Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

Contents lists available at ScienceDirect

Journal of International Financial Markets, Institutions & Money j ou rn al ho me pa ge : w w w . e l s e v i e r . c o m / l o c a t e / i n t f i n

Is there heterogeneity in financial integration dynamics? Evidence from country and industry emerging market equity indexes夽 Michael Donadelli a, Antonio Paradiso b,∗ a b

Research Center SAFE, Goethe University Frankfurt, Grüneburgplatz 1, 60323 Frankfurt am Main, Germany Department of Economics at Ca’ Foscari University of Venice, Cannaregio 873, 30131 Venice, Italy

a r t i c l e

i n f o

Article history: Received 23 April 2014 Accepted 6 June 2014 Available online 17 June 2014 JEL classification: F15 F44 G15 Keywords: Financial integration PCA Industries Systemic banking crises

a b s t r a c t This paper examines the dynamics of the financial integration process across equity markets in one global emerging region (Emerging) and three emerging sub-regions (Asia, Eastern Europe, Latin America) over the last two decades. The proportion of total variation in individual excess returns explained by the first principal component serves as a robust measure of integration. Financial integration is measured in the “national equity market” (market) and in ten different “industrial equity markets” (basic materials, consumer goods, consumer services, financials, healthcare, industrials, oil and gas, telecommunications, technology and utilities). We obtain two main results. First, we observe that the level of integration across emerging equity markets in emerging regions is rather low, both at the country and industry level. Second, the shape of the financial integration process is not homogeneous among different industries. Specifically, J-shaped, U-shaped and increasing trends are observed. Overall, our integration numbers and dynamics simultaneously improve portfolio diversification benefits and reduce risk-sharing opportunities. This is supported by a CAPMbased analysis. © 2014 Elsevier B.V. All rights reserved.

夽 We thank an anonymous referee, Fulvio Corsi and Marcella Lucchetta for helpful discussions and suggestions. All remaining errors are our own. ∗ Corresponding author. Tel.: +39 0412349161.

E-mail addresses: [email protected] (M. Donadelli), [email protected], anto [email protected] (A. Paradiso). http://dx.doi.org/10.1016/j.intfin.2014.06.003 1042-4431/© 2014 Elsevier B.V. All rights reserved.

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

185

1. Introduction Global capital markets are becoming increasingly integrated. Investment and financing decisions by both institutional and private investors are influenced by the perceived degree of integration across international capital markets. Therefore, financial integration has received an enormous amount of attention in the literature, much of it devoted to examining its asset allocation implications. It is popularly known that adding foreign financial assets into a domestic portfolio might allow investors to reduce the overall level of risk, as the domestic and foreign equity market returns tend to drift away from each other. This very simple concept has been discussed in early studies focusing on international portfolio diversification (see Grubel, 1968; Levy and Sarnat, 1970; Lessard, 1976, among many others). The general idea is that in presence of a low correlation between foreign and domestic equity market returns, an investor could smooth portfolio risk without reducing portfolio expected return by adding foreign stocks in her/his domestic portfolio. However, a relatively high degree of financial integration might produce a significantly drop in cross-border portfolio diversification benefits. This drop tends to be stronger in recession times (i.e. in periods in which international equity market returns are strongly correlated). For example, the last 2001 (post-dotcom crisis) and 2008 (sub-prime crisis) recessions were characterized by an unprecedented degree of international synchronization as all major industrialized countries experienced large contractions in equity market prices around the dates of 9/11 terrorist attacks and Lehman bankruptcy. Differently, equity market prices in some emerging countries as well as in specific emerging industries were less affected by these recessions. Most likely, these equity markets are less integrated than others and do not necessarily follow global price indexes (e.g. S&P 500. MSCI World). Therefore, international investors found emerging equity markets more profitable – in terms of risk-return performance – than advanced equity markets. Consequently, international investors re-balanced their portfolios by taking long positions on emerging stocks (see Fig. 1). Nevertheless, from a practical point of view, most of the investments in emerging market equity rely on shares in emerging aggregate regional indexes (e.g. Emerging, Africa, Asia, Eastern Europe, Middle East, Latin America, etc.),1 which implicitly assume that price indexes belonging to the same region might follow similar paths as well as exhibit similar risk-return profiles. While these regional aggregate indexes are very liquid, they do not necessarily provide strong diversification benefits, especially during crisis-periods. This because some country equity indexes embodied in the global emerging aggregate equity index are strongly correlated with other global equity price indexes (e.g. G7 equity price index). However, country equity price indexes within a region might exhibit different patterns (i.e. different integration dynamics). Moreover, industries within a country might also display different dynamics. This might be due to the different ways in which financial shocks affect equity prices in different countries. We argue that the systemic banking crises of the late 1990s and early 2000s did not affect country and industry emerging equity price indexes homogeneously. In this paper, we examine the evolution of the financial integration process in one global emerging region (Emerging) and three emerging sub-regions (Asia, Eastern Europe, Latin America). A dynamic principal component analysis (PCA) is carried out. Following Volosovych (2011, 2013), the percentage of variance explained by the first principal component serves as robust measure of integration, and thus, represents our integration index. Differently from existing studies, which focus exclusively on national equity markets, and thus, employ only aggregate equity indexes (i.e. country equity indexes), this study employs both country and industry equity indexes (see Fig. A.1). First, we employ country equity indexes (Market) to construct the dynamics of the financial integration process across “national equity markets”. Second, following the level 2 classification of Datastream Global Equity Indexes (DGEI), we divide each country equity index in ten different industry equity indexes (basic materials, consumer goods, consumer services, financials, industrials, healthcare, oil and gas, telecommunications, technology, utilities). Thus, we measure integration in these ten “industrial equity markets”. Overall, we analyze integration dynamics in 11 different markets. The ultimate goal is to examine whether there is heterogeneity in the average level of integration as well as in the dynamics of the financial integration process among regions and industries. We stress that such heterogeneity might improve portfolio diversification benefits confirming that 1

E.g. “iShares MSCI Emerging Markets”, “LYXOR UCITS EFT MSCI EMERGING MARKETS”.

186

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

Fig. 1. Stocks traded: industrialized (G7) vs. emerging economies. Notes: stocks traded 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. Panel (a) reports the evolution of the stocks traded (measured in current US$ and as % of world GDP) series across G7 equity markets. Panel (b) reports the evolution of the stocks traded (measured as % of world GDP) series across emerging equity markets. Panel (c) reports the rate of change of the stocks traded series (measured in current US$) both across G7 (solid line) and emerging (dashed line) equity markets. Sample: 1988–2012. Source: World Development Indicators (World Bank).

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

187

country emerging equity indexes within a region do not exhibit similar patterns.2 It turns out that an aggregate regional index might smooth the opportunity to diversify risk. As mentioned above, we focus also on industrial equity markets. The general idea is that equity returns generated by different industries do not necessarily co-move with the country’s equity market index return. As a consequence, the national (market) and industrial equity markets (e.g. basic materials, financials) in each region might have different integration dynamics. This analysis allows us to examine whether some industries are more integrated than others as well as to capture which industry contributed most to shape integration in the national equity markets. Of course, it also allows international investors, and in particular those who are interested in taking position in specific sectors (e.g. Natural Resources&Power, Telecommunications, Media & Technology), to have a better understanding of country- and industry-specific dynamics, and thus, to take positions on industry equity indexes in specific countries rather than exclusively on aggregate emerging aggregate indexes.3 Since the PCA analysis contains information only on the co-movements between equity markets, a static CAPM analysis is carried out to study if there is heterogeneity also in the order of magnitude of the co-movement. While the CAPM R-squared almost confirms our PCA results (i.e. it captures whether country and industry equity indexes follow the aggregate regional index), the estimated betas tell us whether there is a one-to-one relationship (i.e. information on the order of magnitude of the relationship) between specific country/industry equity returns and the return of a related benchmark (e.g. regional aggregate index return). Our main results are as follows. First, we observe that the average level of integration in the global region (Emerging) is lower than in the sub-regions (Asia, Eastern Europe and Latin America).4 However, country equity indexes belonging to the same sub-region do not necessarily follow identical dynamics. Second, we find that the “contagion effect” and the “hearding effect” produced by systemic banking crises (e.g. Asian banking crises) do not affect country and industry equity indexes homogeneously.5 This gives rise to three different integration index patterns: (i) increasing-trend; (ii) J-shaped trend; (iii) U-shaped trend. Overall, our results suggest that the level of integration across equity markets in emerging regions is rather low (i.e. far from one), both at the country and industry level. This simultaneously increases intra-area portfolio diversification benefits and lowers risk-sharing opportunities. Similar conclusions can be drawn from the numbers obtained via a static CAPM analysis. In this simple setting, we observe that the absolute value of 1-ˇ is rarely close to zero, both at the country and industry level. In other words, there is no a one-to-one relationship between country (industry) excess returns and the related country (industry) regional aggregate portfolios. Therefore, both cross-country or cross-industry diversification benefits might be exploited. This result holds across regions. The rest of the paper is organized as follows. Section 2 briefly motivates our study. Section 3 describes data. Section 4 describes the methodology and examines financial integration dynamics in national and industrial equity markets. Section 5, via a standard CAPM, examines whether crosscountry and cross-industry diversification benefits might be effectively exploited. Section 6 concludes. 2. Motivation Emerging equity markets tend to compensate international investors with a higher risk-return performance than advanced equity markets.6 Entries in Table 1 show that the average annualized return in emerging markets is significantly higher than in G7 markets. We stress that this performancegap holds across periods (including the subprime crisis era). Not surprisingly, emerging stocks have received an enormous amount of attention among international investors. This is clear from Fig. 1 which reports the evolution of the stocks traded series (in US$ and as % of world GDP) in the G7 (Panel a) and Emerging (Panel b) countries. Panel (c) reports the

2 For example, the degree of co-movement between Argentina and Brazil might be higher than the one between Argentina and Philippines, even if they are all classified as emerging markets. Moreover, Argentina and Brazil might weakly co-move with Colombia and Peru’, even if they all belong to Latin America. 3 Notice that this represents a common investment strategy in the hedge funds industry. 4 Throughout the paper we use the terms “Global” and “Emerging” interchangeably. 5 For a detailed discussions on the contagion- and hearding effects, see Chiang et al. (2007). 6 See Grootveld and Salomons (2003), Donadelli and Prosperi (2012), Donadelli (2013), among many others.

188

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

Table 1 Regional equity indexes (G7 vs. emerging): average returns (%). Notes: data are annual and expressed in US$ dollars. Equity market

1992–2012

2000–2012

2005–2012

G7 EM Gap (EM-G7)

5.847 11.801 5.954

1.281 12.033 10.752

3.614 15.885 12.270

Source: MCSI (http://www.msci.com/products/indices/performance.html).

rate of change of the stocks traded series (measured in current US$) both across G7 (solid line) and emerging (dashed line) equity markets. We stress that the total value of shares traded in emerging countries is increasing over time. In particular, it sharply increased in the mid-2000s.7 In 2009, we observe simultaneously a significant drop (around 27%) in the trading activity in G7 economies and a large increase (around 47%) in emerging economies (see red vertical bars in Fig. 1). This suggests that equity inflows to emerging economies increased over the subprime crisis period. A similar capital flows took place in 2003 where the stocks traded in the G7 area decreased by 35% and increased by 38% in the emerging world. Overall, the dynamics of the trading activity in G7 and emerging equity markets suggests the following: (i) emerging countries retain less investment restrictions than in the past; (ii) international investors seek for alternative investments during recession periods (during periods of high uncertainty); (iii) economic agents dislike uncertainty on future consumption plans, that is, they are willing to buy assets that payoff in bad states of nature. In other words, emerging markets might improve international portfolio diversification benefits. However, from a practical point of view, long positions on emerging stock markets rely on regional aggregated indexes rather than on “ad hoc” country/industry equity indexes (e.g. ETFs: iShares MSCI Emerging Markets, LYXOR UCITS EFT MSCI EMERGING MARKETS).8 Aggregate emerging equity indexes, such as the ETFs, are widely used among international investors for two main reasons. First, compared to G7 economies, they provide higher risk-return performance. Second, they partially offset country-specific risk and liquidity risk.9 However, an aggregate index implicitly assumes that country equity indexes belonging to the emerging world exhibit similar dynamics. In addition, it does not say anything about industries’ dynamics (i.e. stocks of firms belonging to a specific sector). It might be possible that some industries display different levels of integration as well as different integration dynamics. Therefore, national and industrial equity markets might display different integration dynamics within a region. This paper is aimed at capturing such “integration-heterogeneity”. First, we investigate whether the global emerging region (Emerging) and the three emerging sub-regions (i.e. Asia, Eastern Europe, Latin America) have similar integration dynamics by focusing exclusively on national equity markets. Second, we examine integration dynamics at the industry level. This allows us to identify the sector that has contributed the most to shaping the dynamics of the financial integration index in the national equity markets. Thus, our paper examines deeply the dynamics of the financial integration process in emerging regions at the country and industry level. As robust measure of integration, we use the percentage of variance in equity excess returns explained by the first principal component. To get a dynamic integration index, we perform the PCA in a rolling window context. This integration measure

7 As recently documented, this sharp increase is associated with a massive increase in international trade (see Donadelli, 2013). Notice also that China joined the WTO. 8 iShares MSCI Emerging Markets UCITS ETF (Inc) is an ETF aimed at mimicking the performance of the MSCI Emerging Markets Index. In general, the ETF iShares allow to invest in a wide range of equity markets and asset classes. The MSCI Emerging Markets Index is composed by more 800 stocks listed in the following worldwide exchanges: Argentina, Brasil, Chile, China, Colombia, Czech Republic, Egypt, Hungary, India, Indonesia, Israel, Jordan, South Korea, Malaysia, Mexico, Morocco, Pakistan, Peru, Philippines, Poland, Russia, South Africa, Taiwan, Thailand, Turkey e Venezuela (Source: Borsa Italiana/London Stock Exchange, ISIN: IE00B0M63177). The LYXOR UCITS EFT MSCI EMERGING MARKETS is issued by “LYXOR INTERNATIONAL ASSET MANAGEMENT S.A.” This ETF serves a similar purpose and is also composed by more than 800 stocks belonging to different emerging stock markets (ISIN: FR0010429068). 9 See Bekaert et al. (2007), and Donadelli and Prosperi (2012) for a detailed analysis on the level of liquidity in emerging equity markets.

