Unexpeced Volailiy Shifs and Efficiency of Emerging Sock Marke: The Case of Malaysia Elgilani Elahir Elshareif 1, Hui-Boon Tan 2 and Mei-Foong Wong 3 Absrac This paper analyzed he behavior of Malaysian sock marke during he inervals of high uncerainy. I highlighed he impac of unexpeced volailiy shifs on his small emerging Asian marke, in erms of is efficiency and reurns, during he pas wo decades. The purpose of his sudy was achieved hrough he Ieraed-Cumulaive-Sum-of-Squares-in-Volailiy model (ICSS-EGARCH-M Model), which is one of he new approaches in marke efficiency sudies. The empirical resuls indicaed he rejecion of Efficien Marke Hypohesis for he marke when sudden volailiy shifs were considered. The resuls also provided significan empirical evidences for posiive risk-reurn relaionship in he exchanges. In addiion, he sock marke was found o be more sensiive o global han local evens. The asymmerical responses o good and bad news were also par of he marke behavior. Key words: Efficiency, Volailiy, Malaysian sock marke, EGARCH-M, ICSS algorihm Available online www.bmdynamics.com ISSN: 2047-7031 INTRODUCTION The Asian financial markes have experienced several unexpeced volailiy shifs during he las wo decades, for example, he erupion of he 1997-98 Eas-Asian financial urmoil ha has elevaed he uncerainies in he region. As one of he consequences, he Malaysian equiy marke, among ohers, has also experienced a grea unexpeced shif in he financial volailiy. During ha period, he volailiy of he Kuala Lumpur Sock Exchange (KLSE) increased remendously wih invesors confidence being badly damaged. This was refleced in a subsanial decline in he key benchmark Kuala Lumpur Composie Index (KLCI). This high volailiy shif in he sock marke received a grea deal of aenion from economiss and marke paricipans, including invesors, brokers, dealers, and regulaors. High level of uncerainy reduced invesors confidence and delayed heir invesmen decision. The final impac was he delay of invesmen and economic growh. I has been a well known fac ha undersanding he behavior of sock marke volailiy is imporan o boh policy-makers and marke praciioners. Policy-makers are mainly ineresed in he main deerminans of volailiy, is spillover effec on real economic aciviy and for assessing regulaory proposals o resric inernaional capial flows. On he oher hand, Marke praciioners, are mainly ineresed in he direc effecs ha ime-varying volailiy exers on he pricing of financial asses and hedging sraegies. I is now widely noed ha ime-varying properies of he volailiy of financial asse reurns, which were convenionally measured by is variances and covariances, are no more consan over ime; insead hey evolve over ime. Thus he assumpion of consan variances over ime is no longer valid. One of he mos prominen ools ha emerged o capure such ime varying variances was he Generalized Auoregressive Condiional Heeroskedasiciy (GARCH) family approach, which have long been documened and modeled (see Bollerslev, 1986; and Bollerslev e al., 1992). Volailiy is considered highly persisen if a shock o a given sysem is permanen, and he pas volailiy can be used in consrucing forecass of fuure volailiy. However, ranquil periods where prices are more or less sable could be followed by relaively high volailiy periods characerized by large price changes due o economic, poliical and/or social evens, boh regionally or globally. These sudden changes in uncerainy, or volailiy, should be given subsanial concern as hey can be persisence and have serious impac on asse prices as well as economic fundamenals. In view of his, we have surveyed he relaed lieraure, and found ou ha alhough he sudies on he 1997-Asian-financial crisis and 2007-1Associae Professor Dr. of Finance, Deparmen of Finance and Banking, College of Business Adminisraion, Al Ain Universiy for Science and Technology-Abu Dhabi, Unied Arab Emiraes. E-mail: gsharief@gmail.com 2Universiy of Noingham Malaysia, E-mail: Hui-Boon.Tan@noingham.edu.my 3Tunku Abdul Rahman College, Malaysia, E-mail: wongmf@mail.arc.edu.my
