Frontier stock market integration and the global financial crisis

North American Journal of Economics and Finance 29 (2014) 84–103

Contents lists available at ScienceDirect

North American Journal of Economics and Finance

Frontier stock market integration and the global financial crisis Mei-Ping Chen a,∗, Pei-Fen Chen b, Chien-Chiang Lee c a Department of Accounting Information, National Taichung University of Science and Technology, 129 Sanmin Rd., Sec 3, Taichung 40401, Taiwan b Department of International Business Studies, National Chi-Nan University, Nantou, Taiwan c Department of Finance, National Sun Yat-sen University, Kaohsiung, Taiwan

a r t i c l e

i n f o

Article history: Received 5 August 2013 Received in revised form 7 May 2014 Accepted 8 May 2014 JEL classification: C32 G15 Keywords: Frontier market Stock market integration Financial crisis Granger causality

a b s t r a c t This paper examines the stock market integration between frontier and leading markets, focusing on the periods of pre and post global financial crisis. Using time-series analysis, the results mostly support leading markets can Granger-cause frontier markets. Frontier markets in different regions have distinct relationships with leading markets. Population growth, industry value, interest rate, tax rate, and tariff of the frontier markets significantly influence the integration between both markets. Energy, gross national income, stock traded value, and high-technology exports of leading markets saliently influence the integration. Finally, the global financial crisis impacts the relationship between the frontier and leading markets and changes the determinants of stock market integration. © 2014 Elsevier Inc. All rights reserved.

1. Introduction Countries in the early stages of economic development generally demonstrate long-run growth potential, and frontier markets today are often compared to emerging markets in the late 1990s. Frontier markets, viewed as hugely untapped economic potential and diversification benefits, can relate long-term investment opportunities in recent years. Hence, the topic researching on frontier

∗ Corresponding author. Tel.: +886 4 22196020; fax: +886 4 22196131. E-mail addresses: [email protected] (M.-P. Chen), [email protected] (P.-F. Chen), [email protected] (C.-C. Lee). http://dx.doi.org/10.1016/j.najef.2014.05.004 1062-9408/© 2014 Elsevier Inc. All rights reserved.

M.-P. Chen et al. / North American Journal of Economics and Finance 29 (2014) 84–103

85

equity markets has become extremely imperative because such markets not only facilitate particular countries’ economic development, but also help investors to diversify their portfolios. Compared with the large amount of literature focusing on developed and emerging markets’ integrations (Driessen & Laeven, 2007; Rua & Nunes, 2009, among others), only few empirical researches related to frontier stock markets have been carried out. For example, Berger, Pukthuanthong, and Yang (2011) study the potential diversification benefits of frontier markets. Most previous researches of market integration focus only on financial markets, whereas less attention on the issues of integration determinants of frontier equity markets. According to the theory of stock market co-movement, there is a close link between market prices and country development. For example, Panton, Lessig, and Joy (1976) found that the U.S. is relatively well developed country and has open international capital flows. This paper attempts to discuss the topics of the theory of stock market co-movement, such as exploring the co-movements between frontier markets and regional leading markets, the co-movement change between the subperiods before and after the global financial crisis, and any other factors of influencing the co-movement between frontier and leading equity markets. In this study, a leading country (e.g., the U.S.) is defined as that one country which has the highest GDP within its own sub-region.1 As such, we address the following three issues: firstly, what is the dynamic interrelationship between frontier and leading stock markets (i.e., regional leading market and the U.S.)? Secondly, does the dynamic relationship change between both subperiods before and after global financial crisis? Lastly, if integration relationships do exist, which factors can influence the market integration? We explore all potential dynamic interrelations between frontier and leading equity markets through four hypotheses: (1) Leading market leading hypothesis: a leading market exerts a unidirectional leading effect over market returns. The hypothesis may be sustained by an open economy and large capital in the leading market. (2) Frontier market leading hypothesis: the stock returns of frontier market highlight a unidirectional leading effect over the leading market. The hypothesis may be sustained by the natural resource of the frontier market. (3) Feedback hypothesis: there exists a bidirectional causality relationship between stock returns of frontier and leading markets, representing that frontier and leading markets need to be considered simultaneously. In other words, the change of frontier market returns precedes leading markets, and vice versa. (4) Neutrality hypothesis: no causal relation exists between the stock returns of frontier and leading markets, meaning that frontier markets’ stock returns are not correlated with market returns. Our paper contributes to existing relatively studies with the following three ways. First, the extant empirical research on frontier markets uses the U.S. to explore the stock market integrations (Samarakoon, 2011). Frontier markets might be more correlated with their regional leading markets because of similar culture, natural resources, and geographical factors (e.g., the Middle East is predominantly driven by crude oil production). Since each individual frontier economy is endowed differently, the market integration may be country-specific. Thus, we expand the few related studies by evidencing from the co-movement of frontier markets and regional leading markets, rather than summing up a general conclusion about the existence of stock market integration with the U.S. We not only find that the leading stock markets significantly Granger cause frontier stock markets, but also discuss some possible opposite situations. The lead-lag relationship between the price movements of frontier and leading markets illustrates how one market reflects new information to the other, and how well both markets are linked together. We can draw a conclusion that there is a lead-lag relationship between specific countries, which benefits investors’ portfolio in diversifying their investment by including frontier markets. Second, in contrast to previous researches that simply examine the degree of financial integration of frontier markets, we seek to delve further into what lies behind the response (if any) in the integration between frontier and regional leading markets, such as the factors of economic policy or natural resources. Most previous studies of analyzing financial frontier market integration have focused on the country level (Adjasi, Osei, & Fiawoyife, 2011; Boubaker, Nguyen, & Taouni, 2009) or a specific

1

Source: International Monetary Fund.

86

M.-P. Chen et al. / North American Journal of Economics and Finance 29 (2014) 84–103

region (Moss, Ramachandran, & Standley, 2007), whereas only few studies investigate the integration reasons. Additionally, Carrieri, Errunza, and Sarkissian (2004) mention that financial market development and financial liberalization policies play an important role in integrating G-7 markets, but they do not further examine the feasible determinants of integration between frontier and regional leading markets. Wang, Yang, and Yang (2013) find that country variables are important determinants of future price movements. We thus attempt to apply the Logit regression model on larger frontier market samples to shed light on the possible factors of influencing the dominance of different hypotheses. Third, Marshall, Nguyen, and Visaltanachoti (2011) find that frontier market diversification benefits disappeared during the period of global financial crisis. Zhang, Li, and Yu (2013) find that the global financial crisis has changed permanently the correlations between BRICS and developed U.S. and Europe stock markets. Herber and Rowader (2011) present that the correlation trend between the MSCI frontier index and the world index tends to decrease after the financial crisis, because the peak of the trend is placed right around the crisis years. Since inconclusive results have been found for the financial crisis effect on frontier markets and other markets, we further consider the subsample period effect by investigating the directional changes between frontier and regional leading markets in the pre and post financial crisis subperiods. This can provide more general pictures of frontier markets under different market conditions in the global regions. Finally, this paper extends previous studies (e.g., Gupta & Guidi, 2012; Lucey & Muckley, 2011; Syriopoulos, 2011) with respect to emerging and developed market integrations by employing the vector error–correction model (hereafter VECM). To elaborate on the issue of what factors determine the stock market integration of frontier markets, our time-series data mainly relies on 29 frontier markets and 14 leading markets from five different regions over the period 2000–2011. Our results reveal that the U.S. saliently Granger causes the frontier stock markets, while the condition is comparatively less significant after the crisis. The findings mostly support the leading market leading hypothesis, whereas the results in Africa and Middle East regions after the financial crisis support the frontier market leading and feedback hypotheses. The results in Tunisia, Sri Lanka, Bulgaria, and Latvia four frontier markets support the neutrality hypothesis after the crisis. The results reveal that each individual frontier market has a different relationship with the leading market. The global financial crisis largely influences the causality between frontier and leading markets. We find that high-technology export, tax, and tariff are the key determinants of affecting the frontier markets leading hypothesis through a Logit model. Industry value, gross saving, interest rate, and population growth of the frontier markets and energy factors, gross national income, and stock traded value of the leading markets influence the support to the leading market leading hypothesis. The results are critical for international diversification and economic policymaking (Hsieh, Chen, Lee, & Yang, 2013), since analysts and policymakers can set up their goals based on our main findings. For instance, policymakers in frontier markets should pay much attention on interest rate, tax rate, tariff, gross savings, industry value, and population growth for developing their equity markets; whereas the governments in leading countries should target energy factors (i.e., energy production, energy import, and energy use), gross national income, stock traded value, and high-technology exports to improve their equity markets. In addition, as to the influence of the financial crisis on stock market integration, our findings imply that investors can achieve better portfolio performance while taking financial crisis factors into consideration. The rest of the paper is organized as what follows. Section 2 outlines relevant literatures. Section 3 describes our empirical data. Section 4 discusses the empirical methodology and the results. Section 5 concludes our findings. 2. Literature review 2.1. The importance of frontier markets Nowadays, frontier countries account for 21.6% of the world’s population, 6% of its nominal GDP, but only 3.1% of world capitalization (Speidell, 2011), for which there are still a huge potential investment rooms for international investors in frontier markets. In 2010, the MSCI Frontier Market Index provided 32.4% return. During the last one decade, frontier economies have grown at an average annual rate

M.-P. Chen et al. / North American Journal of Economics and Finance 29 (2014) 84–103

