It should be noted that neither Turkey nor Russia are members of the EU. Thus, their interest
rates follow a different trajectory. While Turkish rates came down from 10.18 percent to 7.5
in February of 2015 they are still quite high compared to the EU rates. This is due to a) higher
inflation rates in Turkey and b) the need to finance a large current account deficit of more
than 6 percent of GDP. Russian rates, on the other hand, have increased due to western

4
Page 4 of 32
embargo following the Ukrainian conflict and annexation of Crimea together with
substantially lower oil prices.

In addition to extremely low rates in Europe, we have also witnessed depreciation of the Euro
from a high of $1.40 per Euro to a low of $1.05 in a period of several months. The
depreciation of the Euro (and appreciation of the US dollar) was largely the result of lower

t
yields in Europe and escape from Euro to dollar denominated assets. The ECB is holding

ip
interest rates while the FED is looking to raise them. The start of the 2016 is likely to see the
first important divergence in monetary policy since the 2008 financial crisis with the FED

cr
pushing through a rate increase while the ECB is expected to cut its deposit rate. Since these
moves have been widely expected, US dollar has appreciated (and the Euro depreciated)

us
throughout 2015. It appears that financial flows (credit expansion and contraction), stock
prices and volatility may move in lockstep across the globe. Nevertheless, it is still important

an
to study co-movements between equity markets over a financial cycle since correlations do
not remain constant over the cycle, presenting diversification opportunities.
M
The central intent of this paper is to explore both price and volatility linkages among five
selected financial markets by utilizing broad equity market index based ETFs. The choice of
the data period in this study (November 9, 2010 to January 30, 2015) is especially appropriate
d

since it covers a turbulent times with many fiscal and monetary policy decisions in the
te

European Union (and in the US) in aftermath of 2008 financial crisis with its global effects.
Since the ETFs used in this study are all equity ETFs representing broad equity market
p

indices the paper uses equity returns and ETF returns interchangeably.
ce

Thus, the paper studies both equity returns and their volatilities in Germany, Austria, Poland,
Russia and Turkey with the main objective of contributing to and expanding upon the
literature on linkages among international equity markets. In examining the return co-
Ac

movements, transmission and persistence of volatilities, we seek to understand if there are

opportunities for international investors/traders to earn a better return for a unit of risk.

2. Literature Review

Much of the earlier research in international stock markets concentrated exclusively on co-
movement between returns (Bekaert, 1995; Kim and Langrin 1996; Rezayat and Yavas 2006;
Yavas and Rezayat 2008; Bekaert et al., 2009). These studies found low correlations across
some country equity markets which provide attractive diversification opportunities. Other
examples include Kiymaz (2002) which utilized cointegration analysis to study the emerging

5
Page 5 of 32
Latin American equity markets and developed equity markets (the USA, the UK, and Japan),
and found no long-run co-movements, implying existence of potential diversification benefits.
Kiymaz (2003) found similar results in the European and African/Middle Eastern frontier
markets. Also using the cointegration methodology, Majid et al. (2009) found the stock
markets in Malaysia, Thailand, Indonesia, the Philippines and Singapore were cointegrated

t
both during the pre- and post-1997 financial crisis. The strength of cointegration was found to

ip
increase during the post-1997 financial crisis. Similarly, Gray (2009) found financial
contagion among emerging EU countries and their linkages strengthened after the 2007 crisis.

cr
Majid and Kassim (2009) supported other studies that found greater degree of integration or
increased co-movement among the stock markets during the crisis period, reducing benefits of

us
diversification.

More recent research (e.g. Diebold & Yilmaz 2011; Kumar 2013; Rey 2013) demonstrated

an
that more information is revealed in the volatility of stock prices, which is measured by the
conditional second moments of the price rather than in the price itself. As such, studying the
transmission of stock market movements is a joint study of the spillover of prices as well as
M
the volatility of prices. The interest in stock market volatilities has also increased after the two
recent stock market crashes (dot.com of 2000 and financial crisis of 2007-2008) which
d

witnessed wide swings in asset prices. Studying volatility is critical in a world of free capital
flows since the degree of equity market volatilities can have impact on the stability of the
te

international monetary system (Harrison and Moore 2009). Finally, analyzing price volatility
can give market participants an assessment of the risk associated with various financial
p

products, and thus facilitate their valuation, along with the development of different hedging
ce

techniques (Ng 2000; Rey 2013).

Academic research on equity market volatility and its transmission have not been conclusive.
Ac

For example, focusing on emerging markets, Schleicher (2001) studied equity markets of
Hungary, Poland, and Czech Republic in terms of return and volatility and found return co-
movements to be significant but not their volatilities. On the other hand, Chou et al. (1999)
found that both volatility and return spillovers from United States to Taiwan were significant.
Li (2007) examined the linkages between Shanghai and Shenzhen stock exchanges of China,
Hong Kong and the United States, and found no spillovers (return and volatility) between the
stock exchanges in China and U.S. markets, although unidirectional volatility spillover from
Hong Kong to those in Shanghai and Shenzhen markets was significant. Other studies
examining the spillover of information both in terms of return and volatility include Hamao et

6
Page 6 of 32
al. (1990); Christofi and Pericli (1999); Kumar and Mukhopadyay (2002); and Kim (2005).
They found intra-regional volatility spillovers to be more significant than the inter-regional
spillovers. Studies like Pretorius (2002) and Johnson and Soenen (2003) have focused also on
the factors affecting the spillover of information across national equity markets. Pretorius
(2002) found that bilateral trade, inflation rate differential, industrial production growth

t
differential, interest rate differential, stock market size and volatility, region etc. are some of

ip
the important factors that can affect the spillover of information among stock markets. Tokat
(2013) found volatility transmissions not to be one-way but interdependent, indicating the

cr
presence of cross-market hedging. As such, it may be important to widen the focus to include
both directions in volatility transmission.

us
Since the main purpose of this paper is to explore price and volatility linkages among the
selected country ETFs, the study contributes to prevalent notable works on country ETFs such

an
as those by Khorana et al. (1998); Tse and Martinez (2007); Hughen and Mathew (2007); and
Levy and Lieberman (2013). Specifically, Khorana et al. (1998) and Tse and Martinez (2007)
investigate the returns on international ETFs and conclude that ETF returns closely track their
M
respective country index. Hughen and Mathew (2007) compares ETFs and CEFs (Closed-end
Funds) in terms of price transmission dynamics. Levy and Lieberman (2013) distinguishes
d

between correlation of a country ETF with its country index on the one hand and S&P 500 on
the other. That is, a distinction is made between when the ETF prices are driven by what goes
te

on in the country vs. what happens in the US stock market. They find the S&P 500 has a
dominant effect on the foreign country ETF during the non-synchronized trading hours. Thus,
p

their findings are in contrast to the findings of Tse and Martinez (2007). The present paper
ce

differs from the above cited ones in that while they aimed at understanding price dynamics
between ETFs and its underlying factors such as NAV, exchange rate and country indices, we
Ac

explore price and volatility dynamics amidst country ETFs.

Other studies similar to the present one include Abbas et al. (2013) which investigates the
presence of volatility transmission among regional equity markets of Pakistan, China, India,
and Sri Lanka in addition to the developed countries (USA, UK, Singapore, and Japan). Their
results show that volatility transmission is present between friendly countries of different
regions with economic links. They also find some evidence of transmission of volatility
between countries which are on unfriendly terms. Another paper by Beirne et al. (2010)
examines global (mature) and regional (emerging) spillovers in local emerging stock markets.
The results suggest that spillovers from regional and global markets are present in the vast

7
Page 7 of 32
majority of emerging markets. Another finding is that while spillovers in mean returns
dominate in emerging Asia and Latin America, spillovers in variance (that is, volatility
spillovers) appear to play a key role in emerging markets in Europe. Li and Majerowska
(2008) examine the linkages between the emerging stock markets in Poland and Hungary on
the one hand and the established markets of Germany and the U.S. on the other and find

t
evidence of returns and volatility spillovers from the developed to the emerging markets.

ip
Sakthivel et al. (2012) study correlation and volatility transmission across stock markets of
USA, India, UK, Japan and Australia. Long and short run market integrations are investigated

cr
through Johansen co-integration and vector error correction models. Long run co-integration
is found across international stock indices. In addition, a bidirectional volatility spillover

us
between US and Indian stock markets and a unidirectional volatility spillover from Japan and
United Kingdom to India are found. Diebold and Yilmaz (2011) provide an empirical analysis

an
of return and volatility spillovers among five equity markets: Argentina, Brazil, Chile,
Mexico and the U.S. The results indicate that both return and volatility spillovers vary widely.
Return spillovers, however, tend to evolve gradually, whereas volatility spillovers display
M
clear bursts that often correspond closely to economic events.

