Regional Spillovers Across Transitioning Emerging and Frontier Equity Markets: A Multi-Time Scale Wavelet Analysis

By

Ginanjar DewandaruCIBAFI, General Council for Islamic Banks and Financial Institutions

Manama, Bahrain

Rumi MasihBank of New York Mellon Asset ManagementOne Wall Street, New York, NY 10286, USA

and

Mansur Masih*INCEIF, The Global University of Islamic Finance

Lorong University A, 59100 Kuala Lumpur, Malaysia

ABSTRACTThe episodic wave of crises experienced across the global financial markets over the past two decades have raised questions as to what extent frontier markets and transitioning emerging markets are vulnerable to exogenous shocks. These markets, by design, have not the institutional or financial architecture supporting their capital base compared to more established markets. We make the initial attempt to examine four such stock markets (Saudi Arabia, UAE, South Africa and Israel) and perform multi-timescale analysis using both time and frequency decompositions of wavelets in order to unveil (i) whether the shocks transmitted were pure contagion or fundamental-based and (ii) also whether the dynamic evolution of stock market integration was mainly short-term or long-term. We find that prior to the 2008/09 subprime crisis, contagion effects generated short-term shocks in contrast to US subprime crisis that reveals evidence supportive of fundamental-based contagion. Results also highlight the vulnerabilities associated with long-term integration. Specifically, we examine the dynamics of weak short-term market integration amongst the equity markets considered, in complement with a stronger long-term integration recently. Some recommendations have been proposed to improve stock market integration, including its stability and resiliency.

*Corresponding Author: Professor of Finance and Econometrics, INCEIF, The Global University of Islamic Finance, Lorong University A, 59100 Kuala Lumpur, Malaysia. Phone: +60-3-2272-3254; Fax: +60-3-7651-4110; Email: mansurmasih@gmail.com

Key Words: contagion, integration, Middle East and South African stock markets, wavelets

JEL Classification Codes: C22, C58, E44, G15

Acknowledgments: The views expressed in this paper are strictly those of the authors and are not necessarily shared by the Bank of New York Mellon, the General Council for Islamic Banks and Financial Institutions (CIBAFI), or any of their affiliated entities.

Regional Spillovers Across Transitioning Emerging and Frontier Equity Markets: A Multi-Time Scale Wavelet Analysis

ABSTRACTThe episodic wave of crises experienced across the global financial markets over the past two decades have raised questions as to what extent frontier markets and transitioning emerging markets are vulnerable to exogenous shocks. These markets, by design, have not the institutional or financial architecture supporting their capital base compared to more established markets. We make the initial attempt to examine four such stock markets (Saudi Arabia, UAE, South Africa and Israel) and perform multi-timescale analysis using both time and frequency decompositions of wavelets in order to unveil (i) whether the shocks transmitted were pure contagion or fundamental-based and (ii) also whether the dynamic evolution of stock market integration was mainly short-term or long-term. We find that prior to the 2008/09 subprime crisis, contagion effects generated short-term shocks in contrast to US subprime crisis that reveals evidence supportive of fundamental-based contagion. Results also highlight the vulnerabilities associated with long-term integration. Specifically, we examine the dynamics of weak short-term market integration amongst the equity markets considered, in complement with a stronger long-term integration recently. Some recommendations have been proposed to improve stock market integration, including its stability and resiliency.

Key Words: contagion, integration, Middle East and South African stock markets, wavelets

JEL Classification Codes: C22, C58, E44, G15

1. Introduction

The shock transmission across countries during financial crises has been an issue of great interest and recently generated a heated policy debate among market participants, central bankers, and governments, as to whether financial shocks in one country can have rapid and large impacts on other countries. Recently, the remarkable US born subprime crisis of 2008-09 that considerably hit the markets all over the world has raised a critical question on the capacity of the global financial system to maintain its financial stability in a meaningful way. The major interest has been its amplification which started out in the floating-rate segment of the US sub-prime mortgage market. Even emerging markets were not spared, with an example of the global crisis impact on Asian region1, which implies that the Asian region still absorbed indirect effects due to the deepening global financial integration (Zhang, Zhang, & Han, 2010). The findings on this particular issue may indicate the impact on the policy makers as to whether they are better off in liberalizing their financial markets (Furman & Stiglitz, 1998; Radelet & Sachs, 1998), which is recognized as a source of financial sector development.

When we look at studies focusing on the Middle East region, Neaime (2012) has empirically highlighted the impact of the recent subprime crisis on the emerging MENA equity markets. He argues that the transmission of external shocks into MENA countries can be attributed to their higher overall trade openness and their mismanagement in terms of domestic financial and macroeconomic policies. In addition, Beirne et al. (2010) also argue that regional spillovers have dominated in Latin America and Middle East while Asia is more exposed to global spillovers. Middle East as a region also represents some markets that represent “frontier equity markets”.

These periods of financial volatility have sparked questions on both the nature of the transmission mechanism across markets, as well as whether the sources of such linkages were due to contagion effects or more fundamental underpinnings. Many prior studies attempted to explain the nature of transmission mechanism across borders during several major financial crises. Dornbusch et al. (2002) and Kaminsky & Reinhart (2000) mentioned the two distinctive characteristics between pure and fundamentals-based contagion. While the former is defined as an excessive transmission of shocks beyond any idiosyncratic disturbances and fundamental linkages (Forbes & Rigobon, 2002; Bae et al., 2003; Eichengreen et al., 1996), the latter is transmitted by way of financial market integration and trade linkages or interdependence (Calvo & Reinhart, 1996; N’Diaye, Zhang, & Zhang, 2010; Zhang, 2008). The strong debate of empirical findings centres on the evidence of these two channels.

Given the presence of contagion effects in emerging countries, along with the different nature of their transmission channels, our study makes the initial attempt to address this issue in the context of equity markets in the Middle East and Africa, namely: Saudi Arabia, UAE, South Africa and Israel. There are three primary issues that this empirical study focuses on : (i) the first is to detect the evidence of contagion not only during the period of the recent subprime crisis but also during the earlier major crises. The finding may demonstrate how the impact of one crisis differs from that of the other crises. (ii) the second is to identify whether each crisis reveals the evidence of fundamental-based or pure contagion. We further examine the lead-lag relationship

1 The credit spreads in Asia, with lower exposures to US sub-prime mortgages, noticeably increased along with those in the United States and Europe (Brana & Lahet, 2010). Valuation losses on CDS in Asia had soared mainly due to global and region-specific risk pricing factors as well as revisions to expected losses from defaults (Kim, Loretan, & Remolona (2010). The Asian interbank markets were also affected by the distress in the US dollar market (Yu & Fung, 2009)

in order to capture the dynamics of transmission. (iii) the third is to investigate the evolution of short-run and long-run integration since well-integrated equity markets tend to have a higher exposure to external shocks. The length of daily observations extend over eighteen years starting from 1994.

As to the methodology, we examine co-movement and lead-lag relationship amongst the equity indices. There are two main reasons of using co-movement to address our research objectives. Firstly, many prior studies identified the evidence of contagion by observing the changing correlations across different markets during crises periods (Candelon, Piplack, & Straetmans, 2008; Chakrabarti & Roll, 2002; and so on). Secondly, the phases of financial integration also can be reflected through the increased cross-country correlations (Goetzmann et al., 2002; Bekaert et al., 2005; Bekaert & Harvey, 1995; Baele, 2005; and so on).