87

of 5.1%, whereas the U.S. economy has only a growth rate of 1.8% annually.2 The possible reasons for why a large number of foreign funds are invested directly into frontier markets may be due to their dramatically increasing GDP growth rates, young and growing populations, a boom in trade and investment, technological catch-up potential, rapid penetration of mobile communications, abundant natural resources, and a growing middle class (Speidell, 2011). The related studies of discussing conditions on frontier markets have that: Speidell and Krohne (2007) pinpoint that investing in frontier markets can be proven to be very rewarding because of their special features (i.e., available information is sparse, local regulations are complex, and research coverage by analysts and brokerage firms is limited). De Groot, Pang, and Swinkels (2012) show that investment strategies based on the frontier stock markets can largely improve the efficiency of investors’ investment portfolio. Therefore, frontier markets are developing in ways that deserve international attention due to their potential growing for production, consumption, and investment. 2.2. The importance of financial integration Many studies have documented the impacts of financial integration (see Table 1). Financial integration has strong implications for financial stability, helping economies to absorb shocks and to foster their development (Lahrech & Sylwester, 2013; Lee, 2013; Yu, Fung, & Tam, 2010). If the financial integration degree is high, the benefits of diversification may be undermined. Conversely, increasing financial integration between markets naturally leads to decreasing cost of capital (Bekaert & Harvey, 2000), greater private investments (Henry, 2000), and higher economic development (Bekaert, Harvey, & Lundblad, 2001; Lee & Hsieh, 2014). The degrees of financial integration provide important insights about the capital flow across countries and enhance awareness of market co-movements. 2.3. Related studies for frontier markets and financial integration Globalization has recently increased financial integration over time, motivating international investors to search for new investment opportunities and improve the risk adjusted returns for their portfolios. Goetzmann, Li, and Rouwenhorst (2005) mention that investors should be willing to keep expanding their investment horizon to new and less integrated equity markets for getting a better diversified portfolio. Speidell and Krohne (2007) indicate that diversification benefits are the chief motivation for investors to include frontier markets in their investment portfolios. Though research explores the integration among frontier markets, emerging markets, and developed markets, little is known about the financial integration of frontier and regional leading countries. If financial integration exists in both frontier and leading markets in the same region, it implies that the benefits of diversification are few for international investors, but the economic growth rate of frontier markets will increase. Therefore, our paper attempts to fill the gap of financial integration between frontier and regional leading countries in the literature. The literature has lately documented frontier market financial integration. For instance, Speidell and Krohne (2007) demonstrate low correlation between frontier and developed markets. Berger et al. (2011) find that those markets exhibit low levels of integration with the world market and offer significant diversification benefits subsequently. Todorov and Bidarkota (2011) find that the impact of the U.S. conditional volatility for most frontier markets is characterized importantly as neither completely segmented nor completely integrated with it from the U.S. Speidell (2011) shows that in Africa, Asia, Europe, Latin America, and the Middle East five frontier regions, there are their own unique opportunities and challenges, implying that each individual frontier economy is endowed differently. Thus, our paper considers the different features of distinct regions by using control variables in the analysis. The extent of frontier market integration merits investigation, as it may have critical implications for policymakers to set up government policies and develop their economics, for international investors to increase their investment opportunities, and for financial institutions to lower the capital

2

IMF World Economic Outlook Database, October 2010.

88

Table 1 Previous works on financial integration. Period

Research purposes

Empirical methods

Country/sample

Results

Carrieri et al. (2004)

1996–1999

Global integration at the industry level.

GARCH.

G-7 countries.

Miles (2005)

1995–2002

VAR.

Carrieri et al. (2007)

1977–2000

Do frontier equity markets exhibit common trends and still provide diversification opportunities? Assess the evolution in market integration.

Frontier markets in Eastern Europe and Africa. Industry portfolios, country fund, ADRs.

Driessen and Laeven (2007)

1985–2002

Regression.

52 countries.

Abad, Chuliá, and Gómez-Puig (2010)

1999–2008

How the benefits of international portfolio diversification differ across countries from the perspective of a local investor. EMU and European government bond market integration.

Country-level integration does not preclude industry-level segmentation. Greater diversification gains can be achieved if local industry investment is country specific. Frontier markets are a good source of diversification opportunities despite a degree of integration. Local risk is a vital factor in explaining the emerging market returns. None of the countries appear to be completely segmented. The benefits of investing abroad are largest for investors in developing countries.

CAPM model.

15 European Union countries.

Chen, Buckland, and Williams (2011)

1993–2009

The change in the dynamic stochastic structure of the industrial sectors of the China and HK share markets.

VECM-MV-GARCH.

China and HK.

Demian (2011)

2001–2009

The impact of EU accession on Central and East European markets.

VAR.

Laopodis (2011)

1990–2009

Dynamic linkages between stock prices and economic fundamentals.

Lucey and Muckley (2011)

1988–2007

Syriopoulos (2011)

2001–2007

The level and evolution of interdependence linkages between the Asian and European markets and the U.S. The risk and return profile of international portfolios allocated by investors to Balkan markets.

Rolling–sample cointegration technique and VAR. GARCH.

Czech Republic, Estonia, Hungary, Poland, Romania, Slovakia. France, Germany, Italy, UK, U.S.

Gupta and Guidi (2012)

1999–2009

The links between the Indian stock market and developed Asian markets.

Cointegration test.

Koulakiotis, Kartalis, and Papasyriopoulos (2012)

1981–2005

Integration effects on volatility of home cross-listed equities.

BEKK-GARCH.

GARCH-in-mean.

VECM.

Note: This table shows the existing literature on financial integration in various empirical methods.

Asian and European stock markets.

U.S., Germany, Greece, Cyprus, Turkey, Romania, Croatia, Bulgaria. India, HK, Japan, Singapore. Germany.

Euro markets are less vulnerable to the influence of world risk factors and more vulnerable to EMU risk factors. Before deregulation there is weak evidence of cointegration between A and B share markets, while both markets appear to be largely cointegrated post deregulation. Integration increases over time, but EU accession plays a minor direct role in the integration. Stock markets appear to move independently of fundamentals in the long run, especially in the post-Euro period. For U.S. investors, European markets provide a superior long-term diversification opportunity relative to that provided by the Asian markets. Both domestic and external forces affect equity market behavior, leading to a long-run equilibrium. There is a short-run relationship and absence of a strong long-run relationship among these markets. Stock price volatility is influenced more by bad and/or good news than by large and/or small news.

M.-P. Chen et al. / North American Journal of Economics and Finance 29 (2014) 84–103

Author

M.-P. Chen et al. / North American Journal of Economics and Finance 29 (2014) 84–103

89

cost and achieve significant profits possibly by international risk sharing. Hence, financial integration is the topic of frontier markets that implicates beyond the fields of corporate finance and investments and deserves further scrutiny (Carrieri, Errunza, & Hogan, 2007). 3. The data 3.1. Data collection and sample composition Our empirical analyses mainly target the leading and frontier markets. We apply data from MSCI, S&P, and Russells’ frontier markets, which cover 47 markets at the earliest stages of economic development. However, due to data availability, this paper includes stock indices of 29 frontier markets and 14 leading markets.3 In order to avoid spurious spillover effects due to non-synchronous trading hours and from the perspective of policy makers concerned with financial stability, we collect weekly data for the stock indices from the DataStream database after matching with S&P Frontier Broad Market Index (BMI) data.4 These markets are located in Africa (Ghana, Kenya, Mauritius, Namibia, Nigeria, Tunisia), Latin America (Argentina, Colombia, Ecuador, Jamaica, Panama, Trinidad and Tobago), Middle East (Bahrain, Jordan, Kuwait, Oman, Pakistan, Qatar, United Arab Emirates), Asia (Bangladesh, Sri Lanka, Vietnam), and Europe (Bulgaria, Croatia, Cyprus, Estonia, Latvia, Lithuania, Malta, Romania, Slovakia, Slovenia). The regional leading countries, chosen according to their leading GDP in each region, include South Africa and Egypt for the African region; China, India and Japan for the Asian region; France, Germany, and the UK for the European region; Brazil, Chile, and Mexico for the Latin America region; and Saudi Arabia and Turkey for the region of Middle East. Among these regions, the Middle East dominates about 57% of the value of the frontier market indices (Watson, 2010),5 especially as the Middle East boasts abundant natural resources. Our samples cover the period 2000–2011, consisting of 12 years data. Following Ding and Jinjarak (2012), we set the pre-financial crisis sub-period from 2000 to 2007, and set the post-financial crisis sub-period since 2008. As suggested that different sample periods can result in contradictory findings, especially for a noted financial crisis, we consider a long enough period frame and split the sample into two subsamples to capture possible time-variant stock market integration across the subperiods before and after the global financial crisis. According to variable category from the World Bank database, the market features that we use to investigate stock market integration are obtained from the World Bank database – including private sector (ease of doing business (DOBUS), export value indices (EXPVI), import value indices (IMPVI), total tax rate (TAX), tariff rate (TARIFF), and international tourism receipts (TOUR)), energy factors (energy production (ENGPOD), energy imports (ENGIMP), and energy use (ENGUSE)), economic policy (foreign direct investment (FDI), gross domestic product per capita (GDP), gross national income per capita (GNI), gross savings (GRSAV), industry value added (INDUST), inflation (INFLAT), and total reserve (RESERVE)), financial sector (listed domestic companies (LISTCO), market capitalization of listed companies (MKTCAP), real interest rate (INTST), total value of stock traded (TRAD), stock traded turnover ratio (TADTUR), and strength of legal rights index (LEGAL), along with other factors (high-technology exports (HI-TECH), land area (LAND), population (POP), and population growth (POPGR)). Table A1 in Appendix presents the detailed descriptions of these variables. Pajuste, Kepitis, and Högfeldt (2000) emphasize the importance of geographic proximity in explaining the level of a country’s financial integration. Levine (2001) mentions that international financial liberalization accelerates economic growth and financial integration. Therefore, this paper also includes geographic distance (DIST) and liberalization (LIB) as control variables in the Logit model.