The final group of papers cited above comes closest to the present paper in coverage and
d

approach in that they focus on investigating stock market returns as well as volatilities.
However, the following differences should be noted. First, they utilize stock market indices as
te

opposed to ETFs used in the present study. Second, the present paper uses daily data as
opposed to the weekly data. While weekly data can have advantages in terms of limiting
p

noise daily data provide larger number of observations. We also study multi-directional
ce

flows whereas most of the literature focuses on unidirectional flows from the developed to
developing markets. Finally, the methodology is somewhat different (vector autoregressive
Ac

model (VAR) as opposed to MARMA) even though the present paper also uses GARCH
methodology like most of the other studies. Also, this paper addresses the questions of
volatility persistence in addition to volatility transmission. In sum, the present paper is
different because it uses country ETFs to study return and volatility transmissions and
volatility persistence and it applies MARMA and GARCH methodologies on selected
European and emerging countries.

The sample includes Germany, Austria, Poland, Turkey and Russiathe last three were
among the fastest growing emerging countries until about 2014 when, in response FEDs
ending the QE and conflict in Ukraine, capital flew out of Turkey and Russia among other

8
Page 8 of 32
emerging markets. One of the main reasons we have assembled this sample is to evaluate
return and volatility spillovers among the countries that have very strong trade relationships.
Germany is Austrias most important economic and trade partner accounting for 31 per cent
of both exports and imports in the services sector alone. For Austria, tourism is the most
important factor in the exchange of services with Germany. With some 51 million overnight

t
stays, Germans are the largest group of foreign visitors, accounting for 38.3 per cent of the

ip
total number of tourists. Economic ties between the two countries are extremely close even
compared with other EU countries. Many German companies have branches, sales offices or

cr
production facilities in Austria, employing a total workforce of nearly 100,000 (Federal
Foreign Office, Berlin, 2015).

us
For more than two decades, Germany has been Polands most important trading partner. More
than a quarter of all Polish exports go to Germany. Poland is also of considerable importance

an
for German foreign trade, ranking tenth among Germanys trading partners for years now. In
terms of both the number of investors and the total amount invested, German companies rank
first among foreign direct investors in Poland. (Federal Foreign Office, Berlin, 2015).
M
Germany is Turkeys most important trading partner. After reaching a new record level in
2013, bilateral trade declined in the first ten months of 2014, by 4.2 percent compared with
d

the previous year, to EUR 27.1 billion. Turkish exports to Germany grew by 7.8 per cent, to
te

EUR 11.1 billion, while Turkish imports from Germany fell by 11.5 per cent, to EUR 16
billion (compared with EUR 33.8 billion for the whole of 2013). Germany is also the biggest
p

foreign investor in Turkey with investments worth more than USD 12 billion since 1980.
(Federal Foreign Office, Berlin, 2015).
ce

German exports to Russia fell by 16 percent in the first half of 2014, to EUR 15.6 billion. In
the same period, Russian exports to Germany grew by 2 per cent, to EUR 20.3 billion.
Ac

Russias annexation of Crimea and its actions in eastern Ukraine overshadow German-
Russian relations as well as Russias relations with the European Union and other partners
(Federal Foreign Office, Berlin, 2015). Table 2 below presents the ranking of Germanys
trading partners in Foreign Trade for 2014.

Table 2: Ranking of Germanys trading partners in Foreign Trade (year 2014)

Rank Country Export (in 1000 EUR) Rank Import (in 1000 EUR)
6 Austria 56,233, 900 12 36,432,271
8 Poland 47,543,819 8 39,761,192
16 Turkey 19,317,543 18 13,286,404

9
Page 9 of 32
13 Russia 29,318,406 10 38,402,723
Source: Statistisches Bundesamt (Federal Statistical Office), Wiesbaden, February 20, 2015.

The inclusion in the sample of Russia and Turkey has also implications for the European
Neighborhood Policy (ENP) which seeks to establish a unified EU policy toward the
countries on the eastern borders of the EU. The idea behind the ENP is to enhance cooperative

t
relationships with neighboring countries to EU so as to promote peaceful, stable, prosperous

ip
and interdependent developments in the region. To that end, policies such as trade promotion,
technological diffusion, human capital mobility and institutional development in the

cr
bordering countries are encouraged (Wesselink and Boschma, 2012). The present paper
studies financial flows among the countries and thus contributes to a greater integration of the

us
bordering countries with the EU consistent with the policy goals of the ENP. In particular, we
study interdependencies among the sample country equity markets in terms both of co-

an
movements of returns and the transmission of market volatilities. As such, we try to
understand linkages among the three EU countries (Austria, Germany and Poland) on the one
hand and two neighboring countries (Russia and Turkey) on the other. As the implementation
M
of the ENP programs progresses, one would expect increased integration among the EU and
the border countries not only in the form of greater trade and technical ties but also of
d

financial flows. This is true even though Russia is not formally part of the ENP countries, the
EU-Russia strategic partnership has similar goals to the ENP. Unfortunately, relations
te

between Russia and the EU deteriorated due to 2008 economic crisis hitting Russia hard and
leading Russia to introduce new protectionist measures, negating the progress made in earlier
p

years. Also, political problems such as the war between Russia and Georgia, The gas dispute
ce

in 2009 led to a further deterioration of political cooperation between the EU and Russia. The
icing on the cake was the Russian involvement in the Ukrainian civil war resulting in the
Ac

annexation of Crimea.

Turkeys membership in the EU is an on-again off-again complicated story but the recent
opening of new chapters in negotiations with the EU mostly reflects an attempt to stem the
growing tide of Syrian immigration to EU. It is also clear that the EU is putting pressure on
Turkey so that the government there would reinvigorate reform efforts to continue
democratization. Turkey is part of Union for the Mediterranean initiative which incorporates
the southern ENP countries, the accession countries Turkey and Croatia, and the Balkan
countries.

10
Page 10 of 32
The next section describes the data followed by a description of the methodologies employed.
We then present the findings and end the paper with the conclusions and suggestions for
future research.

3. Data

This study utilizes Exchange Traded Funds (ETF) instead of market indices mostly used in

t
ip
the literature. ETFs are arguably one of the most versatile of financial instruments that invest
mostly in corporate and sovereign liabilities with the intension of replicating the returns of a

cr
market index.

ETFs are a type of exchange-traded investment product that must register with the SEC under

us
the 1940 Act as either an open-end investment company (generally known as funds) or a
unit investment trust. Since the first domestically offered ETF was created in the 1990s, ETFs

an
have become increasingly popular as investment vehicles for both retail and institutional
investors. (SEC, Investor Bulletin: Exchange-Traded Funds (ETFs), www.investor.gov).
Exchange Traded Funds (ETFs) combine features of mutual funds and stocks. With low
M
overall costs, access to hundreds of companies and trading flexibility, ETFs can be a powerful
addition to investors portfolio (i-shares.com). Similar to mutual funds, ETFs are diversified
d
mixes of stocks or bonds that are managed by experienced professionals. The difference is
that ETFs typically offer a few extra perks, including lower fees, tax benefits and the ability to
te

buy and sell throughout the day as long as the market is open. (i-shares.com). With over 1000
exchange-traded funds (ETFs) globally and more than $1 trillion in assets under management
p

(as of December 31, 2014), iShares helps clients around the world build the core of their
ce

portfolios, meet specific investment goals and implement market views.(i-shares.com).