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

189

Table 2 Emerging regions. Emerging

Asia

East. Europe

Latin America

Argentina Brazil Chile China Colombia Hungary India Malaysia Mexico Pakistan Peru Philippines Poland Russia South Africa Sri Lanka Thailand Turkey

China India Malaysia Pakistan Philippines Sri Lanka Thailand

Hungary Poland Russia Turkey

Argentina Brazil Chile Colombia Mexico Peru

has been recently introduced by Volosovych (2011, 2013), and used to study the dynamics of capital market integration across industrialized economies. As discussed by Volosovych (2011, 2013), the choice of using this dynamic PCA approach is motivated by several factors. The most important is that it is weakly affected by outliers and breaks in the series. 3. Data description Industry equity indexes are from DGEI, a Thomson Datastream database which provides a range of equity indexes across 53 countries, 32 regions and 170 sectors. Stocks are classified according to the “FTSE-Dow Jones Industry Classification Benchmark (IBC)”. The classification consists of six levels of hierarchical structure. Level 1 is the market index. This covers all the sectors in each region or country. Level 2 divides the market into 10 industries and covers all the sectors within each group in each region or country. The subsequent levels subdivide the level 2 classifications into sector classifications in increasing detail. This paper focuses on level 2. In practice, we employ the following ten sector equity indexes: Basic Materials, Consumer Goods, Consumer Services, Financials, Industrials, Oil& Gas, Technology, Telecommunications and Utilities. Country equity indexes (Market) are represented by the Morgan Stanley Capital International (MSCI) indexes. The equity market structure is shown in Fig. A.1. We use sector and country equity indexes for the following emerging market and developing economies: Argentina, Brazil, Chile, China, Colombia, Hungary, India, Malaysia, Mexico, Pakistan, Peru, Philippines, Poland, Russia, South Africa, Sri Lanka, Thailand and Turkey.10 In order to measure financial integration in different regions, we divide our sample in four emerging regions: “Emerging”, which includes all emerging and developing countries in our sample; “Asia”, which includes China, India, Malaysia, Pakistan, Philippines, Sri Lanka, Thailand; “Eastern Europe”, which includes Hungary, Poland, Russia and Turkey; “Latin America”, which consists of the largest equity markets in the region, Argentina, Brazil, Chile, Colombia, Mexico and Peru. Emerging and developing economies, as classified by the IMF, are listed in Table 2.

10 Emerging and developed markets follow the IMF country groups classification. The number of emerging countries selected in this paper is in line with the maximum number of available US$-based sector total return indexes from Thomson Datastream. Our 19 economies are classified as emerging markets also according to the International Finance Corporation country classification. In other words, these countries are characterized by one of the following two features; (i) they belong to low- or middle-income economic region; (ii) their investable market capitalization/GDP ratio is low.

190

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

Both country and industry indexes are total return indexes (i.e. reinvested dividends are included) denominated in U.S. dollars. It is popularly known that local currency indexes produce the purest form of the excess return. However, this implies less interesting results for international investors (Grootveld and Salomons, 2003). Therefore, as standard in the literature, we consider the point of view of a U.S. investor (Bilson et al., 2001; Grootveld and Salomons, 2003; de Jong and de Roon, 2005; Chambet and Gibson, 2008; Pukthuanthong and Roll, 2009; Donadelli and Prosperi, 2012; Donadelli, 2013; Lee et al., 2013; Donadelli and Persha, 2014, among others).11 MSCI data run from January 1994 (or later) to December 2011. DGEI run from January 1994 (or later) to July (2012). Therefore, our analysis relies on a post-liberalizations era.12 Industry (ExRet ci,t ) and country equity (ExRets,t ) excess

returns are computed as follows:13



ExRet ci,t =

DGEI ct

DGEI ct−1



−1

− Rf,t ;

ExRet c,t =

 MSCI t MSCI t−1



− 1 − Rf,t

(3.1)

where Rf,t is the one-month T-bill rate (from Ibbotson Associates), DGEI ct represents the sector equity index at time t in country c, and MSCIt is the country equity index at time t. Summary statistics are reported in Tables A.1 (for country equity excess returns) and A.2 (for sector equity excess returns). In columns 2–5, we report the fourth moments and the Sharpe ratio. The last column identifies structural breaks in the series. Breaks are identified via the Bai–Perron test (details are given in Appendix B). 4. Measuring integration across emerging markets 4.1. Methodology: a review We measure integration across emerging equity markets via a standard PCA. The PCA is a technique aimed at reducing the dimension of a dataset by finding a new set of variables that retains most of the sample’s information. In other words, via the PCA we describe common features of the data. Let X be a vector of p variables x, x1 , x2 , . . ., xp . A linear combination of these variables can be represented as follows Z = ˝ · X    p×1

p×p

(4.1)

p×1

where the first row in Eq. (4.1) takes the form z1 = ˝1 X = ω11 x1 + ω12 x2 + · · · + ω1p xp . The coefficients ω11 , ω12 , . . . ω1p are called loadings, and ˝ in (4.1) represents the loading matrix. Coefficients in z1 are computed so as to have the maximum sample variance of z1 . In this setup, the variance of z1 can be artificially increased by picking larger coefficient values ω1i . All the subsequent components in Z are orthogonal to the previous one (e.g. ω 1 ω2 = 0, ω 2 ω3 = 0, and so forth). In other words, the second principal component is uncorrelated with the first principal component and also has the maximum variance. Hence, z2 explains maximum of the residual variation once z1 is removed from the datamatrix. Specifically, ωki coefficients represent elements of the eigenvector of the sample covariance matrix X corresponding to the kth largest eigenvalue. The PCs can be extracted using both the covariance and correlation matrix. While the former tends to give more weight to the variable with the largest variance, the latter gives each variable an equal weighting in the data matrix (i.e. variables are standardized), independent of their variance, and avoids loading on those variables with the largest standard deviation. In general correlations are not required when variables have the same unit. However, in practice, since “high variance variables” tend to dominate, we perform the PCA on standardized variables.14 The number of papers using the PCs to measure

11 The conversion of local currency equity indexes represents an ubiquitous practice in international finance studies. Returns denominated in this form only retain U.S. inflation, alleviate exchange rate noise, and are consistent across markets. 12 See Henry (2000a,b) and Bekaert et al. (2003) for a detailed discussion on equity market liberalizations in emerging economies. 13 Throughout the paper we use the terms “Industry” and “Sector” interchangeably. 14 For a detailed discussion on the PCA, see Jolliffe (2002).

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

191

Table 3 Systemic banking crisis (SBC) in emerging countries from 1980: Notes: This table reports systemic banking crises dates for each country in our sample. All SBC dates are from Laeven and Valencia (2012). Latin America

SBC

Asia

SBC

Eastern Europe

SBC

Argentina

1980–1982 1989–1991 1995 2001–2003 1990–1994 1994–1998 1981–1985 1982 1998–2000 1981–1985 1994–1996 1983

China India Malaysia Philippines

1998 1993 1997–1999 1983–1986 1997–2001 1989–1991 1983 1997–2000

Hungary

1991–1995 2008–2009 1992–1994 1998 1982–1984 2000–2001

Brazil Chile Colombia Mexico Peru

Sri Lanka Thailand

Poland Russia Turkey

financial integration is limited.15 In the spirit of Volosovych (2011), we measure financial integration across emerging economies by using the percentage of variation in emerging equity market sector and country excess returns explained by the first principal components. This procedure gives rise to a robust measure of integration and solves most of the issues raised by the international finance literature on the use of the correlation-based measures as integration proxies. The first principal component is extracted in a rolling window framework. In practice, the integration index is estimated using a rolling sample of 60 months (5 years). 4.2. Results 4.2.1. Financial crisis vs. integration index Several works have observed that a large increase in the degree of co-movement between international equity market returns as well as in returns volatility is due to contagion effect caused by financial shocks hitting an economy (see Sachs et al., 1996; Corsetti et al., 2005, among many others). Some studies, for example, observe a significant increase in correlation coefficients during the Asian crisis (Baig and Goldfajn, 1999) and Russian crisis (Saleem, 2009; Chkili and Nguyen, 2014). Chiang et al. (2007) also observe that systemic financial crises having international effects define two specific phases. The first phase is characterized by a massive increase in the degree of co-movement between international equity market returns during the crisis (contagion effect). The second one relies on the relatively high cross-country returns correlation observed in the periods following the shock (hearding effect). Table 3 reports systemic banking crisis dates for the period 1980–2012. Entries in Table 3 suggest that most of the crises occurred in the 1990s and early 2000s. Based on dates in Table 3, the nature of the hearding effect and the presence of the 2008–2009 Global Financial Crisis (GFC), we can argue that the 2003–2007 era is “crises-free”. We stress that in this sub-period our integration index will not be contaminated by the contagion and hearding effects. Overall, our sample can be divided in three specific sub-periods. The first period includes pre-2003 observations and is influenced by relevant international financial crises (i.e. Asian and Russian crises). The second period includes 2003–2007 observations and is not contaminated by crises (tranquil times). The third period relies on post-2007 data and includes the recent GFC. We stress that a sub-period analysis allows us to examine whether the presence of banking crises have affected the degree of equity market integration. 4.2.2. Integration across national equity markets Fig. 2 reports the dynamics of the financial integration index – computed by employing country equity indexes – in the four emerging regions: Emerging, Asia, Eastern Europe and Latin America. For

15 An exception is Pukthuanthong and Roll (2009) who extract PCs from a large dataset – composed by country returns – and use them as global risk factors in a multi-factor regression where each country return is regressed on the first ten PCs. The cross-country average adjusted R-squared serves then as robust measure of integration.

192

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218 Eastern Europe - Market

Asia - Market 0.65

0.85

0.6

0.8

0.55 0.75 0.5 0.7 0.45 0.65 0.4 0.6

0.35

0.55

0.3 2000

2002

2004

2006

2008

2010

2000

2012

2002

2004

2006

2008

2010

2012

2010

2012

Emerging - Market

Latin America - Market 0.8

0.7

0.75

0.65

0.7

0.6

0.65

0.55

0.6

0.5

0.55

0.45

0.5

0.4 0.35

0.45 2000

2002

2004

2006

2008

2010

2012

2000

2002

2004

2006

2008

Fig. 2. Integration index: National equity markets. This figure reports the dynamics of the financial integration index across emerging national equity markets in four different emerging regions (Asia, Eastern Europe, Latin America and Emerging). The integration index is represented by the proportion of variance in national equity market excess returns explained by the first principal component. Country equity excess returns are computed as defined in Eq. (3.1). The first principal component is estimated using a rolling window of 60 months. The grey line defines the dynamics of the integration index computed by employing the covariance matrix in the PCA. The black line defines the dynamics of the integration index computed by employing the correlation matrix in the PCA. The smoothed red line – computed using the unobserved component smoother technique explained in Appendix B – represents the trend of the “correlation-based integration index”. Average integration index: Asia = 0.46; Eastern Europe = 0.70; Latin America = 0.62; emerging = 0.49. Shaded vertical bars denote NBER-dated recessions. Sample: January 1994 (or later)–December 2011. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

comparison purposes, we compute the index by employing both the covariance (grey line) and the correlation (black line) matrix in the PCA. However, following the discussion on robustness developed in Section 4.1, we rely exclusively on the integration index computed by employing the correlation matrix in the PCA. The red line in all subplots represents the smoothed estimate of the “correlationbased integration index” using the unobserved component model technique.16 Table 4 provides some summary statistics. Specifically, it reports (i) the average level of integration for the three sub-periods identified in the previous subsection (i.e. pre-2003, 2003–2007, post-2007); (ii) the gap between the post-2007 average integration level and the level of integration measured over the tranquil era (i.e. 2003–2007); (iii) the “jump”, i.e. the difference between the levels of integration measured in January 2010 (post-GFC) and December 2007 (pre-GFC). In addition, it indicates (approximately) the shape of the financial integration process across national equity markets in each region. Based on integration dynamics in Fig. 2 and entries in Table 4, we can draw the following conclusions. The integration index in Asia and Latin America exhibits a J-shaped trend. This is confirmed by entries in Table 4 which show an average level of integration equal to 0.47 (Asia) and 0.61 (Latin America) over the sub-period pre-2003, 0.39 (Asia) and 0.54 (Latin America) over the sub-period

16

Details on the unobserved component model are given in Appendix B.

