Asymmetric return and volatility transmission in conventional and Islamic equities

Umar, Zaghum; Suleman, Tahir Article Asymmetric return and volatility transmission in conventional and Islamic equities Risks Provided in Cooperation with: MDPI – Multidisciplinary Digital Publishing Institute, Basel Suggested Citation: Umar, Zaghum; Suleman, Tahir (2017) : Asymmetric return and volatility transmission in conventional and Islamic equities, Risks, ISSN 2227-9091, MDPI, Basel, Vol. 5, Iss. 2, pp. 1-18, https://doi.org/10.3390/risks5020022 This Version is available at: https://hdl.handle.net/10419/167917 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by/4.0/ risks Article Asymmetric Return and Volatility Transmission in Conventional and Islamic Equities Zaghum Umar 1,* and Tahir Suleman 2 1Suleman Dawood School of Business, Lahore University of Management Sciences Sector U, DHA, Lahore Cantt. 54792, Pakistan 2School of business, Wellington institute of technology, Wellington 5012, New Zealand; [email protected] *Correspondence: [email protected]; Tel.: +92-42-3560-8434 Academic Editor: Mogens Steffensen Received: 13 December 2016; Accepted: 27 March 2017; Published: 29 March 2017 Abstract: This paper analyses the interdependence between Islamic and conventional equities by taking into consideration the asymmetric effect of return and volatility transmission. We empirically investigate the decoupling hypothesis of Islamic and conventional equities and the potential contagion effect. We analyse the intra-market and inter-market spillover among Islamic and conventional equities across three major markets: the USA, the United Kingdom and Japan. Our sample period ranges from 1996 to 2015. In addition, we segregate our sample period into three sub-periods covering prior to the 2007 financial crisis, the crisis period and the post-crisis period. We find weak support for the decoupling hypothesis during the post-crisis period. Keywords: Islamic stock market; conventional stock markets; asymmetric return and volatility spillovers; EGARCH JEL Classification: G01; G10; G15 1. Introduction The resilience of Islamic financial assets during the global financial crisis of 2007 has attracted the attention of academics, investors and policy makers around the world. According to the Islamic financial services industry stability report (IFSB 2015), Islamic financial assets exhibited an impressive compound annual growth of 17% during the period 2009–2013. This phenomenal growth in the Islamic finance assets has inspired researchers to investigate the risk return characteristics of Islamic finance assets. In addition, the performance of Islamic financial assets vis-à-vis conventional financial assets has also attracted a lot of attention and a number of studies have documented the comparative analysis of Islamic and conventional financial assets. The main difference between Islamic and conventional financial assets is that Islamic financial assets must comply with certain restrictions derived from the teachings of the Islamic faith. However, from an investor’s perspective it is important to analyse the transmission of these restrictions on the risk return characteristics of Islamic financial assets. It is also important to analyse how the risk return characteristics of Islamic financial assets differ from the risk of other available (conventional) financial assets. The bulk of the existing literature is focused on the comparative performance of Islamic and conventional financial assets [ 1 – 8 ] The issue of potential risk transmission between Islamic and conventional financial assets is relatively less explored. This paper contributes toward this strand of literature by analysing the risk transmission mechanism between Islamic and conventional equities. In view of the fundamental differences between Islamic and conventional financial assets, one might argue against the potential transmission of risk or volatility across Islamic and conventional equities [ 9 ]. Risks 2017,5, 22; doi:10.3390/risks5020022 www.mdpi.com/journal/risks Risks 2017,5, 22 2 of 18 Majdoub and Mansour [ 10 ] document the weak volatility transmission] between the USA and five emerging Islamic market equity indices. There results were based on BEKK-MGARCH, CCC and DCC models. However, Hammoudeh et al. [ 6 ] report a significant dependence structure between Islamic and conventional equity indices. The results are drawn from copula-based GARCH models. Similarly, Nazlioglu et al. [ 11 ] document evidence of volatility transfer between the Islamic and conventional indexes using the causality-in-variance approach. Rejeb [ 12 ] uses a GARCH model and the quantile regression technique to highlight the existence of strong interdependencies between the conventional stock market and Islamic ones, especially from the conventional developed markets to the emerging Islamic and Arab markets and the Islamic developed markets. Thus, the relatively sparse empirical literature on the issue of volatility transmission between Islamic and conventional equities is showing mixed results. Koutmos and Booth [ 13 ] point out the importance of the quantity (captured by the size of an innovation) and the quality (captured by the sign of an innovation) of news in analysing the transmission mechanism across equity markets. The asymmetric effect of past volatility on current volatility in equity markets is widely documented. In particular, Saadaoui and Boujelbene [ 14 ] investigate the transmission of volatility between the Dow Jones stock index and the Dow Jones emerging