Our study is different from most of the prior studies in that we emphasize the multi-horizon nature of co-movement. We perform multi-timescale analysis using wavelet decompositions as one of the latest techniques in finance to decompose any observed variable on scale-by-scale basis. The decomposition may capture both time series and frequency domain simultaneously. This may provide an ability to distinguish between higher frequencies and lower frequencies. The concept is similar to the prior studies by Bodart & Candelon (2009) and Orlov (2009) that examined contagion by associating high and low frequencies with contagion and interdependence. In addition, to observe the phases of stock market integration, our study uses wavelet coherence with a rolling-window in the multi-horizon nature to evaluate the changing co-movement across the different equity markets.

The paper is organized as follows. Section 2 presents some literature reviews associated with the issue of financial contagion and integration. Section 3 introduces wavelet decomposition analysis of continuous wavelet transform. Section 4 presents and discusses the empirical results. Section 5 provides some overarching interpretations, conclusions and policy implications.

2. Literature review

Our literature review is structured as follows. The first is to present the theoretical underpinnings on financial contagion and interdependence. The second is to show some empirical findings of contagion effects in emerging equity markets. The third is to present some established methods of measuring contagion.

2.1. Contagion

Dornbusch et al. (2002) and Kaminsky & Reinhart (2000) have mentioned the two distinctive characteristics between “fundamentals-based” and “pure” contagion.

The pure contagion or excessive linkage is defined as an excessive transmission of shocks from the crash in origin country into others beyond any idiosyncratic disturbances and fundamental linkages (Forbes & Rigobon, 2002; Bae et al., 2003; Eichengreen et al., 1996). The sentiment shift of investors, unrelated to economic fundamentals (Kumar & Persaud, 2002; Dailami et al., 2008), may lead to a general reversal of funds and eventually trigger financial crises (Forbes & Rigobon, 2002; Kleimeier & Sander, 2003). The role of herd behaviour may burst asset bubbles created by self-fulfilling expectations, moral hazard, or government guarantees, either implied or explicit (Krugman, 1998).

There are mainly three mechanisms of shocks transmission. The first is the transmission of information from markets with more rapid price discovery, where the effect of news may

spread investors’ sentiment across borders (Kaminsky & Schmukler, 1999). Secondly, negative returns in one market may increase the risk premium in other markets, resulting in simultaneous drop of assets prices (Vayanos, 2004; Acharya & Pedersen, 2005; Longstaff, 2008). Thirdly, liquidity shock across countries plays an important role in contagion through a flight-to-quality (Allen & Gale, 2000; Brunnermeier & Pedersen, 2009). On the other hand, the fundamentals-based contagion is defined as a transmission of shocks by way of financial market integration and real linkages in non-crisis and crisis periods, thereby reflecting normal interdependence across borders (Calvo & Reinhart, 1996).

2.2. Empirical studies

Some empirical studies have been done in emerging countries to discover as to whether the certain crisis was spread via purely excessive linkages or merely interdependence. Candelon, Piplack, & Straetmans (2008) find a significant increase of cross-country correlations in five Asian stock markets Singapore, Thailand, South Korea, Taiwan and Malaysia, during Asian crisis. The stability test suggests that the rising co-movement is more of a sudden nature (i.e. contagion or ‘‘Asian Flu’’) rather than gradual (i.e. financial integration). Chung (2005) finds that the changes in the Thai baht exchange rate result in contagious effects on Asian ADR and country fund returns and volatility. Kaminsky & Schmukler (1999) study Hong Kong, Indonesia, Japan, Korea, Malaysia, Philippines, Singapore, Taiwan, and Thailand during the 1997 Asian crisis. They find that some of the large unlikely changes can be explained by apparent substantial news such as international organizations, credit rating agencies, and so on. The herd instincts as an overreaction of investors to bad news have played the major role during turbulence.

During the Argentine default on December 2001, Cifarelli & Paladino (2004) have investigated the crisis on daily spreads on sovereign bonds issued by 10 emerging countries in Latin America and Asia. They find long-lasting contagion effects in Latin America while the changes in daily spreads in Asia due to the Argentinean crisis seem to be less persistent, suggesting a temporary contagious reaction. Some other studies have pointed out the increase of capital inflows into Asia since 2003, which can be attributed to upgraded sovereign ratings in Asia as well as the abundance of world liquidity, the high level of carry trade strategies, and the low level of risk aversion, with the yen as the funding currency (Galati et al., 2007; Gyntelberg & Remolona, 2007). The role of foreign investor risk appetite becomes relevant and may result in a contagion effect if a reversal of investor opinion with regard to emerging market leads to capital outflows. Fernandez-Arias (1996) suggests that the flows are highly unstable if capital inflows are driven by external factors (push factors). Emerging countries are exposed to the evolution of macro and financial conditions in developed countries as well as foreign investor sentiment, becoming a source of fragility.

Looking at the impact of the most recent US-born crisis, credit spreads in Asia have widened even more than they did in United States and Europe, even though Asian region has a minimum exposure to subprime mortgages (Kim, Loretan, & Remolona, 2010). Their study mentions that an amplification mechanism may well explain the contagion effect driven by valuation losses, which can be attributed to the bursting of a global credit bubble. Specifically, it is a global re-pricing which can be considered as the main channel for contagion in Asian credit and equity markets. They find that the valuation losses on credit default swaps (CDS) have increased in part from movements in global and region-specific risk pricing factors. In another study, Longstaff (2010) uses data of ABX subprime indexes and finds evidence of contagion in

the financial markets, which was propagated via liquidity and risk-premium channels. Beirne et al. (2010) study 41 emerging market economies (EMEs) in Asia, Europe, Latin America, and the Middle East using trivariate VAR GARCH (1,1)-in-mean models to capture a transmission channels: spillovers in mean returns, volatility, and cross-market GARCH-in mean effects. They find that spillovers in variance play an important role in emerging Europe while spillovers in mean returns drive the contagion in emerging Asia and Latin America. In addition, regional spillovers have dominated in Latin America and Middle East while Asia is more exposed to global spillovers. In another study, Lagoarde-Segot & Lucey (2006) have used a fixed effect panel data in investigating whether the MENA markets of Turkey, Israel, Jordan, Lebanon, Tunisia, Morocco and Egypt are exposed to joint vulnerability to common exogenous shocks.

Some other studies have found the major role of interdependence during crises periods. Bodart & Candelon (2009) have studied the impact of Mexican and Asian financial crises among eleven emerging countries in Asia and Latin America, and have evidenced a contagion during both crises. Higher interdependence has led to the spread of the crisis in Asia as the spillover effects have been restricted to the region. Contagion is more regional than global (Glick & Rose, 1999; Kaminsky & Reinhart, 2000). By removing heteroskedasticity-bias from correlations, Forbes & Rigobon (2002) document little evidence of a significant increase in market co-movement during the 1994 Mexican and the 1997 Asian crises. This has been a merely interdependence rather than pure contagion. Some recent studies have discussed how the US economic downturn has an impact on the Asia–Pacific economy via the traditional trade linkage (N’Diaye, Zhang, & Zhang, 2010; Zhang, 2008). Chan, Chao, & Chou (2002) have discovered that Asian currency crisis in 1997 did spread among countries which have similar and close trade ties. Recently, Neaime (2012) has empirically highlighted the impact of the recent subprime crisis to the emerging MENA equity markets through measuring the correlation in returns and variances within the MENA region, and between MENA and the more developed financial markets of the US, UK, and FR. The study finds the common factors which have partially driven the volatilities of the MENA equity markets. He argues that the transmission of external shocks into MENA countries can be attributed to their higher overall trade openness and by mismanagement in domestic financial and macroeconomic policies. It seems to suggest the role of fundamental linkages, where the persistence of shocks will ultimately depend on the future domestic fiscal/financial policy responses.