This paper utilizes iShares MSCI Capped/Core Equity ETFs (all Equity ETFs subject to this
Ac

research are issued by iShares). iShares is the largest ETF provider in the world. It is part of
BlackRock, the world's largest asset manager. The iShares family of ETFs is built around
virtually every leading index provider, including Barclays Capital, Cohen & Steers, Dow
Jones, FTSE, JPMorgan, MSCI, NASDAQ, NYSE, Russell and Standard & Poor's
(http://etfdb.com/issuer/ishares/). Selected ETFs seek to track the investment results of a
particular index. For example, ETF-USA (iShares Core S&P 500 ETF - IVV) seeks to track
the investment results of an index composed of large-capitalization U.S. equities S&P 500,
while the selected Italian ETF (iShares MSCI Italy Capped ETF EWI) seeks to track the
investment results of an index composed of Italian equities. The MSCI Index was created by

11
Page 11 of 32
Morgan Stanley Capital International. Each MSCI Index measures a different aspect of global
stock market performance. The MSCI indices are now managed by MSCI Barra.
(useconomy.about.com).

The data period in this paper is from November 9, 2010 to January 30, 2015, a sample of
1063 days on the following ETFs: 1. The iShares MSCI Austria Capped ETF (EWO) seeks to

t
track the investment results of a broad-based index composed of Austrian equities: a)

ip
Exposure to a broad range of companies in Austria; b) Targeted access to Austrian stocks. 2.
The iShares MSCI Germany ETF (EWG) tracks the performance of publicly traded securities

cr
in the MSCI Germany market index. Some of the holdings include Siemens, Bayer, and SAP:
a) Exposure to large and mid-sized companies in Germany; b) Targeted access to 85% of the

us
German stock market; c) Use to express a single country; pair with EWGS for comprehensive
Germany coverage. The German ETF invests 95% of assets in the stocks of the correlating

an
index. It also includes ADRs based on the securities in the index. The stocks in the market
ETF are traded primarily on the Frankfurt Stock Exchange. 3. The iShares MSCI Poland
Capped ETF (EPOL) seeks to track the investment results of a broad-based index composed
M
of Polish equities: a) Exposure to a broad range of companies in Poland; b) Targeted access to
the Polish market. 4. The iShares MSCI Russia Capped ETF (ERUS) seeks to track the
d

investment results of an index composed of Russian equities: a) Exposure to a broad range of

companies in Russia; b) Targeted access to 85% of the Russian stock market. 5. The iShares
te

MSCI Turkey ETF (TUR) seeks to track the investment results of a broad-based index
composed of Turkish equities: a) Exposure to a broad range of companies in Turkey; b)
p

Targeted access to Turkish stocks.

ce

A summary table on the ETFs is provided below:

Table 3: Exchange Traded Funds
Ac

Austria Germany Poland Russia Turkey

(EWO) (EWG) (EPOL) (ERUS) (TUR)
Net Asset as of $54,690,015 $5,073,483,288 $175,070,937 $249,887,995 $548,716,256
06-Feb-2015
Inception date Mar 12, 1996 Mar 12, 1996 May 25, 2010 Nov 9, 2010 Mar 26, 2008
Exchange NYSE Arca NYSE Arca NYSE Arca NYSE Arca NYSE Arca
Asset Class Equities Equities Equities Equities Equities
Benchmark MSCI Austria MSCI MSCI Poland MSCI Russia MSCI Turkey
Index IMI 25/50 Germany Index IMI 25/50 25 / 50 Index Investable
Market Index
Index Ticker M1AT5IM NDDUGR M1PL5IM MSEURU$N MIMUTURN
Shares 3,600,000 177,000,000 7,400,000 20,050,000 10,750,000
Outstanding
12
Page 12 of 32
 Outstanding
Source: iShares by BlackRock, http://www.ishares.com/us/products (06-Feb-2015)

As mentioned previously all equity ETFs utilized in this research are issued by iShares, and
data were collected (daily prices of the sample ETFs) from the iShares official web pages.

By concentrating the analysis on ETF data, we can mitigate if not entirely avoid some

t
substantial problems that arise in traditional academic research such as exchange rates

ip
volatility, divergences in the national tax systems, diversities in stock exchange trading times

cr
and bank holidays, restrictions on cross-border trading and investments, transaction costs.
Designed to mimic the movements of MSCI indices, ETFs provide an easy pool of

us
international diversification products for an investor.

4. Methodology

an
4.1. Multivariate Auto Regressive Moving Average (MARMA)

To study co-movements of daily returns, we utilized the Multivariate Auto Regressive

M
Moving Average (MARMA). MARMA models combine some of the characteristics of the
univariate autoregressive moving average models and, at the same time, some of the
characteristics of regression analysis. A MARMA model deals with an output time series Yt,
d

which is presumed to be influenced by a vector of input time series Xt, and other inputs
te

(factors) collectively grouped and called noise, et. The input series Xt exerts its influence on
the output series via a transfer function, which distributes the impact of Xt over several future
p

time periods. The objective of the transfer function modeling is to determine a parsimonious
model relating Yt, to Xt, and et. (Makridakis et al. 1998). The transfer function model, in
ce

general, may be represented as:

j (L )Yt = w(L )X t + q (L ) t (1)

Ac

where j(L), w(L), q(L) are polynomials of different orders in L. Polynomial j(L) = (1 -j1 L1-j2
L2 -. . . -jp Lp) represents autoregressive part of order p, L denotes lag, L1 Yt represents
Yt 1 , and polynomial q(L) =( 1 -q1 L1-. . . -qp Lq ) represents moving average part of order q.

4.2. Generalized Autoregressive Conditional Heteroskedasticity Model (GARCH)

The literature on measuring volatility provides several models with varying degrees of
difficulty of implementation. From the standpoint of unconditional volatility, the simplest
estimator is standard deviation which gives uniform weight to all observations. However, the

13
Page 13 of 32
standard deviation has two drawbacks; it is symmetric and it is constant. To measure the
dynamic relationship of the volatility of a process, among the models can be used are
exponential smoothing or autoregressive conditional heteroskedastic (ARCH) and generalized
autoregressive conditional heteroskedastic (GARCH) models. ARCH models were introduced
by Engle (1982) and generalized as GARCH by Bollerslev (1986). GARCH models, have

t
become widespread tools for dealing with time series heteroskedasticity and are more widely

ip
used to model the conditional volatility of financial series. Practically, GARCH models are
fitted when errors of AR or ARMA or in general a regression model have variances which are

cr
not independent or the variance of the current error term is related to the value of the previous
periods' error terms as well as past variances. The coefficients of the past periods squared

us
error terms is an indicative of the strength of the shocks in the short term while the coefficient
of the past variances (GARCH effect) measures the contribution of these shocks to long run

an
persistence (Grosvenor and Greenidge, 2010).

The specification of a typical GARCH model is given by:

t2 = + (L ) t21 + (L ) t2
M
(2)

( )
and t2 | t 1 ~ N 0, t21 is the innovation in the asset returns and t-1 ={y t-1 , t-1, y t-2 , t-
d

2), where y t-i, represents the return at time t-i and i is the error resulting from a regression
te

or an ARMA model fitted to returns. This is similar to ARMA models where (L) of order p
is the autoregressive term and polynomial (L) of order q is the moving average term.
p

GARCH processes have commonly tails heavier than the normal distribution. This property
ce

makes the GARCH process attractive because the distribution of asset returns frequently
display tails heavier than the normal distribution. In most empirical applications with finitely
sampled data, the simple ARCH (1) or GARCH (1, 1) is found to provide a fair description of
Ac

the data. ARCH (1) model is as follows:

t2 = + t21 (3)

And a sufficient condition for the conditional variance to be positive is that the parameters of
the model satisfy the following constraints: >0 and >0.

GARCH (1, 1) model is:

t2 = + t21 + t21 >0, 0< 1, 0< 1, + 1 (4)

14
Page 14 of 32
 is the coefficient that measures the extent to which a volatility shock today feeds through
the next period volatility, while + is usually considered to be a measure of persistence of
volatility shock and it measures the rate at which this effect dies over time.