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

193

Table 4 Emerging regions: average integration index level. Notes: Gap is the difference between the average integration index in the GFC (after 2007) and the average integration in the tranquil times (2003–2007). Jump denotes the difference between the value of integration index observed in the aftermath of the GFC (January 2010) and the value observed before the GFC (Dec 2007). The last line reports the shape of the financial integration process (J = J-shaped trend; ⇑ = increasing trend). Region

Asia

East. Eu.

Latin Am.

Emerg.

Market 1. Avg (pre-2003) 2. Avg (2003–2007) 3. Avg (post-2007) Gap (3–2) Jump (January 10–December 07)

0.46 0.47 0.39 0.55 0.16 0.14

0.70 0.61 0.66 0.81 0.15 0.14

0.62 0.61 0.54 0.74 0.20 0.19

0.49 0.43 0.40 0.63 0.23 0.21

Shape

J

⇑

J

J

2003–2007, and 0.55 (Asia) and 0.74 (Latin America) over the post-GFC era. Differently the integration index in Eastern Europe displays a constantly increasing trend. Over the three sub-periods the integration index is equal to 0.61, 0.66 and 0.81, respectively. In all regions, we observe a big jump in the level of integration between the post-GFC (January 2010) and pre-GFC (December 2007). As expected, financial integration dynamics tend to be heavily influenced by systemic banking crisis. In the pre-2003 sample, the crises in Asia and Latin America have artificially produced a large increase in the degree of co-movement between countries’ equity excess returns, and thus, in the integration index. As previously discussed, in the aftermath of the crisis the hearding effect disappears and our integration index exhibits a “natural trend” up to 2007. Afterwards, the GFC affects cross-country equity market returns correlation and our PCA-based integration indexes increase again. Overall, our regional integration dynamics exhibit some degrees of heterogenetity. While the effect of the crises on equity market dynamics in Asia and Latin American disappear in the aftermath of the 1990s and early 2000s crises, in Eastern Europe the hearding effect last for many years and produces a constantly increasing trend. However, the J-shaped trend in Asia and in Latin American has a relatively large impact on the integration process in the global emerging region, which exhibits a J-shaped trend as well (Fig. 2, bottom-right panel). Our results have important international portfolio diversification implications. In particular, the observed degree of heterogeneity in the dynamics of the integration index suggests that by measuring integration exclusively at the global level (i.e. including all emerging markets), we may loose important information which cannot be exploited to build a wisely diversified equity-based portfolio. Another important aspect in the analysis of the regional financial integration pattern is the presence of a country-specific effect (i.e. idiosyncratic component). It might be the case that some countries do not perfectly match the common regional dynamics. Therefore, idiosyncratic components might affect the shape of the regional integration index pattern. As in Volosovych (2011), we capture countryspecific effects by studying the correlations of equity market excess returns with the first principal components (i.e. correlations of the unobserved “emerging region excess return” with country excess return).17 These correlations are reported in Fig. 3 (Latin America, (Panel a), Eastern Europe, (Panel b), and Asia, (Panel c)) and Fig. 4 (Emerging). Not surprisingly, we observe that the correlation in most Asian and Latin American countries is lower in tranquil times (i.e. 2003–2007) than otherwise (i.e. in pre-2003 and post-2007 sub-periods). Therefore, these correlations seem to be influenced by systemic banking crises. An exception is Pakistan that exhibits a relatively low correlation over almost the full sample. We argue that this might be due to liquidity problems in the Pakistani equity market (see Bekaert et al., 2007). In line with the integration dynamics plotted in Fig. 2, correlations in all Eastern European countries are constantly increasing, suggesting that equity price indexes in this region have been constantly affected by the contagion and hearding effects. Differently from Eastern Europe, some countries in Asia and Latin America seem to be more responsible than others for the J-shaped trend

17 E.g. the correlation between the excess return of the Brazilian national equity market and the first principal component extracted by using equity market excess returns of countries belonging to Latin America.

194

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218 Argentina

Brazil

0.38 0.34 0.32 0.3 0.28 0.26 0.24 0.22 0.2

(a)

Chile

0.385 0.38 0.375 0.37 0.365 0.36 0.355 0.35 0.345 0.34 0.335 0.33

0.36

0.37 0.36 0.35 0.34 0.33 0.32 0.31 0.3 0.29 0.28 0.27

1999 2002 2005 2008 2011

1999 2002 2005 2008 2011

Colombia

Mexico

0.36 0.34 0.32 0.3 0.28 0.26 0.24 0.22 0.2 0.18 0.16 0.14

Peru

0.39 0.38 0.37 0.36 0.35 0.34 0.33 0.32 0.31 0.3 0.29 0.28 1999 2002 2005 2008 2011

1999 2002 2005 2008 2011

0.38 0.36 0.34 0.32 0.3 0.28 0.26 0.24 0.22 0.2 1999 2002 2005 2008 2011

1999 2002 2005 2008 2011

Hungary

Russia

0.48

0.45 0.44 0.43 0.42 0.41 0.4 0.39 0.38 0.37 0.36 0.35

0.47 0.46 0.45 0.44 0.43 0.42 0.41 0.4 0.39 2000

2003

2006

2009

2012

2000

2003

Poland 0.48

2009

2012

2009

2012

0.45 0.44 0.43 0.42 0.41 0.4 0.39 0.38 0.37 0.36 0.35 0.34

0.46 0.44 0.42 0.4 0.38 0.36

(b)

2006

Turkey

0.34 2000

2003

2006

2009

China 0.32 0.3 0.28 0.26 0.24 0.22 0.2 0.18 0.16 0.14

2012

2000

0.34 0.33 0.32 0.31 0.3 0.29 0.28 0.27 0.26 0.25 0.24 1999 2002 2005 2008 2011

0.34 0.32 0.3 0.28 0.26 0.24 0.22 0.2 0.18 0.16 0.14 1999 2002 2005 2008 2011

Philippines 0.34 0.33 0.32 0.31 0.3 0.29 0.28 0.27 0.26 0.25 0.24 0.23

1999 2002 2005 2008 2011

2006

Malaysia

1999 2002 2005 2008 2011

Pakistan 0.3 0.25 0.2 0.15 0.1 0.05 0

2003

India

SriLanka 0.25 0.2 0.15 0.1 0.05 0 -0.05

1999 2002 2005 2008 2011

1999 2002 2005 2008 2011

Thailand

(c)

0.35 0.34 0.33 0.32 0.31 0.3 0.29 1999 2002 2005 2008 2011

Fig. 3. Emerging regions: country equity excess return correlation with the 1st principal component. This figure reports the correlations of country equity excess returns with the 1st principal components. Country equity excess returns are computed as defined in Eq. (3.1). The correlation between an observed excess return and an unobserved 1st principal component is computed as explained in Jolliffe (2002). The 1st principal component is obtained using a rolling window of 60 months. Panel (a): Latin America (Argentina, Brazil, Chile, Colombia, Mexico and Peru). Panel (b): Eastern Europe (Hungary, Poland, Russia and Turkey). Panel (c): Asia (China, India, Malaysia, Pakistan, Philippines, Sri Lanka and Thailand). Sample: January 1994 (or later)–December 2011.

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218 Argentina

Brazil

0.2

Chile

0.22 0.215 0.21 0.205 0.2 0.195 0.19 0.185 0.18

0.18 0.16 0.14 0.12 0.1 0.08 2000

2003

2006

2009

2012

0.21 0.2 0.19 0.18 0.17 0.16 0.15 2000

China

2003

2006

2009

2012

0.14 0.12 0.1 0.08 0.06 2009

2012

2000

India

2003

2006

2009

2012

2000

0.2

0.2

0.22

0.18

0.21

0.16

0.16

0.2

0.14

0.14

0.19

0.12

0.12

0.18

0.1

0.1 2006

2009

2012

2003

2006

2009

2012

2003

2006

2000

Peru

2009

2012

2012

2003

2006

2009

2012

0.22 0.2 0.18 0.16 0.14 0.12 0.1 0.08 2000

2003

2006

Poland

2009

2012

2000

Russia

0.22 0.21 0.2 0.19 0.18 0.17 0.16 0.15 0.14 0.13

2009

Philippines

0.21 0.2 0.19 0.18 0.17 0.16 0.15 0.14 0.13 0.12 2000

2006

0.16 2000

Pakistan 0.15 0.14 0.13 0.12 0.11 0.1 0.09 0.08 0.07 0.06

2012

0.17

0.08 2003

2009

Mexico

0.18

2000

2003

Malaysia

0.08

2006

0.22 0.21 0.2 0.19 0.18 0.17 0.16 0.15 0.14

0.16

2006

2003

Hungary

0.2 0.18

2003

2000

Colombia

0.21 0.2 0.19 0.18 0.17 0.16 0.15 0.14 0.13 0.12 0.11 2000

195

2003

2006

2009

2012

SouthAfrica

0.21

0.22

0.2

0.21

0.19

0.2

0.18 0.19 0.17 0.18

0.16

0.17

0.15 0.14 2000

2003

2006

2009

2012

0.16 2000

SriLanka

2003

2006

2009

2012

0.14 0.12 0.1 0.08 0.06 0.04 0.02 0 -0.02 2003

2006

2003

2009

2012

2006

2009

2012

2009

2012

Turkey

0.205 0.2 0.195 0.19 0.185 0.18 0.175 0.17 0.165 0.16 0.155 0.15 2000

2000

Thailand 0.21 0.2 0.19 0.18 0.17 0.16 0.15 0.14 0.13 0.12 0.11 0.1 2000

2003

2006

2009

2012

2000

2003

2006

Fig. 4. Emerging: country equity excess return correlation with the 1st principal component. This figure reports the correlations of country equity excess returns with the 1st principal components. Country equity excess returns are computed as defined in Eq. (3.1). The correlation between an observed excess return and an unobserved 1st principal component is computed as explained in Jolliffe (2002). The 1st principal component is obtained using a rolling window of 60 months. Employed countries: Argentina, Brazil, Chile, China, Colombia, Hungary, India, Malaysia, Mexico, Pakistan, Peru, Philippines, Poland, Russia, South Africa, Sri Lanka, Thailand ad Turkey. Sample: January 1994 (or later)–December 2011.

196

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

observed in Fig. 2. These are Malaysia, Philippines and Sri-Lanka in Asia and Argentina, Chile and Peru in Latin America. Fig. 4, which reports the correlation between the countries’ equity excess returns and the first principal component extracted from the dataset composed by all emerging country excess returns (i.e. Emerging region), presents analogous dynamics. Specifically, (i) Hungary, Poland, Russia and Turkey exhibit increasing patterns; (ii) Argentina, Chile, Malaysia, Peru, Philippines and Sri Lanka display a marked J-shaped trend (or U-shaped trend). 4.2.3. Integration across industrial equity markets This section examines the dynamics of the financial integration index in the four emerging regions in a industry-by-industry context. As previously discussed, we focus on the following sectors: Consumer Goods, Consumer Services, Financials, Industrials, Basic Materials, Oil&Gas, Telecommunications, Utilities, Healthcare, Technology. Our industry-based analysis is twofold. First, it investigates whether there are differences in the dynamics of the integration index among different sectors in emerging regions. Second, it allows to identify those industries that has contributed the most to shaping the financial integration process in the national equity markets (see Fig. 2). To the best of our knowledge, this is the first study that examines financial integration dynamics in ten different industrial equity markets and in four different emerging regions. Consequently, it is hard to take other studies among the international finance literature for comparison purposes. Financial integration dynamics and integration index statistics are reported in Fig. 5 (Panels (A)–(E)) and Table 5, respectively. We find that the integration index in the global emerging region (Emerging) exhibits a J-shaped trend only in the following sectors: basic materials, technology, and telecommunications. In the other sectors, it exhibits a constantly increasing trend (e.g. consumer services, oil and gas). Differently, across emerging sub-regions, integration dynamics embody a relatively high degree of heterogeneity showing increasing, J-shaped and U-shaped trends.18 Overall, as in existing international finance studies, we find that the degree of equity market integration of equity market integration in emerging economies (in most sectors) has increased sharply in the no-crisis period (i.e. 2003–2007). We report here below a detailed discussion on the nature of the financial integration process in the ten different industries. Consumer goods. This industry (automobiles and parts, food and beverages, personal and households goods) is, on average, less integrated than the others. Over the full period, in the Emerging region, the average integration index is equal to 0.31 (see Table 5). In the crises-period (pre-2003) and in tranquil times (2003–2007) the level of integration is equal to 0.24 and 0.23, respectively. Therefore, we do not observe a significant drop in the aftermath of the 1990s and early 2000s’ crises. This gives rise to an increasing trend, ⇑, (see Fig. 5, Panel A, bottom-left plot). Differently, in Latin America, systemic banking crises heavily affected the consumer goods equity price indexes. In fact, we observe an average integration index equal to 0.42 in the pre-2003 era, 0.31 in the 2003–2007 period, and 0.43 in the post-2007 period. This produces a U-shaped trend. We also find that the level of integration in sub-regions (on average) is higher than in the Emerging region. This suggests that the consumer goods equity indexes of countries belonging to the same area tend to follow similar patterns. In particular, in Eastern Europe the average level of integration is close to 60%, and equal to 35% and 38% in Asia and Latin America, respectively. Because of a relative low level of integration, this sector has marginally contributed to the overall level of financial integration. Consumer services. The dynamics of the integration index in the consumer services industry (retail, media, travel and leisure) is similar to that observed in the consumer goods sector in three out of four regions (see Fig. 5, Panel A). An exception is Asia, where the integration index is very smooth (i.e. constant over time). In fact, it presents a small gap as well as a small jump, 0.06 and 0.03, respectively. Notice also that (as in the consumer goods industry) the integration index in Latin America exhibits a U-shaped trend. In fact, the proportion of variation in individual consumer services equity excess returns (in Latin American countries) explained by the first principal component is equal to 0.51, 0.39 and 0.54 in the pre-2003 period, 2003–2007 period and post-2007 period, respectively.

