Dynamic and frequency connectedness across Islamic stock indexes, bonds, crude oil and gold

Nader Trabelsi (Department of Finance and Investment, College of Economics and Administrative Sciences, Al Imam Mohammad Ibn Saud Islamic University (IMSIU), Riyadh, Saudi Arabia and Higher Institute of Computer Science and Management of Kairouan, LARTIGE, Tunisia)

International Journal of Islamic and Middle Eastern Finance and Management

ISSN: 1753-8394

Article publication date: 4 June 2019

Issue publication date: 21 August 2019



This paper aims to investigate the connectedness of Islamic Stock Markets in five regional financial systems, namely, the United States, the United Kingdom, Europe (EU), GCC (Gulf Cooperation Council) and APAC (Asia-Pacific Countries), and across different asset classes (i.e. bonds, gold and crude oil).


This methodology is inspired by Diebold and Yilmaz (2012) and Barunlik and Krehlik (2017) for performing dynamic variance decomposition network and for studying time–frequency dynamics of connectedness at different frequencies.


Results show that the nature of connectedness over the past decade is time–frequency dynamics. The decomposition of the total volatility spillovers is mostly dominated by the long-run component. Furthermore, dominant regions are the largest contributors of spillover index, with the lowest contribution in the system coming from the GCC market. Results also reveal a slightly higher volatility spillover index of Islamic than conventional equity indexes. Finally, the system that encompasses commodities and Islamic finance instruments, generates the much lower volatility spillover.


The findings have significant implications for portfolio managers who are interested in being able to predict asset returns, as well as for policymakers who are concerned with market stability.



Trabelsi, N. (2019), "Dynamic and frequency connectedness across Islamic stock indexes, bonds, crude oil and gold", International Journal of Islamic and Middle Eastern Finance and Management, Vol. 12 No. 3, pp. 306-321. https://doi.org/10.1108/IMEFM-02-2018-0043



Emerald Publishing Limited

Copyright © 2019, Emerald Publishing Limited


The most recent financial crises, i.e. the 2008-2009 global financial crisis (GFC) and the 2010 European sovereign debt crisis (ESDC), have shed light on the significance of systemic risk, and these have made the importance of the evaluation of financial and macroeconomic connectedness grow. In fact, the connectedness has increasingly become central in many areas of research such as risk management and portfolio allocation, and in formulating economic policies. Some researchers have provided evidence on the connection of financial entities (Diebold and Yilmaz, 2015; Barunik and Krehlik, 2017). The goal is to be able to predict which and when a financial entity will spill over. Although volatility connectedness among financial markets have received many studies, most research has focused on connectedness within the same type of asset classes, e.g. stock, bond, commodity futures and FX markets (Aftab et al., 2015; Ansgar et al., 2018; Aviral et al., 2018), but the connectedness within Islamic financial entities, and across Islamic equity indexes and other asset classes, has received little attention. Thus, our goal here is to investigate how Islamic stock indexes of various parts of the world interact with each other over time, and with other markets like bonds, crude oil and gold, by using the spillover index approach and extensions.

The spillover index approach has been recently proposed by DY (2009). The basic spillover index idea is simple but often effective in ranking assets by their systemic importance. Following the original method, the forecast error variance decomposition (FEVD) networks associated to an n -variable vector autoregression (VAR) is used to define weighted and directed networks from market data. The impact of each asset on the network is proportional to its contribution to FEVD of other assets. In DY (2012), authors proposed the generalized variance decompositions of Pesaran and Shin (1998) (Pearson and Shin GFEVD), which are invariant to variable ordering, to identify uncorrelated structural shocks from correlated reduced-form shocks. More recently, Barunik and Krehlik (2017) has been interested in frequency origins of connectedness variables. Precisely, instead of assessing contributions of forecast error variation in an item owing to shock arising in another item, they are hence interested in evaluating shares to forecast error variation in an item at a specific time frequency.

In this paper, we adopt these extensions to assess volatility spillovers across identical and different asset classes. First, we evaluate the time evolution of connectedness across Islamic stock indexes (identical assets) in five international markets. Second, we assess in the US market, the connectedness across Islamic equities and three key US asset classes such as bonds, crude oil and gold. These markets are selected for three major reasons. The major reason is the great interest of these asset classes for financial analysts and investors, who use them for risk-hedging alternatives or as investment opportunities. The second one is that the number and intensity of crises in these markets in the past decades have sharpened. This seeks to provide a valuable insight for Muslim and non-Muslim investors about the influences they have on one another and on the returns of Islamic equity market. The last one is that as the past few years have been a surge in research about bonds, oil and gold, and their connectedness with conventional stock market performance, this will ensure good comparison of our results to others.

We try here to examine whether Islamic finance markets (i.e. stock index and Sukuk) provide a special avenue for the international based investors. We suppose that a weak connectedness of Islamic indexes with other assets can allow a lower forecasting error variance. This measure is important for analysts working in portfolio management for reducing the exposure to future potential losses. Policymakers are also interested in understanding and controlling interactions between markets. In fact, an increase in the total spillover index (TSI) means that there is a rise in the number of links across assets, and hence, a higher level of systemic interconnectedness. Therefore, if policymakers could know better the link among asset classes, they would be able to prevent systemic risk during turmoil periods.