Note that when yt, the rate of return on an asset is not function of a regressors (there is no
regression component in the model), then yt is identical to et indicating a pure GARCH

t
process. In this study we use GARCH (1,1) to analyze the persistence of conditional

ip
volatility of the returns as well as transmission of volatility of returns. Daily ETF returns are
calculated by 100* logarithmic difference of daily closing ETF values. rt = 100 d log(pt ) .

cr
Multivariate GARCH (MGARCH) can identify and use common movements between

us
different asset volatilities. For a survey on Multivariate Garch models see (Bauwens et al.
2006; Bera and Higgind 1993; Matteson and Ruppert 2010; Moon and Yu 2010; Gilenko and
Fedorova 2014; and Kundu and Sarkar 2016).

5. Findings
an
M
To fit a multivariate model to a set of time series data, one has first to evaluate cross
correlations as well as the autocorrelations and partial correlations of data. The results of our
investigation indicated that there are significant cross correlations of lag zero for most of the
d

returns and cross correlations of lag one for some of the returns. Partial correlation and
te

autocorrelation analysis indicated that among the returns only Austria and Russia
demonstrated significant partial correlations of lag one. Consequently, Vector Auto
p

Regressive (VAR) methodology could not be employed since in the VAR methodology the
regressors are the lagged values. Thus, a decision was made in favor of the MARMA model
ce

whereby for each return equation, regressors are the other four ETF returns, its own one-
period lagged returns as well as one-period lagged returns of other ETF returns.
Ac

E.g. yt =f (yt-1, Xit, Xit-1) for i=1.4

We have also utilized the standard Augmented Dickley-Fuller (ADF) test to check whether
the return series are stationary or not (see for details in Brooks, 2014). The results confirmed
the stationarity of the data.

Table 4 presents the co-movements of ETF returns. Since the domestic and foreign stock
market trading is contemporaneous in this paper (all ETFs trade in the US) and also to
understand the flow of information processed, we estimate a MARMA model.

15
Page 15 of 32
Table 4: Co-movements* of daily ETF Returns (Austria, Germany, Poland, Russia and Turkey)
rt ( Austria ) = 0 .673 rt (Germany ) + 0.156 rt ( Poland ) + 0 .054 rt 1(Germany ) + 0 .056 rt ( Russia ) + et

rt (Germany ) = 0.05 + 0.669rt ( Austria ) + 0.207rt ( Poland ) + 0.06rt ( Russia ) 0.036rt 1( Poland ) + 0.034r(Turkey ) + et

rt ( Poland ) = 0.419 rt (Germany ) + 0.148rt (Turkey ) + 0.306 rt ( Austria ) + 0.105rt ( Russia ) + et

t
rt ( Russia ) = 0.253rt (Germany ) + 0.175rt (Turkey ) + 0.216rt (Poland ) + 0.233rt ( Austria ) + 0.06rt 1( Russia ) + et

ip
rt (Turkey ) = 0 .351rt ( Poland ) + 0 .206 rt ( Russia ) + 0 .164 rt (Germany ) + et

cr
*All coefficients are significant at the five percent level (P<0.05); r and e denote returns and error terms

us
respectively.

Because of the existence of significant cross correlation of lag one (t-1) among some of the

an
returns, the one-period lagged returns of some ETF are also presented in Table 4.

First, it is important to note that German market returns (ETF representing Germany- EWG)
affect returns in all of the other sample countries; Austria, Poland, Russia and Turkey.
M
Second, most of the coefficients are positive indicating that the markets move together, i.e.,
they are positively correlated. The only exception is negative one-period lagged Polish returns
d

appearing in the equation for Germany. However, Polish returns of the same time period have
te

a positive coefficient in the same equation indicating co-movement of the two markets. Next,
we note that Turkish returns and Austrian returns do not affect one another; Turkish returns
p

do not appear in the Austrian return equation and vice versa. Investors interested in equity
markets that do not move together (not correlated) would be advised to invest in Turkey and
ce

Austria for portfolio diversification.

Interpreting results and starting with Austria, we observe strong positive relationship with
Ac

German returns - both current as well as one-period lagged returns from Germany are
significant. Also, both Russian and Polish returns appear to be positively affecting Austrian
returns. German returns are positively correlated with returns from the other four countries.
The same holds true for Polish and Russian returns, the latter is also positively correlated with
its own one-period lagged returns.

Turning next to the implications of these findings, we note that many financial advisors
advise their clients to consider investment opportunities in foreign (as opposed to domestic)
markets for diversification purposes. While these recommendations may be specific to the
current market conditions, advances in communication technology, abolition of capital &

16
Page 16 of 32
exchange controls and deregulation in the financial services in recent years seem to have
increased access to foreign markets. Even though some controversy exists among investment
professionals regarding the benefits and costs of investing in foreign equity markets, there is
agreement that global portfolio diversification recommendations are based on the existence of
a low correlation among national stock markets.

t
The findings summarized above in table 4 indicate that even though there are significant

ip
return interdependencies among the global stock markets there are still excellent opportunities
for portfolio diversification. For example, Austrian and Turkish investors can safely diversify

cr
by investing in each others markets. German investors, on the other hand, appear to have
limited opportunities since the German market is positively correlated with all of the other

us
markets included in our samples. Similarly, Russian and Polish investors may not benefit
from diversification into equity markets of the other countries in the sample for they are all

an
positively correlated.

In short, the results of the MARMA analysis confirm findings of many of the earlier studies:
M
that returns on equities move together across the sample countries included in this study. For
example, Rezayat and Yavas, (2006) found significant return co-movements and reduced
diversification benefits. Schleicher (2001) in the study of Hungary, Poland, and Czech
d

Republic also found return co-movements to be significant. Harrison and Moore (2009) on the
te

other hand, found higher returns in emerging markets that had low correlations with
developed markets. Similar results were reported in Yavas and Rezayat (2016).
p

5.1. Volatility
ce

In order to study the volatility and its persistency or transmission using a GARCH-type model
it is a common practice to calculate descriptive statistics on the error terms of ARMA or
Ac

regression and to test whether these errors independent. It is especially important to check the
skewness and kurtosis of the distributions and to test whether the distribution is normal. Table
5 present summary statistics for the ETF return series.

Table 5: Descriptive statistic

AUSTRIA GERMANY POLAND RUSSIA TURKEY
Mean -0.036803 0.013202 -0.040338 -0.076872 -0.035212
Median 0.003542 0.028991 -0.012957 -0.050737 0.093583
Maximum 7.408512 7.377567 7.087931 8.099038 8.154844
Minimum -7.379272 -7.571491 -10.73469 -13.02545 -11.46181

17
Page 17 of 32
Std. Dev. 1.528727 1.584214 1.713058 1.890959 1.874629
Skewness -0.223741 -0.253543 -0.499913 -0.764527 -0.537942
Kurtosis 6.127523 6.087547 6.874766 8.912636 6.401039
Jarque-Bera 441.6875 433.2111 708.5957 1650.404 563.0632
Probability 0.000000 0.000000 0.000000 0.000000 0.000000
Sum -39.08526 14.02056 -42.83867 -81.63763 -37.39517

t
Sum Sq. Dev. 2479.565 2662.828 3113.576 3793.844 3728.604

ip
Observations 1062 1062 1062 1062 1062

cr
Looking at Table 5, we note that during the period under study, mean ETF returns from
Austria, Poland, Russia and Turkey are all negative. The only exception is Germany with

us
positive mean returns (Table 5).

Distributional properties of the return series generally appear to be non-normal. Financial

an
markets tend to have fat tails. This is so because they are subject to more extreme outcomes
in the form of bubbles and crashes. The prime example is nearly 44% fall in 2002, and
M
40.37% in 2008 in the Germans DAX Index. In addition, markets tend to rise at a slower
pace than they fall. After crash in 2008, it took more than five years to the Germans DAX
Index to reach the level before crisis (e.g. 8067.32 on December 24, 2007; 8042.85 on March
d

11, 2013). This gives rise to what is known as negative skewness Consistent with the
te

expectation, all five of the countries in sample have negative skewness. Tay and Zhu (2000),
amongst others, have documented positive and/or negative skewness in Asian equity returns.
p

The kurtosis or degree of excess, in all markets exceeds three (3), indicating a leptokurtic
distribution. Excess kurtosis in equity returns has been well documented by a number of other
ce

studies including Bekaert and Harvey (1997). Accordingly, the Jarque- Bera test statistic (and
corresponding p-value) strongly rejects the null hypothesis of normal distribution for all
Ac

returns in the sample at =0.05 (Table 5).