Islamic stock index using vicariate BEKK-GARCH and DCC-GARCH model and find no evidence of a shock spillover effect between them. Assessing the co-movements among Islamic equity markets versus their conventional counterparts across different regions (Asia–Pacific, USA, Eurozone and United Kingdom), Dewandaru et al. [ 5 ] find incomplete market integration, with Islamic markets demonstrating a higher fundamental integration. Using Engle and Granger’s cointegration technique, El Khamlichi et al. [15] explore the ethical equities potential for diversification in comparison to their conventional counterparts and find an absence of cointegration among two index families (Dow Jones and Standard & Poor’s), therefore indicating diversification opportunities for these indices. Moreover, their work highlights similar tendencies and levels of (in) efficiencies in both Islamic and conventional indices. The purpose of this study is to examine the asymmetric volatility transmission between Islamic and conventional markets. Thus, we test the validity of the decoupling hypothesis of Islamic equities from their conventional counterparts by taking into consideration the asymmetric effects of volatility transmission. In addition, we analyse the standalone regional volatility spillover for both conventional and Islamic equities. One of the drawbacks of the financialization and integration of equity markets is increased interdependence among international markets. This increased dependence has led to a reduction in diversification benefits and an increase in the contagion risk during bad times. Highlighting the financialization of commodity markets, Saadaoui and Boujelbene [ 14 ] find that the subprime crisis contributed to developing a relationship between conventional and emerging Islamic Dow Jones Indexes, and higher correlations between them were witnessed during the financial crisis. The regional spillover dynamics of Islamic and conventional equities helps us to see the degree of integration between the Islamic and conventional markets. In order to capture the asymmetric effect of volatility transmission, we employ a multivariate VAR-EGARCH model. To the best of our knowledge, this is the first paper to analyse the volatility transmission between Islamic and conventional markets by employing this methodology in a multivariate framework. The multivariate VAR-EGARCH model enables us to test the possibility of asymmetric volatility transmission across these equity markets. The results from this paper have a number of implications. From the perspective of investors, it will be useful to analyse the volatility spillover for portfolio diversification and hedging purposes. In particular, it has investment and portfolio implications for institutional investors such as pension funds and insurance companies looking for alternative investment avenues. For investors, the absence of cointegration between conventional and Islamic stock indices signals opportunities for long-term portfolio diversification. Research has shown the presence of mutual risk transmission between the Islamic and conventional stock markets, which indicates the presence of contagion, unaffected by the financial crisis [ 11 ], thereby having important implications for institutional investors. The contagion Risks 2017,5, 22 3 of 18 effect makes returns on investment less certain and questions the return potential of Islamic equities in the diversified portfolio. As far as gains from portfolio diversification are concerned, cointegrated assets exhibit limited gains through portfolio diversification [ 15 ]. Interestingly, research has shown that the Islamic equity market responds to shocks from the risk factors and not from the oil price and the U.S. economic policy uncertainty index [ 11 ] pre- and post-2008 crisis. Therefore, the extent to which Islamic assets can be regarded as a safe investment option during times of financial crisis can be questioned and can hold important implications for investors who aim to benefit through portfolio diversification. It is important to note that Islamic investors must be cautious of structural shocks (such as those ingrained in trade linkages), as these may adversely affect returns [ 5 ]. Notably, investors can receive higher short-term diversification benefits from investing in a mix of EU and U.K. as well as developed and emerging markets [5]. For institutional investors, the lower exposure of Asian Islamic markets to financial leverage can provide a suitable investment hedge. From a strategic investment perspective, investors can maintain a balanced investment portfolio with a strategic asset allocation to Islamic equity as it can ensure a sustainable stream of returns along with a controlled degree of risk across markets [ 5 ]. For policy makers, the empirical evidence on volatility spillovers can be a useful ingredient in formulating policies for market stability. It will also help us analyse whether the decoupling hypotheses between Islamic and conventional finance holds. We employ aggregate Islamic and conventional equity indices for the USA, United Kingdom and Japan. We analyse the volatility transmission across the aggregate Islamic and conventional indices. Our sample period spans from 1996 to 2015. In addition we segregate our sample period into three sub-periods capturing pre-crisis (1996–2007), crisis (2007–2011) and post-crisis (2011–2015). The sub-sample analysis allows us to capture the return and volatility transmission before, during and after the global financial crisis of 2007. Our results show weak support for the decoupling hypothesis for the post-crisis time period. Similarly, we find a lower level of integration for Islamic and conventional equities in the post-crisis period. The rejection of the decoupling hypothesis of Islamic and conventional equities has important implications for investors looking for alternative investment avenues. Similarly, the lower level of integration implies potential diversification and risk reduction opportunities for investors. The remainder of the paper is organized as follows: Section 2describes the methodology employed in this study. Sections 3and 4describe the data and empirical results, respectively, followed by the conclusions in Section 4. 