18 A US-shaped trend is characterized by a sizable gap in the level of integration between the pre-2003 period and 2003–2007 period, and between the post-2007 period and the 2003–2007 period.

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

197

Financials. Not surprisingly, the average level of integration in the financials sector (insurance, real estate, financial services, equity investment instruments) in the area Emerging is (on average) higher than in most of the other sectors. Over the full sample, the integration index is equal to 0.43. The gap in the level of integration between the tranquil era (i.e. 2003–2007) and the post-crisis period (post-2007) is really high (i.e. 0.25), suggesting that both the contagion and hearding effects have heavily influenced financials equity price indexes’ dynamics across emerging markets. As a result, the gap between January 2010–December 2007 is also high (i.e. around 20%). Similar conclusions can be drawn about the average level of integration in Asia, Eastern Europe and Latin America. However, the impact of the systemic banking crises on the integration dynamics in Latin America, which exhibits a marked J-shaped trend, seems to be stronger. We stress that this industry contributed most to the average level of integration across emerging equity markets. Industrials. The average level of integration in this sector (constructions and materials, industrial goods and services, electronic equipment, industrial engineering and transportation, support services) is equal to 0.37. As in the financials industry both the gap and the jump are relatively high, 0.24 and 0.18, respectively. This implies that the recent GFC has heavily affected emerging stocks belonging to this industry. We also observe that the integration index is much higher in Eastern Europe (0.63) than in the other regions (0.39 and 0.43 in Asian and Latin America, respectively). Basic materials. This industry includes chemicals and basic resources firms. The average level of integration (in the Emerging region) is almost as high as in the financials industry (0.41 vs. 0.43). The gap and the jump are also very high suggesting that crises heavily influenced the dynamics of the Basic Materials equity price indexes. As expected, in Asia, Eastern Europe and Latin America the integration index is higher, equal to 0.46, 0.62 and 0.51, respectively. We stress that, as in the other sectors, the level of integration in Eastern Europe is higher than in the other sub-regions. In particular, notice that it is always above 0.45 and close to 0.80 over the post-2007 period. Oil and gas. The average level of integration in this sector, which includes oil and gas producers, oil equipment and distribution, and alternative energy firms, is slightly lower than in the basic materials and financials sectors. However, the oil&gas sector shares with these industries a relatively high difference (i.e. 0.21) in the level of integration between the tranquil period (i.e. 2003–2007) and the post-2007 period. Differently, the GFC-gap (i.e. jump) is considerable lower than in the basic materials, financials and industrials sectors, suggesting that this industry was weakly affected by the subprime crisis. Notice that this jump is even smaller in Asia and Latin America, 0.07 and 0.04, respectively. In the Eastern Europe and Emerging regions the integration index is constantly increasing. Differently, in Asia and Latin America it exhibits a J-shaped trend. Healthcare. Because of the limited number of observations, the dynamics of the integration index in this industry (healthcare equipment and services, pharmaceuticals and biotechnology) cannot be deeply examined. We observe (in the Emerging region), a relative high level of integration (i.e. 0.44). We admit that this might be due to the lower number of countries included in the PCA. For example, only one country (Hungary) and two countries (Chile and Mexico) are available, respectively in Eastern Europe and in Latin America (see Table A.3 for additional details). In fact, Panel E in Fig. 5 does not include the dynamics of the integration index in Eastern Europe. Most broadly, we can only say that equity market integration is increasing over time. Technology. As for the healthcare industry, we do not have enough observations to discuss properly integration dynamics and statistics in this industry (software and computer services, hardware and equipment). Because of this, the average level of integration is high (i.e. 0.46) in the Emerging region. In this case the index is computed by including only 6 out of 18 countries. Notice also that we cannot compute the integration index dynamics in Latin America (see Fig. 5, Panel E). Overall, as for the healthcare industry, we can only say that the integration index tends to increase overtime. Telecommunications. In the Emerging area the integration index is (on average) equal to 0.39. Its dynamics seems to be heavily influenced by the dynamics of the integration index in Latin America which is decreasing for most of the sample (between the two NBER-dated recession periods). We argue that the integration index in Latin America, which exhibits a U-shaped trend (or inverse Jshaped trend), has been mainly driven by the IT bubble. It turns out that the integration index in Latin America in the pre-2003 era is much higher than in the aftermath of the GFC (i.e. 0.61 vs. 0.51).

198

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

Panel(A) Asia - Consumer Goods

Asia - Consumer Services

0.5

0.6

0.45

0.55

0.4

0.5

0.35

0.45

0.3

0.4

0.25

0.35

0.2

2000

2002

2004

2006

2008

2010

2012

0.3

2000

2002

Eastern Europe - Consumer Goods

2004

2006

2008

2010

2012

Eastern Europe - Consumer Services

0.85

0.7

0.8

0.65

0.75 0.6 0.7 0.65

0.55

0.6

0.5

0.55 0.45 0.5 0.4

0.45 0.4

2002

2004

2006

2008

2010

2012

0.35

2005

2006

Latin America - Consumer Goods

2007

2008

2009

2010

2011

2012

Latin America - Consumer Services

0.55

0.65 0.6

0.5

0.55 0.45 0.5 0.4 0.45 0.35 0.4 0.3

0.25

0.35

2000

2002

2004

2006

2008

2010

2012

0.3

2000

2002

Emerging - Consumer Goods

2004

2006

2008

2010

2012

Emerging - Consumer Services

0.55

0.5

0.5 0.45 0.45 0.4 0.4 0.35

0.35

0.3 0.3 0.25 0.25 0.2 0.15

2002

2004

2006

2008

2010

2012

Fig. 5.

0.2

2005

2006

2007

2008

2009

2010

2011

2012

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

199

Panel(B) Asia - Financials

Asia - Industrials

0.6

0.9

0.55

0.8

0.5

0.7

0.45

0.6

0.4

0.5

0.35

0.4

0.3

0.3

0.25

2000

2002

2004

2006

2008

2010

2012

0.2

2000

2002

Eastern Europe - Financials

2004

2006

2008

2010

2012

Eastern Europe - Industrials

0.85

0.8

0.8

0.75

0.75

0.7

0.7 0.65 0.65 0.6 0.6 0.55

0.55 0.5

0.5

0.45

0.45

2004

2005

2006

2007

2008

2009

2010

2011

2012

2002

2004

Latin America - Financials

2006

2008

2010

2012

Latin America - Industrials

0.8

0.75

0.75

0.7

0.7

0.65

0.65

0.6

0.6

0.55

0.55

0.5

0.5

0.45

0.45

0.4

0.4

0.35

2000

2002

2004

2006

2008

2010

2012

2000

2002

Emerging - Financials

2004

2006

2008

2010

2012

Emerging - Industrials

0.7

0.65

0.65

0.6

0.6

0.55

0.55

0.5

0.5 0.45 0.45 0.4 0.4 0.35

0.35 0.3

0.3

0.25

0.25

0.2

2004

2005

2006

2007

2008

2009

2010

2011

2012

0.2

2002

Fig. 5. (Continued )

2004

2006

2008

2010

2012

200

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

Panel(C) Asia - Basic Materials

Asia - OilGas

0.7

0.55

0.65 0.5 0.6 0.55

0.45

0.5 0.4 0.45 0.4

0.35

0.35 0.3 0.3 0.25

2000

2002

2004

2006

2008

2010

2012

0.25

2002

2004

Eastern Europe - Basic Materials 0.85

2006

2008

2010

2012

Eastern Europe - OilGas 0.85 0.8

0.8

0.75

0.75

0.7 0.7 0.65 0.65 0.6 0.6 0.55 0.55

0.5

0.5 0.45

0.45 2000

2002

2004

2006

2008

2010

2012

0.4

2003

2004

2005

Latin America - Basic Materials 0.8

2006

2007

2008

2009

2010

2011

2012

Latin America - OilGas 0.7

0.75

0.65

0.7 0.6 0.65 0.6

0.55

0.55

0.5

0.5 0.45 0.45 0.4

0.4 0.35

2000

2002

2004

2006

2008

2010

2012

0.35

2000

2002

2004

Emerging - Basic Materials

2006

2008

2010

2012

Emerging - OilGas

0.7

0.6

0.65

0.55

0.6

0.5

0.55 0.45 0.5 0.4 0.45 0.35 0.4 0.3

0.35

0.25

0.3 0.25

2000

2002

2004

2006

2008

2010

2012

0.2

2003

Fig. 5. (Continued )

2004

2005

2006

2007

2008

2009

2010

2011

2012

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

201

Panel(D) Asia - Telecommunications

Asia - Utilities

0.55

0.55

0.5

0.5

0.45 0.45 0.4 0.4 0.35 0.35

0.3

0.3 2000

2002

2004

2006

2008

2010

2012

0.25

2002

2004

Eastern Europe - Telecommunications

2006

2008

2010

2012

Eastern Europe - Utilities

0.75

0.8 0.75

0.7

0.7 0.65 0.65 0.6 0.6 0.55 0.55 0.5

0.5

0.45 2004

2005

2006

2007

2008

2009

2010

2011

2012

0.452003

2004

2005

Latin America - Telecommunications

0.6

0.65

0.55

0.6

0.5

0.55

0.45

0.5

0.4

0.45

0.35

2000

2002

2004

2006

2008

2007

2008

2009

2010

2011

2012

Latin America - Utilities

0.7

0.4

2006

2010

2012

0.3

2002

2004

Emerging - Telecommunications

2006

2008

2010

2012

Emerging - Utilities

0.55 0.5

0.45

0.45 0.4

0.4 0.35

0.35

0.3 0.25

0.3 0.2 2004

2005

2006

2007

2008

2009

2010

2011

2012

0.152003

Fig. 5. (Continued )

2004

2005

2006

2007

2008

2009

2010

2011

2012

202

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

Panel(E) Asia - Healthcare

Asia - Technology 0.75

0.6

0.7 0.65

0.55

0.6 0.55

0.5

0.5 0.45

0.45

0.4

2000

2002

2004

2006

2008

2010

2012

0.35

2004

2005

Latin America - Healthcare 0.75

0.75

0.7

0.7

0.65

0.65

0.6

0.6

0.55

2004

2005

2006

2007

2008

2009

2010

2011

2007

2008

2009

2010

2011

2012

Eastern Europe - Technology

0.8

0.55

2006

2012

0.5

2005

2006

Emerging - Healthcare

2007

2008

2009

2010

2011

2012

Emerging - Technology

0.55

0.65

0.6

0.5

0.55 0.45 0.5 0.4 0.45 0.35

0.3

0.4

2004

2005

2006

2007

2008

2009

2010

2011

2012

0.35

2005

2006

2007

2008

2009

2010

2011

2012

Fig. 5. Integration index: industrial equity markets. This figure reports the dynamics of the financial integration index across emerging industrial equity markets in four different emerging regions (Asia, Eastern Europe, Latin America and Emerging). The integration index is represented by the proportion of variance in industrial equity market excess returns explained by the first principal component in each emerging region. The first principal component is estimated using a rolling window of 60 months. Sector equity excess returns are computed as defined in Eq. (3.1). The grey line defines the dynamics of the integration index computed by employing the covariance matrix in the PCA. The black line defines the dynamics of the integration index computed by employing the correlation matrix in the PCA. The smoothed red line – computed using the unobserved component smoother technique explained in Appendix B – represents the trend of the “correlation-based integration index”. Shaded vertical bars denote NBER-dated recessions. Sample: January 1994 (or later)–July 2012. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

Differently, the dynamics is steadily increasing in Eastern Europe and exhibits a weak J-shaped trend in Asia. Utilities. In the Emerging region, the average level of integration in this sector – composed by electricity and gas, water and multiutilities listed companies – is lower than in most of the other industries (i.e. basic materials, consumer services, financials, healthcare, industrials, telecommunications). As in

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

203

Table 5 Emerging regions: average integration index level (sectors). Notes: Gap is the difference between the average integration index in the GFC (post-2007) and the average integration in the tranquil times (2003–2007). Jump denotes the difference between the value of integration index observed in the aftermath of the GFC (January 2010) and the value observed before the GFC (December 2007). The last line reports the shape of the financial integration process (U = U-shaped trend; J = J-shaped trend; ⇑ = increasing trend). Region

Asia

East. Eu.