So far, these empirical findings still involve an intensive debate as to whether shock transmission channels have been transmitted excessively or fundamental-based linkages. Our study therefore attempts to identify a presence of contagion and mainly to unveil the shocks transmission channels between markets in Middle East and South Africa.

2.3. Measuring financial contagion

The impact of crisis was extensively discussed in theoretical and empirical studies on how to test financial contagion across border. The contagion between two markets may happen: (i) from crash in origin country into another country; or (ii) from one country into another country due to a common shock attributable to crisis in origin country. Specifically, the contagion is detected if the change of correlation between the two markets, between pre-crisis and crisis periods, is statistically significant. Well-documented empirical studies have tested the

change of co-movement amongst the different financial markets by using econometric techniques2. However, there remain two major technical problems.

Firstly, Forbes & Rigobon (2002) pointed out that previous studies did not correct the correlation measure for heteroscedasticity so that the testing of contagion would become biased. They found little evidence of contagion in major crises periods when they used a heteroscedasticity-corrected correlation measure. Others studies have followed their methods and came up with more or less similar findings (Collins & Biekpe, 2003; Lee et al., 2007). However, this method has recently been criticized in terms of their assumed model, thereby the results would remain biased (Corsetti et al., 2005; Bartram & Wang, 2005; Pesaran & Pick, 2007).

Secondly, the standard time-domain instruments, using econometric techniques, have major difficulty in distinguishing fundamentals-based contagion from other shock transmissions. This is due in part to the problematic task of obtaining good proxies of the influence that comes from macroeconomic fundamentals.

Recently, Bodart & Candelon (2009) and Orlov (2009) proposed an alternative method to examine contagion by associating high and low frequencies with contagion and interdependence, respectively. Our study applies the same concept by using wavelet analysis. Wavelets are localized in both time and scale (frequency band) and can be used to decompose any observed variable on scale-by-scale (different frequency bands) basis in order to analyse the dynamics of co-movement across different time horizons without losing any information. There are two main advantage of using this method. Firstly, this approach will avoid the heteroscedasticity bias mentioned by Forbes & Rigobon (2002), given volatility should affect both low and high timescale correlations. Secondly, financial contagion and interdependence can be associated to its corresponding frequency bands, which is similar to Bodart & Candelon (2009) and Orlov (2009). Specifically, the study by Bodart & Candelon (2009) is focused on a partition based on frequency domain analysis, where the frequency ranges play an important role in distinguishing contagion from interdependence, with an outlier test.

On the other hand, wavelet serves as an interesting alternative to both time series and frequency domain methods since wavelet transforms the original series into different frequency components, with a resolution matched to its scale. In other words, wavelet provides a multi-resolution decomposition property which accommodate both time and frequency domains simultaneously, thereby decomposing into time-scale components. This characteristic is especially useful when we deal with signals which are both non-stationary, containing some outliers, and portraying the changing frequencies over time, as what we normally discover in financial data. Wavelet also provides an alternative representation of the variability as well as association structure of the certain stochastic process in a time series on a scale-by-scale basis. In that case, the ‘‘energy-preserving’’ property of the wavelet decomposition allows for a scale-based decomposition of the energy in a time series (Gallegati, 2010; Ashgarian and Nossman, 2011). Moreover, the use of wavelet assists in addressing the issues raised by Bodart & Candelon (2009) since we can observe the causal structure rather than on static correlation. This allows us to take into account the underlying dynamic structure, and deals with the problem of asymmetry. There are some recent studies which have applied wavelet to differentiate pure contagion from fundamental-based contagion or interdependence. 2 (King & Wadhwani, 1990), cointegrating relationships (Longin & Solnik, 1995), probit/logit models (Eichengreen et al., 1996; Baig & Goldfajin, 1999; Kaminsky & Reinhart, 2000), regime switching (Gallo & Otranto, 2008), the factor model (Corsetti et al., 2005), the copula approach (Rodriguez, 2007), principal components models (Kaminsky & Reinhart, 2000)

For example, using wavelet, Fernandez (2005) studied spillover in different time scales for markets in G7 countries, emerging Asia, Western Europe, Eastern Europe and the Middle East, the emerging Far East, Latin America, North America, and the Pacific region for the period 1990-2002. The results show evidence of price spillovers from the G7 countries to other sample countries, but a weaker relationship from other regions to G7 at different time scales. Gallegati (2010) has applied wavelet analysis on the stock market indices of G7 countries, including Brazil and Hong Kong. He found that: (i) international financial contagion between countries can be evidenced during the US subprime crisis, and (ii) these financial contagion effects are scale dependent, where the effects are not homogenous across scales. In another study, Ashgarian and Nossman (2011) filtered the integration variables by excluding all the scales related to the details. The effect of filtering at this point is to eliminate all the variations which belong to frequencies above 26 weeks, which is approximately one year.

2.4. Stock market integration

There are numerous empirical studies which attempt to measure the level of stock market integration across countries. In Middle East, some studies have mentioned that this emerging market still lags behind as compared to developed markets. Darat & Hakim (1997) has assessed price linkages among Arab stock markets (Amman, Cairo, and Casablanca) and their integration with global markets. They come up with the finding that these markets are integrated only within the region but not at the international level. In another study, Bulter & Malaikah (1992) find market inefficiency when they study individual stock returns in Kuwaiti and Saudi Arabian markets. The MENA region, as a part of the emerging markets, are typically much smaller, less liquid, and more volatile than developed markets (Domowitz et al., 1998). Recently, Neaime (2012) suggest that the MENA equity markets of Egypt, Jordan, Morocco, Tunisia and the UAE have been already integrated with the rest of the world's markets.

As an unclear debate pertaining to integration progress, another main contribution of our study is to investigate the evolution of short-run and long-run stock market integration of different markets in Middle East and South Africa by observing co-movement thoroughly at time-frequency space to further identify the dynamic linkages.

3. Methodology

We follow the study of Grinsted et al. (2004) and Aguiar-Conraria, Martins & Soares (2012) which apply wavelet coherency in the form of continuous wavelet transform. The continuous wavelet transform of a time series xt with respect to ѱ is a function of two variables given by the following convolution

where the bar denotes the complex conjugate, τ is the time position or translation parameter controlling its location, s is the scale or dilation parameter that controls the width of the wavelet,

and is a normalization factor to make sure that wavelet transforms are comparable across scales (frequency bands) and time series.

As to the mother wavelet, the most frequent choice is the Morlet wavelet which is given by

The Morlet wavelet is a complex sine wave within a Gaussian envelope while ω0 is the wave number (see, for example, Adisson (2002) for further details). Commonly, the results are obtained with a particular choice ω0 equal to 6 as it provides a good balance between time and frequency localization (see, for example, Grinsted et al. (2004)). Aguiar-Conraria & Soares (2014) mention that the Morlet wavelet is frequently used mainly due to four properties: (1) the three sensible ways in converting wavelet scales into frequencies are equal; (2) it has optimal joint time-frequency concentration; (3) the time radius and the frequency radius are equal; (4) it is an analytic wavelet.