global financial crisis were exensive, sudies peraining o he effec of sudden changes in volailiy on he efficiency of Asian emerging sock marke during he wo crises were scarce and did no receive much aenion. In his case sudy, a grea ineres was focused on he effec of sudden volailiy change on he sock price level of a small emerging Asian sock marke - he Malaysian marke. This sudy concenraed on he Kuala Lumpur Sock Exchange Composie Index (KLCI) and four secor indices. These secor indices were he planaion (PLT), Properies (PROP), Indusrial (IND), and Finance (FIN). As an emerging sock marke, he Kuala Lumpur sock exchange (KLSE) has received a grea aenion from researchers and invesors. I has been regarded as one of he poenial invesmen alernaives and is developmen is deemed rapidly growing (see Hooy e al., 2004) in he Asia-Pacific region. I is imporan o noe a few significan milesones of he marke. Firs, he marke has undergone financial liberalizaion since 1980 s. Secondly, i experienced a gradual improvemen in erms of coss reducion and reurns incremen of he equiies, and hirdly, is marke efficiency has been subsequenly influenced and uplifed by he auhoriies during he pas wo decades (see Kawakasu and Morey, 1999). Besides, he marke has adoped inernaional sandard and pracices for is financial and non-financial disclosure, which in urn may have improved he informaional efficiency of he marke. Obviously he Eas Asian emerging sock markes have been rocked by several sudden shifs in he volailiies. These sudden shifs encompass he 1997-Asian Financial Crisis and 2007- Global Financial Crisis, which had recorded a serious impac on hese markes. In view of his, more empirical findings should be learned and documened from he crises for fuure benefis. As one of he affeced counries, Malaysia was chosen as a case o provide empirical insigh on he maer. There were wo sudies, namely, Law (2006) and Goh e al. (2005) have looked ino he volailiy and srucure breaks of he Malaysian sock marke. While Law (2006) only invesigaed he incremen of volailiy of he Malaysian sock prices during he Asian Financial Crisis period, Goh e al. (2005) only examined he srucural breaks in he marke and relaed i o he res of he ASEAN markes. Thus, more in-deph sudies relaed o his issue are sill imely and required, in paricular hose address he impac of 2008-2009 Global Financial Crisis and 2010-2011 European Deb Crisis. By focusing on one of he Asian emerging sock markes, he Bursa Malaysia, his paper examined he effec of sudden shifs in uncondiional variance of five Malaysian sock prices on he efficiency of he marke. Wih he similar purpose, he effec on he risk-reurns relaionship in he marke was also analyzed. These sudden changes in volailiy were idenified endogenously using he ieraed cumulaive sums of squares (ICSS) algorihm developed by Inclan and Tiao (1994). To our knowledge, his approach has no been employed in research works peraining o he Malaysian sock marke, paricularly in invesigaing he effec of large sudden shifs of volailiy on he performance of he marke. For oher Asian sock markes, he Japanese and Korean markes were sudied, by Wang and Moore 4 (2009). In anoher recen sudy, Kang e al. (2009) have invesigaed five Cenral European sock markes. According o heir empirical esimaions, when sudden shifs of volailiy were included in he GARCH model, he persisence of volailiy was reduced significanly in each of he five markes. Based on he resuls, hey have suggesed ha many previous sudies may have oversaed he degree of volailiy persisence in financial ime series, and highlighed he imporance of capuring volailiy shifs in he model. The reminder of his paper is organized as follows. Secion 2 presens an overview of he sock marke efficiency hypohesis. Secion 3 describes he daa and mehodology. Secion 4 repors he empirical resuls and secion 5 is he concluding remarks. STOCK MARKET EFFICIENCY The Efficien Marke Hypohesis (EMH) describes an efficien marke as one where securiy prices fully and speedily reflec available informaion (see Fama, 1970). The conen of any new informaion becoming available will be quickly digesed by marke paricipans and if he informaion causes hem o change heir opinion of he securiy s inrinsic value, heir subsequen acion will rapidly cause an 4However, heir sudy, based on GARCH approach, is incapable of inegraing sudden changes and is also unsuiable for examinaions of volailiy persisence (see Nwogugu, 2006).