2. Methodology In this section, we describe the methodology employed in this study. We start with the Bivariate VAR model, which will help us to test the intramarket spillover between Islamic and conventional equities and thus test the decoupling hypothesis of Islamic and conventional equities. Thereafter, we present the methodology for multivariate VAR-EGARCH to test the intermarket spillovers of Islamic and conventional equities. 2.1. Bivariate VAR-EGARCH Model In order to capture the return and volatility spillover between Islamic and conventional equities, we employ a Bivariate VAR-EGARCH model. This technique helps us to analyse the persistence of shocks to conditional variance. In addition, it requires no parameter restriction to ensure the non-negativity of the conditional variance (see [ 16 ]). The asset return dynamics can be captured by a first-order vector autoregressive (VAR) model as follows: RC,t RI,t!= βC,o βI,o!+ βC,1 βI,1 βC,2 βI,2 ! RC,t−1 RI,t−1!+ εC,t εI,t!(1) Risks 2017,5, 22 4 of 18 εt|ψt−1= εc,t εI,t!∼N(0, Σt)(2) and Σt= hcC,t hIC,t hCI,t hII,t!, (3) where Rc,t and RI,t represent the returns of conventional and Islamic equity indices, respectively; εt denotes the error term conditional on the past information set ψt−1 ; hCC,t , hII,t are the variance of conventional and Islamic indices, respectively; and hCI,t represents the covariance between these two indices. The impact of conventional equities on the Islamic equities returns and vice versa is measured by βC,Iand βI,C, respectively. As mentioned above, we employ a bivariate version of the EGARCH model proposed by Nelson. The bivariate EGARCH model is written as follows: loghC,t=γC+γCClog σ2 C,t−1+γCI log σ2 I,t−1+gC(ZC,t−1)(4) loghI,t=γI+γIClog σ2 C,t−1+γII log σ2 I,t−1+gI(ZI,t−1), (5) where the subscripts Iand Cstand for Islamic and conventional, respectively. The sign and size effect of the lagged innovation are determined by the following functions: gC(ZC,t−1)=(|ZC,t−1|−E|ZC,t−1|)+τCZC,t−1(6) gI(ZI,t−1)=(|ZI,t−1|−E|ZI,t−1|)+τIZI,t−1(7) and σC,I,t=ρC,IσC,tσI,t. (8) The standardized innovation in the above equation is Zt=εt/σt . The correlation in Equation (8) is assumed to be time-invariant, an assumption that reduces the number of parameters to be predicted. The parameters γCI and γIC measure the impact of conventional and Islamic markets on Islamic and conventional stock returns, respectively. The size effect is measured by the first two terms and the third term captures the sign effect in Equations (6) and (7) for conventional and Islamic stocks, respectively. Furthermore, the asymmetry impact on the volatility is measured by the parameters τC and τI for both markets. Asymmetry is present in the returns if τC and τI< 0 and is statistically significant. The extent to which negative innovations increase volatility more than positive innovation is defined as |−1+τ|/(1+τ) . The parameter vector θ(β,λ,γ,τ) is estimated by maximum likelihood. The log likelihood function for the bivariate EGARCH model is written as L(θ)=−T log (2π)−0.5ΣT t=1log (|Ht(θ))−0.5ΣT t=1εt(θ)0H−1 t(θ)εt(θ), (9) where Tis the number of observations, εt is the 1 × 2 vector of innovation at time t, Σt is the time varying 2 ×2 variance-covariance matrix and θis the vector of parameters to be estimated. 2.2. Multivariate VAR-EGARCH Model In order to analyse the regional spillover dynamics of Islamic and conventional equities, we employ a multivariate VAR-EGARCH extension of Nelson’s [ 17 ] E-GARCH. The multivariate EGARCH imposes no parameter and sign restrictions, permits volatility asymmetry and is more robust to deviation to standard error. In addition, the multivariate VAR-EGARCH model is also free from a priori restrictions on the structure of relationship among the variables under consideration [ 18 ]. Following Koutmos [19], we use the following specification of the multivariate EGARCH model: Ri,t=βi,0 +∑3 j=1βi,jRj,t−1+εi,t, for i,j=1, 2, 3; (10) Risks 2017,5, 22 5 of 18 σ2 i,t=exp {αi,0 +∑3 j=1αi,jfjzj,t−1+γiln (σ2 i,t−1)}, for i,j=1, 2, 3; (11) fjzj,t−1=Zj,t−1−EZj,t−1+τjZj,t−1, for i,j=1, 2, 3; (12) σi,j,t=ρi,jσi,tσj,t, for i,j=1, 2, 3; and i6=j, (13) where Ri,t represents return at time t for the markets iwhere, i= 1, 2, 3, (1 = USA, 2 = UK and 3 = Japan). The system of the above equation and all system parameters are conditioned upon the information set denoted by Ωt−1 , which carries all information till time t − 1. σi,t is the conditional variances. In Equation (13) σi,j,t is the conditional covariance between markets iand jand εi,t is the innovation at time t and z i,t is the standard innovation (i.e., zi,t=εi,t/σi,t ). Equation (10) describes the return in each market as the function of its own previous returns and also of cross-market returns. If βi,j is significant then market ileads market j. Equation (11) is conditional variance, which is