Similar with Fourier analysis, several interesting features can be obtained in the wavelet domain. We can define the wavelet power spectrum as

It measures the relative contribution at each time and at each scale (frequency band) to the time series’ variance. Similarly, we can define the cross-wavelet spectrum as

where Wx(τ,s) and Wy(τ,s) are wavelet transforms of two time series x(t) and y(t). The cross-wavelet spectrum can be decomposed into real and imaginary parts since the mother wavelet is in general complex. We define the cross wavelet power, as |Wxy(τ, s)|, which depicts the local covariance between two time-series at each time and frequency. On the other hand, the wavelet coherency has a major advantage of being normalized by the power spectrum of the two time-series. Similar in Fourier analysis, we can define wavelet coherency of given two time-series x(t) and y(t) as:

where S denotes a smoothing operator in both time and scale (frequency band). We can use wavelet squared coherency to measure co-movement of two time series over time and across frequencies. Such a wavelet-based measure allows for a richer description on the co-movement between two variables of interest. Similar to the squared coefficient of correlation, R2(τ,s) is in the range between 0 and 1 with a high (low) value indicating a strong (weak) co-movement. Hence, by observing the contour plot of the above measure, we can identify the regions in the time–frequency space where the two time series move together and, specifically, assess both time and frequency varying features of the co-movement. We will rely on Monte Carlo simulations for statistical inference.

The use of a complex-valued wavelet is a major advantage that we can compute the phase of the wavelet transform of each series. By computing the phase-difference, we can obtain information about the possible delays of the oscillations of the two series as a function of time and scale (frequency band). The phase difference can be defined as

The information on the signs of each part is to determine the value of ϕxy ϵ [-π, π]. A phase-difference of zero indicates that the time-series co-move at the specified frequency; if ϕxy ϵ [0, π/2], then the series move in phase with time-series x leads y; ϕxy ϵ [-π/2, 0] indicates the leading role of y; a phase-difference is π (or –π) will be an anti-phase relation; if ϕxy ϵ [π/2, π] , then y is leading; time-series x is leading if ϕxy ϵ [-π, -π/2].

4. Data and Empirical Results

We apply wavelet analysis to stock indices in four different countries in Middle East and Africa regions, namely Saudi Arabia (Tadawul All Share), United Arab Emirate (DFM General), Israel (Tel Aviv 25) and South Africa (FTSE/JSE Africa). Specifically, we use daily data from 28 February 1994 to 31 August 2011 for Saudi Arabia and Israel, from 31 July 1995 to 31 August 2011 for South Africa, and from 30 January 2004 to 31 August 2011 for UAE. The continuous wavelet transform will be used to examine the multi-horizon nature of co-movements and lead-lag relationships. All data are sourced from Datastream International.

4.1. The evolution of volatility at time-frequency space

Figure 1 shows the continuous wavelet power spectrum for each index. The definition of wavelet power spectrum is the absolute value squared of the wavelet transform, which provides a measure of the time series variance at each time and at each scale (frequency). We measure the statistical significance of wavelet power against the null hypotheses of a stationary process with a background power spectrum. As estimated by Monte Carlo simulations using phase randomized surrogate series, the thick black contour in regions designates the 5% significance level (95% confidence level) against red noise. The cone of influence (COI), signalling distortion of the picture by edge effects, is displayed with a lighter shade. We must neglect the areas outside the cone of influence given there is no statistical confidence. The range of power is from blue (low power) to red (high power).

CWT: Tel Aviv 25 CWT: Tadawul All Share

CWT: JSE Africa CWT: DFM General

Figure 1. Wavelet power spectrum of Islamic equity indices The thick black contour designates the 5% significance level against red noise and the cone of influence (COI) where edge effects might distort the picture is shown as a lighter shade

Our plots of wavelet power spectrum have revealed the evolution of variance for all stock indices over entire observations. We can clearly evidence some high-power areas for each stock index temporarily, thereby suggesting high-volatility regimes due to market turbulence at certain periods. In the year 1994, the wavelet power spectrum of Israel stock index shows a small high-power area at 0.25 year frequency-band, which can be linked to the 1994 Mexican crisis. We also can see large high-power areas for South Africa at frequency-band below one year in the year 1997 till 1999. These can be attributed to the 1997 Asian financial crisis, 1998 Russian crisis as well as 1999 Brazilian crisis. In the following periods, the wavelet power spectrum of Israel stock index have revealed high-power areas at frequency-band below 0.5 year in the year 2000 till 2002, which can be referred to the overlapping impacts of dot-com crisis, Argentinean crisis, WTC terrorist attack and American stock market crash after the Enron and WorldCom bankruptcies in the year 2002. South Africa also has a minimum exposure to these major crises.

Overall, whole high-power areas during these crises periods have involved higher frequency bands (below one year), which therefore may infer short-term shocks to volatility attributed to investors’ sentiment (excessive).

On the other hand, our plots show significant high-power areas at both higher (below one year) and lower frequency-band (above one year) mostly in the year 2008. Given that the lower frequency-band may represent economic fundamental, it may imply that each index has been exposed to long-term shocks to volatility, thereby suggesting a structural problem during the recent global crisis. It seems that stock markets in Middle East are only exposed to the US-born subprime crisis, whereas South Africa is mostly affected by the crises before 2007. As to the Israeli stock market, it has absorbed the impacts from all concerned major crises more equally.

4.2. Stock market co-movements

Following the study by Dewandaru et al. (2015) in the context of Asian equity markets, in this section our study also uses wavelet coherence in the continuous form to decompose the series. As stated by Dewandaru et al. (2015), the advantage is that we may extend our decomposition to obtain more frequency bands (timescales) in longer horizon. This may provide robustness in discovering the evidence of: (i) pure contagion without relation to a sudden stop described in our previous empirical results; and (ii) fundamental-based contagion in multiple frequency bands rather than only in one smooth timescale. Wavelet coherence also applies a rolling-window in a multi-horizon nature. In particular, to investigate stock market comovements, we will decompose the stock index return mostly up to frequency band 4 years in order to discover the shock transmission channels either excessive or interdependence.

Figure 2 presents wavelet squared coherency and wavelet phase-difference between equity indices3. The wavelet coherency is displayed through a contour plot. The horizontal axis denotes time component while the vertical axis represents frequency component, which is converted to time units (years). The coherency ranges from blue (low coherency) to red (high coherency) to measure the degree of co-movement. Regions of high coherency between two equity indices indicate strong local correlation. The thick black line in the coherency plots will designate the statistical significant area at 5% significance level estimated from a Monte Carlo simulation. Therefore, the cross-wavelet coherency has a power to investigate varying characteristics of the relationship between index returns in the time–frequency domain.

Wavelet phase-difference, on the other hand, will attempt to discover the structure of financial linkages by observing the lead-lag relationships between the stock markets. We follow Grinsted et al. (2004) to measure wavelet phase-difference at a specific frequency (presented in the form of pointing arrows). To interpret the graph, for example for TADAWUL-JSE, the relative phase relationship is shown as pointing arrows: Right: in-phase; Left: anti-phase; Down: TADAWUL leading JSE by 90°; Up: JSE leading TADAWUL by 90°.

Comparatively, while the advantage of Grinsted et al. (2004) is for those who are interested in measuring phase-difference at a specific frequency, an additional advantage of Aguiar-Conraria, Martins & Soares (2012) is that it also can measure phase-difference at a particular frequency band on average presented in the form of graph. Our overall findings show that the results from the two confirm each other. Looking at Figure 2 on the right column, the wavelet phase-difference can be shown as graph at frequency band 4~8 years. A phase-

3The results of wavelet coherency are derived using Wavelet coherence package toolbox developed by Grinsted et al. (2004)

difference of zero indicates that the time-series co-move at the specified frequency4; if ϕxy ϵ [0, π/2] , then the series move in phase with time-series x leads y; ϕxy ϵ [-π/2, 0] indicates the leading role of y; a phase-difference is π (or –π) will be an anti-phase relation; if ϕxy ϵ [π/2, π] , then y is negatively leading; time-series x is negatively leading if ϕxy ϵ [-π, -π/2].