Looking at the standard deviations, the highest volatility during the period of our study is
exhibited by Russia (1.89) followed by Turkey (1.87). The Austrian market (and the German
market) has the lowest volatility (1.52 and 1.58). These results are hardly surprising since
volatility is higher in emerging markets such as Russia and Turkey, emerging markets that are
the most affected by the 2007-08 financial crisis and the subsequent actions of the Federal
Reserve Board of the US (FED) and the European Central Bank (ECB) as well as EU-
imposed austerity policies in countries like Greece, Spain, Italy and Portugal. In addition,
other studies in the literature found similar results: those volatilities tend to be higher in

18
Page 18 of 32
emerging, less developed markets than developed ones (Abbas et al. 2013; Frankel and
Roubini, 2001).

5.2. Volatility Persistence

Volatility persistence deals with the nature of volatility and whether the current periods
volatility is affected by past periods volatility, and if so, to what extent. Volatility persistence

t
ip
is one of the statistical properties of stock returns and exchange rates. Volatility occurs in
clusters in that major swings in asset prices (and exchange rates) do not suddenly stop after

cr
major news breaks, and instead they tend to persist. If volatility is persistent, it implies that
todays volatility arising out of new information today is likely to influence tomorrows

us
volatility, and future volatilities. Malhotra et al (2013) conducted a study on bonus stock
announcements in the Indian stock market and noticed that volatility persisted for some time,
and eventually, faded away. Other recent studies such as Hai et al (2103) find volatility

an
persistence in GDP growth series in Hong Kong, Taiwan and Singapore. Engle and Patton
(2001) indicated that the other noteworthy statistical property of volatility is its tendency to
M
revert to the mean. Mean reversion in volatility is generally interpreted as meaning that there
is a normal level of volatility to which volatility will eventually return.

To analyze persistence in volatility, GARCH (1, 1) specification is commonly used. The

d

literature referred to above indicates that the sum of the ARCH and GARCH effects is a
te

measure of volatility persistence. If that sum is closer to one, it means that effects of shocks
fade away very slowly. The lower the values of GARCH & ARCH effects, the faster the
p

effects fade away.

ce

Table 6: Volatility Persistence (GARCH (1, 1))

Coefficient Austria Germany Poland Russia Turkey
AR(1) 0.0731 0.0629
Ac

(0.028) (0.027)
Constant 0.0127 0.0187 0.0347 0.040 0.2152
() (0.015) (0.011) (0.003) (0.027) (0.000)
ARCH(-1) 0.0371 0.0507 0.0586 0.0590 0.1211
() (0.000) (0.000) (0.000) (0.000) (0.000)
GARCH(-1) 0.9559 0.9404 0.9295 0.9261 0.8221
() (0.000) (0.000) (0.000) (0.000) (0.000)
+<1 0.9931 0.9911 0.9880 0.9855 0.9432
AIC 3.444726 3.487435 3.715784 3.827465 3.993567

19
Page 19 of 32
SIC 3.463452 3.501469 3.729818 3.846190 4.007601
ARCH-LM test statistic 0.800509 0.370397 0.218996 0.660560 0.266805
(Obs*R-squared)
Prob. Chi-Square(1) 0.3709 0.5428 0.6398 0.4164 0.6055

The parameters shown in table 6 lie within the normal range. For daily data, Arch reaction

t
ip
parameter, usually ranges between 0.05 (for a market that is relatively stable) and about 0.1
(for a market that is jumpy). In other words, measures the extent to which shocks to todays

cr
returns feed through into volatility of next period, and + measures the rate in which this
effect dies over time. As shown in the tables, Turkey (0.12) exhibits values of more than

us
0.1, indicating very jumpy market. Next are Russia (0.059), Poland (0.058) and Germany
(0.05). Austria has the lowest Arch coefficient (.03) indicating stable short term volatility.
Long term (cumulative) effect of past shocks on returns is measured by the GARCH

an
parameter , which usually ranges between 0.85 and 0.98. In this study, ranges from a low
value of 0.82 in Turkey to 0.9559 in Austria. Finally looking at both Arch and Garch effects,
M
Austria and Germany have + values close to 1.0 indicating that the effects of the volatility
shocks fade away slowly. Turkey stands out in the sample because it has a highly volatile,
jumpy market. However, as far as the long term persistence is concerned, volatilities in the
d

Turkish market fade quickly.

te

Figures 1-3 below depict variances of ETF returns in the sample countries. Note that while
Austria and Germany (and to a certain extend Poland) are relatively stable, Russia and Turkey
p

exhibit high levels of volatility. Volatility in all of the markets in 2011 may have been due the
ce

ever-evolving Greek crisis. Russian military intervention in Ukraine and the subsequent
EU/US embargo may have been responsible for the volatility in the Russian market where
both the ruble and the equity markets tumbled at the end of 2013 and early 2014. Turkish
Ac

market volatility in 2014 was probably fueled by the political uncertainty along with exodus
of financial assets from Turkey.

Post estimation tests are also presented in table 6. ARCH LM test confirms that there are no
ARCH effects left in the standardized residuals indicating that the variance equation is
correctly specified. Furthermore, all Q-statistics for standardized and squared residuals are not
significant which implies that both the mean and the variance equations are correctly specified
for all sample countries. The post estimation tests confirm that the model is robust.

20
Page 20 of 32
Figure 1: Variances of ETF returns: All
24

20

16

12

t
8

ip
4

cr
0
IV I II III IV I II III IV I II III IV I II III IV I
2011 2012 2013 2014

us
Aus tria Germany P oland
Rus s ia Turk ey

Figure 2: Variances of ETF returns: Germany, Austria and Poland

20

16

12 an
M
8
d
4
te

0
IV I II III IV I II III IV I II III IV I II III IV I
2011 2012 2013 2014
p

Germany A us tria Poland

Figure 3: Variances of ETF returns: Germany, Russia and Turkey

ce

24

20
Ac

16

12

0
IV I II III IV I II III IV I II III IV I II III IV I
2011 2012 2013 2014

Germany Russ ia Turk ey

21
Page 21 of 32
To confirm the validity of the GARCH (1, 1) model as the appropriate model for volatility
persistence analysis, we utilized GARCH-in-mean model. Engle, Lilien and Robins (1987)
suggested an ARCH in mean model (ARCH-M), where the conditional variance of asset
returns enters into the conditional mean equation. Bollerslev, Engle and Wooldridge (1988)
extend the ARCH-M model to the GARCH-M (GARCH-in-mean) as an extension of GARCH

t
model. GARCH-M model is given by specification (Brooks, 2014)

ip
yt = + t 1 + ut , ut ~ N (0, t2 ) (5)

cr
t2 = 0 + 1ut21 + t21 (6)

If is positive and statistically significant, then increased risk, given by an increase in the

us
conditional variance, leads to a rise in the mean return. Thus can be interpreted as a risk
premium (Brooks, 2014). The results of the maximum likelihood (ML) estimates of the

an
GARCH-M parameters are presented in table 7.