Latin Am.

Emerg.

Basic materials 1. Average (pre-2003) 2. Average (2003–2007) 3. Average (post-2007) Gap (3–2) Jump (January 10–December 07)

0.46 0.41 0.39 0.59 0.20 0.19

0.62 0.55 0.55 0.76 0.21 0.20

0.51 0.52 0.41 0.63 0.22 0.08

0.41 0.34 0.31 0.57 0.26 0.19

Shape

J

⇑

J

J

Consumer goods 1. Average (pre-2003) 2. Average (2003–2007) 3. Average (post-2007) Gap (3–2) Jump (January 10–December 07)

0.34 0.31 0.29 0.42 0.13 0.11

0.57 0.51 0.48 0.68 0.20 0.17

0.38 0.42 0.31 0.43 0.12 0.12

0.31 0.24 0.23 0.42 0.19 0.16

Shape

J

J

U

⇑

Consumer services 1. Average (pre-2003) 2. Average (2003–2007) 3. Average (post-2007) Gap (3–2) Jump (January 10–December 07)

0.41 0.40 0.39 0.45 0.06 0.03

0.54 NA 0.45 0.60 0.15 0.13

0.48 0.51 0.39 0.54 0.15 0.17

0.36 NA 0.26 0.43 0.17 0.14

Shape

⇑

⇑

U

⇑

Financials 1. Average (pre-2003) 2. Average (2003–2007) 3. Average (post-2007) Gap (3–2) Jump (January 10–December 07)

0.41 0.33 0.35 0.53 0.18 0.11

0.66 NA 0.53 0.78 0.25 0.17

0.57 0.55 0.48 0.67 0.19 0.18

0.43 NA 0.31 0.56 0.25 0.18

Shape

⇑

⇑

J

⇑

HealthCarea

0.50

NA

0.68

0.44

1. Average (pre-2003) 2. Average (2003–2007) 3. Average (post-2007) Gap (3–2) Jump (January 10–December 07)

0.49 0.49 0.52 0.03 0.01

NA NA NA NA NA

NA 0.66 0.70 0.04 0.02

NA 0.38 0.50 0.12 0.09

Shape

⇑

–

⇑

⇑

Industrials 1. Average (pre-2003) 2. Average (2003–2007) 3. Average (post-2007) Gap (3–2) Jump (January 10–December 07)

0.39 0.30 0.32 0.53 0.21 0.12

0.63 0.48 0.57 0.73 0.16 0.08

0.43 0.41 0.37 0.50 0.13 0.13

0.37 0.25 0.27 0.51 0.24 0.18

Shape

⇑

⇑

J

⇑

Oil and gas 1. Average (pre-2003) 2. Average (2003–2007) 3. Average (post-2007)

0.37 0.33 0.32 0.43

0.66 NA 0.53 0.79

0.45 0.41 0.41 0.52

0.38 NA 0.27 0.48

204

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

Table 5 (Continued ) Region

Asia

East. Eu.

Latin Am.

Emerg.

Gap (3–2) Jump (January 10–December 07)

0.11 0.07

0.26 0.13

0.11 0.04

0.21 0.12

Shape

J

⇑

J

⇑

Telecommunications 1. Average (pre-2003) 2. Average (2003–2007) 3. Average (post-2007) Gap (3–2) Jump (January 10–December 07)

0.43 0.40 0.40 0.49 0.09 0.07

0.58 NA 0.53 0.63 0.10 0.08

0.54 0.61 0.51 0.51 0.00 0.09

0.39 NA 0.34 0.44 0.10 0.10

Shape

J

⇑

U

J

Technologya 1. Average (pre-2003) 2. Average (2003–2007) 3. Average (post-2007) Gap (3–2) Jump (January 10–December 07)

0.48 NA 0.46 0.49 0.03 0.09

0.64 NA 0.59 0.67 0.08 0.14

NA NA NA NA NA NA

0.46 NA 0.41 0.50 0.09 0.16

Shape

⇑

J

−

J

Utilities 1. Average (pre-2003) 2. Average (2003–2007) 3. Average (post-2007) Gap (3–2) Jump (January 10–December 07)

0.38 0.32 0.35 0.45 0.10 0.04

0.59 NA 0.52 0.66 0.14 0.16

0.43 0.39 0.36 0.51 0.15 0.11

0.33 NA 0.25 0.42 0.17 0.12

Shape

⇑

J

J

⇑

a

Integration indexes are computed by using a relatively low number of countries (details on data availability are given in Table A.3)

other industries, the difference in the level of integration between the pre-2003 period and the nocrises era (i.e. 2003–2007) is relatively high (0.17). However, it is lower than in the basic materials (0.26), financials (0.25) and industrials (0.24) sectors. We confirm that, in Asia, Eastern Europe and Latin America the average integration index is higher than in the Emerging region. Finally, notice that the dynamics of the integration index in the utilities sector follows a J-shaped trend in Eastern Europe and Latin America and is almost constantly increasing in Asia and Emerging. 4.3. Integration, co-movement and diversification: a CAPM analysis Portfolio diversification benefits do not rely exclusively on the degree of co-movement between equity indexes, but also on the magnitude of the co-movements. If a stock and the aggregate benchmark index (i.e. market) move together, diversification benefits do not necessarily disappear. In a CAPM framework, this depends on the quantity of risk (i.e. beta). In other words, a far from unity beta might still provide diversification benefits. In the spirit of Mauro et al. (2002), to examine the benefits of holding a portfolio of stocks issued by a variety of emerging market countries rather than by only one country (e.g. aggregate regional equity index) we run a set of univariate regressions. In practice, for each emerging equity market, we estimate a univariate regression with the excess return on the base country (industry) equities on the left-hand side and the excess return on a regional market country (industry) equally weighted portfolio of equities on the right-hand side. Formally, ExRet c,t = ˛c + ˇc (GCI − Rf,t ) + c,t , ExRet c,t = ˛c + ˇc (RCI − Rf,t ) + c,t ,

(4.2a)

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

205

Table 6 Beta coefficients on country equity market excess returns. Notes: beta coefficients are estimated by regressing each country’s monthly excess return on the excess return of a global emerging market equally weighted portfolio (Panel A) and a regional emerging market equally weighted portfolio (Panel B). Excess returns are computed as defined in Eq. (3.1). Global and regional equally weighted portfolios are computed as defined in Eq. (4.3a). Standard errors are Newey and West (1987, 1994). In presence of structural breaks, a dummy is included (see Table A.1). Structural breaks are identified via the Bai–Perron test. The crosscountry average beta is a simple average. Region

Panel A

Country

Mkt: GLOBAL (January 1995–December 2011)

Panel B Mkt: sub-region (January 1995–December 2011)

ˇ

SE

R

|1 − ˇ|

ˇ

SE

R2

|1 − ˇ|

Latin America Argentina Brazil Chile Colombia Mexico Peru Average

1.061 1.230 0.737 0.806 0.966 0.883

0.087 0.057 0.042 0.078 0.056 0.105

0.452 0.674 0.615 0.366 0.639 0.486 0.539

0.061 0.230 0.263 0.194 0.034 0.117 0.150

1.191 1.271 0.727 0.861 0.978 0.971

0.079 0.064 0.048 0.065 0.048 0.074

0.612 0.772 0.641 0.449 0.702 0.631 0.635

0.191 0.271 0.273 0.139 0.022 0.029 0.154

Asia China India Malaysia Pakistan Philippines Sri Lanka Thailand Average

0.954 1.210 0.641 0.654 0.911 0.585 1.138

0.075 0.113 0.070 0.137 0.058 0.123 0.091

0.442 0.422 0.380 0.177 0.487 0.165 0.484 0.365

0.046 0.210 0.359 0.346 0.089 0.415 0.138 0.229

0.969 1.409 0.787 0.745 1.009 0.704 1.354

0.098 0.091 0.075 0.123 0.062 0.124 0.073

0.452 0.568 0.522 0.229 0.594 0.239 0.681 0.469

0.031 0.409 0.213 0.255 0.009 0.296 0.354 0.224

1.251 1.090 1.613 1.318

0.100 0.076 0.116 0.139

0.068 0.058 0.089 0.093

0.727 0.619 0.651 0.651 0.662

0.086 0.225 0.184 0.127 0.156

0.057

0.251 0.090 0.613 0.318 0.318 0.078

0.914 0.775 1.184 1.127

0.922

0.617 0.555 0.547 0.402 0.530 0.661

Eastern Europe Hungary Poland Russia Turkey Average South Africa

2

ExRet ci,t = ˛ci + ˇic (GII − Rf,t ) + ci,t , ExRet ci,t = ˛ci + ˇic (RII − Rf,t ) + ci,t ,

(4.2b)

where the excess returns are defined as in Eq. (3.1), and the country and industry global and regional equally weighted portfolios are computed as follows: 1

MSCI i,t , N N

GCI =

i=1

1

DGEI i,t , N i=1

(4.3a)

i=1

N

GII =

1

MSCI i,t , K K

RCI =

1

DGEI i,t , K K

RII =

(4.3b)

i=1

where MSCIi,t is the Morgan Stanley Capital International Total Return Index of country i at time t, DGEIi,t is the Datastream Global Equity Index of country i at time t, N is the total number of emerging countries in the emerging word region (GLOBAL or Emerging), and K is the total number of countries in each sub-region (Asia, Eastern Europe, Latin America). Table 6 reports the estimation results of the regressions defined in Eq. (4.2a). Panel A reports the estimated values for the global region (Emerging). Panel B reports the estimated values for the other emerging sub-regions. Industry-by-industry estimation results (see Eq. (4.2b)), for both the global region (Panel A) and the sub-regions (Panel B) are reported in Table 7.

206

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

Table 7 Beta coefficients on industrial equity market excess returns. Notes: Beta coefficients are estimated by regressing each industry monthly excess returns (in each country) on the excess return of a global emerging market equally weighted portfolio (Panel A) and a regional emerging market equally weighted portfolio (Panel B). Excess returns are computed as defined in Eq. (3.1). Global and regional industry equally weighted portfolios are computed as defined in (4.3b). Standard errors are Newey and West (1987, 1994). In presence of structural breaks, a dummy is included (see Table A.2). Structural breaks are identified via the Bai–Perron test. The cross-country average beta is a simple average. Regions

Panel A

Panel B

Country

BasMats: GLOBAL (August 1996–July 2012)

BasMats: sub-region (August 1996–July 2012)

ˇ

SE

R2

|1 − ˇ|

ˇ

SE

R2

|1 − ˇ|

Latin America Argentina Brazil Chile Colombia Mexico Peru Average

0.862 1.252 0.723 0.638 1.199 0.590

0.086 0.084 0.058 0.098 0.113 0.077

0.305 0.668 0.494 0.260 0.577 0.324 0.438

0.138 0.252 0.277 0.362 0.199 0.410 0.273

1.060 1.299 0.759 0.831 1.305 0.666

0.127 0.060 0.064 0.117 0.105 0.075

0.455 0.706 0.535 0.356 0.671 0.406 0.522

0.060 0.299 0.241 0.169 0.305 0.334 0.235

Asia China India Malaysia Pakistan Philippines Thailand Average

1.297 1.025 0.784 0.497 1.246 1.325

0.119 0.110 0.139 0.085 0.171 0.169

0.438 0.477 0.307 0.151 0.313 0.410 0.349

0.297 0.025 0.216 0.503 0.246 0.325 0.269

1.183 0.860 0.809 0.426 1.377 1.345

0.115 0.162 0.103 0.071 0.125 0.168

0.502 0.460 0.449 0.152 0.527 0.582 0.445

0.183 0.140 0.191 0.574 0.377 0.345 0.302

Eastern Europe Hungary Poland Turkey Average

0.884 1.167 1.331

0.097 0.097 0.136

0.341 0.515 0.356 0.404

0.116 0.167 0.331 0.205

0.762 0.911 1.327

0.067 0.071 0.076

0.494 0.609 0.692 0.598

0.238 0.089 0.327 0.218

1.105

0.107

0.477

0.105

Africa South Africa Regions

ConsGds: GLOBAL (August 1996–July 2012)

ConsGds: sub-region (August 1996–July 2012)