equivalen change in he securiy s marke price. To examine he adjusmen of securiy prices o a precise definiion of informaion, he EMH akes hree forms depending upon he exen of he informaion deemed available o marke paricipans. These forms are defined as he weak form efficiency, Semisrong efficiency and srong-form efficiency. The EMH has araced researchers, invesors and policy makers, paricularly in emerging markes. In major markes like he London and New York sock exchanges, he evidence has in favor of he EMH, especially a he weak and semi-srong form levels (see Fama, 1991). The emerging markes, on he oher hand, indicae significan deparures from he Efficien Marke Hypohesis (see Ome e al., 2002). Given he large volume of research already exised for maure sock markes, i is ineresing o examine wheher Malaysian securiy prices fully and speedily reflec he available informaion in he emerging sock markes, especially in he presence of sudden changes in uncondiional variance shifs. DATA AND METHODOLOGY Daa The daa employed by his sudy consised of he weekly closing sock price index of Kuala Lumpur Composie Index (KLCI), planaion (PLT), Properies (PROP), Indusrial (IND), and Finance (FIN). Weekly raher han daily daa were chosen o avoid he poenial biases associaed wih micro-srucural issues, non-rading, he bid-ask spread effec in daily daa, and problems of hin rading which were ofen associaed wih mos emerging markes. All sock prices series were colleced from DaaSream and spanned from June 1990 o June 2011, which include boh he Asian Financial Crisis and Global Financial Crisis. The sock marke reurns were calculaed according o he following expression: r (log S log S 1) 100 (1) Where log S was he naural logarihm of he index a ime (week). Dividends were assumed away (see Campbell e al., 1997). Mehodology I is widely recognized ha he GARCH family models are sandard and widely used for sock marke volailiy analyses. However, he sandard GARCH models overesimae he persisence of volailiy when sudden shifs of uncondiional variance are observed in he daa (See Lasrapes, 1989). Therefore, sudden variance changes, which characerize mos of Malaysian sock prices, mus be aken ino accoun when modeling volailiy by employing he GARCH family models. In examining he volailiy behavior of he Malaysian sock prices, i is observed ha all Malaysian sock prices exhibied high flucuaion which was followed by low flucuaion in several periods. In addiion, i was found ha several srucural breaks, ouliers, exreme observaions were mosly associaed wih sudden changes in uncondiional variance in he sample period. All hese have undesirable effecs on Maximum Likelihood esimaion. Our survey on he lieraure found a srand of research ha highlighed he exisence of severe excess kurosis in heir esimaed residuals. This excess kurosis can be originaed from aberran observaions like srucural changes and ouliers. Franses and Ghijsels (1999) proposed a mehodology o deec and correc addiive ouliers (AO) wihin he GARCH framework. On he oher hand, Charles and Darné (2005) exended he Franses-Ghijsels mehodology by incorporaing innovaive ouliers (IO) ino he GARCH family models. Anoher school of hough employed Ieraive Cumulaive Sum of Squares (ICSS) algorihm advocaed by Inclan and Tiao (1994) o deec sudden shifs in uncondiional variance. The deeced sudden shifs were incorporaed ino he variance equaion of GARCH model using appropriae dummy variables. For example, Wilson e al. (1996) used his approach for oil fuures series; Malik (2003) employed i for exchange raes; Hammoudeh and Li (2008) employed i for he Gulf Arab equiy markes; Kang e al. (2009) employed i for he Japanese and Korean equiy markes; and Wang and Moore (2009) employed i for he European equiy markes. I is imporan a his poin o noe ha even hough Nelson s (1991) exponenial GARCH, or EGARCH model is robus in capuring asymmeric effecs of volailiy, he model, however, by iself may no be able o capure all of he variance effecs when here is(are) sudden shif(s). I was also noed by several previous sudy ha condiional heeroskedasic models end o overesimae he persisence of volailiy when here are insabiliies a he uncondiional second momen. Therefore, i is crucial o include he