a function of conditional variance at previous lags and is used to accommodate the asymmetric relation between stock returns and volatility changes. The function fjzj,t−1 is made to account for both the magnitude and sign of zj . The component of fjzj,t−1 , i.e., Zj,t−1−EZj,t−1 represents magnitude effect and Zj,t−1 sign effect, so that if Zj,t−1< 0 the slope of the function will be equal to − 1 +τj whereas for Zj,t−1> 0 the slope becomes 1 +τj ; for a shock to be positive, the value of Zj,t−1 must be greater than its own expectation and vice versa. The coefficient of fjzj,t−1 , that is, αi,j , measures cross-market spillover, which may be either symmetric or asymmetric depending upon τj , which is the coefficient of Zj,t−1 . The persistence of the conditional variance is measured by γi and for unconditional variance to be finite γ< 1 must hold. Equation (12) hence allows for standardized own and cross-market innovation to influence the conditional variance in each market asymmetrically. To, estimate these parameters we assume that they are normally distributed, taking the log of the probability density function of the parameters of system the likelihood for multivariate VAR-EGARCH model can be written as L(Θ)=−0.5(NT)ln(2π)−1 2∑T t=1ln(|St|)+εtS−1 tεt(14) where Nis the number of equations (in this case we have three); Tis the total number of observations Θ is the 33 × 1 parameter vector to be estimated; ε1=|ε1,t,ε2,t,ε3,t,| is the 1 × 3 vector of innovation at time t;S t is the 3 × 3 time varying conditional variance-covariance matrix with diagonal elements given by Equation (2) for i= 1, 2, 3; and cross-diagonal elements are given by Equation (4) for i, j = 1, 2, 3 and i6=j. 3. Data In this section, we report the details of the data series employed and the estimation results for return and volatility spillover among Islamic and conventional equity indices. Table 1shows the sample statistics along with the mnemonic codes for each data series. We obtained all the data series from Thompson Reuters DataStream. The full sample period encompasses daily observations from January 1996 to December 2015. We divide the total sample period into three sub-sample periods encompassing the pre-crisis period (January 1996–June 2007), crisis period (July 2007–June 2011) and post-crisis period (July 2011–December 2015). Following the extant literature, we use the Dow Jones total return Islamic indices for the USA, United Kingdom and Japan. Dow Jones Islamic indices include companies that fulfil certain Sharia requirements such as acceptable products, business activities, debt levels, and interest income and expenses. Equities are included following a screening methodology that is based upon input from an independent Sharia supervisory board. These indices exclude companies involved in industries such as alcohol, pork-related products, conventional financial services, entertainment, tobacco, weapons and defence. In addition, the financial screening ensures exclusion of companies for which any of the following three parameters are above 33%: •The total debt divided by trailing 24-month average market capitalization, Risks 2017,5, 22 6 of 18 • The sum of a company’s cash and interest-bearing securities divided by trailing 24-month average market capitalization, •The accounts receivables divided by trailing 24-month average market capitalization. We employ Dow Jones global total return indices for the USA, United Kingdom and Japan as our conventional equity indices. The sample means for all equities are positive, with the highest mean for U.K. Islamic. Table 1presents descriptive statistics for the three markets for both Islamic and conventional equities. All the returns show negative skewed and high kurtosis, establishing higher leptokurtic behaviour. Significant statistics for the Jarque–Bera test reject the normality assumption for all the return series, which motivates us to use non-linear models. Further autocorrelation of simple and squared returns confirms the presence of linear and non-linear dependences. The ARCH test also displays significant results, which further confirms the presence of heteroskedasticity. Finally, we also conduct Engle and Ng’s test for asymmetric response of variance to past shocks. We find significant coefficients for the sign-based test, which reveals that positive and negative shocks have different effects on the residuals. From these results we confirm the presence of an asymmetric effect and thus the appropriateness of the multivariate VAR-EGARCH model for studying the relationship between Islamic and conventional equities. Table 1. Sample statistics. Statistics Islamic Conventional Japan USA UK Japan USA UK Mean 0.007 0.014 0.010 0.002 0.010 0.010 Median 0.006 0.016 0.022 0.004 0.032 0.017 Std. Dev. 0.620 0.543 0.562 0.608 0.424 0.592 Skewness −0.053 −0.133 −0.158 −0.017 −0.402 −0.102 Kurtosis 6.703 9.642 11.510 7.244 10.246 9.518 Jarque-Bera 2982 *** 9606 *** 15768 *** 3916 *** 11554 *** 9245 *** AC(10) Residual 0.0100 0.0240 0.0170 0.0100 0.0170 0.0250 AC(10) Squared Residual 0.1180 0.1830 0.2020 0.1580 0.1430 0.1910 Arch 0.1701 *** 0.2131 *** 0.1902 *** 0.1635 *** 0.2210 *** 0.1931 *** Size bias −0.0031 −0.0933 * −0.1202 * −0.0125 −0.0342 −0.0981 * Negative sign bias −0.3527 *** −0.7095 *** −0.6834 *** −0.3736 *** −0.5264 *** −0.7262 *** Positive sign bias 0.2875 *** 0.1603 *** 0.2543 *** 0.2835 *** 0.2382 *** 0.2897 *** The residual diagnostics based on simple and squared residuals at the 10th lag are also reported in the table. The significance level is *, **, *** at 10, 5 and 1 %, respectively. 