Country

ˇ

SE

R

|1 − ˇ|

ˇ

SE

R2

|1 − ˇ|

Latin America Argentina Brazil Chile Colombia Mexico Peru Average

1.467 1.231 0.796 0.927 0.979 0.342

0.198 0.117 0.100 0.120 0.112 0.066

0.284 0.419 0.360 0.250 0.251 0.108 0.279

0.467 0.231 0.204 0.073 0.021 0.658 0.276

1.692 1.161 0.714 1.073 1.036 0.324

0.166 0.094 0.092 0.108 0.120 0.066

0.480 0.470 0.366 0.426 0.356 0.123 0.370

0.692 0.161 0.286 0.073 0.036 0.676 0.321

Asia China India Malaysia Pakistan Philippines Sri Lanka Thailand Average

1.032 0.853 0.983 0.572 0.626 0.468 1.406

0.115 0.105 0.216 0.134 0.087 0.088 0.15

0.132 0.300 0.280 0.145 0.213 0.079 0.404 0.222

0.032 0.147 0.017 0.428 0.374 0.532 0.406 0.277

1.402 0.794 1.244 0.557 0.775 0.541 1.525

0.255 0.135 0.166 0.133 0.086 0.110 0.189

0.273 0.286 0.498 0.150 0.327 0.119 0.478 0.304

0.402 0.206 0.244 0.443 0.225 0.459 0.525 0.358

Eastern Europe Hungary Poland Turkey Average

1.081 1.009 1.767

0.175 0.093 0.175

0.305 0.417 0.385 0.369

0.081 0.009 0.767 0.286

0.893 0.627 1.480

0.102 0.063 0.107

0.546 0.420 0.708 0.558

0.107 0.373 0.48 0.320

2

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

207

Table 7 (Continued ) Regions

Panel A

Panel B

Country

ConsGds: GLOBAL (August 1996–July 2012)

ConsGds: sub-region (August 1996–July 2012)

ˇ

SE

R2

|1 − ˇ|

ˇ

Africa South Africa

1.329

0.111

0.414

0.329

|1 − ˇ|

R2

SE

Regions

ConsSvs: GLOBAL (November 1999–July 2012)

ConsSvs: sub-region (November 1999–July 2012)

Country

ˇ

SE

R2

|1 − ˇ|

ˇ

SE

R2

|1 − ˇ|

Latin America Argentina Chile Colombia Mexico Average

0.906 0.818 0.900 0.795

0.155 0.074 0.141 0.101

0.186 0.454 0.201 0.434 0.319

0.094 0.182 0.100 0.205 0.145

1.300 0.705 1.311 0.686

0.141 0.086 0.124 0.090

0.471 0.406 0.526 0.390 0.448

0.300 0.295 0.311 0.314 0.305

Asia China Malaysia Pakistan Philippines Sri Lanka Thailand Average

1.172 0.614 1.021 0.858 0.393 0.684

0.196 0.067 0.225 0.109 0.120 0.071

0.313 0.383 0.181 0.358 0.055 0.352 0.274

0.172 0.386 0.021 0.142 0.607 0.316 0.274

1.433 0.611 1.463 0.987 0.639 0.839

0.235 0.070 0.309 0.106 0.112 0.097

0.460 0.369 0.368 0.463 0.152 0.487 0.383

0.433 0.389 0.463 0.013 0.361 0.161 0.303

Eastern Europe Hungary Poland Russia Turkey Average

1.728 0.992 1.465 1.703

0.308 0.087 0.319 0.169

0.298 0.404 0.345 0.450 0.374

0.728 0.008 0.465 0.703 0.476

1.303 0.586 1.028 1.083

0.223 0.069 0.211 0.085

0.515 0.425 0.514 0.543 0.499

0.303 0.414 0.028 0.083 0.207

Africa South Africa

0.926

0.094

0.343

0.074

Regions

Fin: GLOBAL (May 1998–July 2012)

Country

ˇ

SE

R2

|1 − ˇ|

ˇ

SE

R2

|1 − ˇ|

1.077 1.152 0.589 0.836 0.941

0.093 0.079 0.060 0.0806 0.089

0.358 0.572 0.480 0.377 0.489 0.455

0.077 0.152 0.411 0.164 0.059 0.173

1.287 1.158 0.585 0.961 0.983

0.085 0.075 0.058 0.084 0.093

0.569 0.640 0.524 0.516 0.591 0.568

0.287 0.158 0.415 0.039 0.017 0.183

Asia China India Malaysia Pakistan Philippines Sri Lanka Thailand Average

0.772 1.161 0.514 0.615 0.921 0.340 1.173

0.112 0.139 0.142 0.114 0.101 0.108 0.130

0.214 0.443 0.165 0.143 0.472 0.071 0.399 0.272

0.228 0.161 0.486 0.385 0.079 0.66 0.173 0.310

0.943 1.305 0.825 0.875 1.063 0.517 1.470

0.097 0.180 0.112 0.127 0.106 0.108 0.192

0.270 0.473 0.365 0.248 0.532 0.144 0.531 0.366

0.057 0.305 0.175 0.125 0.063 0.483 0.470 0.240

Eastern Europe Hungary Poland Russia Turkey Average

1.459 1.080 1.818 1.633

0.175 0.100 0.218 0.138

0.577 0.603 0.391 0.493 0.516

0.459 0.080 0.818 0.633 0.498

0.924 0.675 1.335 1.067

0.103 0.061 0.171 0.092

0.651 0.663 0.596 0.593 0.626

0.076 0.325 0.335 0.067 0.201

0.891

0.070

0.431

0.109

Fin: sub-region (May 1998–July 2012)

Latin America Argentina Brazil Chile Colombia Mexico Average

Africa South Africa

208

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

Table 7 (Continued ) Regions

Panel A

Panel B

Country

Healthcare: GLOBAL (August 1998–July 2012)

Healthcare: sub-region (August 1998–July 2012)

ˇ

SE

R2

|1 − ˇ|

ˇ

SE

R2

|1 − ˇ|

Latin America Chile Mexico Average

1.003 1.023

0.121 0.263

0.392 0.260 0.326

0.003 0.023 0.013

0.828 1.172

0.105 0.105

0.553 0.713 0.633

0.172 0.172 0.172

Asia India Pakistan Thailand Average

0.721 0.933 0.811

0.120 0.162 0.125

0.357 0.346 0.343 0.349

0.279 0.067 0.189 0.178

0.813 1.194 0.993

0.109 0.142 0.095

0.421 0.527 0.476 0.475

0.187 0.194 0.007 0.129

Eastern Europe Hungary Average

1.409

0.202

0.533 0.533

0.409 0.409

Africa South Africa

1.100

0.087

0.502

0.100

Regions

Industrials: GLOBAL (October 1996–July 2012)

Industrials: sub-region (October 1996–July 2012)

Country

ˇ

SE

R

|1 − ˇ|

ˇ

SE

R2

|1 − ˇ|

Latin America Argentina Brazil Chile Colombia Mexico Peru Average

0.818 1.082 0.726 0.336 1.146 0.771

0.136 0.093 0.094 0.067 0.133 0.302

0.188 0.456 0.353 0.070 0.531 0.047 0.274

0.006 0.025 0.274 0.664 0.146 0.229 0.224

1.006 1.025 0.746 0.337 1.021 1.867

0.188 0.245 0.151 0.114 0.243 0.763

0.298 0.426 0.388 0.070 0.438 0.313 0.322

0.006 0.025 0.254 0.663 0.021 0.867 0.306

Asia China India Malaysia Pakistan Philippines Sri Lanka Thailand Average

1.004 1.284 0.581 1.627 1.120 0.440 1.405

0.185 0.250 0.122 0.999 0.201 0.133 0.139

0.198 0.289 0.198 0.113 0.355 0.087 0.446 0.241

0.004 0.284 0.419 0.627 0.12 0.56 0.405 0.346

0.869 1.134 0.463 2.211 0.813 0.283 1.227

0.300 0.380 0.174 1.263 0.296 0.150 0.145

0.233 0.355 0.197 0.339 0.293 0.054 0.535 0.287

0.131 0.134 0.537 1.211 0.187 0.717 0.227 0.449

Eastern Europe Hungary Poland Turkey Average

1.055 1.057 1.557

0.170 0.114 0.169

0.355 0.422 0.357 0.378

0.055 0.057 0.557 0.223

0.848 0.740 1.411

0.069 0.067 0.082

0.566 0.508 0.723 0.599

0.152 0.26 0.411 0.274

0.992

0.110

0.428

0.008

Africa South Africa

2

Regions

Oil and gas: GLOBAL (March 1998–July 2012)

Oil and gas: sub-region (March 1998–July 2012)

Country

ˇ

SE

R

|1 − ˇ|

ˇ

SE

R2

|1 − ˇ|

Latin America Argentina Brazil Chile Colombia Average

0.956 1.488 0.626 0.540

0.194 0.108 0.091 0.125

0.213 0.589 0.297 0.079 0.295

0.044 0.488 0.374 0.460 0.342

1.149 1.351 0.565 0.935

0.275 0.146 0.081 0.161

0.396 0.619 0.309 0.316 0.410

0.149 0.351 0.435 0.065 0.250

1.215

0.099

0.408

0.215

1.396

0.126

0.522

0.396

Asia China

2

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

209

Table 7 (Continued ) Regions

Panel A

Country

Oil and gas: GLOBAL (March 1998–July 2012)

India Malaysia Pakistan Philippines Sri Lanka Thailand Average Eastern Europe Hungary Poland Russia Turkey Average Africa South Africa

Panel B Oil and gas: sub-region (March 1998–July 2012)

ˇ

SE

R2

|1 − ˇ|

ˇ

SE

R2

|1 − ˇ|

1.060 0.552 0.689 0.983 0.225 1.108

0.104 0.114 0.147 0.169 0.110 0.078

0.389 0.235 0.123 0.196 0.022 0.453 0.261

0.060 0.448 0.311 0.017 0.775 0.108 0.276

1.169 0.630 1.023 1.310 0.361 1.111

0.145 0.118 0.175 0.199 0.117 0.116

0.456 0.298 0.269 0.340 0.062 0.440 0.341

0.169 0.37 0.023 0.310 0.639 0.111 0.288

1.298 1.010 1.449 1.573

0.149 0.097 0.164 0.112

0.460 0.463 0.448 0.437 0.452

0.298 0.010 0.449 0.573 0.3325

1.088 0.703 1.060 1.230

0.074 0.056 0.111 0.084

0.687 0.554 0.593 0.662 0.624

0.088 0.297 0.06 0.23 0.169

1.230

0.091

0.583

0.230

Regions

Tech: GLOBAL (June 1996–July 2012)

Country

ˇ

SE

R2

|1 − ˇ|

ˇ

Tech: sub-region (June 1996–July 2012) SE

R2

|1 − ˇ|

Latin America China India Thailand Average

0.424 1.097 0.679

0.103 0.130 0.115

0.148 0.591 0.308 0.349

0.576 0.097 0.321 0.331

0.714 1.321 0.897

0.109 0.199 0.172

0.366 0.636 0.461 0.488

0.286 0.321 0.103 0.237

Eastern Europe Hungary Poland Turkey Average

1.185 0.982 1.532

0.187 0.074 0.184

0.463 0.595 0.534 0.531

0.185 0.018 0.532 0.245

1.015 0.711 1.273

0.096 0.089 0.071

0.608 0.555 0.670 0.611

0.015 0.289 0.273 0.192

Regions

Telec: GLOBAL (December 1998–July 2012)

Telec: sub-region (December 1998–July 2012)

Country

ˇ

SE

R2

|1 − ˇ|

ˇ

SE

R2

|1 − ˇ|

Latin America Argentina Brazil Chile Mexico Peru Average

1.175 1.136 0.624 0.965 0.970

0.120 0.063 0.084 0.089 0.227

0.360 0.562 0.283 0.557 0.146 0.382

0.175 0.136 0.376 0.035 0.03 0.150

1.269 1.019 0.589 0.775 1.348

0.110 0.086 0.129 0.118 0.360

0.573 0.614 0.343 0.487 0.394 0.482

0.269 0.019 0.411 0.225 0.348 0.254

1.245 0.505 0.734 0.636 0.787

0.193 0.104 0.112 0.089 0.092

0.393 0.187 0.192 0.306 0.320 0.280

0.245 0.495 0.266 0.364 0.213 0.317

1.539 0.631 1.026 0.818 0.960

0.163 0.141 0.114 0.077 0.092

0.544 0.266 0.343 0.435 0.432 0.404

0.539 0.369 0.026 0.182 0.040 0.231

1.089 0.942 1.137 1.760

0.117 0.087 0.190 0.249

0.512 0.394 0.306 0.405 0.404

0.089 0.058 0.137 0.760 0.261

0.780 0.706 0.997 1.516

0.107 0.085 0.151 0.129

0.546 0.463 0.494 0.629 0.533

0.22 0.294 0.003 0.516 0.258

1.270

0.155

0.468

0.270

Asia India Malaysia Pakistan Philippines Thailand Average Eastern Europe Hungary Poland Russia Turkey Average Africa South Africa

210

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

Table 7 (Continued ) Regions

Panel A

Country

Utilities: GLOBAL (March 1998–July 2012)