3 All indices are in U.S. dollars, providing additional comparability across markets and implicitly taking care of the currency market effects for international investors. 4 The S&P frontier BMI is a broad benchmark index that tracks the equity return performance of 496 stocks publicly traded in 35 frontier countries. 5 The specific differences of frontier markets include: market capital, GNP per person, turnover, GDP growth, and corruption (Watson, 2010).

90

M.-P. Chen et al. / North American Journal of Economics and Finance 29 (2014) 84–103

Table 2 Stock returns used and summary statistics for each country. Country

Period

Panel A: Africa 2000/1/1–2011/12/31 Ghana 2000/1/1–2011/12/31 Kenya 2000/1/1–2011/12/31 Mauritius 2000/2/11–2011/12/31 Namibia 2000/1/1–2011/12/31 Nigeria 2000/1/1–2011/12/31 Tunisia 2006/7/28–2011/12/31 S. Africa 2006/7/28–2011/12/31 Egypt Panel B: Asia 2000/1/1–2011/12/31 Bangladesh 2000/1/1–2011/12/31 Sri Lanka 2006/7/28–2011/12/31 China 2000/1/1–2011/12/31 India 2000/1/1–2011/12/31 Japan Panel C: Europe 2000/1/1–2011/12/31 Bulgaria 2000/1/1–2011/12/31 Croatia 2000/1/1–2011/12/31 Estonia 2000/1/1–2011/12/31 Latvia 2000/1/1–2011/12/31 Lithuania 2000/1/1–2011/12/31 Romania 2000/1/1–2011/12/31 Slovakia 2000/1/1–2011/12/31 Slovenia 2000/1/1–2011/12/31 France 2000/1/1–2011/12/31 Germany 2000/1/1–2011/12/31 UK Panel D: Latin America 2006/7/28–2011/12/31 Argentina 2000/1/1–2011/12/31 Colombia Ecuador 2000/1/1–2011/12/31 2000/1/1–2011/12/31 Jamaica 2007/2/9–2011/12/31 Panama Trinidad and Tobago 2000/1/1–2011/12/31 Brazil 2000/1/1–2011/12/31 Chile 2006/7/28–2011/12/31 Mexico 2000/1/1–2011/12/31 Panel E: Middle East Bahrain 2000/4/28–2011/12/31 2000/1/1–2011/12/31 Jordan Kuwait 2005/1/14–2011/12/31 2000/4/28–2011/12/31 Oman 2000/1/1–2011/12/31 Pakistan Qatar 2005/1/7–2011/12/31 2005/1/7–2011/12/31 UAE 2000/1/1–2011/12/31 Saudi Arabia 2000/1/1–2011/12/31 Turkey U.S. 2000/1/1–2011/12/31

Kurtosis

Jarque-Bera

Index

Mean (%)

Std. Dev.

Skewness

S&P S&P S&P S&P S&P S&P S&P S&P

−0.0034 0.2267 0.1964 0.1418 0.1845 0.1242 0.1141 −0.1217

2.6365 3.8262 2.7919 3.0035 3.8254 2.0967 4.8469 4.3702

−0.0524 0.3853 0.1716 −0.1482 −0.3574 0.0551 0.0304 −1.6347

19.5167 16.2606 17.4016 12.1092 5.6793 12.4369 7.1730 8.8347

7115.885 4602.072 5412.888 2149.311 200.571 2323.146 206.106 529.330

S&P S&P S&P FTSE S&P

0.2554 0.2301 0.1584 0.1412 −0.0788

3.5445 3.5492 4.6168 4.1582 2.8497

0.5219 1.3333 −0.3214 −0.5798 −0.3475

17.6517 13.2987 5.2616 5.8529 4.9173

5627.777 2951.979 65.416 247.367 108.479

S&P S&P S&P S&P S&P S&P S&P S&P S&P S&P S&P

0.1473 0.1099 0.1916 0.1838 0.1136 0.1595 0.3057 0.1880 −0.0285 −0.0115 −0.0315

4.8990 3.6903 4.2111 4.2655 3.5495 5.3850 4.1321 3.0857 3.5599 3.8575 3.1974

1.4326 −0.4671 −0.2218 0.5366 −0.3354 −0.6749 −0.5265 0.8230 −1.0157 −0.8981 −1.3008

26.3144 12.7799 9.4498 31.7081 16.4341 13.1281 43.6792 15.2233 9.0645 8.0611 14.7672

14,391.99 2517.524 1090.191 21,526.70 4719.134 2723.088 43,191.55 3967.748 1066.910 752.261 3788.212

S&P S&P S&P S&P S&P S&P FTSE S&P FTSE

0.0807 0.3705 0.3088 0.1766 0.1116 0.1606 0.1923 0.2368 0.2050

5.3184 3.6222 4.5924 3.8896 1.3811 1.9362 5.5608 4.2679 4.3903

−1.6689 −1.5516 −2.4772 2.7286 −0.1838 2.6811 −0.7137 −1.9568 −0.6813

11.2948 12.1033 105.223 30.7662 4.5088 23.9115 7.3579 16.5444 10.1179

946.009 2354.893 273,196.8 20,886.08 25.7227 12,155.96 548.489 2352.107 1369.905

S&P S&P S&P S&P FTSE S&P S&P S&P FTSE S&P

−0.0110 0.1439 0.0210 0.1630 0.1145 0.1088 −0.1494 0.1636 −0.0048 −0.0000

2.1594 2.9552 3.4242 2.7007 4.1531 4.2443 4.7499 3.7184 7.5613 2.8511

−0.9074 0.0736 −1.1164 −1.3152 −0.9224 −0.5684 −1.2758 −1.4295 −1.3092 −0.7862

9.9147 5.7833 10.0593 13.5524 6.1735 8.1916 9.9793 11.0918 20.4982 8.7705

1298.960 202.624 833.714 3006.088 351.458 429.563 839.831 1921.065 8165.197 933.032

Note: This table provides the relevant descriptive statistics of the stock returns in each country. We apply weekly stock indices of 29 frontier markets and 14 leading markets from DataStream, expressed in U.S. dollars. The length of the data samples is not uniform, and we are constrained by data availability. Stock returns are calculated as 100 times the difference in the log of the stock indices.

3.2. Descriptive statistics Table 2 provides the relevant descriptive statistics of the stock returns in each country. Following the previous way, stock return is the percentage of the difference of the stock price index in log. Specifically, we report information on the mean, standard deviation, skewness coefficient, kurtosis coefficient, and the Jarque–Bera normality test (Jarque & Bera, 1980) for stock returns. As for the market

M.-P. Chen et al. / North American Journal of Economics and Finance 29 (2014) 84–103

91

Table 3 Causality test between frontier markets and leading markets. F-statistic

All period

Null hypothesis

L→F

Panel A: Africa Ghana

Kenya

Mauritius

Namibia

Nigeria

Tunisia

Panel B: Asia Bangladesh

Sri Lanka

Panel C: Europe Bulgaria

Croatia

Estonia

Latvia

Lithuania

Romania

Slovakia

Pre-crisis

Post-crisis

F→L

L→F

F→L

0.00 0.92 21.61*** 0.29 0.04 17.82*** 0.30 0.34 5.29*** 15.04*** 1.59 100.94*** 2.59 4.00*** 15.20*** 0.71 1.24 5.39***

0.26 0.26 0.07 1.18 2.34 1.91 0.95 1.18 1.19 0.20 4.67** 0.21 1.78 0.53 3.23** 2.00 0.28 0.09

0.71 0.92 44.78*** 0.02 1.88 18.48*** 2.27 4.61*** 9.56*** 2.27 2.71 5023.2*** 2.60 7.41*** 2.45 0.89 1.71 12.39***

0.50 0.34 0.29 0.15 0.22 0.32 1.94 0.18 0.91 0.93 0.17 0.31 0.02 0.66 0.41 0.45 0.71 1.26

China India Japan U.S. China India Japan U.S.

4.52*** 13.57*** 10.14*** 17.63*** 0.36 8.04*** 14.66*** 15.70***

0.46 0.53 1.71 0.29 0.00 0.34 0.03 2.98**

3.17** 12.04*** 7.56*** 178.80*** 0.31 360.56*** 14.66*** 19.42***

0.75 0.14 2.63 1.09 0.33 1.28 0.09 4.42**

0.90 1.98 2.81*** 2.68 1.07 1.14 1.50 1.76

France Germany UK U.S. France Germany UK U.S. France Germany UK U.S. France Germany UK U.S. France Germany UK U.S. France Germany UK U.S. France Germany UK U.S.

68.63*** 68.70*** 0.00 69.50*** 1.59 1.39 6.16*** 5.78*** 1.66 1.66 6.51*** 7.75*** 9.16*** 10.86*** 19.39*** 20.82*** 1.43 1.02 1.24 1.85 12.15*** 10.92*** 20.06*** 18.73*** 9.89*** 8.95*** 21.40*** 22.24***

0.39 0.30 0.03 0.22 2.45 1.94 4.10** 2.04 2.31 2.91 5.22*** 5.50*** 1.45 1.80 1.26 1.83 0.73 0.51 1.07 0.15 0.40 1.26 1.59 0.72 5.39*** 3.00 2.81 0.73

46.11*** 46.05*** 45.94*** 46.93*** 14.91*** 12.12*** 29.99*** 21.84*** 13.35*** 10.43*** 27.88*** 17.58*** 25.43*** 26.04*** 40.59*** 32.19*** 8.63*** 7.14*** 17.41*** 10.98*** 38.89*** 33.43*** 42.58*** 38.15*** 21.31*** 16.07*** 231.90*** 24.10***

0.78 0.55 0.29 0.27 2.11 0.44 2.19 0.82 0.36 0.18 0.19 0.03 2.27 2.18 3.20** 5.26*** 0.62 0.97 0.49 1.59 0.12 0.60 0.50 4.51** 1.95 0.66 0.96 0.84

0.49 0.12 1.82 1.92 9.41*** 8.07*** 2.64** 1.44 4.61*** 4.43*** 1.58 1.90 0.79 0.23 0.31 1.17 7.54*** 6.93*** 2.71** 2.58 2.16 2.90** 1.61 2.67** 3.43** 2.52 3.92*** 4.55***

South Africa Egypt U.S. South Africa Egypt U.S. South Africa Egypt U.S. South Africa Egypt U.S. South Africa Egypt U.S. South Africa Egypt U.S.