4. Results and Discussion In this section, we report the empirical results and analysis of our study. We start this section with the empirical results for the intra-market spillover among Islamic and conventional equities, followed by the results for inter-market spillovers. 4.1. Intra-Market Spillover among Islamic and Conventional Equities We employ the bivariate VAR-EGARCH model to analyse the intra-market return and volatility spillover among Islamic and conventional equities. The first, second and third panels of Table 2 reports the estimation results of the bivariate VAR-EGARCH model for Japan, the USA and the United Kingdom, respectively, using Equations (1), (4) and (5). The coefficients β1,2 and β2,1 show the return spillover from Islamic to conventional and conventional to Islamic equity indices, respectively. The volatility spillover between Islamic and conventional equity indices and vice versa is measured through γ12 and γ21 , respectively. τCand τI are the coefficients of asymmetry for the conventional and Islamic equity indices, respectively. Risks 2017,5, 22 7 of 18 Table 2. Estimation from the bivariate VAR-EGARCH (full sample). Japan United States United Kingdom β1,0 −0.0125 ** β1,0 −0.0077 β1,0 0.0045 β1,0 0.0061 β1,0 0.0109 * β1,0 0.0101 β1,1 0.3426 *** β2,1 −0.4983 *** β1,1 0.0168 β2,1 0.0519 ** β1,1 −0.0546 β2,1 0.1367 *** β1,2 0.2752 *** β2,2 0.4157 *** β1,2 0.1456 *** β2,2 −0.0808 ** β1,2 0.0453 β2,2 0.1075 ** γ10 −0.0213 *** γ20 −0.0187 *** γ10 −0.0401 *** γ20 −0.0317 *** γ10 −0.0234 *** γ20 −0.0196 *** γ11 0.9768 *** γ21 −0.0469 *** γ11 0.9791 *** γ21 0.0182 ** γ11 0.9829 *** γ21 0.0051 * γ12 0.1249 *** γ22 0.9784 *** γ12 0.0186 *** γ22 0.9763 *** γ12 0.0426 *** γ22 0.9839 *** τC−0.1305 *** τI−0.1023 *** τC−0.5661 *** τI−0.6544 *** τC−0.5986 *** τI−0.4983 Correlation Coefficients ρ1,2 0.6560 *** ρ2,1 0.6560 *** ρ1,2 0.8651 *** ρ2,1 0.8651 *** ρ1,2 0.8589 *** ρ2,1 0.8589 *** Residual Diagnostics AC(10) Residual 0.01628 0.02231 AC(12) Residual 0.00093 0.01600 AC(10) Residual −0.01268 −0.01423 AC(10) Squared Residual 0.00526 0.00174 AC(12) Squared Residual 0.03174 0.02518 AC(10) Squared Residual 0.01978 0.01112 Note: This table reports the results from the bivariate VAR-EGARCH equation, as explained in the methodology section. The parameters β1,2 and β2,1 display the return spillover from Islamic to conventional and conventional to Islamic equity markets, respectively, whereas γ12 and γ21 are the volatility spillover from Islamic to conventional and conventional to Islamic equity markets. The asymmetry parameter for τ and γ is volatility persistence. The cross-correlation of the returns is denoted by ρ . The residual diagnostics based on simple and squared residuals at the 10th lag are also reported in the table. The significance level is *, **, *** at 10, 5 and 1%, respectively. Risks 2017,5, 22 8 of 18 The coefficient for return spillover is statistically significant for all countries except the coefficient for return spillover between Islamic and conventional equities for the United Kingdom. The coefficients of second moment (volatility spillover) are statistically significant for all three countries. This implies that the conditional variance of conventional and Islamic equities is influenced by their past innovation. Similarly, the coefficient of asymmetry is statistically significant for all countries, which confirms that negative shocks have more impact on volatility as compared to positive shocks of the same magnitude. These results show that the risk and return of both Islamic and conventional equities are interlinked. The results discussed above are based on a relatively long time period, characterized by different intervals of the economic business cycle. Therefore, we segregate our sample period into three intervals: pre-crisis, crisis and post-crisis. The pre-crisis period ranges from 1995 to June 2007, the crisis period ranges from July 2007 to June 2011 and the post-crisis period ranges from July 2011 to 2015. Tables A1–A3 report the results of the bivariate VAR-EGARCH model for pre-crisis, crisis and post-crisis periods. Here again, we are interested in the coefficients of return and volatility spillover along with the coefficient of asymmetry. The results for the pre-crisis period are similar to the full sample period. However, the results for the crisis and post-crisis periods are slightly different. During the crisis period, one of the volatility spillover coefficients is statistically insignificant for each country. Similarly, one return spillover coefficient is statistically insignificant for both Japan and the United Kingdom. During the post-crisis period, at least one return spillover coefficient is statistically insignificant for each country, whereas one of the volatility spillover coefficients is statistically insignificant for Japan and the USA. The sub-sample analysis gives us some interesting insights into the decoupling hypothesis. Our results show that the 2007 financial crisis resulted in a reduction in the interdependence between Islamic and conventional equities. Thus, we find weak support for the decoupling hypothesis during and after the crisis period. 