Latin America Argentina Brazil Chile Colombia Peru Average Asia China India Malaysia Pakistan Philippines Thailand Average Eastern Europe Hungary Russia Turkey Average

Panel B Utilities: sub-region (March 1998–July 2012)

ˇ

SE

R2

|1 − ˇ|

ˇ

SE

R2

|1 − ˇ|

1.020 1.320 0.741 0.510 0.440

0.139 0.119 0.076 0.110 0.085

0.292 0.448 0.426 0.141 0.205 0.302

0.02 0.320 0.259 0.490 0.56 0.330

1.295 1.628 0.833 0.719 0.505

0.165 0.113 0.066 0.116 0.090

0.446 0.644 0.507 0.269 0.246 0.422

0.295 0.628 0.167 0.281 0.495 0.373

0.957 1.049 0.326 0.707 1.003 0.887

0.129 0.204 0.157 0.187 0.119 0.113

0.273 0.323 0.198 0.113 0.308 0.313 0.255

0.043 0.049 0.674 0.293 0.003 0.113 0.196

1.118 1.118 0.499 0.939 1.267 0.919

0.106 0.107 0.085 0.136 0.106 0.095

0.391 0.385 0.280 0.213 0.484 0.353 0.351

0.118 0.118 0.501 0.061 0.267 0.081 0.191

0.964 2.157 1.758

0.132 0.307 0.222

0.371 0.496 0.451 0.439

0.036 1.157 0.758 0.650

0.580 1.359 1.060

0.067 0.106 0.145

0.495 0.727 0.604 0.609

0.42 0.359 0.06 0.280

The absolute difference between one and a country’s or industry’s beta coefficient is always larger than zero (i.e. the coefficients tend to be different from one). For the Emerging region, |1 − ˇ| ranges from a minimum of 0.034 (Mexico) to a maximum of 0.613 (Russia). This implies that the benefits of holding a portfolio of equities issued by a region rather than by only one country are still relatively high. The R2 is rarely close to one. At the country level, it ranges from a minimum of 0.165 (Sri Lanka) to a maximum of 0.674 (Brazil). This suggests that the global emerging equity portfolio (i.e. emerging aggregated index) account for a relatively small fraction of the fluctuation in individual country returns. By focusing on emerging sub-regions, we can draw similar conclusions. We stress that results differ only quantitatively. In fact, we observe (i) slightly higher R2 s; (ii) slightly lower absolute differences, |1 − ˇ|. This is due to the fact that the sub-regional portfolios are composed by country equity indexes belonging to the same area (e.g. the country equity indexes of Hungary, Poland, Russia and Turkey are used to build the Eastern Europe equally weighted portfolio). At the industry level, we obtain similar results (see Table 7). We conclude by arguing that a high degree of heterogeneity across betas and R2 s, both at the country and industry level, in different emerging regions increases international portfolio diversification benefits.19 5. Concluding remarks We use a dynamic PCA to examine the financial integration process in different emerging regions. As in Volosovych (2011, 2013), the degree of integration is captured by the proportion of total variation in individual excess returns explained by the first principal component. We focus on national equity markets as well as on ten different industrial equity markets. Financial integration is separately measured in one global emerging region (Emerging) and in three sub-regions (Asia, Eastern Europe

19 As discussed in existing empirical studies, betas tend to vary over time. For robustness, a sub-sample CAPM analysis is carried out. In line with these studies, we observe a higher average betas and R2 over the last 10 or 5 years. Still, a relatively high degree of heterogeneity across country and industry betas is found. Results are available upon request.

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

211

and Latin America). Our approach is unique and allows to examine whether there is heterogeneity in the level of integration across different equity markets in different regions. Of course, this has strong asset allocation implications. Our main results are as follows. First, we observe that the average level of integration in the global region is lower than in the sub-regions. However, equity indexes belonging to the same regions do not necessarily follow identical dynamics. Second, we find that the contagion- and hearding-effects produced by systemic banking crises do not affect country equity indexes and industry equity indexes homogeneously. This gives rise to three different integration index patterns: (i) increasing-trend; (ii) J-shaped trend; (iii) U-shaped trend. It turns out that this integration measure tend to be influenced by crises. In addition, we observe that (i) the integration index is rather low, suggesting that both national and industrial emerging equity markets are not perfectly integrated; (ii) the average level of integration varies across industries suggesting that some sectors (e.g. basic materials, financials) have largely contributed to the improvements in the degree of integration. A CAPM analysis is also conducted. We find that |1 − ˇ| is rarely close to zero. This confirms that both cross-country and cross-industry diversification benefits can be exploited. We conclude by arguing that our results have strong asset allocation and consumption smoothing implications. On one hand, the observed average low level of financial integration, the high degree of heterogeneity in the level of integration across industries and the weak co-movement between the return of a market-weighted portfolio and individual returns improve both cross-country and cross-industry diversification benefits. On the other hand, a low level of financial integration tends to produce an inefficient international risk-sharing environment.

Appendix A. Summary statistics See Tables A.1–A.3 and Fig. A.1

Table A.1 Country equity market excess returns: descriptive statistics. Notes: This table reports the mean, standard deviation, skewness and kurtosis values for the excess return of 18 emerging national equity markets. Country equity market excess returns are computed as defined in Eq. (3.1). Means and standard deviations are expressed in annual terms. ShR represents the Sharpe Ratio (i.e. excess return per unit of risk). The last column reports the presence of structural breaks in the series. Structural breaks are detected using the Bai–Perron test. Sample: January 1994 (or later)–December 2011. Market

Mean

St.Dev.

ShR

Skew

Kurt

BP-test

Argentina Brazil Chile China Colombia Hungary India Malaysia Mexico Pakistan Peru Philippines Poland Russia South Africa Sri Lanka Thailand Turkey

10.89 20.43 9.95 1.66 18.49 18.08 12.84 3.98 10.67 7.76 19.88 1.30 7.35 28.06 11.02 7.25 5.11 18.53

40.70 40.30 24.86 37.27 34.20 41.55 47.73 29.18 32.34 39.25 33.31 33.51 40.51 56.85 29.27 37.18 41.93 54.63

0.27 0.51 0.40 0.04 0.54 0.44 0.27 0.14 0.33 0.20 0.60 0.04 0.18 0.49 0.38 0.20 0.12 0.34

−0.05 −0.43 −0.80 0.33 −0.24 −0.56 0.21 0.03 −1.28 −0.07 −0.48 0.19 −0.39 −0.03 −0.79 1.56 0.43 0.18

5.60 2.77 2.90 3.10 1.60 4.82 2.33 3.80 4.14 1.69 5.66 2.60 2.35 2.63 4.84 6.86 4.15 2.16

No No No No No No No 1998:M10 No No No No No No No No No No

Source: Datastream.

212

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

Table A.2 Industry equity market excess returns: descriptive statistics. Notes: This table reports the mean, standard deviation, skewness and kurtosis values of the industrial equity market excess returns in the 18 emerging economies. Industry equity excess return are computed as defined in Eq. (3.1). Means and standard deviations are expressed in annual terms. ShR represents the Sharpe Ratio (i.e. excess return per unit of risk). The last column reports the presence of structural breaks in the series. Structural breaks are detected using the Bai–Perron test. Sample: January 1994 (or later)–July 2012. ShR

Skew

Kurt

Mean

St.Dev.

BP-test

BasMats Argentina Brazil Chile China Colombia Hungary India Malaysia Mexico Pakistan Peru Philippines Poland South Africa Thailand Turkey

16.88 19.42 16.13 21.66 28.00 12.83 18.25 5.80 24.08 10.97 17.82 9.27 17.74 19.08 15.38 32.43

41.76 41.05 28.21 52.02 44.22 40.64 39.49 37.36 42.78 33.81 29.00 58.66 46.23 42.77 54.61 62.04

0.40 0.47 0.57 0.42 0.63 0.32 0.46 0.16 0.56 0.32 0.61 0.16 0.38 0.45 0.28 0.52

1.85 0.01 0.33 1.03 1.70 −0.37 0.47 0.58 −0.20 0.15 0.21 0.98 −0.30 0.75 3.02 0.36

12.15 1.33 0.63 3.69 6.51 3.16 2.18 5.36 2.05 0.49 2.01 3.90 2.19 6.33 21.72 0.63

No No No No 2002:M10 / 2006:M2 No No No No No No 2003:M3 No No No No

ConsGds Argentina Brazil Chile China Colombia Hungary India Malaysia Mexico Pakistan Peru Philippines Poland South Africa Sri Lanka Thailand Turkey

3.49 24.16 11.34 23.83 11.31 9.31 16.38 9.51 2.12 14.47 11.27 10.32 12.40 22.94 14.32 12.31 22.08

51.86 35.35 25.86 50.96 34.34 38.50 28.67 33.65 43.30 34.71 25.57 28.11 29.47 39.75 33.49 45.05 55.65

0.07 0.68 0.44 0.47 0.33 0.24 0.57 0.28 0.05 0.42 0.44 0.37 0.42 0.58 0.43 0.27 0.40

0.27 −0.14 −0.22 3.23 −0.22 0.39 −0.04 0.71 0.93 0.55 0.61 0.36 −0.29 0.53 0.54 1.02 0.29

2.25 1.27 2.06 22.58 3.02 6.74 0.55 5.56 6.51 1.06 9.29 2.24 1.16 2.87 2.45 8.49 1.20

No No No No No No No No No 1998:M11 No 2009:M4 No No No 1998:M1 No

ConsSvs Argentina Chile China Colombia Hungary Malaysia Mexico Pakistan Philippines Poland Russia South Africa Sri Lanka Thailand Turkey

15.91 16.10 11.84 15.77 −1.13 7.91 10.41 6.28 5.65 19.33 48.94 15.24 6.75 10.93 29.15

43.15 28.33 45.49 45.99 68.48 29.82 32.11 57.17 39.66 40.60 54.51 34.28 34.26 32.82 69.73

0.37 0.57 0.26 0.34 −0.02 0.27 0.32 0.11 0.14 0.48 0.90 0.44 0.20 0.33 0.42

0.39 0.45 0.72 1.68 1.34 0.40 −0.52 1.99 0.78 0.37 2.96 −0.40 0.88 0.91 1.24

3.76 1.30 3.26 7.63 5.28 4.13 2.23 7.20 4.23 1.52 21.74 1.47 2.24 4.62 4.39

No No No No No No No No No No No No No 2008:M12 No

Financials Argentina Brazil

7.97 17.21

46.31 37.40

0.17 0.46

0.08 0.26

0.94 1.09

No No

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

213

Table A.2 (Continued ) ShR

Skew

Kurt

Mean

St.Dev.

Chile China Colombia Hungary India Malaysia Mexico Pakistan Philippines Poland Russia South Africa Sri Lanka Thailand Turkey

15.46 19.11 13.35 31.58 16.42 13.18 16.75 16.00 10.12 10.92 44.61 14.96 10.76 8.37 29.38

23.61 42.98 34.86 49.09 42.82 37.76 39.11 40.23 34.83 40.91 74.28 32.15 31.61 48.54 60.66

0.66 0.44 0.38 0.64 0.38 0.35 0.43 0.40 0.29 0.27 0.60 0.47 0.34 0.17 0.48

0.94 1.00 0.45 0.12 0.55 1.56 −0.20 −0.11 0.29 −0.39 1.39 −0.36 0.92 1.61 0.54

8.10 2.25 1.73 4.68 1.71 11.73 3.17 0.34 1.29 2.63 8.84 1.75 1.65 8.95 1.09

Healthcare Chile Hungary India Mexico Pakistan South Africa Thailand

19.11 21.21 12.48 31.66 7.63 15.98 12.96

33.62 43.04 26.19 42.09 36.08 31.68 31.70

0.57 0.49 0.48 0.75 0.21 0.50 0.41

1.18 0.71 −0.13 0.21 −0.41 −0.39 0.09

4.77 6.37 0.19 2.96 1.28 1.19 2.37

Industrials Argentina Brazil Chile China Colombia Hungary India Malaysia Mexico Pakistan Peru Philippines Poland South Africa Sri Lanka Thailand Turkey

10.58 23.59 14.13 18.07 29.67 16.81 30.97 4.29 8.36 22.08 28.34 13.06 5.89 13.08 12.71 19.48 28.45

42.99 37.61 29.07 49.44 42.66 39.60 52.58 30.11 41.93 101.88 72.29 41.64 37.48 33.19 33.83 48.11 65.91

0.25 0.63 0.49 0.37 0.70 0.42 0.59 0.14 0.20 0.22 0.39 0.31 0.16 0.39 0.38 0.40 0.43

1.49 −0.15 0.22 1.70 3.53 −0.09 2.32 −0.14 −0.25 11.83 7.60 0.70 −0.22 −0.49 1.03 0.89 0.83

7.60 0.77 0.59 7.78 20.58 1.57 15.61 3.63 3.42 165.08 90.98 5.63 1.13 1.51 2.63 2.69 3.07

OilGas Argentina Brazil Chile China Colombia Hungary India Malaysia Pakistan Philippines Poland Russia South Africa Sri Lanka Thailand Turkey

7.99 26.75 14.07 21.26 29.30 24.41 10.65 12.21 18.65 10.13 15.74 26.59 17.35 24.08 22.36 26.94

43.66 44.91 26.84 46.92 46.25 44.06 38.45 27.56 42.14 49.97 36.22 50.00 34.85 33.03 39.26 65.41

0.18 0.60 0.52 0.45 0.63 0.55 0.28 0.44 0.44 0.20 0.43 0.53 0.50 0.73 0.57 0.41

2.12 0.18 0.38 0.94 1.03 0.11 0.52 −0.02 0.49 1.48 0.17 −0.44 0.13 1.26 0.61 0.29

19.32 0.91 0.52 2.99 7.02 1.01 2.54 5.51 2.78 5.79 1.72 2.43 1.41 4.91 2.03 0.85

BP-test No No 2002:M11 No No No No No No No No No No No No No No No No No No No No No No No 1996:M12 No No No No No No No No No No No No No No No No No No No No No No No No No No No No

214

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

Table A.2 (Continued ) St.Dev.