L→F 2.80** 0.68 3.35** 9.63*** 1.30 6.89*** 7.03*** 0.83 7.76*** 23.91*** 2.38 7.49*** 3.06** 2.71* 4.32*** 1.62 1.77 0.26

F→L 0.54 0.78 0.57 1.57 0.78 4.80*** 2.06 1.78 4.57** 1.28 14.26*** 0.08 4.41** 1.79 4.29** 1.50 0.75 0.95 0.07 0.52 0.10 0.09 0.07 0.82 0.81 0.29 0.26 0.20 0.17 0.67 2.13 2.30 4.33** 0.62 5.19*** 6.62*** 8.98*** 10.38*** 0.18 0.16 0.02 0.03 1.37 1.61 1.28 2.09 0.08 0.31 0.29 0.65 3.53** 2.63 1.49 0.35

92

M.-P. Chen et al. / North American Journal of Economics and Finance 29 (2014) 84–103

Table 3 (Continued )

F-statistic

All period

Null hypothesis

L→F

Slovenia

France Germany UK U.S. Panel D: Latin America Brazil Argentina Chile Mexico U.S. Colombia Brazil Chile Mexico U.S. Brazil Ecuador Chile Mexico U.S. Brazil Jamaica Chile Mexico U.S. Brazil Panama Chile Mexico U.S. Brazil Trinidad and Tobago Chile Mexico U.S. Panel E: Mid-East Saudi Arabia Bahrain Turkey U.S. Saudi Arabia Jordan Turkey U.S. Saudi Arabia Kuwait Turkey U.S. Oman Saudi Arabia Turkey U.S. Saudi Arabia Pakistan Turkey U.S. Qatar Saudi Arabia Turkey U.S. Saudi Arabia UAE Turkey U.S.

2.58 458.69*** 85.25*** 3.45*

Pre-crisis F→L 2.50 2.61 1.42 1.81

2328.90*** 1036.68*** 3333.95*** 1288.96*** 3.288* 0.05 5.29*** 7.02 30.84*** 5.14*** 44.08*** 44.09*** 13.02*** 0.34 22.86*** 12.40*** 3193.78*** 3.17** 7607.05*** 5566.88*** 3.48** 1.03 5.83*** 1.86

0.22 1.65 0.13 0.45 1.09 0.63 0.70 0.31 0.04 0.27 0.15 0.38 0.58 1.60 0.18 0.71 1.06 0.023 1.54 0.02 0.06 2.04 0.05 1.06

5625.61*** 865.98*** 6891.63*** 22.21*** 2.03 121.86*** 3094.03*** 1448.16*** 3625.19*** 4782.55*** 915.89*** 5984.46*** 18.37*** 2.39 17.50*** 2159.08*** 1276.76*** 2746.54*** 2198.76*** 1113.96*** 2269.74***

0.01 0.41 0.66 1.71 1.77 1.52 0.06 0.04 0.23 0.09 0.67 1.31 0.37 0.78 0.26 0.58 0.30 0.14 0.11 0.34 0.13

L→F 7.22*** 4.81*** 17.17*** 15.95*** 4987.48*** 3702.94*** 10,303.05*** 4677.20*** 5.33*** 23.57*** 13.05*** 1.19 32.26*** 4.87*** 37.86*** 36.53*** 7.20*** 0.29 412.11*** 26.97*** 5961.57*** 9443.26*** 18,558.04*** 23,946.71*** 3.85*** 0.64 6.48*** 2.36** 7550.18*** 747.46*** 11,144.38*** 6.44*** 0.85 6.22*** 49.98*** 47.60*** 50.30*** 62,621*** 797.17*** 10,458.61*** 13.87*** 2.57 18.99*** 2432.97*** 1746.01*** 4.222.90*** 2500.31*** 1906.06*** 4650.61***

Post-crisis F→L 0.52 0.02 0.08 2.56 0.44 2.84 0.14 0.25 1.43 0.80 0.31 0.56 0.03 1.28 0.09 0.65 0.42 0.00 0.31 0.22 2.21 2.17 3.65** 0.10 0.83 0.11 0.48 0.22 0.01 0.38 0.93 1.83 2.18 0.93 0.01 0.03 0.06 0.02 0.60 1.39 0.10 1.19 0.33 0.29 0.29 0.12 0.03 0.05 0.10

L→F 7.97*** 7.07*** 7.91*** 6.69***

F→L 1.94 2.63 2.83 3.38**

132.37*** 5.28*** 105.76*** 9.76*** 2.31 5.32*** 2.54 3.39** 1.28 3.22** 4.01*** 5.35*** 13.35*** 1.30 10.12*** 0.76 8.41*** 3.43** 8.62*** 3.47** 0.06 0.06 0.49 0.30

0.18 3.95** 1.05 11.15*** 0.00 0.04 0.62 0.20 0.06 0.01 0.15 0.14 2.69 1.03 1.63 0.83 2.46 0.19 1.68 1.66 3.14** 2.11 2.76 4.09**

19.08*** 4.33*** 3.92*** 20.81*** 4.61*** 4.01*** 18.26*** 5.52*** 2.33 29.69*** 6.55*** 6.66*** 6.22*** 0.81 5.21*** 43.96*** 11.36*** 38.90*** 34.29*** 4.90*** 7.25***

1.41 1.06 0.70 0.79 0.53 4.26*** 2.81 3.32** 2.72 3.81** 1.31 7.25*** 0.69 1.21 0.60 4.69*** 0.75 2.55 2.4 2.40 1.11

Notes: The results are for all coefficients jointly to test for VECM-based Granger Causality in the all period, pre-crisis, and postcrisis in the case of a two period lags. **, *** denote significance at 5% and 1% levels. L → F denotes leading market Granger causes frontier market. F → L denotes frontier market Granger causes leading market. Following Ding and Jinjarak (2012), we set pre-financial crisis sub-period from 2000 to 2007, and set post-financial crisis sub-period after year 2008.

M.-P. Chen et al. / North American Journal of Economics and Finance 29 (2014) 84–103

93

returns, Colombia has the highest mean (0.37%), whereas the lowest mean is that of UAE (−0.149%). As for risk measured by standard deviation, Turkey’s stock market has the highest risk level (7.56), whereas Panama records the lowest risk level with a standard deviation of 1.38. The length of the data samples is not uniform, and we are constrained by data availability. For instance, the available USD index for South Africa and Egypt starts only from July 2006.6 As shown in Table 2, more than half of stock returns are negatively skewed, indicating a left fat tail phenomenon. Likewise, most of the series are leptokurtic (flat distribution tails) as the kurtosis statistics indicate (>4). In fact, the Jarque–Bera tests tend to reject most of the null hypothesis of normality for all series at the 1% level. 4. The results There are four main steps involved in the methodology procedure. We first begin with testing the existence of a unit root for the investigated variables by using the Augmented Dickey–Fuller (ADF) test. The second step is testing cointegration by the Johansen approach. The presence of a cointegrating relationship predicts that frontier and leading markets have a long-run equilibrium relationship, which is related to our causality tests. The fourth step conducts the Logit regression analysis to identify the possible factors of influencing the nexus. 4.1. Unit-root test and cointegration test We examine the stationarity of stock price series via Augmented Dickey–Fuller (ADF) tests, in which the null hypothesis is based on the existence of a unit root. According to our examinations, all of stock price indices have a unit root at the 5% significance level, but they are all stationary after the first difference. In other words, all stock price series are integrated of order one (indicated as I(1)). Therefore, a further cointegraiton test is necessary. In fact, their cointegration relationships hold via the Johansen cointegration test.7 As a result, we need to incorporate error correction terms, estimated from a cointegration regression, into causality tests. 4.2. Causality test Following a procedure suggested in Wang, Yang, et al. (2013), Lee, Chen, and Chang (2013), and Wang, Yao, and Fang (2013), we use Granger causality to assess the lead-lag relationships between the frontier and leading markets. According to Engle and Granger (1987), a cointegrating regression raises an error-correction representation, for which an error-correction term (ECT) must be incorporated into the model. Thus, we specify a VECM system for our causality tests. For the two-variable model with one cointegrating relationship, the VECM with (p-1) lags can be expressed as follows: Ft = ˛1 + 11 ECTt−1 +

p−1 

ˇ1j Ft−j +

j=1

Lt = ˛2 + 21 ECTt−1 +

p−1  j=1

p−1 

1j Lt−j + εt ,

(1)

2j Ft−j + εt ,

(2)

j=1

ˇ2j Lt−j +

p−1  j=1

where j is the lag length, the ECT represents the error-correction terms derived from the long-run cointegrating relationship,  is the first-difference operator. ␧t are Gaussian residuals. Eq. (1) is used to test the null hypothesis that Lt (leading market stock return) does not cause Ft (frontier market stock return), for which we estimate the significance of the lagged dynamic terms–that is, the null

6 Observations for all series full length of the whole data sample are 627. The observations for the pre-crisis and post-crisis sub-periods are 418 and 209, respectively. 7 The results of unit-root and cointegration test are omitted due to the space limitation, and are available on request.