It should be noted that the number of studies analyzing the nature of the relationship between Islamic equity indexes and across Islamic assets and other asset classes is very small. The present study is meant to narrow this literature gap in three ways. First, the spillover measure proposed by DY (2012), and based on the VAR model, enables us to quantify the contribution of each index to global market uncertainty and to infer which market is the major or lowest contributor to total connectedness. Second, the paper also captures the time-varying connectedness among markets using a rolling-window approach. Third, the BK (2017)frequency spillover index enables us to decompose the contribution of each market to global connectedness at different frequencies.

Our underlying data are daily and cover the period November 07, 2005, to Marsh, 31, 2015, taken from DataStream. The data include Dow Jones (DJ) Shari’a-compliant indexes for the United States (US), the United Kingdom (UK), the EU, GCC and APAC. Daily information on US Dow Jones Sukuk index is also available.

Our results exhibit a time-varying pattern of connectedness of conventional and Islamic stock indexes among examined regions, with increases observed during periods of recent financial crisis, implying that the GFC and the ESDC greatly influenced the connectedness among international financial markets. We have also found that both the UK and the EU markets are the largest contributors of volatility spillovers. However, the financial shock spillover “From” and “To” GCC market is indeed infrequent, suggesting that the GCC market has relatively little influence in international markets. Once the frequency-domain analysis is carried out, we have found that the strong volatility spillover is driven by higher frequencies. Finally, examining for US investors’ volatility spillovers of three creative systems, we have found that the 100 per cent Islamic finance system can exhibit substantially lower TSI. In making a comparison between Islamic and conventional financial systems, we support the hypothesis in the Islamic financial system that traders are Islamic bonds (i.e. Sukuk) holders.

The remainder of this paper proceeds as follows: we begin by presenting a literature review on Islamic finance during the recent crisis periods, with a particular emphasis on the relationship between Islamic equity indexes to their conventional ones or to other traditional asset classes in Section 2. Then, we describe our methodology on how we calculate the average (i.e. total) spillovers and to identify connectedness frequency in Section 3. After that, we present our data and substantive results in Sections 4 and 5, respectively. To finish, we conclude in Section 6.

2. Literature review

Despite the large body of literature measuring systemic risk both theoretically and empirically, we will here concentrate on selected studies focusing on the vulnerabilities and resistance of Islamic assets when facing distressful periods. First, we will mention the study of Herwany and Febrian (2013) who test whether Indonesian financial stocks are immune from global financial distress. This study using co-integration tests and VAR framework has found that Shari’a-compliant stock prices are less fluctuating, less associated with macroeconomic indicators and less risky and therefore relatively immune to the financial crisis. This evidence from Indonesian financial market was also found by Shaista and Rizvi (2013), especially for the Asia Pacific and emerging markets by applying the continuous wavelet technique. Authors suggest that Islamic indexes have proved to be more stable owing to their rigid screening criteria. Dewandaru et al. (2014) use the similar technique to examine the contagion between Islamic and conventional equity indexes in five regions during the major crisis. They observe incomplete market integration and fundamentals-based contagion during the subprime crisis. Rizvi et al. (2015) also use the continuous wavelet approach and find that most global shocks were transmitted via excessive linkages from the US to APAC. The subprime crisis reveals a fundamentals-based contagion, while Islamic markets show traces of reduced exposure, owing to low leverage effect. This reduces the risk exposure of Islamic markets during the recent large-scale financial turmoil as further proved by many further studies. Mahjoub and Mansour (2014) generalized autoregressive conditional heteroskedasticity (GARCH) family models and show low-volatility spillovers between the US Stock Market and emerging Islamic Stock Markets. Dewandaru et al. (2015) conclude, with a study sampling between 2008 and 2012, that the differences in betas between Islamic and conventional indexes at most of the time frequencies (longer or shorter horizons) are not statistically significant. A few exceptions show equal returns with lower risks for Islamic indexes mostly at longer horizons in some countries. The results from this study are also held by Ho et al. (2014). By comparing the risk-adjusted performances of 12 global Islamic indexes and their conventional indexes and using several statistics derived from the capital asset pricing model (i.e. beta, Sharpe ratio, Treynor index and Jensen alpha), Ho et al. (2014) find out that Islamic indexes outperform conventional ones in a bear market owing to their lower volatility and betas’ measures. Results also reveal the presence of lower correlations for some Islamic sector-pairs (financials, utilities, and consumer services) at the short-term horizon. On the other side, the existence of contagion effects during time of financial turmoil between global conventional equity and bond indexes and the Morgan Stanley Capital International (MSCI) Islamic stock market indexes of the G7, BRICS, the EU and EMU, as well as the MSCI World Islamic Stock index and the Dow Jones Sukuk index, has been documented by the study of Dimitris et al. (2016). Their results from the asymmetric DCC model reveal no strong contagion evidence between conventional and Islamic equity and bond indexes. Undertaking a time-frequency analysis between compliant Shari’a stocks and Sukuk in the GCC markets, Alaoui et al. (2015) provide evidence of strong correlations for different time frequencies with a contagion effect for closer markets. They further find out significant evidence of flight-to-quality to the Sukuk market in promoting financial stability during the subprime crisis. In Alaoui et al. (2015), the estimation results based on dynamic conditional correlation (DCC)-GARCH model provide evidence of significant behavioral shifts in the Sukuk/Shari’a stock relationship which can be explained by market liquidity, US CDS spreads and crude oil prices.