Technology China Hungary India Poland Thailand Turkey

12.30 −6.24 39.49 8.69 16.45 45.04

35.87 57.76 56.02 43.83 54.21 78.66

0.34 −0.11 0.70 0.20 0.30 0.57

0.46 0.69 2.61 0.39 2.17 1.05

0.83 3.23 16.84 1.80 13.71 4.28

No No 1997:M1 / 2000:M3 No No No

Telecom Argentina Brazil Chile Hungary India Malaysia Mexico Pakistan Philippines Peru Poland Russia South Africa Thailand Turkey

7.60 15.07 8.65 7.96 10.96 9.64 17.99 4.23 8.36 23.35 9.89 23.18 28.29 10.12 27.13

46.71 39.94 30.68 38.31 45.44 31.80 33.14 43.75 31.27 57.70 36.93 55.59 47.98 43.46 67.23

0.16 0.38 0.28 0.21 0.24 0.30 0.54 0.10 0.27 0.40 0.27 0.42 0.59 0.23 0.40

0.41 0.02 0.22 −0.15 0.74 0.70 −0.21 0.13 0.21 4.23 0.23 0.90 0.35 1.24 1.39

1.47 0.29 0.51 0.20 2.54 5.70 0.53 0.98 0.12 38.06 0.71 7.15 2.78 7.36 4.80

No No No No No No No No 2002:M11 No No No No 1998:M1 No

Utilities Argentina Brazil Chile China Colombia Hungary India Malaysia Pakistan Peru Philippines Russia Thai Turkey

1.58 14.99 12.08 19.71 21.85 8.51 13.97 4.61 9.97 11.45 11.91 25.03 13.66 29.46

37.70 41.36 24.51 37.63 30.13 36.23 38.44 28.95 43.13 22.51 39.06 65.15 35.40 67.94

0.04 0.36 0.49 0.52 0.73 0.23 0.36 0.16 0.23 0.51 0.31 0.38 0.39 0.43

0.02 −0.04 0.21 0.14 2.11 0.08 0.21 0.27 0.43 −0.05 0.57 0.25 0.91 1.78

2.23 0.95 0.20 1.12 8.84 2.21 0.65 5.09 1.59 1.24 1.76 3.78 4.99 9.76

Source: Datastream.

ShR

Skew

Kurt

Mean

BP-test

No No No No No No No 1998:M9 No 1999:M5 2003:M6 No No No

Table A.3 Financial integration index: countries and time-horizons. Notes: This table reports the number of countries included (×) in the estimation of the dynamics of the financial integration process across emerging equity markets in each region. The integration index is computed as described in Section 3. Sector equity indexes are from level 2 of DGEI. Country equity indexes are represented by MSCI. Region

Emerging and developing economies Latin America

Market

ConsGoods

ConsService

Financials

Industrials

BasicMaterials

Oil and gas

Sample

Arg

Brazil

Chile

Col

Mex

Peru

LatAm Asia EastEu Emerg

×

×

×

×

×

×

×

×

×

×

×

×

LatAm Asia EastEu Emerg

×

×

×

×

×

×

×

×

×

×

×

×

LatAm Asia EastEu Emerg

×

×

×

×

LatAm Asia EastEu Emerg

×

×

×

×

×

×

LatAm Asia EastEu Emerg

×

×

×

×

×

×

×

×

×

×

×

LatAm Asia EastEu Emerg

×

×

×

×

×

×

×

×

×

×

LatAm Asia EastEu Emerg

×

×

×

×

×

×

×

×

× ×

×

×

×

×

×

×

×

Asia

Sample

China

India

Mal

Pak

Phil

SriLan

Thai

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

Eastern Europe

Sample

Hun

Pol

Rus

Tur

× ×

× ×

× ×

× ×

01/95–12/11

× ×

× ×

× ×

08/96–07/12

× ×

× ×

× ×

× ×

11/99–07/12

× ×

× ×

× ×

× ×

05/98–07/12

× ×

× ×

× ×

10/96–07/12

× ×

× ×

× ×

04/94–07/12

× ×

× ×

× ×

03/98–07/12

Africa

Sample

RSA

01/94–12/11 01/94–12/11 ×

01/95–12/11

×

08/96–07/12

×

11/99–07/12

×

05/98–07/12

×

10/96–07/12

×

08/94–07/12

×

03/98–07/12

08/94–07/12 01/94–07/12

01/94–07/12 01/94–07/12

08/94–07/12

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

01/94–07/12

08/94–07/12 01/94–07/12

08/94–07/12

×

×

01/94–07/12

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

Sector

08/94–07/12 12/96–07/12 × ×

215

216

Sector

Region

Emerging and developing economies Latin America

Telec

Utilities

HealthCare

Technology

LatAm Asia EastEu Emerg LatAm Asia EastEu Emerg LatAm Asia EastEu Emerg LatAm Asia EastEu Emerg

Sample Mex

Peru

×

×

×

×

×

×

×

×

×

Arg

Brazil

Chile

×

×

× ×

×

×

× ×

×

Col

×

×

×

Sample India

Mal

Pak

Phil

SriLan

Thai

Eastern Europe

Sample

Hun

Pol

Rus

Tur

× ×

× ×

× ×

× ×

12/98–07/12

× ×

× ×

03/98–07/12

Africa

Sample

RSA

08/94–07/12 ×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

×

12/94–07/12 ×

12/98–07/12

09/96–07/12

× ×

Asia China

08/95–07/12 × ×

03/98–07/12

08/98–07/12

×

×

×

×

×

×

×

×

×

×

×

×

×

01/94–07/12 ×

×

08/98–07/12

07/98–07/12 × ×

× ×

× ×

06/99–07/12 06/99–07/12

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

Table A.3 (Continued )

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

217

Fig. A.1. Emerging market structure.

Appendix B. Technical appendix B.1. The trend-cycle decomposition The trend-cycle decomposition is carried out via Unobserved Component Model (UCM) technique. In contrast to other methodologies (e.g. Hodrick–Prescott filter), this methodology is not affected by the end-of-sample issues (see St-Amant and van Norden, 1998, among others). The trend-cycle decomposition model is defined as follows: xt = t +

t

+ t

where t ∼NID(0, 2 ), t and t are the trend and cycle components, respectively, and xt represents our integration index. To get the long-run behavior of xt , we extract and isolate the trend component. The trend is imposed to be smooth by having a fixed level and a stochastic slope; the cycle is modeled as a stationary trigonometric cycle and the irregular is a standard white noise sequence (see Harvey and Trimbur, 2003). B.2. The Bai–Perron test The Bai and Perron (2003)’ break test is used to identify breaks in the mean of the market excess returns of various countries (Eq. (3.1)). Our analysis builds on Russell and Banerjee (2008) who employ the test Bai and Perron’s methodology to identify breaks in the average US inflation rate. The estimated model is: ExRet t = k+1 + t where ExRett represent excess returns and  k+1 is a series of k + 1 constants that estimates the mean rate of excess returns in each of k + 1 regimes and t is a random error. The model uses a trimming rate of 10 per cent of the sample. References Bai, J., Perron, P., 2003. Computation and analysis of multiple structural change models. Journal of Applied Econometrics 18, 1–22. Baig, T., Goldfajn, I., 1999. Financial market contagion in the Asian crisis. IMF Staff Papers 46 (2), 167–195.

218

M. Donadelli, A. Paradiso / Int. Fin. Markets, Inst. and Money 32 (2014) 184–218

Bekaert, G., Harvey, C.R., Lundblad, C., 2003. Equity market liberalization in emerging markets. Journal of Financial Research 26 (3), 275–299. Bekaert, G., Harvey, C.R., Lundblad, C., 2007. Liquidity and expected returns: lessons from emerging markets. Review of Financial Studies 20, 1783–1831. Bilson, C.M., Brailsford, T.J., Hooper, V.J., 2001. Selecting macroeconomic variables as explanatory factors of emerging stock market returns. Pacific-Basin Finance Journal 9, 401–426. Chambet, A., Gibson, R., 2008. Financial integration, economic instability and trade structure in emerging markets. Journal of International Money and Finance 27, 654–675. Chiang, T.C., Jeon, B.N., Li, H., 2007. Dynamic correlation analysis of financial contagion: evidence from Asian markets. Journal of International Money and Finance 26 (7), 1206–1228. Chkili, W., Nguyen, D.K., 2014. Exchange rate movements and stock market returns in a regime-switching environment: evidence for BRICS countries. Research in International Business and Finance 31, 46–56. Corsetti, G., Pericoli, M., Sbracia, M., 2005. Some contagion, some interdependence: more pitfalls in tests of financial contagion. Journal of International Money and Finance 24 (8), 1177–1199. de Jong, F., de Roon, F.A., 2005. Time varying market integration and expected returns in emerging markets. Journal of Financial Economics 78, 583–613. Donadelli, M., 2013. Global integration and emerging stock market excess returns. Macroeconomics and Finance in Emerging Market Economies 6 (2), 1–36. Donadelli, M., Prosperi, L., 2012. On the role of liquidity in emerging markets stock prices. Research in Economics 66 (4), 320–348. Donadelli, M., Persha, L., 2014. Understanding emerging market equity risk premia: industries, governance and macroeconomic policy uncertainty. Research in International Business and Finance 30, 284–309. Grootveld, H., Salomons, R., 2003. The equity risk premium: emerging vs. developed markets. Emerging Markets Review 4 (2), 121–144. Grubel, H., 1968. Internationally diversified portfolios: welfare gains and capital flows. American Economic Review 58, 1299–1314. Harvey, A.C., Trimbur, T., 2003. General model-based filters for extracting cycles and trends in economic time series. Review of Economics and Statistics 85, 244–255. Henry, P., 2000a. Stock market liberalization, economic reform and emerging market equity prices. Journal of Finance 55, 529–564. Henry, P., 2000b. Do stock market liberalization cause investment booms? Journal of Financial Economics 58, 301–334. Jolliffe, I.T., 2002. Principal component analysis series. Springer Series in Statistics, 2nd ed. Springer, New York. Laeven, L., Valencia, F., 2012. Systemic Banking Crises Database: An Update. Washington, IMF Working Paper No. 12/163. Lee, C., Chen, M., Chang, C., 2013. Dynamic relationships between industry returns and stock market returns. North American Journal of Economics and Finance 26, 119–144. Lessard, D.R., 1976. World, country and industry factors in equity returns: implications for risks reduction through international diversification. Financial Analysis Journal 32, 28–32. Levy, H., Sarnat, M., 1970. International diversification of investment portfolios. American Economic Review 60, 668–675. Mauro, P., Sussman, N., Yafeh, Y., 2002. Emerging markets spreads: then versus now. Quarterly Journal of Economics 117 (2), 695–733. Newey, W.K., West, K.D., 1987. A simple, positive semi-definite, heteroskedasticity and autocorrelation consistent covariance matrix. Econometrica 55 (3), 703–708. Newey, W.K., West, K.D., 1994. Automatic lag selection in covariance matrix estimation. Review of Economic Studies 61 (4), 631–654. Pukthuanthong, K., Roll, R., 2009. Global market integration: an alternative measure and its application. Journal of Financial Economics 92 (2), 214–232. Russell, B., Banerjee, A., 2008. The long-run Phillips curve and non-stationary inflation. Journal of Macroeconomics 30, 1792–1815. Sachs, J., Tornell, A., Velasco, A., 1996. Financial crises in emerging markets: the lessons from 1995. Brookings Papers on Economic Activity 1, 146–215. Saleem, K., 2009. International linkage of the Russian market and the Russian financial crisis: a multivariate GARCH analysis. Research in International Business and Finance 23, 243–256. St-Amant, P., van Norden, S., 1998. Measurement of the output gap: a discussion of recent research at the Bank of Canada. Bank of Canada Technical Report, No. 79. Volosovych, V., 2011. Measuring financial market integration over the long run: is there a U-shape? Journal of International Money and Finance 30, 1535–1561. Volosovych, V., 2013. Learning about financial market integration from principal components analysis. CESifo Economic Studies 59 (2), 360–391.