94

M.-P. Chen et al. / North American Journal of Economics and Finance 29 (2014) 84–103

H0 all ␥1j = 0 by the Wald test. When we reject the null hypothesis, it means that leading markets Granger-cause frontier markets. On the other hand, Eq. (2) is used to test whether Ft does not cause Lt - that is, to test null H0 all ␥2j = 0, employing the Wald test. Note that since the Johansen procedure is sensitive to the choice of lag length, we determine the lag length according to the Schwarz information criterion (SIC; Schwarz, 1978) in the VAR model. Given the results of cointegration tests, we perform the VECM-based causality tests. From Table 3 we find evidence that the U.S. market mostly Granger causes the frontier markets in the pre-period of the financial crisis, whereas a frontier market leading relationship is seldom observed in this model. The U.S. market leading condition is comparatively less after the crisis in Asian and European frontier markets. The conditions that frontier markets influence the leading markets rarely in the pre-period of the crisis, however the feedback and the frontier market leading hypotheses are more significant in several markets in the post-period of the crisis, signifying the impact of the financial crisis did alter the causality between frontier and leading markets. In the regions of Asia, Europe, and Latin America, regional leading markets do have strong Granger causality with frontier markets before the crisis.8 To sum up, the findings we find mostly support the leading market leading hypothesis, but some findings support the feedback and frontier market leading hypotheses fairly well in Africa and Middle East regions after the crisis. The results of four frontier markets (i.e., Tunisia, Sri Lanka, Bulgaria, and Latvia) support the neutrality hypotheses after the crisis as well, compared to the condition that markets supported the leading market leading hypotheses before the crisis. Hence, the findings show two implications. The first implication is that each individual frontier economy is endowed differently, and thus financial integration might differ from country to country. The findings of De Groot et al. (2012) depict an increasing correlation during the financial crisis period can explain the phases in most frontier countries, except for Tunisia, Sri Lanka, Bulgaria, and Latvia. The second is that the global financial crisis largely influences the causality between frontier and leading markets, signifying the level of change in potential international diversification opportunities. The possible reasons for that financial crisis changed the relationship between markets are that the crisis changed the dependence structure (Bhatti & Nguyen, 2012) and/or its spillover effects (Lin, 2012). Nevertheless, it is difficult to pinpoint the factors for explaining our findings with respect to causality tests. We thus further conduct the Logit regression in the next section to detect the determinants of the market integration. 4.3. Logit regression analysis As suggested in Marshall et al. (2011), their future research can focus on the sources of frontier market diversification benefits. The benefits are perhaps driven by frontier market economies having more different exposures to commodities, e.g., crude oil. Many frontier markets are oil exporters that benefit from increasing oil prices, while most developed countries are net oil importers, by which increasing oil price curtails economic growth. To examine which factors affect the relationship between frontier and leading markets returns, we perform the Logit regression analysis with a dummy variable, D, for which it equal to one if the frontier (leading) market return leads the leading (frontier) market returns or if a bidirectional relation exists, and zero otherwise. We include the above-mentioned factors as independent variables in our Logit models. We test the independent variables, influenced by the mentioned hypotheses above, via the following model specification: Di = ı0 + Di = 0 +



ı country factor Fi +  · dummyi + εi ,

(3)

 country factor Li +  · dummyi + εi ,

(4)



Subscript i denotes the number of countries. All the independent variables are the average values of yearly data over our sample period (e.g., a country’s pre-crisis FDI is the average value of FDI yearly

8 Following the reviewer’s suggestion about structural breaks, we perform Chow tests for Estonia and Trinidad & Tobago. Trinidad & Tobago shows significant p-value = 0.012 in 2008/01/04; Estonia shows significant p-value = 0.09 in 2007/11/23, 0.06 in 2007/9/21, while p-value = 0.22 in 2008/01/04. Therefore, there exist structural breaks for Estonia and Trinidad & Tobago. We specify the same structural break date 2008/1/4 as the start point of the global financial crisis across countries.

Table 4 Binary Logit models on determinants of the stock markets’ leading relationship.

Independent Variables

All period

Pre-crisis

L→F

F→L Std. error

Coeff.

Panel A: Private sector Constant DOBUS F EXPVI F IMPVI F TAX F TARIFF F TOUR F AFRICA U.S. Pseudo R2

0.688 0.010 −0.007 0.012 −0.008 −0.004 0.000 −2.034** 1.364** 0.203

1.467 0.009 0.005 0.008 0.012 0.084 0.000 0.827 0.659

−6.993 0.216 0.030 −0.088 0.260** −3.090* 0.000 14.584* 1.286 0.453

Constant DOBUS L EXPVI L IMPVI L TAX L TARIFF L TOUR L AFRICA U.S. Pseudo R2

1.729 −0.003 −0.005 0.003 0.025 0.011 0.000 −1.908** 6.116 0.180

1.901 0.022 0.005 0.009 0.043 0.089 0.000 0.766 4.037

−0.898 −0.078 −0.018 0.029 0.018 −1.278 0.000 0.705 −6.586 0.245

Panel B: Energy Constant ENGPOD F ENGIMP F ENGUSE F Pseudo R2

0.782** 0.000 −0.005 0.000 0.070

0.306 0.000 0.003 0.000

−2.617*** 0.000 0.011 0.000 0.086

Constant ENGPOD L ENGIMP L ENGUSE L Pseudo R2

1.043*** 0.000** −0.011* 0.000*** 0.114

0.384 0.000 0.006 0.000

−2.939*** 0.000 0.003 0.000 0.046

L→F Std. error 6.524 0.194 0.046 0.094 0.131 1.773 0.000 7.512 1.050

Coeff.

F→L Std. error

Coeff.

L→F Std. error

F→L

Coeff.

Std. error

Coeff.

Std. error

−0.186 0.004 −0.006** 0.012*** −0.007 −0.073 0.000 0.251 0.029 0.145

1.169 0.008 0.003 0.004 0.009 0.069 0.000 0.692 0.499

−0.663 0.004 −0.002 −0.001 0.013 −0.215** 0.000 1.347* 0.818 0.106

1.304 0.009 0.003 0.004 0.010 0.090 0.000 0.743 0.536

3.873* 0.009 −0.002 −0.004 −0.037 −0.129 0.000 0.561 −0.122 0.086

2.030 0.018 0.003 0.007 0.044 0.089 0.000 0.792 1.780

−0.977 0.002 −0.001 0.001 0.000 −0.048 0.000 0.851 0.162 0.043

2.159 0.020 0.004 0.007 0.049 0.101 0.000 0.749 2.014

1.156

2.536

5.304

7.220

0.001 0.012 0.016 −0.229*** 0.000

0.014 0.016 0.020 0.065 0.000

−0.044 0.011 0.187** −0.827 0.000 *

0.042 0.032 0.091 0.552 0.000

1.605* 0.286

0.894

2.244* 0.468

1.224

2.165

27.808

1706.076

−0.011 −0.020 0.028 −0.073* 0.000

0.013 0.016 0.020 0.042 0.000

−0.478 0.206 −0.291 1.210 0.000

8.191 9.178 18.317 20.226 0.000

−1.910 0.131

4.847

−44.552 0.238

1433.656

0.565 0.000 0.007 0.000

1.992*** 0.000 −0.009* 0.000 0.066

0.383 0.000 0.005 0.000

−3.152*** 0.000 0.013 0.000 0.102

1.217 0.000 0.019 0.000

0.262 0.000 0.000 0.000 0.055

0.332 0.000 0.000 0.000

−1.591*** 0.000 0.000 0.000 0.032

0.385 0.000 0.001 0.000

0.655 0.000 0.008 0.000

1.525*** 0.000** −0.013** 0.000** 0.008

0.363 0.000 0.006 0.000

−3.806*** 0.000 0.002 0.000 0.059

0.851 0.000 0.011 0.000

0.659** 0.000 −0.004 0.000 0.021

0.313 0.000 0.003 0.000

−1.840*** 0.000 −0.002 0.000 0.340

0.412 0.000 0.003 0.000

30.728 0.533 0.031 0.210 0.494 4.847 0.000 1.350 30.392

5.297**

M.-P. Chen et al. / North American Journal of Economics and Finance 29 (2014) 84–103

Coeff.

Post-crisis

95

96

Table 4 (Continued )

Independent Variables

All period

Pre-crisis

L→F Coeff.

Constant FDI L GDP L GNI L GRSAV L INDUST L INFLAT L RESERVE L Pseudo R2

−3.776* 0.000 0.000 0.000** −0.055 0.135 0.000 −3.776 0.105

Panel D: Financial sector 2.566** Constant 0.001 LISTCO F 0.000 MKTCAP F INTST F −0.204* 0.013 TRAD F −0.004 TADTUR F LEGAL F −0.087 1.138* U.S. AFRICA 0.107 Pseudo R2 Constant LISTCO L

1.250 0.000

Std. error 1.841 0.000 0.000 0.000 0.083 0.063 0.107 0.000 0.875 0.807 2.120 0.000 0.000 0.000 0.144 0.094 0.000 2.120

Coeff.

Std. error

−1.996 0.000 0.000 0.000 −0.121 0.061 −0.140 0.000

3.492 0.000 0.000 0.000 0.152 0.116 0.246 0.000

0.331 0.091

1.276

−1.758 0.000 0.000 0.000 −1.835 0.682 −1.352

6917.090 0.000 0.090 0.000 338.179 115.724 800.402

0.312

1.140 0.001 0.000 0.108 0.045 0.009 0.142 0.670

−1.473 −0.001 0.000 −0.351 −0.031 −0.007 0.167 0.811

1.838 0.002 0.000 0.231 0.107 0.017 0.205 0.841

0.147 0.980 0.000

−1.668 0.000

Post-crisis

L→F

1.196 0.000

Coeff.