Compared to the above results supporting the presence of a negative relationship between Islamic and conventional indexes, there is also a strand in the literature where the opposite relationship is supported. For instance, using an iterative cumulative sum of squares (ICSS) algorithm to identify volatility structural changes of several major Islamic and conventional indexes, Charles et al. (2011) shows that during GFC, both indexes were affected negatively. A slightly higher volatility of Islamic indexes as compared to their financial counterpart has been also documented. This empirical evidence is also confirmed by Guyot (2012), in which the performance comparison of several DJ Islamic indexes and their regional stock counterparts shows that Islamic indexes become more sensitive to geopolitical issues like, 9-11 attack, and subprime crises in comparison with the entire sample period. In a more recent literature, Merdad et al. (2015) test for the existence of an Islamic-effect by looking at differences in stock returns between Islamic and conventional firms in Saudi Arabia from 2003 to 2011. Their empirical results capture the presence of a negative relationship between Saudi Islamic firms and average stock returns. Further, using the automatic portmanteau test and variance ratio tests, Charles et al. (2015) show that Islamic portfolios outperform conventional ones during the crisis period. In other related literature, using the dynamic conditional beta approach, Ahmet (2015) suggests that Islamic equity markets have only lower distress risk than conventional ones in short term, with no significant difference in the levels of systemic risk during the GFC.

In line with the previous literature, few types of research studies have been conducted on Islamic equity interdependence with other asset classes (e.g. commodity, precious metals, currency, etc.). In a seminal paper, Hussin et al. (2012) argues that the Islamic stock return in GCC and Malaysia countries reacted mostly positively to oil price increases. Beside that Khan and Masih (2014), conclude that the correlations between commodity and Islamic equities evolve and are highly volatile. Moreover, Abdullah et al. (2015) show that the Singapore Islamic index is leading the other commodities. Recently, Nagayev et al. (2016) concluded that crude oil possibly indicates a robust negative connection with Islamic equity over the 2001-2003 periods in the high scale (256 days). They also show that there is a robust positive link from 2007 to 2013 in the medium and high scales. The researchers also conclude that there is a negative correlation between gas and Islamic equity. Zhang and Li (2016) argue that hikes in the oil–equity correlations can be a long-run phenomenon. More recently, Metadjer and Boulila (2018) investigated the causal relationship between Islamic bonds (Sukuk), oil and precious metals “silver and gold” prices in the Asia Pacific region. This study used the VAR model, which relies on daily data. The findings provide substantial evidence in favor of the relation between Sukuk and the commodity market variables (i.e. oil, gold, silver, etc.).

3. Methodology

We apply the variance decomposition matrix of DY (2012, 2015) to analyze the directional connectedness across different markets. This requires three steps.

3.1 Step 1

We estimate VAR coefficients. For that, we consider the n-variate process Yt = (yt,1, … , yt,n) described by the structural VAR(p) at t = 1, … , T.

(1) Φ(L)Yt=εt
where Φ(L)=pΦpLp is n × n p-th order lag-polynomial and εt is white noise (possibly non-diagonal) with zero mean and covariance matrix ∑. Yt is however the volatility of different asset classes.

3.2 Step 2

To use VAR coefficients to generate the h-step Pearson–Shin GFEVD, we retain the following moving average representation of VAR process:

(2) Yt=Ψ(L)εt
where Ψ(L) is an n × n infinite lag polynomial matrix of coefficients.

Let us define our own variance shares as the fractions of the h-step-ahead error variances in forecasting yj that are due to shocks to yj, for j = 1, 2 … , n, and across variance shares, or spillovers, as the fractions of the h-step-ahead error variances in forecasting yj that are due to shocks to yk, for k = 1, 2, … , n, such that jk This can be written in the form:

(3) (θh)j,k=((Σ)k,k)1h=0H((ΨhΣ)j,k)2/h=0H(ΨhΣΨh)j,j

3.3 Step 3

As hHθj,k1, DY propose in measuring pairwise-directional connectedness Cjk(h) a standardized GFEVD, given by:

(4) (θ˜h)j,k=(θh)j,k/k(θh)j,k

Then, the total directional connectedness from an index k to the other indexes is defined as:

(5) Cj(h)=100×jk,j=1nCj,k(h)/j,k=1nCj,k(h)

Similarly, the total directional connectedness of other indexes to index j is given by:

(6) Ck(h)=100×jk,k=1nCj,k(h)/j,k=1nCj,k(h)

The total aggregation of the variance decomposition across all indexes measures is as following:

(7) Ch=100×jk(θ˜h)j,k(θ˜h)j,k=100×(1Tr{θ˜h}θ˜h)

Tr{.} is the trace operator.

3.4 Frequency connectedness

After assessing overall error variation in asset j owing to shock arising in an asset k, we follow BK (2017) to evaluate shares of forecast error variation at a specific frequency band.

Formally, let us have a frequency band d = (a, b): a, b ∈ (−π,π), a < b. The generalized variance decompositions on frequency band “d” are defined as:

(8) (θd)j,k=12πΓ(ω)(f(ω))j,k

Using the spectra representation of GFEVD, it is straightforward to define connectedness measures on band “d” as:

(9) (θ˜d)j,k=(θd)j,k/k(θ)j,k

The frequency connectedness on the frequency band “d” is then defined as:

(10) CdF=100×(θ˜dθ˜Tr{θ˜d}θ˜d)

θ˜d measure reflects the sum of all elements of the θ̂d matrix.