We also present cross wavelet power at Figure 3, where the power ranges from blue (low power) to red (high power). The red color will simply mean that the two series have high common power. Intuitively, the two markets share common high volatility regime.

To interpret the result of wavelet phase-difference in the field of finance and economics, we should be aware that the leading role of one market over another market does not necessarily mean that there is a specific causality between the two. We should interpret with caution that the two markets, in fact, co-move with one market taking a leading role over another. To explore further whether it implies any causation, we commonly need to investigate several channels of transmission, according to the documented theoretical and empirical studies, and are estimated using Granger causality in a multivariate framework. Since we focus on measuring market co-movements in a bivariate framework, this is therefore beyond the scope of our study, which may be a subject for future research.

TADAWUL-JSE: coherency plot TADAWUL-JSE: 1~4 years frequency band

TADAWUL-DFM: coherency plot TADAWUL-DFM: 1~2 years frequency band

4The results are derived using ASToolbox by Aguiar-Conraria & Soares (2011)

JSE-DFM: coherency plot JSE-DFM: 1~2 years frequency band

TEL AVIV-TADAWUL: coherency plot TEL AVIV-TADAWUL: 1~4 years frequency band

TEL AVIV-JSE: coherency plot TEL AVIV-JSE: 1~4 years frequency band

TEL AVIV-DFM: coherency plot TEL AVIV-DFM: 1~2 years frequency band

Figure 2.Wavelet squared coherency and phase-difference for overall stock marketsThe 5% significance level against red noise is shown as a thick contour. The relative phase relationship is shown as pointing arrows: Right: in-phase; Left: anti-phase; Down: first series leading second series by 90°; Up: second series leading first series by 90°. 4.2.1. Exploring periods relating to “contagion”

We consider that the higher frequency bands (below 1 year) may reveal information related to short-run market linkages, i.e. liquidity preferences, cross-border asset listing and trading, market panic/sentiment, and so on. On the other hand, the lower frequency bands (above 1 year) may represent fundamental linkages closely related to the real economy. This frequency range also may reflect long-run equilibrium relationship (convergence) amongst the different stock markets. The concept is similar to Bodart & Candelon (2009) and Orlov (2009) who have associated high and low frequencies with contagion and interdependence.

Looking wavelet squared coherency in Figure 2, we notice that all of our index-pairs mostly have been dominated by low coherency areas in the short term (higher frequency band). However, some co-movements increase temporarily during several major crises periods, and come back later to lower coherency after the crises are over. Using these plots, we therefore will define contagion as a temporary increase of co-movement immediately during certain crisis period, which is reflected as a higher coherency area in our coherency plots. The study of UAE only starts in 2004 onwards as its observations begin on 30 January 2004.

We can find a temporary increase to high coherency in 1997 and 1998 for index-pairs of Saudi Arabia-South Africa and Israel-South Africa at frequency-band below one year, as well as for Israel-Saudi Arabia at frequency-band below 0.5 year. These periods can be linked to the Asian and Russian financial crises. In 1997, we find the leading role of South Africa over Saudi Arabia, as well as Saudi Arabia and South Africa over Israel. Meanwhile, the pointing arrows in 1998 show that Saudi Arabia leads South Africa, Israel leads Saudi Arabia, and South Africa leads Israel. In the subsequent periods, we also can see high-coherency areas in the year 1999 for Saudi Arabia-South Africa at frequency-band below 0.25 year, and for Israel-South Africa at frequency-band below one year. This period can be referred to Brazilian crisis, with the leading role Saudi Arabia over South Africa and South Africa over Israel.

Our coherency plots also show high-coherency areas in 2000 till 2001 for index pairs of Saudi Arabia-South Africa and Israel-Saudi Arabia at frequency-band near 0.5 year, and for Israel-South Africa at frequency-band below one year. These can be attributed to the overlapping impacts of dot-com crisis, Argentinean crisis and WTC terrorist attack. It seems that the pair of Israel-Saudi Arabia is the least affected by these major crises. As to the lead-lag relationship, the pointing arrows show the leading role of South Africa and Israel over Saudi Arabia, as well as South Africa over Israel.

Overall, there is a tendency that all contagions prior to the US-born crisis period seem to involve channels unrelated to fundamental. The underlying reason could be the immediate increase of co-movement temporarily in the short-run frequency ranges, whereas the long-run frequency bands tend to show either no comovement or a more of gradual process persistently, indicating the evolution of long-run integration. This finding can be supported further by our results of cross wavelet power. When we look at Figure 3, cross wavelet power plots reveal common high-power areas temporarily at higher frequency band (below one year) during concerned major crises prior to the global crisis. This may imply that the excessive contagions have generated a common short-run high volatility regime amongst stock markets temporarily due to extreme price swings, thereby indicating substantial exposures to common short-term shocks. Another finding also illustrates that the degree of short-term contagions appear mostly at frequency bands above 0.25 years. This may imply that the reversal of funds is less likely to take place immediately when the crisis begins and could be attributed, amongst other factors, to a weak information linkage, slower adjustment of absorbing negative news, substantial transaction cost, and real-sector stocks.

Finally, the remarkable evidence exists in the year 2007/10, where all index-pairs are exposed to relatively very large high-coherency areas, which can be linked to US subprime crisis and followed then by the EMU sovereign debt crisis. The finding likely shows the evidence of fundamental-based contagion in response to structural shocks since the immediate increases of co-movements occur at both higher and mostly lower frequency bands. This has been in line with our cross wavelet power results at Figure 3 showing that all index-pairs have experienced both common short-term and long-term shocks to volatility temporarily, signifying market turbulence related to fundamental. As to the lead-lag relationships, the pointing arrows generally show the leading role of Saudi Arabia over South Africa and UAE, as well as Israel over UAE. Besides, bi-directional relationships can be observed in South Africa-UAE, Israel-Saudi Arabia, and Israel-South Africa.

Figure 3. Cross Wavelet TransformThe 5% significance level against red noise is shown as a thick contour. The relative phase relationship is shown as pointing arrows: Right: in-phase; Left: anti-phase; Down: first series leading second series by 90°; Up: second series leading first series by 90°. The red colour denotes high power spectrum.

4.2.2. Market integration

Many prior studies, as described earlier, suggested that the deepening market integration has increased the vulnerability of the market to external shocks. In this section, we therefore attempt to measure markets’ co-movements as a proxy of their integration level. Previous studies by Bekaert et al. (2005) and Goetzmann et al. (2005) have documented a positive causality of market integration into market dependence. Given the well-integrated equity markets, the asset prices will be explained mostly by common factors so that a local market return would be determined by their covariances with other markets (Bekaert & Harvey, 1995; Bekaert et al., 2002; Baele, 2005; Bekaert et al., 2005).

Wavelet coherency with a rolling-window may provide evaluation on the episodes of market integration. Following Dewandaru et al. (2014), from our coherency plots at Figure 2, we can see that the increase of co-movement at lower frequency bands (above 1 year) occurs gradually and tends to be more persistent. This frequency band may represent fundamental linkages, i.e. long-run convergence, macro-fundamental integration, closer trade ties, long-term equilibrium relationships, and so on.