Table 7. Estimation results of GARCH-M (1, 1)

M
Austria Germany Poland Russia Turkey
@SQRT(GARCH) 0.1260 0.1089 0.1840 0.0811 -0.1294
() (0.252) (0.306) (0.065) (0.427) (0.3439)
d

Constant 0.0135 0.0194 0.0249 0.0496 0.2097

() (0.014) (0.011) (0.091) (0.051) (0.008)
te

ARCH(-1) 0.0387 0.0521 0.0536 0.0683 0.1179

() (0.000) (0.000) (0.000) (0.000) (0.000)
p

GARCH(-1) 0.9539 0.9386 0.9377 0.9204 0.8246

ce

() (0.000) (0.000) (0.000) (0.000) (0.000)

+<1 0.9925 0.9908 0.9914 0.9887 0.9424
AIC 3.451376 3.489444 3.655595 3.834304 3.945921
Ac

SIC 3.474765 3.512833 3.683662 3.862371 3.973988

The ML estimates of the coefficient () of the conditional standard deviation (i) in the mean
equation are found positive for all markets except the Turkish market. However, the ML
estimates of are significant at 10% level only for Polish market indicating that an increase in
volatility there leads to a rise in future returns. ARCH LM test for Polish market confirms that
there are no ARCH effects left in the standardized residuals (Obs*R-squared 0.439210, Prob.
Chi-Square (1) 0.5075) and all Q-statistic for standardized and squared residuals are not
significant; indicating that both the mean and the variance equations are correctly specified. In
22
Page 22 of 32
the other markets the GARCH-M failed to confirm the hypothesis that an increase in volatility
leads to a rise in future returns (coefficients () are not significant). When compared with
other studies our results are not substantially different. Engle, Lilien, and Robins (1987) found
that an increase in risk (variance) tends to result in higher expected returns in share prices.
Fabozzi, et al., (2004) also found that higher risks result in higher returns for the Shenzhen

t
and the Shanghai exchange. Panait and Slavescu (2012) showed that GARCH-M failed to

ip
confirm the hypothesis that an increase in volatility leads to a rise in future returns in
Romanian market. Tah (2013) presented a similar conclusion for the Nairobi Stock Exchange

cr
(of Kenya). On the other hand, using the same methodology (GARCH-M) Tah found negative
and significant relationship between expected returns and conditional variance in Zambian

us
market.

To assess the robustness of the GARCH (1, 1) and GARCH-M (1, 1) models for Polish

an
market we compared AIC and SIC criterion. As shown in Table 7. GARCH-in-mean (1, 1)
AIC and SIC criterion are slightly lower for Polish market than for GARCH (1, 1) (Table 6.),
indicating that for the Polish market the GARCH-in-mean (1, 1) is slightly better model than
M
the GARCH (1, 1).

5.3. Volatility Transmission:

d

The transmission of shocks from the returns of one market to another was well-documented
te

by Ewing (2002). Co-movements across volatilities (co-volatility) due to common

information that simultaneously affects expectation in these markets and information
p

spillovers caused by cross-market hedging are some of the reasons for volatility
ce

transmissions. In addition to endogenous events or variables, exogenous variables,

deterministic events (macroeconomic announcement), all may have an influence on the
volatility process.
Ac

To detect transmission of volatility between stock markets, we use the Augmented GARCH
model as developed by Duan (1997):

t2 = + t21 + t21 + X t (7)

Where Xt is to be the residual squared of ARMA model and is the term that measures the
magnitude of volatility transmission across the markets. Zouch et al. (2011) and Edwards
(1998) used this method and detected the presence of capital transmission effect from Mexico
to Chile for Mexican bonds during 1994 crisis. Augmented GARCH model is considered to

23
Page 23 of 32
be very versatile univariate volatility model. Its superiority over standard GARCH model in
the presence of ARCH effects has been very well documented in Specht (2000) and Specht
and Gohout (1998).
This study employs one-period lagged squared returns for Xt, except for returns of Austria and
Russia. This is so because, as mentioned earlier, we could not fit AR or ARMA to any of the

t
returns except for those two country ETFs. Therefore, we report transmission of volatility of

ip
past period returns as opposed to transmission of conditional volatilities, except for Austria
and Russia.

cr
Table 8: Volatility transmission

t2( Austria ) = 0 .011 + 0.960 t21( Austria ) + 0.032 rt 21(Germany )

us
t2(Germany) = 0.013 + 0.056et21(Germany) + 0.944 t21(Germany) 0.008rt21( Poland) + 0.002rt21(Turkey)

an
t2( Poland) = 0.010 + 0.018et21( Poland) + 0.946 t21( Poland) + 0.031rt21(Germany) + 0.011rt21(Turkey)
t2( Russia) = 0.311+ 0.125et21( Russia) + 0.775 t21( Russia) + 0.065rt21(Germany) 0.015rt21(Turkey) 0.041rt21( Poland)
M
t2(Turkey ) = 0 .215 + 0.121et21(Turkey ) + 0.822 t21(Turkey )
All coefficients are significant at the five percent level (p<0.05)
d

Among the sample of countries, the only market not experiencing volatility spillovers from
te

other markets is Turkey. Past period volatility of the German market affects volatilities in the
Austrian, Russian and Polish markets. Since the coefficient of the German volatility spillover
p

term is positive, it indicates that an increase in the German market volatility increases
volatilities in the Austrian, Russian and Polish markets. In addition to volatility spillovers
ce

from Germany and its own past period volatility the Polish market volatility is also affected
by the volatility of the Turkish market. Finally, the Russian market has volatility spillovers
Ac

from Germany, Poland and Turkey, but not from Austria. However, volatility coefficients
associated with Poland and Turkey are negative, indicating that a decreasing volatilities in
Poland and Turkey in effect increases volatility in the Russian market.

Many of the findings above are corroborated by some of the studies summarized in the
literature review section. For example, Christofi and Pericli (1999) and Kumar &
Mukhopadyay (2002) found both inter and intra-regional volatility spillovers to be significant.
Also, similar to Tokat (2013) this study finds volatility transmissions to be multi-directional

In summary, it may be concluded that during the period covering this study (2010-2014),
there is some evidence of cross-transmission of volatility between the stock markets. These
24
Page 24 of 32
results are corroborated by Beirne et al. (2010) that finds spillovers in variance (volatility)
appear to play a key role in emerging Europe. Schleicher (2001), on the other hand, found
return co-movements significant but not their volatilities in Hungary, Poland, and Czech
Republic.

6. Conclusions

t
ip
This paper studied the transmission of equity returns and volatility among five equity markets
using daily ETF data from November 9, 2010 to January 30, 2015. A multivariate

cr
autoregressive moving average (MARMA) model along with a generalized autoregressive
conditional heteroskedasticity (GARCH) model are used to identify the source and magnitude

us
of return and volatility spillovers.

The findings of this study indicate that co-movements between daily ETF returns representing

an
the countries under study are significant. This finding implies that there are diminishing
opportunities for investors to diversify their portfolios. Nevertheless, one could still find
significant diversification possibilities for investors. For example, Austrian and Turkish
M
investors can safely diversify by investing in each others markets. In general, significant co-
movements of markets indicates that investors should not rely only on current domestic news
d

to guide their investment decisions, but also take into consideration international news for
there are substantial spillovers. In addition, investors might be able to ride financial cycles by
te

following closely monetary policies of the FED and ECB and resulting credit expansion or
p

contraction since recent empirical research indicates (and as corroborated in this study) equity
prices are sensitive to the FED and ECB policy moves which result in changes in capital
ce

flows, credit expansion/contraction and interest rates.

Another finding of the study indicates that among the sample countries, emerging markets
Ac

(Turkey, Russia and Poland) are more volatile than the developed markets (Germany and
Austria). Once again, these results are similar to findings of other studies (Abbas et al., 2013;
Beirne et al., 2010; Frankel and Roubini, 2001; Yavas and Rezayat, 2016).

The results of the GARCH-in-mean (GARCH-M (1,1)) model indicated that an increase in
volatility leads to a rise in future returns only in the context of the Polish market. That is, the
GARCH-M model failed to confirm the hypothesis that an increase in volatility leads to a rise
in future returns in the markets other than the Polish Market.