The frequency connectedness decomposes the overall connectedness defined in equation (3) into distinct parts that, when summed, give the original connectedness measure C.

4. Data and descriptive statistics

In this paper, we concentrate on Dow Jones Shari’a-compliant indexes of the US, the UK, the EU, APAC and GCC markets (known to us as DJIM US, DJIM UK, DJIM EU, DJIM APAC and DJIM GCC respectively). Our data cover also daily information on Dow Jones Sukuk index, US 10-year Treasury bond yield, gold and US Benchmark oil market (i.e. West Texas Intermediate [WTI] Crude Oil). The data span the period November 7, 2005, to March 31, 2015, with a total of 1,942 daily observations. For the computation of volatility, we restrict the analysis to daily absolute returns as Wang et al. (2016a, 2016b). Tables I and II summarize the descriptive statistics of the data.

A comparison between daily return and volatility of conventional and Islamic DJ indexes is documented by Panel “a” and Panel “b” of Table I. Let us first observe from Panel “a” that Islamic DJ indexes and unscreened conventional counterparts have almost similar values, with high average return related to US DJ Shari’a-compliant index. From Panel “b,” one can see that the UK and the EU markets are more volatile than other geographical regions’ markets. The skewness indicates that daily returns of conventional and Islamic equity indexes have a left tail. Furthermore, the coefficients of kurtosis are significantly deviated than three for all series. Finally, the results for the augmented Dickey–Fuller test (ADF test) indicate that the time series are stationary.

Following Table II, it is observed that the average daily return is positive and varies within assets. The energy market gives the greatest average of daily return to other markets. It has also the very low average of daily volatility. The gold market seems the riskier. More interestingly, we can observe that the skewness is positive for all markets, except for the case of Sukuk. This implies that there is a greater chance that bonds, gold and WTI have been gone up rather than down during the past decade. Compared with the normal distribution, Kurtosis of all daily returns is higher. This is consistent with the presence of fat tails in data series. Finally, the results of the augmented Dickey–Fuller test (ADF test) indicate that series are stationary.

5. Empirical results

5.1 Directional spillover index

We use VAR (2) approach with 100-days ahead forecasting horizon (h) to construct Table III. Panel “a” shows the spillover index for Islamic stock indexes. Panel “b” is, however, relative to spillover index of conventional counterpart indexes. The TSI, expressed by equation (7), appears to have been relatively larger, which indicates that, on average, roughly half of the forecast error variances across all five markets come from spillovers. In addition, a comparison of Panel “a” with Panel “b” shows that Islamic and conventional equity spillover indexes are approximately equivalent, while bilateral spillovers are significantly different, ranging mostly from 1 to 40 per cent. For instance, we can learn from Panel “a” that innovations to the US index returns are responsible for 20.79 per cent of the error variance in forecasting 100-days-ahead EU error variance, but only 1.52 per cent of the error variance in forecasting 100-days-ahead GCC error variance. That is, volatility spillovers from the US to the EU are larger than from the US to GCC. As another example, we see that total volatility spillovers from the EU and the UK to others are much larger than total return spillovers from others to the EU and the UK, with net directional spillover being roughly 10 per cent (i.e. 74.52 − 61.99 = 12.53 per cent for the EU and 67.78 − 61.30 = 6.48 per cent for the UK). The latter suggests that the EU and the UK markets are more shock transmitters to other markets for the past decade. They are then the most influential markets. Whereas GCC is the tiny and negligible spillover transmitter and recipient of shocks, and hence less maker of further uncertainty (i.e. directional spillovers from and to GCC markets have varied between 0 and 3 per cent over our sample period). It is then interesting to note that GCC equities are less susceptible to global shocks. Concerning the sensitivity of APAC is, in general, more significant.

It also appears that TSI of Islamic DJ equity indexes is slightly higher than TSI of their conventional counterparts. In other terms, although DJ filtering criteria remove a large number of Shari’a-noncompliant firms, we observe a similar responsiveness for shocks which indicates a spillover effect among them. This result can support the conclusion of earlier studies that supported Islamic equities are a transmission risk of “From” and “To” conventional equity markets (Trabelsi and Naifar, 2017).

5.2 Rolling-window total spillover index

We now turn to estimate volatility dynamic connectedness using 200-day rolling window samples. The resulting time series of spillover index are presented in Figure 1. Our first observation reveals that Islamic and conventional volatility spillover indexes tend to move very much in harmony. In addition, the spillover shape clearly shows that while there are periods of increased market interdependence, there are other periods during which the spillovers of uncertainty were less important. Major events associated with peaks are indicated in Figure 1. In particular, we can discuss the total spillover plot following three cycles. The first one starts with the first signs of subprime worries until GFC. During this period, we show no trend but clear bursts associated with a set of events that led to the subprime crisis.

The second cycle started with the collapse of the Lehman Brothers at the end of 2008 and lasted till mid-2014. During this long episode, the spillover index related to Islamic stocks rose quickly from 40 per cent to reach close to 60 per cent at the end of 2008 and stayed high as the sub-sample window is moved in time to include the EU debt crisis starting in 2010 and the period of general economic decline around the world. Analogous results have been observed with conventional stock indexes, which is an indicative evidence of financial contagion (Mensi et al., 2017).