F→L Std. error

3.295 0.000 0.001* 0.000 0.231 −0.363* 0.199 0.000 −3.049*** 2.867** 0.404

3.228 0.000 0.001 0.000 0.158 0.217 0.129 0.000 1.174 1.306

4.707 0.000 0.000 0.000 0.533 −0.441 −0.018 0.000 0.385

6.512 0.000 0.000 0.000 0.328 0.367 0.076 0.000

4.436** 0.001 0.000 −0.064 −0.025 −0.003 −0.220 1.172** −2.470 0.248

2.040 0.001 0.000 0.095 0.026 0.007 0.244 0.926 0.963

0.168 0.000

1.590 0.001

Coeff.

L→F Std. error

−2.241 0.000 0.000 0.000 0.284 −0.399* 0.295 0.000

3.536 0.000 0.000 0.000 0.263 0.210 0.188 0.000

1.792* 0.273

1.063

−48.718 0.000 −0.002 0.000 −3.148 3.210 0.279 0.000 0.207

0.828 0.001 0.000 −0.462 −0.749 0.027 −0.115 1.865*

Coeff.

Std. error

Coeff.

Std. error

2.169** 0.000 0.000 0.000 −0.048* −0.029 −0.076 0.000

1.090 0.000 0.000 0.000 0.028 0.025 0.064 0.000

−1.418 0.000 0.000 0.000 0.007 0.033 −0.032 0.000

1.186 0.000 0.000 0.000 0.044 0.033 0.084 0.000

0.132

0.206

2.262 0.000 0.000 0.000** −0.385*** 0.178** −0.187** 0.000 0.248

1.728 0.000 0.000 0.000 0.138 0.072 0.080 0.000

−0.085 0.000 0.000 0.000 −0.073 0.013 0.013 0.000 0.090

1.634 0.000 0.000 0.000 0.133 0.065 0.084 0.000

2.883 0.001 0.000 0.356 0.680 0.107 0.243 1.074

1.924** 0.000 0.000 −0.099 −0.011 −0.008 −0.178 −0.093 0.259 0.116

0.844 0.001 0.000 0.109 0.025 0.009 0.137 0.497 0.641

−2.198** −0.004 0.000 −0.153 −0.008 −0.001 0.267* 0.811 0.623 0.116

1.020 0.003 0.000 0.116 0.034 0.014 0.154 0.541 0.673

1.072 0.000

−0.027 0.000*

0.809 0.000

−1.711* 0.000

0.911 0.000

62.044 0.000 0.003 0.000 3.967 4.186 0.915 0.000

0.355 −1.408 0.000

F→L

M.-P. Chen et al. / North American Journal of Economics and Finance 29 (2014) 84–103

Panel C: Economic policy Constant −1.399 0.000 FDI F 0.000 GDP F 0.000 GNI F 0.210** GRSAV F INDUST F −0.140** 0.166 INFLAT F RESERVE F 0.000 −1.949** AFRICA 1.907** U.S. Pseudo R2 0.266

F→L

0.000 −0.295 −0.031* 0.012 0.075 −5.247 0.139

Panel E: Others 0.289 Constant 0.000 HITECH F LAND F 0.000 0.000 POP F POPGR F 0.929** AFRICA −2.494*** DIST 0.000* 2.973*** U.S. Pseudo R2 0.267 Constant HITECH L LAND L POP L POPGR L AFRICA DIST U.S. Pseudo R2

0.000 0.214 0.018 0.023 0.186 15.307

0.518 0.000 0.000 0.000 0.747 −2.023*** 0.000** 2.623** 0.202

0.000 −0.036 −0.002 0.005 0.035 −0.073

0.000 0.232 0.020 0.027 0.222 17.809

0.080

0.000 0.040 −0.035 0.023 0.004 −17.087

0.000 0.050 0.029 0.026 0.287 20.441

0.335

0.577 0.000 0.000 0.000 0.398 0.811 0.000 1.091

−2.556*** 0.000 0.000 0.000 −0.616 0.585 0.000 1.098 0.109

0.870 0.000 0.000 0.000 0.576 1.343 0.000 1.161

0.635 0.000 0.000 0.000 0.487 0.694 0.000 1.199

−0.135 0.000** 0.000 0.000 −2.798 0.224 0.000 5.181 0.204

2.844 0.000 0.000 0.000 1.929 1.295 0.000 7.333

0.000 −0.001 0.000 −0.001 0.011 −0.312

0.000 0.025 0.012 0.009 0.180 6.435

0.000 −0.001 −0.008 0.012 0.128 7.442 −0.107 0.037

0.000 0.026 0.009 0.008 0.122 6.129 0.625

0.000 −0.006 −0.004 0.003 0.032 −2.155 0.680 0.043

0.000 0.030 0.008 0.008 0.118 6.388 0.670

−0.056 0.000 0.000 0.000 0.374** −0.022 0.000 0.596 0.182

0.322 0.000 0.000 0.000 0.153 0.389 0.000 0.485

−1.111*** 0.000 0.000* 0.000 −0.008 0.385 0.000 0.186 0.075

0.307 0.000 0.000 0.000 0.057 0.381 0.000 0.421

1.266** 0.000 0.000 0.000* −0.126 −0.061 0.000 0.025 0.123

0.573 0.000 0.000 0.000 0.363 0.381 0.000 0.803

−0.590 0.000 0.000 0.000 −0.295 0.456 0.000 −0.550 0.081

0.622 0.000 0.000 0.000 0.407 0.386 0.000 0.918

0.053

0.551 0.000 0.000 0.000 0.201

−2.141** 0.000 0.000 0.000 −1.080

0.930 0.000 0.000 0.000 0.664

1.334 0.168

0.823

1.693* 0.223

1.001

−0.275 0.000** 0.000 0.000 0.773

1.154 0.000 0.000 0.000 0.681

−1.720 0.000 0.000 0.000 −0.146

1.294 0.000 0.000 0.000 0.754

−4.340 0.148

2.686

0.907 0.070

2.357

0.764 0.000** 0.000* 0.000 0.201

Notes: This table presents the Logit regression results in Eqs. (3) and (4). *, **, and *** show significance at 10%, 5%, and 1%, respectively. The dependent variable in the regressions is an indicator set equal to one if the leading market (L) Granger causes the frontier market (F), signified by L → F, or the frontier market (F) Granger causes the leading market (L), signified by F → L. Independent variables with ‘L’ represent leading market variables; while independent variables with ‘F’ represent frontier market variables. Variable definitions: DOBUS, ease of doing business that a low numerical rank means that the country is conducive to business operation; EXPVI, the current value of exports; IMPVI, the current value of imports; TAX, total tax rate; TARIFF, tariff rate; TOUR, international tourism receipts; ENGPOD, energy production; ENGIMP, energy import; ENGUSE, energy use; FDI, foreign direct investment; GDP, gross domestic product; GNI, gross national income; GRSAV, gross saving; INDUST, industry value; INFLAT, inflation; RESERVE, total reserve; LISTCO, number of listed domestic companies; MKTCAP, market capitalization of listed companies; INTST, real interest rate; TRAD, stocks traded value; TADTUR, stocks traded turnover ratio; HITECH, high-technology exports; LAND, land area; POP, population; POPGR, population growth. Coefficient estimates with the t-statistics in parentheses. Standard errors are corrected using the White (1980) procedure.

M.-P. Chen et al. / North American Journal of Economics and Finance 29 (2014) 84–103

MKTCAP L INTST L TRAD L TADTUR L LEGAL L U.S. AFRICA Pseudo R2

97

98

M.-P. Chen et al. / North American Journal of Economics and Finance 29 (2014) 84–103

data in a country from 2000 to 2007), and no lagged terms are used. Independent variables with ‘L’ represent the leading market variables; while independent variables with ‘F’ represent the frontier market variables. Table 4 shows the estimated results from the Logit regression model. Panels A, B, C, D, and E test the factors of private sector, energy, economic policy, financial sector, and others that affect the market leading relationships, respectively. For the private sector of Panel A in Table 4, the frontier market variables are the main factors causing the frontier and leading market returns nexus, compared with the variables of the leading markets. For the whole period, a higher TAX is the prominent factor affecting frontier market leading hypothesis, implying a higher TAX in a frontier market causes capital flow to the other market, as well as a lower TARIFF in a frontier market attracts leading market capital to a frontier market. The result is consistent with Lane and Milesi-Ferretti (2003) that a tax policy may influence the level of international crossholdings, because firm assets may be shifted into countries with low corporate income tax rates. A lower TARIFF is the prominent factor affecting the frontier market leading hypothesis, representing that a lower TARIFF attracts capital infusion from abroad. In the pre-period of the financial crisis, TAX and TOUR have significant influences on the frontier market leading causality; TARIFF of the frontier and leading markets has a salient impact on the leading market leading causality. While in the postcrisis, a lower EXPVI and higher IMPVI in the frontier market contribute to support the leading market leading hypothesis. Moreover, for the energy factors of Panel B in Table 4, the leading market variables are the main factors which lead to support the leading market leading hypothesis. For the whole period and the post-period of crisis, a higher ENGPOD and ENGUSE plus a lower ENGIMP of leading markets are the prominent factors affecting the leading market leading hypothesis. Though some frontier markets own abundant natural resources, the energy factors of frontier markets seem not to be important for the nexus. Speidell (2011) documents that the frontier markets vary greatly in their exposures to individual commodities because they may be exporters of some commodities and importers of others. Even a strong crude oil exporter, like Nigeria, it may have to import downstream products due to its lack of internal refining capacity. The major resource producers are nationally owned or global companies listed in the developed countries. Consistent with Speidell (2011), frontier markets should be viewed simply as incidental beneficiaries or victims of commodity fluctuations rather than commodity-driven markets. For the discussion of economic policy of Panel C in Table 4, a lower INDUST and a higher GRSAV in a frontier market are the main factors supporting the leading market leading hypothesis, implying the shortage of INDUST and a high GRSAV in a frontier market drive capital flows from a leading market to a frontier market. In the pre-period of the crisis, a lower INDUST of frontier market is the key factor affecting feedback hypothesis. For a frontier market, a higher GDP is the prominent factor affecting the leading market leading hypothesis as well. As for the post-crisis period, a higher GNI and INDUST, a lower INFLAT and GRSAV of leading markets lead to support strongly for the leading market leading hypothesis. For the financial sector of Panel D in Table 4, a lower INTST of frontier markets and a lower TARD of leading markets are the main factors impacting the leading market leading hypothesis, while no factors are salient in explaining the frontier market leading hypothesis during the whole period. This phase signifies that a lower INTST of frontier markets causes capital inflow to frontier markets, and a low TRAD of leading markets drives the support for the leading market leading hypothesis. In addition, the U.S. is the prominent factor affecting the feedback hypothesis, especially for the pre-crisis period. As for the post-crisis period, a higher LISTCO of leading markets provide support for the leading market leading hypothesis; while LEGAL of frontier markets positively support the frontier market leading hypothesis. Lastly, for the discussion of other factors of Panel E in Table 4, POPGR of a frontier market is the main factor positively supporting the leading market leading hypothesis for the whole period. This phenomenon can be explained by the fact that the frontier markets have young and growing populations and growing middle class, and therefore they attract foreign direct investments (Speidell, 2011). Higher HITECH in a leading market causes the financial integration between leading and frontier markets. AFRICA, DIST, and U.S. saliently cause the leading market leading phenomenon. Before the crisis, larger land and higher HITECH of a frontier market and higher HITCH of a leading market can cause