25
Page 25 of 32
There is strong evidence of volatility spillovers. The results show that with the exception of
Turkey, other equity markets experience volatility spillovers from other markets in the
sample. Given that volatilities can proxy for risk, there are implications for both individual
and institutional investors in terms of further examining pricing securities, hedging and other
trading strategies as well as framing regulatory policies. Clearly, volatility transmission and

t
the time-varying nature of volatility have implications for investors and portfolio managers

ip
who assess such information and rebalance their portfolios continually to achieve efficient
portfolio diversification.

cr
As hedging becomes another area of interest for investors, its importance is growing as a
vehicle almost as important as asset allocation. Since volatilities indicate risks, volatility

us
transmissions open up a new area for financial products that are tailor-made to allow investors
to benefit from (or hedge against) sudden changes in market volatility. Some of these new

an
financial products such as new forms of ETFs are synthetic. This means that the returns are
generated from a swap with counterparty, and that the money is not used to buy the
underlying securities. While this exposes the investors to counterparty risk, when yields are
M
low and uncertainty is high, there is strong demand for products that are created to take more
risk but limit potential losses. Equity market volatility can be used by investors in such a
d

strategy
te

New financial products based on VIX (a measure of volatility) are being introduced as a
hedge against a risk of market meltdown. As an example, a new ETF invests in VIX futures
p

contracts but shifts from longterm to short term contracts (and vice versa) when the VIX
moving average reaches a certain threshold.
ce

There are implications also with respect to European Neighborhood Policies. If the
implementation of various programs under the ENP continues to bear fruit one result may be
Ac

greater integration (and hence higher correlation) of the equity markets of the EU and the
border countries. As such, investors desiring to diversify their portfolios may have to look
elsewhere to find markets that do not move together. Similarly, ENP may give rise to
increased volatility spillovers among the highly connected markets

26
Page 26 of 32
REFERENCES
Abbas, Q, Khan, S and Shah, Z. 2013. Volatility transmission in regional Asian stock
markets. Emerging Markets Review, 16, issue C: 66-77.
Bauwens, L., Laurent, S., and Rombouts, J. 2006. Multivariate GARCH Models: a Survey.
Journal of Applied Econometrics, 21:79-109.
Beirne, J., Caporale, G., Schulze-Ghattas, M and Spagnolo, N. 2010. Global and regional
spillovers in emerging stock markets: A multivariate GARCH-in-mean analysis. Emerging

t
Markets Review 11, 3: 250-260.

ip
Bekaert, G. 1995. Market Integration and Investment Barriers in Emerging Equity Markets.
World Bank Economic Review, 9: 75-107.

cr
Bekaert, G. and Harvey, C. R. 1997. Emerging Equity Market Volatility. Journal of Financial
Economics, 43:29-77.

us
Bekaert G., Hodrick, Robert J., Zhang, Xiaoyan 2009. International Stock Return
Comovements. Journal of Finance, 64(6): 2591-2626, December.
Bera, N. and Higgins, M. L. 1993. ARCH Models: Properties, Estimation and Testing.
Journal of Economics Survey, 7(4):305-366.

an
Bollerslev, T. 1986. Generalized Autoregressive Conditional Heteroskedasticity. Journal of
Econometrics, 31(3): 307-327.
Bollerslev, T., R.F. Engle and J.M. Wooldridge 1988. A Capital Asset Pricing Model with
M
Time-Varying Covariances. Journal of Political Economy, 96, pp. 116-131
Brooks, C. 2014. Introductory Econometrics for Finance, 3rd Edition, Cambridge University
Press.
d

Christofi, A., and Pericli, A 1999. Correlation in Price Changes and Volatility of Major Latin
American Stock Markets. Journal of Multinational Financial Management, 9:7993.
te

Chou, R. Y., Lin, J., and Wu, C. 1999. Modeling the Taiwan Stock Market and International
Linkages. Pacific Economic Review, 4: 305-320.
p

Diebold, F.X. and Yilmaz, K. 2011. Equity Market Spillovers in the Americas. In R. Alfaro
(ed.) Financial Stability, Monetary Policy, and Central Banking. Santiago: Bank of Chile
ce

Central Banking Series, Volume 15, 199-214, 2011.

Duan, J. C. 1997. Augmented GARCH (p, q) Process and its Diffusion Limit. Journal of
Econometrics 79: 97-127.
Ac

Edwards, S. 1998. Interest Rate Volatility, Capital Controls, and Contagion. NBER Working
Paper Series.
Engle, R F. 1982. Autoregressive Conditional Heteroskedasticity with Estimates of the
Variance of United Kingdom Inflation. Econometrica 50:9871007.
Engle, R.F., David M. Lilien, Russel P. Robins 1987. Estimating Time Varying Risk Premia
in the Term Structure: The ARCH-M Model. Econometrica, Vol. 55, No. 2 (March, 1987),
391-407
Engle, R. F. and Patton, A.J 2001. What Good is Volatility Model? Quantitative Finance
1:237245.
Ewing, B. T. 2002. The transmission of shocks among S&P indexes. Applied Financial
Economics, 12(4): 285-290.

27
Page 27 of 32
Fabozzi, F.J., Tunaru, R., Wu, T. 2004. Modeling Volatility for the Chinese Equity Markets.
Analysis of Economics and Finance 5, 79-92.
Frankel, J.A and Roubini, N 2001. The Role of Industrial Country Policies in Emerging
Market Crises. IMF Working Papers 09/104.
Gilenko, E. and Fedorova, E. 2014. Internal and external spillover effect for the BRIC
countries: Multivariate GARCH-in-mean approach. Research in International Business and
Finance, Vol. 31, p32-45.

t
Gray, D. 2009. Financial contagion among members of the EU-8: a cointegration and Granger

ip
causality approach. International Journal of Emerging Markets, Vol. 4 No. 4, pp. 299314.
Grosvenor, T., and Greenidge, K. 2010. Stock Market Volatility from Developed Markets to

cr
Regional Markets. Research and Economic Analysis Department Central Bank of Barbados.
Hai, V., Thanhi, A. K., and Zhang, T. Z. 2013. Measuring Asymmetry and Persistence in
Conditional Volatility in Real Output: Evidence from Three East Asian Tigers Using a

us
Multivariate GARCH Approach. Applied Economics, 45(20): 2909-2914
Hamao, Y., Masulis, R. W., and Ng, V. 1990. Correlations in Price Changes and Volatility
across International Stock Markets. The Review of Financial Studies, 3 (2): 281-307.

an
Harrison, B., and Moore, W. 2009. Spillover Effects from London and Frankfurt to Central
and Eastern European Stock Markets. Applied Financial Economics, 19(18): 1509-1521.
Hughen J.C. and Mathew, P.G 2007. The Efficiency of International Information Flow:
M
Evidence from the ETF and CEF Prices. http://ssrn.com/abstract=1029844 or
http://dx.doi.org/10.2139/ssrn.1029844
Investor Bulletin: Exchange-Traded Funds (ETFs), SEC, August, 2012 www.investor.gov;
http://www.sec.gov/investor/alerts/etfs.pdf
d

Johnson R. A., and Soenen, L. 2003. Economic Integration and Stock Market Co movement
in the Americas. Journal of Multinational Financial Management, 13: 85-100.
te

Kim S. J., 2005. Information Leadership in the Advanced Asia-Pacific Stock Markets: Return,
Volatility and Volume Information Spillovers from the US and Japan. Journal of the Japanese
p

and International Economies, 19(3):338-365.

Kim, S. W., and Langrin, R. B 1996. Stock Price Movements Spillovers under Foreign
ce

Exchange Liberalization: The Case of Jamaica, Trinidad and Tobago, and the United States.
University of the West Indies, Mona, Jamaica. Presented at the XXVIIth Annual Conference
on Monetary Studies.
Ac

Kiymaz, H. 2002. An analysis of linkages among equity markets: study of selected Latin
American and Developed countries. Journal of Economics and Business Studies, Vol. 1 No. 1,
pp. 1633.
Kiymaz, H. 2003. Are there diversification benefits from investing in frontier equity markets?
Journal of Accounting and Finance Research, Vol. 11 No. 2, pp. 6475.
Khorana, A., Nellis, E., and Trester, J.J. 1998. The Emergence of Country Index Funds: An
Examination of World Equity Benchmark Shares (WEBS). The Journal of Portfolio
Management, 24(4): 78-84
Kumar, K. K., and Mukhopadyay, C. 2002. A Case of Us and India. NSE Research Paper
Kumar, M. 2013. Returns and volatility spillover between stock prices and exchange rates:
Empirical evidence from IBSA countries. International Journal of Emerging Markets, Vol. 8
(2), pp.108 128