The third cycle started in mid-2014 and lasted until the first quarter of 2015. During this period, shocks to future uncertainty were transmitting less across the studied indexes. Moreover, although GCC economies have been adversely affected by the recent fall of crude oil prices, the spillover index dropped back and entered into a calm period, meaning that the GCC market had little impact on stability in other financial markets. This result indicates an increase in portfolio diversification.

5.3 Measuring spillover index into the different frequency domains

The frequency domain is the natural place to study the long-run, medium-run, or short-run connectedness shifts. In this study, we retain three time–frequency bands: Freq1 band: (pi/2, pi) which roughly corresponds to periods less than four days; Freq2 band: (pi/4, pi/2) roughly corresponds to periods between four days to ten days; and Freq3 band: (0, pi/4) roughly corresponds to periods more than ten days. Figures 2 and 3 present the frequency decomposition of connectedness over the studied period.

Similar results have also been observed in both systems (i.e. conventional and Islamic systems). Referring to Figure 2, the connectedness among markets picks during specific turmoil episodes: episode starting in mid-2007 and end in late 2009 when ESDC entered; episode between Q1 2010 and mid-2011; episode early 2012 to end 2013; and episode beginning at 2015 to Q3 2015. During these episodes, the large global system connectivity is driven mostly by low-frequency components or long-term periods (d is longer than 10 days). At this situation, the high connectedness is translated into long-term uncertainty driving, hence, increasing systemic risk in these periods.

While there are periods in which connectedness is created at low frequency, there are other periods in which the high-frequency connectedness was more important; we show in particular an important short-term connectedness over the periods. Q1 2009-2010, Q2 2011-2012 and more recently Q3 2013-2015. In other terms, these periods in which connectedness is created at high frequencies are periods when the studied financial markets reflect information rapidly. In this situation, shocks creating incertitude will directly impact investor’s behavior by altering their expectations within the next few days. This indicates that market participants are more certain about the long-term stability of the system. However, if the high connectedness is picked up by a lower frequency of the cross-spectral density, shocks may not be transmitted rapidly. This can affect long-term stock returns expectations and affect, in turn, the long-term investor’s behavior. As BK (2017), the increases in the long-term connectedness and the declines in short-term connectedness can be attributed to change in beliefs of market participants about the created information. In fact, when market participants interpret that the information generated a volatility shocks would quickly propagate and impact the stability of the system, this can create short-term connectedness. Moreover, the belief in the surge volatility persistence can amplify the uncertainty and therefore enhances long-term connectedness.

5.4 Cross-class assets’ analysis

In this sub-section, we analyze three systems: the first one (S1) is the benchmark, which included besides gold and WTI, US DJ and US Treasury Note 10Y; the second (S2) is an asset class mix which included instead of US DJ its Islamic counterpart; and the third (S3) is the 100 per cent Islamic set that involved besides Islamic DJ equity index, the DJ Sukuk index instead of conventional bonds. Exploring these systems allows us to give a global view of spillovers across these asset classes in the Islamic and conventional system.

The results reported in Table IV suggest the following. When the DY (2012) methodology is used, the results indicate that the four analyzed markets are not closely linked with each other. In particular, the total connectedness obtained from the three systems ranges from 12 to 18 per cent, with a remarkable lower level of spillover in the creative Islamic system. Furthermore, the results also suggest that the equity and crude oil markets are, on average, mostly net transmitters of return volatility shocks to other markets, while gold and bond market are, on average, net receivers. These results are also shared by Wang et al. (2016a, 2016b) and Aviral et al. (2018).

For the rolling window, the results of total connectedness obtained through DY (2012) and BK (2017) methods are presented in Figures 4 and 5, respectively. In Figure 4, we show the overall connectedness index ranging from 5 to 50 per cent, with a substantial variation over the period. Such a variation is expected because the studied period has included many turbulent times (see major events in Figure 1). Four bursts are obviously observed over the sample period. Moreover, spillover plots of three considered systems are quite different. This means that shocks have been transmitted across these systems with different strengths. Precisely, S3 has exhibited the high global connectedness index at the onset of the GFC with a remarkable jump to above 50 per cent. This can reflect that Islamic financial system has suffered more than conventional one during this period. The ESDC is a multi-year debt crisis with interactions between sovereign debts and banking problems. At the end of 2009, the benchmark system (i.e. S1) has been profoundly affected by this debt crisis. After that, the net spillover index has stayed high throughout the several stages of the ESDC with two clear bursts at the end of 2011 and 2012 (see events in Figure 2). At these picks, S1 and S2 appear to be highly connected systems. There is, however, a strong connectedness during periods of the crisis between different markets that could be interpreted as the result of contagion. This finding is in line with Mensi et al. (2016, 2017), Wang et al. (2016a, 2016b) and Sangar et al. (2018) who find evidence of substantial time-varying volatility spillovers during recent crisis.

Let us now consider these findings following the time-frequency version analysis.