M.-P. Chen et al. / North American Journal of Economics and Finance 29 (2014) 84–103

99

the leading market leading phenomenon. After the crisis, larger land in a frontier market causes the frontier market leading condition, while more POP in a leading market and larger POPGR in frontier market can cause the leading market leading condition. The findings signify that the population, LAND and HITECH are important factors of influencing the frontier-leading market return nexus. An in-depth discussion about these determinants is not pursued here due to space limitation, and it may be a valuable topic to further investigate in the future. In sum, the U.S. saliently contributes to the leading market leading hypothesis, while the non-African region contributes to the frontier and leading market nexus. The POPGR, GRSAV, INDUST, and INTST of frontier markets, the energy factors, and TRAD of leading markets drive the leading market leading hypothesis. The TAX and TARIFF of frontier markets and HITECH of leading markets drive the frontier market leading hypothesis. Surprisingly, the natural resources factors of frontier markets do not show any importance in the nexus. The global financial crisis in 2008 not only influences the financial integration relationships between frontier and leading markets, but also affects the factors of financial integration. In addition, we do find the importance of DIST, like the same findings in Pajuste et al. (2000). Regions factors are also included as control variables, and African market shows a strongly negative influence on the nexus. 5. Conclusions Because the financial empirical literature on frontier markets remains surprisingly a lack of information, perceptions of excessive risk or unknown variables discourage investors from bringing their capital into frontier markets. Different from conventional studies that mainly explore the stock market integration between frontier and other stock markets, we first examine the market return nexus (i.e., frontier market leading, leading market leading, feedback, and neutrality hypotheses) between frontier and leading stock markets (i.e., U.S. and regional leading markets) and the factors driving the hypotheses. Our empirical findings enrich the thin body of the empirical financial literature focusing on frontier economies and offer investors an ‘early-mover advantage’. In this paper we embark on systematic study of the factors of frontier and leading markets that explain the relationship between frontier and leading markets over the period 2000–2011. By and large, the results herein suggest that the U.S. saliently Granger causes frontier markets, while this is comparatively less after the crisis than Asian and European frontier markets. Our findings mostly support the leading market leading hypothesis, while the feedback and frontier market leading hypotheses are fairly presented in Africa and Middle East regions after the crisis. The results of Tunisia, Sri Lanka, Bulgaria, and Latvia four frontier markets support neutrality hypotheses after the crisis, compared to the condition that markets supported the leading market leading hypothesis before the crisis. The causality test results indicate two implications. First, each individual frontier economy is endowed differently, and thus the financial integration for other countries might differ from country to country. Second, the global financial crisis largely influences the causality between frontier and leading markets. The population growth, industry value, grass savings, and real interest rate of a frontier market are the factors of determining the leading market leading condition. For energy factors, gross national income and stocks traded value of a leading market drive the leading market leading condition, whereas tax rate and tariff rate of a frontier market and high-technology exports of a leading market drive the frontier market leading condition. Surprisingly, the natural resource factors of frontier markets do not show any importance in the nexus. The global financial crisis in 2008 not only influences the financial integration relationships between frontier and leading markets, but also changes the factors of financial integration. Our findings in this paper are most important to investors, scholars, and policymakers. There is an implication that frontier markets in different regions have a distinct lead-lag relationship with leading markets. The impact of the global financial crisis should be heeded when looking at financial integration in general. For policymakers of frontier markets, the finding suggests that population growth, industry value, real interest rate, gross savings, tax rate, and tariff rate should be looked at when trying to achieve financial integration with leading markets. For policymakers of leading markets, they should pay more attention on energy factors, gross national income, stock traded value, and high-technology exports in order to improve stock market integration. For investors, diversifying

100

M.-P. Chen et al. / North American Journal of Economics and Finance 29 (2014) 84–103

among frontier and leading markets presents arbitrage opportunities. Investors taking into account the above-mentioned variables should be able to benefit from international investment. To conclude, the frontier markets follow a common path of growth and different integration with different markets. The linkages between frontier and leading markets are anticipated to strengthen in the future. A future study can take transaction cost and the investment risks of frontier markets into further consideration. The International Monetary Fund (IMF) World Economic Outlook Database (2010) mentions immature markets offer upside potentials. Investing in frontier markets – that are smaller, less liquid, and has more investment restrictions than emerging markets does, indeed entail risks beyond those encountered in emerging markets. In addition, future research could analyze the optimal Logit model developed by Acosta-González and Fernández-Rodríguez (2014), or a probit model in Fernandez-Perez, Fernández-Rodríguez, and Sosvilla-Rivero (2014) to identify most important variables. Acknowledgement We would like to thank the Editor Hamid Beladi and two anonymous referees for their highly constructive comments. Appendix. See Table A1

Table A1 Descriptions of the variables. Variable Economic policy FDI

GDP

GNI

GRSAV INDUST INFLAT

RESERVE

Energy ENGPOD

ENGIMP ENGUSE

Description Foreign direct investment (net inflows, U.S.$) are the net inflows of investment to acquire a lasting management interest in an enterprise operating in an economy other than that of the investor. It is the sum of equity capital, reinvestment of earnings, other long-term capital, and short-term capital. GDP per capita (U.S.$) is gross domestic product divided by mid-year population. GDP is the sum of gross value added by all resident producers in the economy plus any product taxes and minus any subsidies not included in the value of the products. GNI per capita based on purchasing power parity (PPP). PPP GNI is gross national income converted to international dollars using purchasing power parity rates. GNI is the sum of value added by all resident producers plus any product taxes (less subsidies). Gross savings (% of GDP) are calculated as gross national income less total consumption, plus net transfers. Industry, value added (% of GDP) It comprises value added in mining, manufacturing (also reported as a separate subgroup), construction, electricity, water, and gas. Inflation as measured by the consumer price index that reflects the annual percentage change in the cost to the average consumer of acquiring a basket of goods and services that may be fixed or changed at specified intervals, such as yearly. Total reserves (U.S.$) comprise holdings of monetary gold, special drawing rights, reserves of IMF members held by the IMF, and holdings of foreign exchange under the control of monetary authorities. Energy production (kt of oil equivalent) refers to forms of primary energy – petroleum (crude oil, natural gas liquids, and oil from non-conventional sources), natural gas, solid fuels (coal, lignite, and other derived fuels), and combustible renewables and waste – and primary electricity, all converted into oil equivalents. Energy import, net (% of energy use). Energy use (kt of oil equivalent) refers to use of primary energy before transformation to other end-use fuels, which is equal to indigenous production plus imports and stock changes, minus exports and fuels supplied to ships and aircraft engaged in international transport.

M.-P. Chen et al. / North American Journal of Economics and Finance 29 (2014) 84–103

101

Table A1 (Continued )

Variable Financial sector LISTCO

MKTCAP

INTST TRAD

TADTUR LEGAL

Private sector DOBUS

EXPVI IMPVI TAX

TARIFF

TOUR

Other factors HITECH LAND POP POPGR Control variable DIST LIB

Description Listed domestic companies are domestically incorporated companies listed on the country’s stock exchanges at the end of the year. This indicator does not include investment companies, mutual funds, or other collective investment vehicles. Market capitalization of listed companies (U.S.$) is the share price times the number of shares outstanding. Listed domestic companies are domestically incorporated companies listed on the country’s stock exchanges at the end of the year. Real interest rate is the lending interest rate adjusted for inflation as measured by the GDP deflator. Stocks traded, total value (% of GDP) refers to the total value of shares traded during the period. This indicator complements the market capitalization ratio by showing whether market size is matched by trading. Stocks traded, turnover ratio (%) is the total value of shares traded during the period divided by the average market capitalization for the period. Strength of legal rights index measures the degree to which collateral and bankruptcy laws protect the rights of borrowers and lenders and thus facilitate lending. The index ranges from 0 to 10, with higher scores indicating that these laws are better designed to expand access to credit. Ease of doing business ranks economies from 1 to 183, with first place being the best. A high ranking (a low numerical rank) means that the regulatory environment is conducive to business operation. Export value index is the current value of exports (f.o.b.) converted to U.S. dollars and expressed as a percentage of the average for the base period (2000). Import value indices are the current value of imports (c.i.f.) converted to U.S. dollars and expressed as a percentage of the average for the base period (2000). Total tax rate (the percentage of commercial profits) measures the amount of taxes and mandatory contributions payable by businesses after accounting for allowable deductions and exemptions as a share of commercial profits. Tariff rate, most favored nation, simple mean, primary products (%). Simple mean most favored nation tariff rate is the unweighted average of most favored nation rates for all products subject to tariffs calculated for all traded goods. International tourism receipts (current U.S.$) are expenditures by international inbound visitors, including payments to national carriers for international transport. These receipts include any other prepayment made for goods or services received in the destination country. High-technology exports are products with high R&D intensity, such as in aerospace, computers, pharmaceuticals, scientific instruments, and electrical machinery. Data are in current U.S. dollars. Land area (sq. km) is a country’s total area, excluding area under inland water bodies, national claims to continental shelf, and exclusive economic zones. Population, total refers to the total population. Population growth (annual %) is the exponential rate of growth of mid-year population from year t − 1 to t, expressed as a percentage. Distance is the great circle distance between principal cities. Liberalization is the official liberalization dates.