4.2. Inter-Market Spillover Effects In this section we analyse the inter-market return and volatility spillover for conventional equity indices across the USA, United Kingdom and Japan. We analyse the inter-market spillover effects of Islamic and conventional indices across these markets on a standalone basis, i.e., we analyse the spillover effects of Islamic and conventional equities separately. This analysis will help us to see the level of integration across markets for both these indices. A lower level of integration may translate into higher diversification opportunities and lower contagion effects. We employ the multivariate MVR-EGARCH model given by Equations (10) and (11) for the USA, United Kingdom and Japan. Similar to the analysis in the previous section, we estimate our results for both the full sample period and three sub-sample periods. We start our analysis by reporting the results for the full sample period. Table 3shows the estimated MVR-EGARCH results for Islamic equities, whereas Table 4gives the results for convention equities. The AR coefficients ( β1,1 , β2,2 , β3,3 ) are negative and significant for all three equity market indices, indicating a negative serial correlation in returns. Focusing on the first moment interdependencies, there is significant spillover from the United Kingdom to Japan and the USA to the United Kingdom but not from Japan to the USA and United Kingdom. However, there are significant spillovers from both the USA and the United Kingdom to the Japanese equity market ( β3,1 and β3,2 ). Moving to the volatility spillovers (second moment interdependencies), the results are stronger. The conditional variance for every country is influenced by innovations from the other two countries. There are significant volatility spillovers from the United Kingdom and Japan to the USA ( α12 and α13 ), from the USA and Japan to the United Kingdom ( α21 and α23 ) and also from the USA and United Kingdom to Japan ( α31 and α32 ). Furthermore, the volatility transmission is asymmetrical for all three equity indices. Risks 2017,5, 22 15 of 18 Table A4. Estimation from the multivariate MVR-EGARCH model (Islamic equities—pre-crisis). United States United Kingdom Japan β1,o 0.0103 β2,o 0.0153 ** β3,o −0.0039 β1,1 −0.0077 β2,1 0.3144 *** β3,1 0.3951 *** β1,2 0.0232 β2,2 −0.1020 ** β3,2 0.1284 *** β1,3 −0.0129 β2,3 −0.0353 ** β3,3 −0.0188 α10 −0.0280 *** α20 −0.0277 *** α30 −0.0238 *** α11 0.0812 *** α21 0.0315 *** α31 0.0278 ** α12 0.0631 *** α22 0.0934 *** α32 0.0615 *** α13 0.0443 ** α23 0.0155 α33 0.1469 *** τ1−0.0340 *** τ2−0.3050 ** τ3−0.1127 * γ10.9788 *** γ20.9817 *** γ30.9764 *** Correlation Coefficients ρ1,2 0.0683 *** ρ2,3 0.1567 *** ρ1,3 0.3352 *** Residual Diagnostics AC(10) Residual 0.01663 AC(10) Residual 0.01055 AC(10) Residual 0.01739 AC(10) Squared Residual 0.00790 AC(10) Squared Residual 0.00353 AC(10) Squared Residual 0.00378 Note: This table reports the results from the multivariate VAR-EGARCH equation, as explained in the methodology section (Equations (10)–(13)). The parameters β1,2 and β1,3 display the return spillover from the United Kingdom and Japan to the USA’s Islamic equity returns. β2,1 and β2,3 display the return spillover from the USA and Japan to the United Kingdom and β3,1 and β3,2 display the return spillover from the USA and the United Kingdom to Japan. α12 and α13 are the volatility spillover from the United Kingdom and Japan to the USA’s Islamic equity returns. α2,1 and α2,3 display the volatility spillover from the USA and Japan to the United Kingdom and α3,1 and α3,2 the volatility spillover from the USA and the United Kingdom to Japan. The asymmetry parameter for τ and γ is volatility persistence. The cross-correlation of the returns is denoted by ρ . The residual diagnostics based on simple and squared residuals at the 10th lag are also reported in the table. The significance level is *, **, *** at 10, 5 and 1%, respectively. Table A5. Estimation from the multivariate MVR-EGARCH model (conventional equities—pre-crisis). United States United Kingdom Japan β1,o 0.0101 * β2,o 0.0141 * β3,o −0.0098 β1,1 0.3394 *** β2,1 0.5718 *** β3,1 0.6517 *** β1,2 −0.0832 *** β2,2 −0.2604 *** β3,2 −0.0402 β1,3 −0.0762 *** β2,3 −0.1167 *** β3,3 −0.1128 *** α10 −0.0512 *** α20 −0.0553 *** α30 −0.0240 *** α11 0.0636 *** α21 0.0464 *** α31 0.0357 *** α12 0.0790 *** α22 0.1105 *** α32 0.0524 *** α13 0.0121 α23 −0.0056 α33 0.1108 *** τ1−0.4675 *** τ2−0.1168 τ3−0.1832 ** γ10.9768 *** γ20.9697 *** γ30.9779 *** Correlation Coefficients ρ1,2 0.6284 ** ρ2,3 0.3870 *** ρ1,3 0.1874 *** Residual Diagnostics AC(10) Residual 0.01087 AC(10) Residual 0.02988 AC(10) Residual 0.02535 AC(10) Squared Residual 0.00696 AC(10) Squared Residual 0.00885 AC(10) Squared Residual 0.00421 Note: This table reports the results from the multivariate VAR-EGARCH equation, as explained in the methodology section (Equations (10)–(13)). The parameters β1,2 and β1,3 display the return spillover from the United Kingdom and Japan to the USA’s conventional equity returns. β2,1 and β2,3 display the return spillover from the USA and Japan to the United Kingdom and β3,1 and β3,2 display the return spillover from the USA and the United Kingdom to Japan. α12 and α13 are the volatility spillover from the United Kingdom and Japan to the USA’s conventional equity returns. α2,1 and α2,3 display the volatility spillover from the USA and Japan to the United Kingdom and α3,1 and α3,2 the volatility spillover from the USA and the United Kingdom to Japan. The asymmetry parameter for τ and γ is volatility persistence. The cross-correlation of the returns is denoted by ρ . The residual diagnostics based on simple and squared residuals at the 10th lag are also reported in the table. The significance level is *, **, *** at 10, 5 and 1%, respectively. Risks 2017,5, 22 16 of 18 Table A6. Estimation from the multivariate MVR-EGARCH model (Islamic equities—crisis). United