It is evident from Figure 5 that the return volatility spillovers are time-frequency varying. Let us start with the subprime crisis (mid-2007). At the beginning of the period, the short- and long-term component of volatility forecast error variance steadily increased closely, but never exceeded 10 per cent. Things changed dramatically since the downfall of Lehman Brothers. At this time, connectedness jumped to the highest level attributed to long-term component. More precisely, the collapse of Lehman Brothers has confirmed to US investors the trigger of the worst crisis after the Second War World. At that time, the exceptional volumes of data and information and the extraordinary connections among markets created a large portion of future uncertainty which was transmitted with increasing persistence through the system. Similar behavior can be also observed during mid-2011 to 2013, where the increase in volatility of US Stock Markets sparked by ESDC spilled over to other markets. Compared to Figure 1, there was, however, a poor responsiveness of stock markets to gold and crude oil shocks (see events in Figure 4). On the other hand, it might be worth noting that there was, as shown in Figure 5, two major periods (i.e. [i] 2010-mid 2011 and [ii] early 2013 to mid-2014), when strong total connectedness was arising owing to an increase in short-term component. These periods coincide with visible troughs in the total connectedness index among markets (Figure 4). In this case, market participants were interpreting that the information creating volatility shocks would quickly propagate and impact the stability of the whole system, resulting in short-term connectedness. The results partly agree with recent studies, such as those conducted by Aviral et al. (2018), who also reported a time–frequency varying of the degree of connectedness across four global markets, namely, equity, bonds, CDS and foreign exchange markets.

The shapes of the three considered systems seem similar, with the lowest level related to Islamic one, especially throughout the several stages of the GFC and the ESDC. The Islamic finance can, hence, contribute significantly to financial stability through the reducing of uncertainty. It can be also a solution to the problem of financial contagion. All these findings are important to portfolio managers who are looking for tools to monitor the accumulation of risk.

6. Conclusion

This paper investigates the connectedness of Islamic stock return volatilities within various international markets (i.e. US, the UK, the EU, APAC, and GCC), and across different asset classes (i.e. bonds, crude oil and gold). Our methodology is inspired by DY (2012) and BK (2017) to perform dynamic variance decomposition network and to study time–frequency dynamics of connectedness at different frequencies.

We have found some evidence on the time-varying nature of the connectedness across regionals with spikes observed during hectic market periods over the past decade. The UK and EU markets are the largest contributors to return volatility spillovers. Whereas the GCC market is the smallest net recipient and transmitter return volatility shocks to other markets. We have also found that there is a slightly higher volatility spillover index of Islamic than conventional equity indexes. Following the frequency-domain analysis, our study concludes that total volatility spillover from higher frequency contributes most to the total connectedness, while the contribution from lowest frequency does exist but for a limited time. This following surge in the long-term connectedness and the decline in short-term connectedness are attributed to changes in market participants’ beliefs about the stability of the system. Finally, the system that encompasses gold, crude oil and Islamic finance markets, is not well integrated, and therefore, it is less resilient to systemic risk.

Our findings have policy and practical implications. First, risk management and policies can be achieved through the control and management of connectedness index. Second, Islamic financial markets can improve diversification opportunities and promote financial stability. This works when actual screening filters from various index providers will be monitored.


Islamic and conventional TSIs using 200-day rolling window

Figure 1.

Islamic and conventional TSIs using 200-day rolling window

Islamic spillover indexes using 200-day rolling window and for different business cycles

Figure 2.

Islamic spillover indexes using 200-day rolling window and for different business cycles

Conventional spillover indexes using 200-day rolling window and for different business cycles

Figure 3.

Conventional spillover indexes using 200-day rolling window and for different business cycles

TSIs using 200-day rolling window among different asset classes

Figure 4.

TSIs using 200-day rolling window among different asset classes

TSIs among different asset classes and for different business cycles

Figure 5.

TSIs among different asset classes and for different business cycles

Descriptive statistics of conventional and Islamic DJ equity indexes

Panel a. Daily log returns statistics of conventional and Islamic DJ equity index
Mean 0.03 0.02 −0.02 0.02 0.02 0.04 0.03 0.00 0.02 0.02
Median 0.08 0.09 0.02 0.06 0.06 0.05 0.09 0.03 0.05 0.06
Maximum 6.18 9.49 6.57 11.95 9.81 5.00 10.97 7.44 11.28 11.45
Minimum −5.57 −9.51 −7.86 −10.34 −10.08 −5.45 −9.80 −9.13 −9.57 −9.91
SD 1.23 1.37 0.93 1.49 1.50 0.98 1.41 1.07 1.52 1.46
Skewness −0.22 −0.62 −1.59 −0.10 −0.21 −0.33 −0.43 −1.60 −0.13 −0.11
Kurtosis 6.34 10.32 16.55 10.48 8.83 6.82 10.25 16.94 8.10 9.37
ADF −13.00 −12.50 −11.50 −13.30 −12.90 −13.10 −11.70 −11.60 −13.50 −12.70
Panel b.  Daily Volatility statistics of conventional and Islamic DJ equity indexes
Mean 0.84 0.92 0.50 1.02 1.01 0.68 0.96 0.57 1.08 1.01
Median 0.54 0.66 0.26 0.72 0.70 0.48 0.69 0.28 0.79 0.71
Maximum 6.18 9.51 7.86 11.95 10.08 5.45 10.97 9.13 11.28 11.45
Minimum 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
SD 0.89 1.00 0.76 1.08 1.09 0.69 1.02 0.90 1.06 1.04
Skewness 2.10 3.07 3.74 3.08 2.75 2.28 3.04 3.67 2.61 2.85
Kurtosis 8.47 18.41 22.75 19.88 15.32 10.44 18.93 21.52 15.31 17.42
ADF −6.10 −5.80 −7.00 −6.00 −5.80 −4.90 −5.80 −7.00 −6.00 −5.80