Note: All data are employed at the annual frequency. The data source of DIST is from CEPII database http://www.cepii.fr/ anglaisgraph/bdd/distances.htm. The data source of LIB is from Bekaert and Harvey (2000). The rest of the data is from World Bank database.

References Abad, P., Chuliá, H., & Gómez-Puig, M. (2010). EMU and government bond market integration. Journal of Banking and Finance, 34, 2851–2860. Acosta-González, E., & Fernández-Rodríguez, F. (2014). Forecasting financial failure of firms via genetic algorithms. Computational Economics, 43, 133–157. Adjasi, C. K. D., Osei, K. A., & Fiawoyife, E. U. (2011). Explaining underpricing of IPOs in frontier markets: Evidence from the Nigeria stock exchange. Research in International Business and Finance, 25, 255–265. Bekaert, G., & Harvey, C. R. (2000). Foreign speculators and emerging equity markets and economic development. Journal of Development Economics, 66, 465–504.

102

M.-P. Chen et al. / North American Journal of Economics and Finance 29 (2014) 84–103

Bekaert, G., Harvey, C. R., & Lundblad, C. (2001). Does financial liberalization spur growth? Cambridge, MA: National Bureau of Economic Research. Working paper 8245 Berger, D., Pukthuanthong, K., & Yang, J. J. (2011). International diversification with frontier markets. Journal of Financial Economics, 101, 227–242. Bhatti, M. I., & Nguyen, C. C. (2012). Diversification evidence from international equity markets using extreme values and stochastic copulas, Journal of International Financial Markets. Institutions and Money, 22, 622–646. Boubaker, A., Nguyen, D. K., & Taouni, I. (2009). Rational speculative bubbles: Theory and empirics in a frontier emerging markets. ICFAI Journal of Applied Finance, 15, 49–61. Carrieri, F., Errunza, V., & Hogan, K. (2007). Characterizing world market integration through time. Journal of Financial and Quantitative Analysis, 42, 915–940. Carrieri, F., Errunza, V., & Sarkissian, S. (2004). Industry risk and market integration. Management Science, 50, 207–221. Chen, J., Buckland, R., & Williams, J. (2011). Regulatory changes, market integration and spillover effects in the Chinese A, B and Hong Kong equity markets. Pacific-Basin Finance Journal, 19, 351–373. De Groot, W., Pang, J., & Swinkels, L. (2012). The cross-section of stock returns in frontier emerging markets. Journal of Empirical Finance, 19, 796–818. Demian, C. V. (2011). Cointegration in Central and East European markets in light of EU accession. Journal of International Financial Markets, Institutions and Money, 21, 144–155. Ding, D., & Jinjarak, Y. (2012). Development threshold, capital flows, and financial turbulence. North American Journal of Economics and Finance, 23, 365–385. Driessen, J., & Laeven, L. (2007). International portfolio diversification benefits: Cross-country evidence from a local perspective. Journal of Banking and Finance, 31, 1693–1712. Engle, R. F., & Granger, C. W. J. (1987). Co-integration and error correction: Representation, estimation, and testing. Econometrica, 55, 251–276. Fernandez-Perez, A., Fernández-Rodríguez, F., & Sosvilla-Rivero, S. (2014). The term structure of interest rates as predictor of stock returns: Evidence for the IBEX 35 during a bear market. International Review of Economics and Finance, 31, 32–33. Goetzmann, W. N., Li, L., & Rouwenhorst, K. G. (2005). Long-term global market correlations. Journal of Business, 78, 1–38. Gupta, R., & Guidi, F. (2012). Cointegration relationship and time varying co-movements among Indian and Asian developed stock markets. International Review of Financial Analysis, 21, 10–22. Henry, P. B. (2000). Do stock market liberalization cause investment booms? Journal of Financial Economics, 58, 301–334. Herber, P., & Rowader, N. (2011). The case for frontier markets now. Forward management. http://www.forwardfunds.com/pdfs/ market insights/forward whitepaper 12 15 2010.pdf Hsieh, M. F., Chen, P. F., Lee, C. C., & Yang, S. J. (2013). How does diversification impact bank stability? The role of globalization, regulations, and governance environments. Asia-Pacific Journal of Financial Studies, 42(5), 813–844. Jarque, C. M., & Bera, A. K. (1980). Efficient test for normality, homoskedasticity and serial dependence of regression residuals. Economical Letters, 6, 255–259. Koulakiotis, A., Kartalis, N., & Papasyriopoulos, N. (2012). Asymmetric and threshold effects on comovements among Germanic cross-listed equities. International Review of Economics and Finance, 24, 327–342. Lahrech, A., & Sylwester, K. (2013). The impact of NAFTA on North American stock market linkages. North American Journal of Economics and Finance, 25, 94–108. Lane, P., & Milesi-Ferretti, G. M. (2003). International financial integration. International Macroeconomics. Discussion paper series No. 3769. Laopodis, N. T. (2011). Equity prices and macroeconomic fundamentals: International evidence. Journal of International Financial Markets, Institutions and Money, 21, 247–276. Lee, C. C. (2013). Insurance and real output: The key role of banking activities. Macroeconomic Dynamics, 17, 235–260. Lee, C. C., Chen, M. P., & Chang, C. H. (2013). Dynamic relationships between industry returns and stock market returns. North American Journal of Economics and Finance, 26, 119–144. Lee, C. C., & Hsieh, M. F. (2014). Bank reforms, foreign ownership, and financial stability,. Journal of International Money and Finance, 40, 204–224. Levine, R. (2001). International financial liberalization and economic growth. Review of International Economics, 9, 688–702. Lin, C. H. (2012). The comovement between exchange rates and stock prices in the Asian emerging markets. International Review of Economics and Finance, 22, 161–172. Lucey, B. M., & Muckley, C. (2011). Robust global stock market interdependencies. International Review of Financial Analysis, 20, 215–224. Marshall, B. R., Nguyen, N. H., & Visaltanachoti, N. (2011). Frontier market diversification and transaction costs. Massey University. Working paper. Miles, W. (2005). Do frontier equity markets exhibit common trends and still provide diversification opportunities? International Economic Journal, 19, 473–482. Moss, T., Ramachandran, V., & Standley, S. (2007). Why doesn’t Africa get more equity investment? Frontier stock markets, firm size and asset allocations of global emerging market funds. Center for Global Development. Working Paper. Pajuste, A., Kepitis, G., & Högfeldt, P. (2000). Risk factors and predictability of stock returns in Central and Eastern Europe. Emerging Markets Quarterly, 4, 7–25. Panton, D., Lessig, V., & Joy, O. (1976). Comovements of international equity markets: A taxonomic approach. Journal of Financial and Quantitative Analysis, 11, 415–432. Rua, A., & Nunes, L. (2009). International comovement of stock market returns: A wavelet analysis. Journal of Empirical Finance, 16, 632–639. Samarakoon, L. P. (2011). Stock market interdependence, contagion, and the US financial crisis: The case of emerging and frontier markets,. Journal of International Financial Markets, Institutions, and Money, 21, 724–742. Schwarz, G. (1978). Estimating the dimension of a model. Annals of Statistics, 6, 461–464. Speidell, L. S., & Krohne, A. (2007). The case for frontier markets. Journal of Investing, 16, 12–22.

M.-P. Chen et al. / North American Journal of Economics and Finance 29 (2014) 84–103

103

Speidell, L. S. (2011). Frontier market equity investment: Finding the winners of the future Frontier Market Asset Management. Research Foundation of CFA Institute. Syriopoulos, T. (2011). Financial integration and portfolio investments to emerging Balkan equity markets. Journal of Multinational Financial Management, 21, 40–54. Todorov, G., & Bidarkota, P. (2011). Time-varying financial spillovers from the US to frontier markets. Working paper. www.ssrn.com Wang, A. T., Yang, S. Y., & Yang, N. T. (2013). Information transmission between sovereign debt CDS and other financial factors – The case of Latin America. North American Journal of Economics and Finance, 26, 586–601. Wang, X., Yao, L. J., & Fang, V. (2013). Stock prices and the location of trade: Evidence from China-backed ADRs. North American Journal of Economics and Finance, 26, 677–688. Watson, T. (2010). Frontier markets equity. Towerswatson.com. White, H. A. (1980). Heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica, 48, 817–838. Yu, I., Fung, K. P., & Tam, C. S. (2010). Assessing financial market integration in Asia-equity markets. Journal of Banking and Finance, 344, 2874–2885. Zhang, B., Li, X., & Yu, H. (2013). Has recent financial crisis changed permanently the correlations between BRICS and developed stock markets? North American Journal of Economics and Finance, 26, 725–738.