As to the short-term comovements (high-frequency bands), we can clearly see that the appearances of high-coherency areas are mostly transitory at certain crises periods, which confirm the role of contagions. Blue areas seem to dominate higher-frequency bands for each index pair. This evidence tends to indicate very weak short-term integrations amongst different markets. The domination of blue areas also can be found at lower-frequency bands that imply weak long-term integrations mostly from the beginning of our observations. However, some index-pairs have experienced a gradual increase of long-run convergence. Specifically, the strong long-term integration for the index-pairs of Israel-Saudi Arabia has started in the year 1994, with the leading role of Israel, before they become weakly integrated in 2000. In the subsequent periods, the strong long-term integration for index-pairs of Israel-Saudi Arabia and Israel-South Africa emerges in the year 2003, whereby Israel has a leading role over the two markets. For the index-pair of Saudi Arabia-South Africa, the strong long-term integration starts recently in the end of the year 2006. Finally, high-coherency areas become considerably large for all index-pairs in the year 2007 onwards due to structural shocks during the recent global crisis.

5. Conclusions and policy implications

Our empirical results unveil a series of findings that deserve scrutiny at the policy level as well as implications for global equity investors. We show the presence of common shocks whereby equity markets in the Middle East and South Africa are significantly adversely influenced by historical periods of some major crises around the globe.

Prior to the recent subprime crisis, contagion effects have generated short-term shocks that may potentially involve, amongst other factors, excessive channels. The sentiment can be due to investors’ behaviour during a period of greater uncertainty, i.e. herding, financial panic, loss of confidence, and so on. The shift in investor sentiment should lead to a general reversal of funds and cause capital outflows, becoming a source of fragility. This finding has been substantiated by another empirical result that shows short-term volatility shocks temporarily amongst different stock markets, signalling a short-run high volatility regime due to extreme price swings during turmoil. In terms of the degree of vulnerabilities to volatility shocks, we find that stock markets in Middle East are mostly exposed to the US-born subprime crisis, whereas

South Africa is mostly affected by the crises before 2007. As to the Israeli stock market, it has absorbed the impacts from all concerned major crises more equally. From causal perspective, leading roles seem to vary from one crisis to another but we find that Saudi Arabia and South Africa generally play dominant positions.

On the other hand, the recent US subprime crisis and EMU sovereign debt crisis unveil a very different story where results support the evidence of fundamental-based contagion for all concerned index-pairs. The immediate increase of comovements, as well as common high volatility regime temporarily, can be observed at high and low frequency bands, which tends to indicate the presence of shocks permeating over both short and long horizons. It would be an indication of structural problem that substantially affects real economy, where policy makers should not rely merely on monetary adjustment to sustain. This assertion of fundamental channels also can be attributed to our finding that confirms a gradual increase of comovements recently for several index-pairs at lower frequency bands (above 1 year) towards the end of the sample. The rising comovement at this frequency range tends to signify the improvement of macro-fundamental linkages, higher economic integration and trade ties, and long-run stock market convergence. This is understandable as even though the direct impact from US subprime assets was limited in Middle East and South Africa, the global crisis can be well-explained by a higher overall trade integration, especially between Middle East with the rest of the world through the World Trade Organization, the Euro Mediterranean Free Trade Agreements, and other bilateral free trade agreements. This may increase an exposure to fundamental contagion and possibly explain the severity of the recent crisis’ impact on the concerned markets, i.e. a significant drop of stock market capitalization due in part to the plunge in the real estate market, the decline in oil prices and production, higher CDS spreads, liquidity shortages in global financial market, and so on (Khamis et al., 2010). As to the causal perspective, we generally find the leading roles of Saudi Arabia, as well as some bi-directional relationships.

As to the evolution of short-term integration, we find that weak comovements at higher-frequency bands dominate from the beginning till the end of observations. The less co-movements at this frequency band indicates a low degree of cross-border assets listing and trading, lack of opportunity for liquidity across border, less foreign investments into compliant equities, and so on. This has been confirmed by another result whereby the increasing co-movements at whole ranges of frequency bands are only transitory during economic turbulence, and they generally get back to their equilibrium (weak integration in nature) during crisis recovery. This is understandable as Middle East and South Africa equity market are still relatively less developed. For instance, in a previous study of Middle East region, Abdmoulah (2010) examines the evolving efficiency of the GCC markets in relation to the regulatory efforts to develop those markets. Abdmoulah (2010) finds insignificant improvement in the efficiency of those markets despite the efforts by regulators. Although the GCC markets implemented important regulations with regard to disclosure and transparency enhancement, the GCC markets are lacking important steps to fix other important aspects that affect market efficiency. For example, the GCC markets are still dependent on individual investors, who are more likely to be less informed, subject to high transaction cost, providers of limited liquidity, and subject to behavioral biases. Masih et al. (2010) have suggested to integrate the GCC markets and to facilitate cross listing, which can enhance liquidity, mitigate thin trading, and provide more depth to the market.

To this end, it is imperative to improve stock market development and resiliency in Middle East and South Africa. Firstly, as to the Middle East region where shocks are transmitted

via fundamental linkages during the recent crisis, it emphasizes the need to diversify away its economy from oil sector into the industrial and services sectors, which eventually will minimize their exposure to external shocks due to their heavy reliance on oil.

Secondly, markets in this region need to enhance the degree of regional short-term stock market integration at regional level by improving equity market infrastructure and its depth: (i) facilitating cross listing can enhance liquidity, mitigate thin trading, and provide more depth to the market, (ii) restructuring the trading mechanism and introducing the market maker role in the equity market can reduce transaction cost, reduce excessive volatility, and keep prices closer to their true value, (iii) expanding equity markets by encouraging public offering by family and public firms to attract more local and regional savings and to promote our concerned equity market as diversification haven for regional investors, and (iv) improving their investment climate through effective financial reforms in order to attract the regional funds.

Thirdly, the high degree of regional stock market integration would reduce the cost of capital. Firms may rely on the local/regional markets rather than the world market and therefore may reduce the exposures to external shocks around the globe. A well-functioning stock market through lower capital-raising costs can promote economic growth by fuelling the engine of growth through faster capital accumulation, as well as by improving economic efficiency and productivity through better resource allocation. More importantly, the enhancement of local capital market and investment climate as well as regional integration should prioritize on long-term capital flows; therefore, it may potentially minimize the exposures to short-term speculative flows which have been considered as the source of contagions at regional level.

References

Abdmoulah, W. (2010). Testing the Evolving Efficiency of Arab Stock Markets. International Review of Financial Analysis, 19, 25-34.Acharya, V., & Pedersen, L. (2005). Asset pricing with liquidity risk. Journal of Financial Economics, 77, 375–410Adisson, P. (2002). The Illustrated Wavelet Transform Handbook. London: The Institute of PhysicsAguiar-Conraria, L., & Soares, M. J. (2011). Business cycle synchronization and the Euro: A wavelet analysis.Journal of Macroeconomics, 33,

477–489Ahlgren, N., & Antell, J. (2010). Stock market linkages and financial contagion: A cobreaking analysis. The Quarterly Review of Economics and

Finance, 50, 157–166Ait-Sahalia, J., Cacho-Diaz, J., & Laeven, R.J. A. (2010). Modeling financial contagion using mutually exciting jump processes (NBER Working

Paper, No. 15850, p. 150, 2010)Allen, F., & Gale, D. (2000).Financial contagion. Journal of Political Economy, 108, 1–33Bae, K., Karolyi, G., & Stulz, R. (2003). A new approach to measuring financial contagion. Review of Financial Studies,16, 716–763Baek, I.-M., & Jun, J. (2011). Testing contagion of the 1997–98 crisis in Asian stock markets with structural breaks and incubation periods.