States United Kingdom Japan β1,o 0.0134 β2,o −0.0044 β3,o −0.0103 β1,1 −0.0606 * β2,1 0.5033 *** β3,1 0.4296 *** β1,2 0.0091 β2,2 −0.2592 β3,2 0.1248 *** β1,3 0.0082 β2,3 0.0535 ** β3,3 −0.1781 *** α10 −0.0288 *** α20 −0.0153 *** α30 −0.0629 *** α11 0.0581 ** α21 0.0290 ** α31 0.0260 ** α12 0.0407 ** α22 0.0740 *** α32 0.0206 α13 0.0296 ** α23 0.0254 * α33 0.1786 *** τ1−0.2173 *** τ2−0.4677 *** τ3−0.4659 *** γ10.9799 *** γ20.9831 *** γ30.9542 *** Correlation Coefficients ρ1,2 0.0494 ** ρ2,3 0.1199 *** ρ1,3 0.6357 *** Residual Diagnostics AC(10) Residual 0.01948 AC(10) Residual 0.02327 AC(10) Residual 0.01360 AC(10) Squared Residual 0.00698 AC(10) Squared Residual 0.00593 AC(10) Squared Residual 0.00898 Note: This table reports the results from the multivariate VAR-EGARCH equation, as explained in the methodology section (Equations (10)–(13)). The parameters β1,2 and β1,3 display the return spillover from the United Kingdom and Japan to the USA’s Islamic equity returns. β2,1 and β2,3 display the return spillover from the USA and Japan to the United Kingdom and β3,1 and β3,2 display the return spillover from the USA and the United Kingdom to Japan. α12 and α13 are the volatility spillover from the United Kingdom and Japan to the USA’s Islamic equity returns. α2,1 and α2,3 display the volatility spillover from the USA and Japan to the United Kingdom and α3,1 and α3,2 the volatility spillover from the USA and the United Kingdom to Japan. The asymmetry parameter for τ and γ is volatility persistence. The cross-correlation of the returns is denoted by ρ . The residual diagnostics based on simple and squared residuals at the 10th lag are also reported in the table. The significance level is *, **, *** at 10, 5 and 1%, respectively. Table A7. Estimation from the multivariate MVR-EGARCH model (conventional equities—crisis). United States United Kingdom Japan β1,o 0.0059 β2,o −0.0004 β3,o −0.0112 β1,1 0.3678 *** β2,1 0.8221 *** β3,1 0.6923 *** β1,2 −0.1676 *** β2,2 −0.5247 *** β3,2 −0.0398 β1,3 −0.0558 *** β2,3 −0.0682 ** β3,3 −0.2819 *** α10 −0.0224 *** α20 −0.0120 *** α30 −0.0860 *** α11 0.0852 *** α21 0.0580 *** α31 0.0471 ** α12 0.0051 α22 0.0244 *** α32 0.0007 α13 0.0138 α23 0.0166 α33 0.1969 *** τ1−0.1841 *** τ2−0.2023 ** τ3−0.4570 *** γ10.9857 *** γ20.9875 *** γ30.9352 *** Correlation Coefficients ρ1,2 0.8588 *** ρ2,3 0.2504 *** ρ1,3 0.1331 *** Residual Diagnostics AC(10) Residual 0.01305 AC(10) Residual 0.01883 AC(10) Residual 0.01063 AC(10) Squared Residual 0.00590 AC(10) Squared Residual 0.00850 AC(10) Squared Residual 0.00407 Note: This table reports the results from the multivariate VAR-EGARCH equation, as explained in the methodology section (Equations (10)–(13)). The parameters β1,2 and β1,3 display the return spillover from the United Kingdom and Japan to the USA’s conventional equity returns. β2,1 and β2,3 display the return spillover from the USA and Japan to the United Kingdom and β3,1 and β3,2 display the return spillover from the USA and the United Kingdom to Japan. α12 and α13 are the volatility spillover from the United Kingdom and Japan to the USA’s conventional equity returns. α2,1 and α2,3 display the volatility spillover from the USA and Japan to the United Kingdom and α3,1 and α3,2 the volatility spillover from the USA and the United Kingdom to Japan. The asymmetry parameter for τ and γ is volatility persistence. The cross-correlation of the returns is denoted by ρ . The residual diagnostics based on simple and squared residuals at the 10th lag are also reported in the table. The significance level is *, **, *** at 10, 5 and 1%, respectively. Risks 2017,5, 22 17 of 18 Table A8. Estimation from the multivariate MVR-EGARCH model (Islamic equities—post-crisis). United States United Kingdom Japan β1,o −0.0110 β2,o −0.0307 ** β3,o 0.0242 β1,1 −0.0239 β2,1 0.2790 *** β3,1 0.3962 *** β1,2 0.0943 * β2,2 0.0325 β3,2 0.1008 * β1,3 0.0374 β2,3 0.0092 β3,3 −0.2105 ** α10 −0.1844 α20 −0.1529 α30 −0.0538 * α11 0.0082 α21 0.0039 α31 0.0024 α12 0.0121 α22 0.0215 α32 0.0080 α13 0.1671 ** α23 0.0659 α33 0.0801 ** τ1−0.2581 τ2−0.4031 τ3−0.2497 γ10.9121 *** γ20.9182 *** γ30.9673 *** Correlation Coefficients ρ1,2 −0.0129 ρ2,3 0.0907 * ρ1,3 0.5256 *** Residual Diagnostics AC(10) Residual 0.03585 AC(10) Residual 0.03963 AC(10) Residual 0.02418 AC(10) Squared Residual 0.00605 AC(10) Squared Residual 0.00042 AC(10) Squared Residual 0.00374 Note: This table reports the results from the multivariate VAR-EGARCH equation, as explained in the methodology section (Equations (10)–(13)). The parameters β1,2 and β1,3 display the return spillover from the United Kingdom and Japan to the USA’s Islamic equity returns. β2,1 and β2,3 display the return spillover from the USA and Japan to the United Kingdom and β3,1 and β3,2 display the return spillover from the USA and the United Kingdom to Japan. α12 and α13 are the volatility spillover from the United Kingdom and Japan to the USA’s Islamic equity returns. α2,1 and α2,3 display the volatility spillover from the USA and Japan to the United Kingdom and α3,1 and α3,2 the volatility spillover from the USA and the United Kingdom to Japan. The asymmetry parameter for τ and γ is volatility persistence. The cross-correlation of the returns is denoted by ρ . The residual diagnostics based on simple and squared residuals at the 10th lag are also reported in the table. The significance level is *, **, *** at 10, 5 and 1%, respectively. Table A9. Estimation from the multivariate MVR-EGARCH model (conventional equities—post-crisis). United States United Kingdom Japan β1,o 0.0001 