Descriptive statistics of other asset classes

Statistics Sukuk Bonds Gold WTI
Daily return statistics
Mean 0.02 0.29 0.93 1.66
Median 0.02 0.22 0.67 1.21
Maximum 7.81 3.63 9.35 17.83
Minimum −14.44 0.00 0.00 0.00
SD 0.47 0.27 0.92 1.67
Skewness −10.95 2.67 2.48 2.84
Kurtosis 513.92 19.40 15.03 16.80
ADF −42.32 −45.30 −43.00 −47.43
Daily Volatility statistics
Mean 0.28 0.92 1.65 0.09
Median 0.22 0.66 1.21 0.04
Maximum 3.63 9.35 17.83 14.44
Minimum 0.00 0.00 0.00 0.00
SD 0.27 0.91 1.66 0.46
Skewness 2.66 2.47 2.85 20.88
Kurtosis 19.38 14.96 16.95 548.02
ADF −41.52 −36.52 −40.75 −34.89

Spillover table of equity market volatilities, 07/11/2005-31/03/2015

To US UK EU GCC APAC Contribution from others Net
Panel a. DJ Islamic stock return volatilities
US 46.94 20.40 24.57 0.73 7.33 53.05 1.15
UK 18.52 38.69 32.68 0.89 9.19 61.30 6.48
EU 20.79 31.10 38.00 0.72 9.35 61.99 12.53
GCC 1.52 0.87 0.95 93.97 2.66 6.02 −1.88
APAC 13.35 15.39 16.30 1.78 53.15 46.84 −18.30
Contribution to others 54.20 67.76 74.52 4.14 28.54 106.48
Contribution with own 101.14 106.15 112.52 98.11 81.69 TSI = 45.84
Panel b. DJ conventional stock return volatilities
US 63.06 13.80 16.19 0.86 6.06 36.93 −3.27
UK 10.77 42.63 33.84 1.44 11.30 57.36 9.87
EU 12.16 33.75 42.81 1.24 10.02 57.18 12.11
GCC 1.33 1.92 2.00 91.48 3.25 8.51 −2.73
APAC 9.39 17.74 17.25 2.23 16.30 30.65 0.00
Contribution to others 33.66 67.23 69.29 5.78 30.65 190.63
Contribution with own 96.72 109.86 112.1 97.26 46.95 TSI = 41.32

Spillover table between different asset classes

Asset classes From to Net
Panel a. System 1: S1
DJ equity index 23.73 29.08 5.35
US Treasury Note 10Y 16.79 16.35 −0.44
Gold 14.48 9.81 −4.67
WTI 20.20 19.96 −0.24
TSI 18.80
Panel b. System 2: S2
DJ Islamic index 22.45 23.82 1.37
US Treasury Note 10Y 14.66 16.30 1.64
Gold 14.24 10.73 −3.51
WTI 18.71 19.22 0.51
TSI 17.52
Panel c. System 3: S3
DJ Islamic index 16.56 17.68 1.12
Sukuk 3.87 2.08 −1.70
Gold 12.49 11.23 −1.26
TSI 12.04


Abdullah, A.M., Saiti, B. and Masih, M. (2015), “The impact of crude oil price on Islamic stock indices of South East Asian countries: evidence from MGARCH-DCC and wavelet approaches”, Borsa Istanbul Review, Vol. 16 No. 4, pp. 219-232.

Aftab, M., Ahmad, R. and Ismail, I. (2015), “Dynamics between currency and equity in Chinese markets”, Chinese Management Studies, Vol. 9, pp. 333-354.

Ahmet, S. (2015), “Systemic risk in conventional vs Islamic equity markets”, Research Department of Borsa İstanbul, Working Paper Series No 28.

Ansgar, B. and Irina, D. (2018), “International spillovers in global asset markets”, Economic Systems, Vol. 42 No. 1, pp. 3-17.

Aviral, T., Juncal, C., Rangan, C. and Mark, W. (2018), “Volatility spillovers across global asset classes: evidence from time and frequency domains”, The Quarterly Review of Economics and Finance, doi: 10.1016/j.qref.2018.05.001.

Barunik, J. and Krehlik, T. (2017), “Measuring the frequency dynamics of financial connectedness and systemic risk”, available at: https://ssrn.com/abstract=2627599.

Charles, A., Olivier, D. and Pop A. (2015), “Risk and ethical investment: empirical evidence from Dow jones Islamic indexes”, Research in International Business and Finance, p. 35.

Charles, A., Pop, A. and Darné, O. (2011), “Is the Islamic finance model more resilient than the conventional finance model? Evidence from sudden changes in the volatility of Dow jones indexes”, International Conference of the French Finance Association (AFFI), 11-13 May 2011, available at: http://ssrn.com/abstract=1836751.

Dewandaru, G., Batcha, O.I., Masih, M. and Masih, R. (2015), “Risk-return characteristics of Islamic equity indices: multi-timescales analysis”, Journal of Multinational Financial Management, Vol. 29, pp. 115-138.

Dewandaru, G., Rizvi, S.A.R., Bacha, O.I. and Masih, M. (2014), “What factors explain stock market retardation in Islamic countries”, Emerging Markets Review, Vol. 19, pp. 106-127.