Journal of Asian Economics, 22 (5), 356-368Baele, L., & Inghelbrecht, K. (2010).Time-varying integration, interdependence and contagion. Journal of International Money and Finance, 29,

791–818

Baig, T., & Goldfajn, I. (1999).Financial market contagion in the Asian crises (IMF staff papers 46, 167-195) Washington, DC: IMFBartram, S. & Wang, Y. (2005). Another look at the relationship between cross market correlation and volatility. Finance Research Letters, 2 (2),

75-88Beirne, J., Caporale, G. M., Schulze-Ghattas, M., & Spagnolo, N. (2010). Global and regional spillovers in emerging stock markets: A

multivariate GARCH-in-mean analysis. Emerging Markets Review, 11, 250–260Bertero, E., & Mayer, C. (1989). Structure and Performance: Global Interdependence of Stock Markets around the Crash of October 1987

(C.E.P.R. Discussion Papers, Vol. 307)Billio, M., & Caporin, M. (2010). Market linkages, variance spillovers, and correlation stability: Empirical evidence of financial contagion.

Computational Statistics and Data Analysis,54, 2443-2458Bjerve, S., & Doksum, K. (1993). Correlation curves: measures of association as functions of covariate values. Annual Statistics, 21, 890–902Bodart, V., & Candelon, B. (2009). Evidence of interdependence and contagion using a frequency domain framework. Emerging Markets

Review, 10, 140–150Brana, S., & Lahet, D. (2010). Determinants of capital inflows into Asia: The relevance of contagion effects as push factors. Emerging Markets

Review, 11, 273–284Brunnermeier, M., & Pedersen, L. (2009).Market liquidity and funding liquidity. Review of Financial Studies,22, 2201–2238Bulter, K.C., & Malaikah, S.J. (1992). Efficiency and inefficiency in thinly traded stock markets: Kuwait and Saudi Arabia. Journal of Banking

and Finance, 16, 197–210.Calvo, S., & Reinhart, C. (1996). Capital flows to Latin America: is there evidence of contagion effects? In: Calvo, G., Goldstein, M., Hochreiter,

E. (Eds.), Private Capital Flows to Emerging Markets After the Mexican Crisis. Washington, DC: Institute for International EconomicsCalvo, G. A., & Mendoza, E. G. (2000). Rational contagion and the globalization of securities markets. Journal of international economics, 51

(1), 79–113Candelon, B., Piplack, J., & Straetmans, S. (2008). On measuring synchronization of bulls and bears: The case of East Asia. Journal of Banking

& Finance, 32, 1022–1035Cashin, P., Kumar, M. S., & McDermott, C. J. (1995). International integration of equity markets and contagion effects (IMF Working Papers

No. 95/110)Collins, D. & Biekpe N. (2003). Contagion: a fear for African equity markets? Journal of Economics and Business, 55, 285-297Corsetti, G., Pericoli, M., & Sbracia, M. (2005).Some contagion, some interdependence: more pitfalls in tests of financial contagion. Journal of

International Money and Finance,24, 1177–1199Chakrabarti, R., & Roll, R. (2002). East Asia and Europe during the 1997Asian collapse: a clinical study of a financial crisis. Journal of

Financial Markets, 5, 1–30Chan, K. S., Chao, C., & Chou, W. L. (2002).Trade similarities and contagion among the Asian crisis economies. Journal of Asian Economics,

13, 271-283Chiang, 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, 1206−1228Chiang, M.-H., & Wang, L.-M. (2011). Volatility contagion: A range-based volatility approach. Journal of Econometrics165 (2), 175-189Chou, R.Y.-T., Ng, V., & Pi, L.K. (1994). Cointegration of international stock market indices(IMF Working Papers No. 94/94)Chung, H. (2005). The contagious effects of the Asian financial crisis: some evidence from ADR and country funds. Journal of Multinational

Financial Management, 15, 67–84Cifarelli, G., & Paladino, G. (2004).The impact of the Argentine default on volatility co-movements in emerging bond markets. Emerging

Markets Review, 5, 427–446

Click, R.W., & Plummer, M.G. (2005). Stock market integration in ASEAN after the Asian financial crisis. Journal of Asian Economics, 16, 5–28.

Costinot, A., Roncalli, T., & Teletche, J. (2000). Revisiting the dependence between financial markets with copulas (Working Paper Credit Lyonnais)

Dailami, M., Masson, P., & Padou, J.J. (2008). Global monetary conditions versus country-specific factors in the determination of emerging market debt spreads. Journal of International Money and Finance, 27 (8), 1325–1336

Darat, A.F., & Hakim, S.R. (1997). Price linkages, efficiency, and integration of emerging stock markets in the Middle East. In: Conference on Regional Trade, Finance and Labor Markets in Transition, Beirut

Dewandaru, G., Masih, R. & Masih, A.M.M.(2015). Why is No Financial Crisis a Dress Rehearsal for the Next? Exploring Contagious Heterogeneities across Major Asian Stock Markets, Physica A: Statistical Mechanics and its Applications, 419, 241 - 259

Domowitz, I., Glen, J., & Madhavan, A. (1998). International cross-listing and order flow migration: evidence from an emerging market. Journal of Finance, 53, 2001–2027.

Dornbusch, R., Park,Y. C. & Claessens, S. (2002). Contagion: understanding how it spreads. World Bank Research Observer, 15, 177–197Dungey M., Fry, F., Gonzalez-Hermosillo.B.,& Martin V. L. (2005). Empirical Modeling of Contagion: A Review of Methodologies.

Quantitative Finance, 5 (1), 9-24Eichengreen, B., Rose, A. K., & Wyplosz, C. (1996).Contagious currency crises: first tests. Scandinavian Journal of Economics, 98, 463–484Favero, C.A., & Giavazzi, F. (2002).Is the International Propagation of Financial Shocks Non-linear? Evidence from the ERM. Journal of

International Economics, 57 (1), 231-46Fernandez-Arias, E. (1996). The new wave of private capital inflows: push or pull? Journal of Development Economics, 48 (2), 389–418Forbes, K. J., & Rigobon, R. (2002). No contagion, only interdependence: Measuring stock market comovements. Journal of Finance, 57, 2223–

2261Furman, J., & Stiglitz, J. E. (1998). Economic crises: evidence and insights from East Asia. Brookings Papers on Economic Activity, 2, 1–113.Galati, G., Heath, A., & McGuire, P. (2007). Evidence of carry trade activity (BIS Quarterly Review)Gallegati, M. (2010). A wavelet-based approach to test for financial market contagion. Computational Statistics and Data Analysis,

doi:10.1016/j.csda.2010.11.003Gallo, G.M., & Otranto, E. (2008). Volatility spillovers, interdependence and comovements: a Markov switching approach. Computational

Statistics and Data Analysis, 52, 3011–3026Glick, R., & Rose, A.K. (1999). Contagion and crisis. Why are currency crises regional? Journal of International Money and Finance, 18, 603–

617Goetzmann, W.N., Li, L., & Rouwenhorst, K.G. (2002). Long-term global market correlations (Yale ICF working paper 00-60, 52)Goldstein, M. (1998). The Asian financial crisis: causes, cures and systemic implications. In: Policy Analysis in International Economics, 55.