β2,o −0.0106 β3,o 0.0051 β1,1 0.2802 *** β2,1 0.6161 *** β3,1 0.6846 *** β1,2 −0.0926 *** β2,2 −0.3553 *** β3,2 −0.0433 β1,3 −0.0303 *** β2,3 −0.0247 β3,3 −0.2857 *** α10 −0.0712 *** α20 −0.0520 α30 −0.1316 ** α11 0.1030 *** α21 0.0748 *** α31 0.0605 *** α12 −0.0141 α22 0.0094 α32 −0.029 α13 −0.0048 α23 −0.0163 α33 0.1538 *** τ1−0.7891 *** τ2−0.3245 * τ3−0.4148 ** γ10.9676 *** γ20.9679 *** γ30.9236 *** Correlation Coefficients ρ1,2 0.8404 *** ρ2,3 0.1763 *** ρ1,3 0.2879 *** Residual Diagnostics AC(10) Residual 0.01134 AC(10) Residual 0.03414 AC(10) Residual 0.01297 AC(10) Squared Residual 0.00926 AC(10) Squared Residual 0.03226 AC(10) Squared Residual 0.00569 Note: This table reports the results from the multivariate VAR-EGARCH equation, as explained in the methodology section (Equations (10)–(13)). The parameters β1,2 and β1,3 display the return spillover from the United Kingdom and Japan to the USA’s conventional equity returns. β2,1 and β2,3 display the return spillover from the USA and Japan to the United Kingdom and β3,1 and β3,2 display the return spillover from the USA and the United Kingdom to Japan. α12 and α13 are the volatility spillover from the United Kingdom and Japan to the USA’s conventional equity returns. α2,1 and α2,3 display the volatility spillover from the USA and Japan to the United Kingdom and α3,1 and α3,2 the volatility spillover from the USA and the United Kingdom to Japan. The asymmetry parameter for τ and γ is volatility persistence. The cross-correlation of the returns is denoted by ρ . The residual diagnostics based on simple and squared residuals at the 10th lag are also reported in the table. The significance level is *, **, *** at 10, 5 and 1%, respectively. Risks 2017,5, 22 18 of 18 References 1. O. Al-Khazali, H.H. Lean, and A. Samet. “Do Islamic stock indexes outperform conventional stock indexes? A stochastic dominance approach.” Pac.-Basin Financ. J. 28 (2014): 29–46. [CrossRef] 2. M.E. Arouri, H.B. Ameur, N. Jawadi, F. Jawadi, and W. Louhichi. “Are Islamic finance innovations enough for investors to escape from a financial downturn? Further evidence from portfolio simulations.” Appl. Econ. 45 (2013): 3412–3420. [CrossRef] 3. A. Charles, O. Darnéb, and A. Pop. “Risk and ethical investment: Empirical evidence from Dow Jones Islamic indexes.” Res. Int. Bus. Financ. 35 (2015): 33–56. [CrossRef] 4. G. Dewandarua, O.I. Bachab, A.M.M. Masihb, and R. Masihc. “Risk-return characteristics of Islamic equity indices: Multi-timescales analysis.” J. Multinatl. Financ. Manag. 29 (2015): 115–138. [CrossRef] 5. G. Dewandaru, S. Rizvi, R. Masih, M. Masih, and S. Alhabshi. “Stock market co-movements: Islamic versus conventional equity indices with multi-timescales analysis.” Econ. Syst. 38 (2014): 553–571. [CrossRef] 6. S. Hammoudeh, W. Mensi, J.C. Reboredo, and D.K. Nguyen. “Dynamic dependence of the global Islamic equity index with global conventional equity market indices and risk factors.” Pac.-Basin Financ. J. 30 (2014): 189–206. [CrossRef] 7. C.S.F. Ho, N.A. AbdRahman, N.H.M. Yusuf, and Z. Zamzamin. “Performance of global Islamic versus conventional share indices: International evidence.” Pac.-Basin Financ. J. 28 (2014): 110–121. [CrossRef] 8. M.K. Yilmaz, A. Sensoy, K. Ozturk, and E. Hacihasanoglu. “Cross-sectoral interactions in Islamic equity markets.” Pac.-Basin Financ. J. 32 (2015): 1–20. [CrossRef] 9. M.M. Hasan, and J. Dridi. The Effects of Global Crisis on Islamic and Conventional Banks: A comparative Study. International Monetary Fund Working Paper No. 10/201; Washington, DC, USA: International Monetary Fund, 2010. 10. J. Majdoub, and W. Mansour. “Islamic equity market integration and volatility spillover between emerging and US stock markets.” N. Am. J. Econ. Financ. 29 (2014): 452–470. [CrossRef] 11. S. Nazlioglu, S. Hammoudeh, and R. Gupta. “Volatility transmission between Islamic and conventional equity markets: evidence from causality-invariance test.” Appl. Econ. 47 (2015): 4996–5011. [CrossRef] 12. A. Rejeb. “Volatility Spillover between Islamic and conventional stock markets: Evidence from Quantile Regression analysis.” Available online: https://mpra.ub.uni-muenchen.de/73302/ (accessed on 25 February 2017). 13. G. Koutmos, and G.G. Booth. “Asymmetric volatility transmission in international stock markets.” J. Int. Money Financ. 14 (1995): 747–762. [CrossRef] 14. A. Saadaoui, and Y. Boujelbene. “Volatility Transmission Between Dow Jones Stock Index and Emerging Islamic Stock Index: Case of Subprime Financial Crises.” EMAJ Emerg. Mark. J. 5 (2015): 41–49. [CrossRef] 15. A. El Khamlichi, K. Sarkar, M. Arouri, and F. Teulon. “Are Islamic Equity Indices More Efficient Than Their Conventional Counterparts? Evidence from Major Global Index Families.” J. Appl. Bus. Res. 30 (2014): 1137. [CrossRef] 16. T. Bollerslev, R.Y. Chou, and K.F. Kroner. “Arch modeling in finance: A review of the theory and empirical evidence.” J. Econom. 52 (1992): 5–59. [CrossRef] 17. D.B. Nelson. “Conditional heteroskedasticity in asset returns: A new approach.” Econom. J. Econom. Soc. 59 (1991): 347–370. [CrossRef] 18. A. Antoniou, G. Pescetto, and A. Violaris. “Modelling international price relationships and interdependencies between the stock index and stock index futures markets of three EU countries: A multivariate analysis.” J. Bus. Financ. Account. 30 (2003): 645–667. [CrossRef] 19. G. Koutmos. “Modeling the dynamic interdependence of major European stock markets.” J. Bus. Financ. Account. 27 (1996): 975–988. [CrossRef] © 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).