28
Page 28 of 32
Kundu, S. and Sarkar, N. 2016. Return and volatility interdependences in up and down
markets across developed and emerging countries. Research in International Business and
Finance Vol. 36, pp.297-311.
Levy, A. and Lieberman, F. 2013. Overreaction of Country ETFs to US Market Returns:
Intraday vs. Daily Horizons and the Role of Synchronized Trading. Journal of Banking and
Finance: 37(5):1412-1421
Li, H. 2007. International Linkages of the Chinese Stock Exchanges: A Multivariate GARCH

t
Analysis. Applied Financial Economics, 17: 285-295.

ip
Li, H., Majerowska, E. 2008. Testing stock market linkages for Poland and Hungary: A
multivariate GARCH approach. Research in International Business and Finance, Vol. 22,
Issue 3, pp. 247-266.

cr
Majid, M.S.A. and Kassim, S.H. 2009. Impact of the 2007 US financial crisis on the emerging
equity Stock returns, volatility spillover, and other financial issues in emerging markets.

us
International Journal of Emerging Markets, Vol. 4 No. 4, pp. 299314
Majid, M.S.A., Meera, A.K.M., Omar, M.A. and Aziz, H.A. 2009. Dynamic linkages among
ASEAN-5 emerging stock markets. International Journal of Emerging Markets, Vol. 4 No. 2,
pp. 16084

an
Malhotra, M. S., Thenmozhi, M., and Kumar, A. G. 2013. Evidence on Changes in Time
Varying Volatility around Bonus and Rights Issue Announcements. International Journal of
Emerging Markets, 8(2), 2-2.
M
Makridakis, S. G., Wheelwright, S., and Hyndman, R. 1998. Forecasting Methods and
Applications, Wiley, 3rd Ed.
Matteson, D. and Ruppert, D. 2010. GARCH Models of Dynamic Volatility and Correlation.
d

www.Orie, Cornell.edu
Moon, G.H, and Yu, W.C. 2010. Volatility Spillovers between the US and the China Stock
te

Markets: Structural Break Test with Symmetric and Asymmetric GARCH Approaches.
Global Economic Review, 39(2):129-149.
Ng, A. 2000. Volatility Spillover Effects from Japan and the US to the Pacific-Basin. Journal
p

of International Money and Finance, 19: 207-233.

ce

Panait, I., Slavescu, E.O. 2012. Using Garch-in-Mean Model to Investigate Volatility and
Persistence at Different Frequencies for Bucharest Stock Exchange during 1997-2012.
Theoretical and Applied Economics, Vol. XIX, No. 5(570), p55-76.
Ac

Pretorius, E. 2002. Economic Determinants of Emerging Stock, Market Interdependence.

Emerging Markets Review, 3: 84-105.
Rey, H. 2013. Dilemma not Trilemma: The Global Financial Cycle and Monetary Policy
Independence. www.kc.frb.org/publicat/sympos/2013/2013Rey.pdf
Rezayat, F. and Yavas, B.F. 2006. International Portfolio Diversification: A Study of
Linkages among the U.S., European and Japanese Equity Markets. Journal of Multinational
Financial Management, 16: 440-458.
Sakthivel, P., Bodkhe, N., and Kamaiah, B 2012. Correlation and Volatility Transmission
across International Stock Markets: A Bivariate GARCH Analysis. International Journal of
Economics and Finance, 4, 3: 253-262.
Schleicher, M. 2001. The Co-movements of Stock Markets in Hungary, Poland and the Czech
Republic. International Journal of Finance and Economics, 6: 27-39.

29
Page 29 of 32
Specht, K., and Gohout, W. 1998. Volatilities Analyze mit-dem Augmented GARCH-Model.
Allgemeines Statistisches Archiv 82: 339351.
Specht, K. 2000. Modelle zur Schtzung der Volatilitt, Wiesbaden: Gabler Verlag.
Tah, K.A. 2013. Relationship between Volatility and Expected Returns in Two Emerging
Markets. Business and Economics Journal, Vol. 2013: BEJ-84 http://astonjournals.com/bej
Tay, N., and Zhu, Z. 2000. Correlations in Returns and Volatilities in Pacific-Rim Stock
Markets. Open Economies Review 11:27-47

t
ip
Tokat, A.H. 2013. Understanding Volatility Transmission Mechanism among the CDS
markets: Europe and North America vs. Brazil and Turkey. Economia Aplicada, 17(1): 1-111
Tse Y., and Martinez, V. 2007. Price Discovery and informational efficiency of International

cr
Ishares funds. Global Finance Journal, 18: 1-15.
Wesselink, E., Boschma, R. 2012. Overview of the European Neighbourhood Policy: Its

us
History, Structure and Implemented Policy Measures, WP1/04 Search Working paper,
http://www.ub.edu/searchproject/wp-content/uploads/2012/02/WP-1.4.pdf
Yardeni, E. and Quintana, M 2014. Performance 2014: Global Stock Markets. Yardeni
Research Inc. NY. www.yardeni.com

an
Yavas, B. F. and Rezayat, F. 2008. Integration among Global Equity Markets: Portfolio
Diversification using Exchange-Traded Funds. Investment Management & Financial
Innovations. 5(3):30-43.
M
Yavas, B. F. and Rezayat, F. 2016. Country ETF Returns and Volatility spillovers in Emerging
Stock Markets, Europe and the United States. International Journal of Emerging Markets,
Forthcoming, 11(3).
Zouch, M.A., Abbes, M.B., and Boujelbene, Y. 2011. Subprime crisis and Volatility
d

Spillover. Int. J. Monetary Economics and Finance, 4(1):1-20

te

Internet sources
p

Bloomberg, http://www.bloomberg.com/markets/rates-bonds/government-bonds/germany/
(March 18, 2015)
ce

ECB and European Commission https://www.ecb.europa.eu/stats/money/long (March 18,

2015).
Federal Foreign Office, Berlin, http://www.auswaertiges-amt.de (March 18, 2015).
Ac

http://performance.morningstar.com/funds/etf/total-returns.action?t=eem
OECD Stat Extracts http://stats.oecd.org/
Statistisches Bundesamt (Federal Statistical Office), Wiesbaden, February 20, 2015.
www.i-shares.com (February 6, 2015)
http://etfdb.com/issuer/ishares
www.useconomy.about.com
http://valueandopportunity.com/2014/06/10/the-german-dax-at-10-000-looking-back/ (May
31, 2015)

30
Page 30 of 32
Graphical Abstract
14

12

10

t
4

ip
2

0
IV I II III IV I II III IV I II III IV I II III IV I

cr
2011 2012 2013 2014

Germany Austria
20

us
16

12

an
4

0
IV I II III IV I II III IV I II III IV I II III IV I
M
2011 2012 2013 2014

Germany Poland

24

20
d

16
te

12

4
p

0
IV I II III IV I II III IV I II III IV I II III IV I
ce

2011 2012 2013 2014

Germany Russia
20

16
Ac

12

0
IV I II III IV I II III IV I II III IV I II III IV I
2011 2012 2013 2014

Germany Turkey

31
Page 31 of 32
20

16

12

t
ip
0
IV I II III IV I II III IV I II III IV I II III IV I
2011 2012 2013 2014

cr
Germany Aus tria Poland

24

us
20

16

an
12

4
M
0
IV I II III IV I II III IV I II III IV I II III IV I
2011 2012 2013 2014
d
Germany Russ ia Turk ey
te

Highlights
p

Co-movements between daily ETF returns representing the countries under study are
ce

significant

Emerging markets (Turkey, Russia and Poland) are more volatile than the developed
markets (Germany and Austria). However, volatilities in the Turkish market fade quickly
Ac

while volatility shocks fade away slowly in Austria and Germany.

Austria and Germany have + values close to 1.0 indicating that the effects of the
volatility shocks fade away slowly. Turkey stands out in the sample because it has a highly
volatile, jumpy market. However, as far as the long term persistence is concerned,
volatilities in the Turkish market fade quickly

There is strong evidence of volatility spillovers.

The results show that with the exception of Turkey, other equity markets experience
volatility spillovers from other markets in the sample

32
Page 32 of 32