Washington, DC: Institute for International Economics Gonzalez-Hermosillo, B. (2008). Investor's Risk Appetite and Global Financial Market Conditions (IMF/WP/08/85)Grinsted, A., Moore, J. C., & Jevrejeva, S. (2004). Application of the cross wavelet transform and wavelet coherence to geophysical time series.

Nonlinear Processes in Geophysics, 11, 561-566Guo, F., Chen, C. R., & Huang, Y. S. (2011). Markets contagion during financial crisis: A regime-switching approach. International Review of

Economics and Finance, 20, 95–109Gyntelberg, J., & Remolona, E. (2007).Risk in carry trades: a look at a target currencies in Asia and the Pacific (BIS Quaterly Review)Hamao, Y., Masulis, R., & Ng, V. (1990).Correlations in price changes and volatility across international stock markets. The Review of Financial

Studies, 3, 281-307Inci, A. C., Li, H. C., & McCarthy, J. (2011). Financial contagion: A local correlation analysis. Research in International Business and

Finance,25, 11–25

Jeon, J., Oh, Y., & Yang, D.Y. (2006). Financial market integration in East Asia: Regional or global. Asian Economic Papers, 5 (1), 73–89.Kaminsky, G.L., & Schmukler, S.L. (1999).What triggers market jitters? A chronicle of the Asian crisis, Journal of International Money and

Finance, 18 (4), 537–560Kaminsky, G., & Reinhart, C. (2000). On crises, contagion, and confusion. Journal of International Economics, 51, 145–168Kaminsky, G., Reinhardt, C., & Vegh, C. (2003).The unholy trinity of financial contagion. Journal of Economic Perspectives,17, 51–74Kawai, M. (2005). East Asian economic regionalism: Progress and challenges. Journal of Asian Economics, 16, 29–55.Khamis, M., Senhadji, A., Hasan, M., Kumah, F., Prasad, A., & Sensenbrenner, G. (2010). Impact of the Global Financial Crisis on the Gulf

Cooperation Council Countries and Challenges Ahead (Middle East and Central Asia Department)Kim, S., Kim, S.H., & Wang, Y. (2006). Financial integration and consumption risk sharing in East Asia. Japan and the World Economy, 18,

143–157.Kim, S., Kim, S.H., & Wang, Y. (2007). Saving, investment and international capital mobility in East Asia. Japan and the World Economy, 19

(2), 279–291.Kim, D. H., Loretan, M., & Remolona, E. M. (2010). Contagion and risk premia in the amplification of crisis: Evidence from Asian names in the

global CDS market. Journal of Asian Economics, 21, 314–326King, M., & Wadhwani, S. (1990).Transmission of volatility between stock markets. Review of Financial Studies, 3, 5–33Kiyotaki, N., & Moore, J. (2002). Evil is the root of all money. American Economic Review: Papers and Proceedings, 92, 62–66Kleimeier, S., & Sander, H. (2003). Contagion and causality: an empirical investigation of four Asian crisis episodes. Journal of International

Financial Markets, Institutions and Money, 13, 171–186Kumar, M.S., & Persaud, A. (2002). Pure contagion and investors' shifting appetite: analytical issues and empirical evidence. International

Finance, 5 (3), 401–436Krugman, P. (1998). What happened to Asia? (Mimeo). Massachusetts: Massachusetts Institute of TechnologyLagoarde-Segot, T., & Lucey, B.M. (2006). Financial vulnerability in emerging markets. Evidence from the Middle East and North Africa . (The

Institute for International Integration Studies Discussion Paper Series, iiisdp114)Lee, S.B., & Kim, K.W. (1993). Does the October 1987 crash strengthen the comovements among national stock markets? Review of Financial

Economics, 3, 1993: 89–102Lee, H.-Y., Wu, H.-C., & Wang, Y.-J. (2007). Contagion effects in financial markets after the South-East Asia tsunami. Research in

International Business and Finance, 21, 281-296Longin, F., & Solnik, B. (1995). Is the correlation in international equity returns constant: 1960–1990? Journal of International Money and

Finance,14, 3–26Longstaff, F. A. (2008). Flight-from-leverage in distressed asset markets (UCLA Working Papers)Longstaff, F. A. (2010). The subprime credit crisis and contagion in financial markets. Journal of Financial Economics, 97, 436–450Loretan, M., & English, W. (2000).Evaluating Correlation Breakdowns during periods of market volatility (Working Paper Board of Governors

of the Federal Reserve System)Lowell, J., Neu, C. Richard, & Tong, D. (1998). Financial Crises and Contagion in Emerging Market Countries.(RAND,MR-962).

Masih, M., Alzahrani, M., & Al-Titi, O. (2010). Systematic risk and time scales: New evidence from an application of wavelet approach to the emerging Gulf stock markets. International Review of Financial Analysis, 19, 10-18.

Masson, P. (1999a). Contagion : macroeconomic models with multiple equilibria. Journal of International Money and Finance, 18 (4), 587–602Masson, P. (1999b). Multiple equilibria, contagion and the emerging market crises (IMF Working Paper WP/99/164)Neaime, S. (2012). The global financial crisis, financial linkages and correlations in returns and volatilities in emerging MENA stock markets.

Emerging Markets Review, doi:10.1016/j.ememar.2012.01.006N’Diaye, P., Zhang, P., & Zhang, W. (2010). Structural reform, intra-regional trade, and medium-term growth prospects of East Asia and the

Pacific - Perspectives from a multi-region model. Journal of Asian Economics, 21, 20–36Orlov, A. (2009). Co-spectral analysis of exchange rate comovements during Asian financial crisis. Journal of International Financial Markets,

Institutions and Money, 19,742–758Pesaran, M. H., & Pick, A. (2007). Econometric issues in the analysis of contagion. Journal of Economic Dynamics & Control, 31, 1245–1277Radelet, S., & Sachs, J. (1998). The East Asian financial crises: diagnosis, remedies, prospects. Brookings Papers on Economic Activity, 1, 1–74.Rodriguez, J. C. (2007). Measuring financial contagion: A Copula approach. Journal of Empirical Finance, 14, 401–423Saleem, K. (2009). International linkage of the Russian market and the Russian financial crisis: A multivariate GARCH analysis. Research in

International Business and Finance, 23, 243–256Samarakoon, L.P. (2011). Stock market interdependence, contagion, and the U.S. financial crisis: The case of emerging and frontier markets.

Journal of International Financial Markets, Institutions and Money, 21 (5), 724-742Vayanos, D. (2004). Flight to quality, flight to liquidity, and the pricing of risk(Working paper, London School of Economics)Wang, K.-M., & Thi, T.-B.N. (2007).Testing for contagion under asymmetric dynamics: Evidence from the stock markets between US and

Taiwan. Physica A, 376, 422–432Yu, I. W., & Fung, L. (2009). A study on the transmission of money market tensions in EMEAP economies during the credit crisis of 2007–2008

(Hong Kong Monetary Authority Working Paper, 09/2009)Yu, I., Fung, K., & Tam, C. (2010). Assessing financial market integration in Asia – Equity markets. Journal of Banking and Finance, 34, 2874–

2885 Yuhn, K.H., (1997). Financial integration and market efficiency: Some international evidence from cointegration tests. International Economic

Journal, 11, 103–116Zhang, Z. (2008). Can demand from China shield East Asian economies from global slowdown? (Hong Kong Monetary Authority Working

Paper)Zhang, W., Zhang, Z., & Han, G. (2010). How does the US credit crisis affect the Asia–Pacific economies? Analysis based on a general

equilibrium model. Journal of Asian Economics, 21, 280–292