Diebold, F.X. and Yilmaz, K. (2012), “Better to give than to receive: predictive directional measurement of volatility spillovers”, International Journal of Forecasting, Vol. 28 No. 1, pp. 57-66.

Diebold, F.X. and Yilmaz, K. (2015), Financial and Macroeconomics Connectedness: A Network Approach to Measurement and Monitoring, Oxford University Press, New York, NY.

Dimitris, K., Naifar, N. and Dimitriou, D. (2016), “Islamic financial markets and global crises: contagion or decoupling?”, Economic Modelling, Vol. 57, pp. 36-46.

Guyot, A. (2012), “Efficiency and dynamics of Islamic investment: evidence of geopolitical events on Dow Jones Islamic market indexes”, Emerging Markets, Finance and Trade, Vol. 47 No. 6, pp. 24-45.

Herwany, A. and Febrian, E. (2013), “Global stock price linkages around the US financial crisis: evidence from Indonesia”, Global Journal of Business Research, Vol. 7 No. 5, pp. 35-45.

Ho, C.S.F., Rahman, N.A.A., Yusuf, N.H.M. and Zamzamin, Z. (2014), “Performance of globalIslamic versus conventional share indices: international evidence”, Pacific-Basin Finance Journal, Vol. 28, pp. 110-121.

Hussin, M.Y.M., Muhammad, F., Noordin, K., Marwan, N.F. and Razak, A.A. (2012), “The impact of oil price shocks on Islamic financial market in Malaysia”, Labuan eJournal Muamalat and Society, Vol. 6, pp. 1-13.

Khan, A. and Masih, M. (2014), “Correlation between Islamic stock and commodity markets: an investigation into the impact of financial crisis and financialization of commodity markets”, INCEIF, 15th Malaysian Finance Association Conference.

Mahjoub, J. and Mansour, W. (2014), “Islamic equity market integration and volatility spillover between emerging and US stock markets”, The North American Journal of Economics and Finance, Vol. 29, pp. 452-470.

Mensi, W., Boubaker, F.Z., Al-Yahyaee, K.H. and Kang, H.S. (2017), “Dynamic volatility spillovers and connectedness between global, regional, and GIPSI stock markets”, Finance Research Letters, doi: 10.1016/j.frl.2017.10.032.

Mensi, W., Hammoudeh, S., Nguyen, D.K. and Kang, S.H. (2016), “Global financial crisis and spillover effects among the US and BRICS stock markets”, International Review of Economics Finance, Vol. 42, pp. 257-276.

Merdad, H.J., Hassan, M.K. and Hippler, W.J. (2015), “The Islamic risk factor in expected stock returns: an empirical study in Saudi Arabia”, Pacific-Basin Finance Journal, Vol. 34, pp. 293-314.

Metadjer, W. and Boulila, H. (2018), “Causal relationship between Islamic bonds, oil price and precious metals: evidence from Asia pacific”, Journal of Islamic Economics, Vol. 10 No. 2, pp. 285-298.

Nagayev, R., Disli, M., Inghelbrecht, K. and Ng, A. (2016), “On the dynamic links between commodities and Islamic equity”, Energy Economics. Vol. 58, pp. 125-140.

Pesaran, M.H. and Shin, Y. (1998), “Generalized impulse response analysis in linear multivariate models, Economics Letters, Vol. 58 No. 1, pp. 17-29.

Rizvi, S.A.R., Arshad, S. and Alam, M. (2015), “Crises and contagion in Asia Pacific Islamic v/s conventional markets”, Pacific Basin Finance Journal, Vol. 34, pp. 308-319.

Shaista, A. and Rizvi, S.R. (2013), “The impact of global financial shocks to Islamic indices: speculative influence or fundamental changes?”, Journal of Islamic Finance, Vol. 2, No. 1, pp. 1-11.

Trabelsi, N. and Naifar, N. (2017), “Are Islamic stock indexes exposed to systemic risk? Multivariate GARCH estimation of CoVaR”, Research in International Business and Finance, Vo. 42, pp. 727-744.

Wang, G.J., Xie, C., Jiang, Z.Q. and Stanley, H.E. (2016a), “Extreme risk spillover effect in world gold markets and the global financial crisis”, International Review of Economics and Finance, Vol. 46, pp. 55-67.

Wang, G.J., Xie, C., Jiang, Z.Q. and Stanley, H.E. (2016b), “Who are the net senders and recipients of volatility spillovers in china’s financial markets?”, Finance Research Letters, Vol. 18, pp. 255-262.

Zhang, B. and Li, X.-M. (2016), “Recent hikes in oil-equity market correlations: transitory or permanent?, Energy Economics, Vol. 53, pp. 305-315.

Further reading

Aloui, C., Hammoudeh, S. and Ben Hamida, H. (2015), “Co-movement between sharia stocks and Sukuk in the GCC markets: a time-frequency analysis”, Journal of International Financial Markets Institutions and Money, Vol. 34, pp. 69-79.

Diebold, F.X. and Yilmaz, K. (2009), “Measuring financial asset return and volatility spillovers, with application to global equity markets”, The Economic Journal, Vol. 119 No. 534, pp. 158-171.

Corresponding author

Nader Trabelsi can be contacted at: nadertrabelsi2003@yahoo.fr or nhtrabelsi@imamu.edu.sa