breakpoins of sudden shifs in variance ha can capure some of he ime-varying volailiy or breaks in he volailiy process. In order o esimae he number of shifs in variance and he poin in ime of each variance shif, he ICSS algorihm procedure proposed by Inclan and Tiao (1994) is employed. The algorihm focuses on idenifying an uncondiional variance change due o a sudden shif ha changes he variance during a ime inerval. The approach was proven o be an effecive ool in deecing sudden changes in variance of a ime series. To begin wih he algorihm, le r be he sock price reurn series wih uncondiional variance. The variances wihin each inerval were given by,. Here, was he oal number of variance changes in T observaions and were he change poins. I can be clearly denoed as for ; for ;, for The above variance changes could be idenified via he following seps. Firs, he Ieraive-Cumulaive- Sum-of-Squares of he series were produced using for. Secondly, a dummy variable was se up as wih (2) If here was no significan shif in he variance, oscillaed around zero and would exhibi a horizonal line when ploed agains k. If here was a significan shif in he variance, on he oher hand, would drif upward and downward around zero when ploed agains k. In his case, sudden changes in variance were deeced using he criical values which are given in able 1 of Inclan and Tiao (1994). The significan changes were deeced using he criical values obained from he disribuion of, where was muliplied by o sandardize he disribuion. To idenify significan changes we examined he sandardize series insead of he series. The null hypohesis of homogeneous variance can be rejeced, if he maximum of exceeds he criical value. Le denoed he value a which max k reached. If max k falls ouside he predeermined boundary, hen is aken as a poin in uncondiional variance shif. Frequenly funcion alone is insufficien o deec muliple variance shifs a differen inerval due o he so called masking effec. The masking effec means ha moderae variance shifs may no be deeced due o he exisence of large variance shifs. This problem is resolved by he ICSS algorihm by evaluaing he over differen ime inervals. The algorihm is hus based on successive evaluaions of a differen pars of he series, dividing consecuively afer a possible change poin is found. Once he breakpoins have been idenified using he algorihm, he nex sage is o consruc he EGARCH-M model which incorporaed he idenified variance shifs. The exended Ieraed-Cumulaive-Sum-of-Squares-in-Volailiy model (ICSS-EGARCH-M Model), which incorporaes sudden shifs of volailiy in he EGARCH(p, q)-m model was formulaed as: (3) (4) where r represened respecive weekly sock reurn, was he marke innovaion or residuals, h was he respecive condiional variance of he reurns process based on he informaion se of relevan and available pas daa; ω, β, γ and α were parameers o be esimaed, were dummy variables aking a value of 1 for each breakpoin of sudden shif of variance and 0 oherwise. The condiional variance was asymmerical, if he esimaed leverage-effec erm, γ, was significanly differen from zero (γ 0). Parameers in he variance equaion (Equaion 5) were obained hrough he maximum likelihood esimaion, namely he Marquard mehod wih robus sandard errors. The adequacy of he EGARCH-M model was examined by employing hree sandard diagnosic ess, namely he Ljung Box Q-saisic and Jarque-Berra saisic in deecing any possible auocorrelaion and non-normaliy of he sandardized residuals; he ARCH LM es, in addiion, was employed o deec wheher he sandardized residuals exhibied any auoregressive condiional heeroskedasiciy (ARCH) effec. (5)
EMPIRICAL RESULTS Based on Table 1 which repored he descripive saisics for he five indices, a number of observaions can be made. Firs, during our sample period from June 1990 o June 2011, all he Malaysian sock prices have a posiive mean reurn, wih he Properies Index demonsraed a higher reurn han he ohers. Secondly, he sandard deviaion, which measured sock reurn volailiy, indicaed ha he Properies Index has he highes value of 1.56%, among all. Thirdly, Reurns from KLCI and all he oher four secors displayed posiive skewness. All he coefficiens of skewness repored in Table 1 indicaed ha sock reurns in Bursa Malaysia were posiively skewed and no normally disribuion. Fourhly, he kurosis values for all he five indices exceeded 3, indicaing a lepokuric disribuion. The Jarque-Bera (JB) saisic repored in he fifh column of Table 1 confirmed he significan deparure of hese sock reurns from normal disribuion, which was similar o he findings of Kang e al. (2009), Wang and Moore (2009), even hough hese previous analyses were done on oher markes. In view of his, any hypohesis esing on he process of reurns generaion in his emerging marke will be limied as he reurns were no normally disribued. Inser able 1 here In Table 2, he deeced sudden shifs in he sock marke were lised. I was found ha mos of he sudden shifs of variance for all indices, eiher he general main lising or secor lising, occurred around similar break posiions over ime. In his sudy, we managed o idenify he evens ha migh have caused he break posiions, or sudden shifs in volailiy for reurns in he main marke and secoral marke. I was found ha sudden changes in volailiy of hese ransacions were caused by boh global and regional evens, since he Malaysian sock marke was quie inegraed wih ohers globally. These evens included: huge porfolio capial inflows in 1993, massive capial fligh in early 1994, he 1997 98 Asian Financial Crisis; he 11 Sepember 2001 caasrophe; he 2003-04 US-Iraq war and Severe Acue Respiraory Syndrome (SARS) oubreak, he 2005-06 poliical crisis in Thailand and Eas Timorese Crisis, he 2007 sub-prime morgage crisis; he US financial crisis of 2008; he 2008-09 Global Financial Crisis; and he recen European Deb Crisis of 2010-11. In general, hese sudden shifs in variance indicaed asymmerical responses of he marke o posiive and negaive news. Inser able 2 here Table 3 repored he esimaed ICSS-EGARCH-M model esimaion. To assess he descripive validiy of he model, he Ljung-Box Q saisics were calculaed for serial correlaion in sandardized residuals, and squared sandardized residuals. All series were found o be free of serial correlaion using he squared sandardized residuals. The ARCH-LM es suggesed in general he absence of heeroskedasiciy in he residuals. The absence of serial correlaion in he sandardized squared residuals implied ha he model was well specified. The significan across any indices (in column 2 5 of Table 3) implied a deparure from he Efficien Marke Hypohesis, where pas informaion can be used o improve fuure predicabiliy of prices or reurns. This resul was consisen wih ha of Marshall and Cahan (2005). I indicaed ha small emerging Malaysian marke, which is quie in isolaion from major markes, is no aracive o hedge funds and oher professional invesors, and hus inefficiencies were no rade away, as hose happened in larger and developed markes. In addiion, prohibiion of shor-selling in Malaysia during he sample period also formed an environmen where inefficiencies were more likely o persis. The deparure from he Efficien Marke Hypohesis during he sample period also refleced he chaoic financial environmen in he marke. The marke was affeced significanly by various news evens, in paricular hose peraining o he crises. Under hese chaoic marke condiions, invesors end o overreac no only o local news, bu also o news originaed from oher inernaional markes. I was observed ha he esimaed risk coefficien φ (S. Dev.) of condiional variance was significan in all he five secors. In oher words, a posiive and significan associaion was found beween risk and reurn in all cases of hese sock secors. This implied ha, in general, an invesor in he Malaysian sock marke, who bears risk, will be compensaed wih a higher reurn. The γ coefficien, on he oher hand, revealed wheher asymmeric news impacs have been observed. The esimaed resuls repored in Table 3 indicaed ha he esimaed γ coefficiens of Kuala Lumpur Composie index and indusrial index were saisically significan, indicaing he presence of
asymmerical news effecs across he main lising and indusrial secor. I seemed ha negaive price shocks led o considerably higher volailiy in sock reurns as compared o posiive shocks. In addiion, he esimaed value of coefficien (i = 1, 2) of all he five indices displayed raher low persisence o volailiy shocks. These findings were consisen wih hose repored by Kang e al. (2009) for he case of anoher small emerging Asian sock marke - he Korean sock marke. Inser able 3 here CONCLUDING REMARKS This paper examined he efficiency of an emerging sock marke, namely he Malaysian sock exchanges, in erms of is reurns and condiional volailiy in he presence of sudden shifs in uncondiional variance vis-à-vis an Ieraed-Cumulaive-Sum-of-Squares-algorihm-in-volailiy (ICSS-EGARCH-M) model. Our empirical resuls, afer accouning for sudden shifs of volailiy, sill indicaed a significan deparure from he Efficien Marke Hypohesis for he main and all he four sock secors. Among all, he general index for he main marke KLCI and he secor index Indusrial, were he mos inefficien. The resuls also provided significan empirical evidence for posiive risk-reurn relaionship in he exchange. Moreover, his sudy also found ha he sock marke, across all secors, was more sensiive o global news evens as compare o he local ones. The asymmerical responses o good and bad news were also an imporan characerisic of he Malaysian marke behavior. REFERENCES Bollerslev, T. (1986). Generalized Auoregressive Condiional Heeroskedasiciy. Journal of Economerics, 31, 307-327. Bollerslev, T., Chou, R.Y., & Kroner, K.F. (1992). ARCH modeling in finance: a review of he heory and empirical evidence. Journal of Economerics, 52, 5 59. Campbell, J., Lo, A., & Mackinlay, A. (1997). The Economerics of Financial Markes. Princeon Universiy Press. Charles, A., & Darne, O. (2005). Ouliers and GARCH models in financial daa. Economics Leers, 86, 347-352. Fama, E. (1970). Efficien capial marke: a review of heory and empirical work. Journal of Finance, 25, 383-417. Fama, E. (1991). Efficien capial markes: II. Journal of Finance, 46, 1575-1615. Franses, P.H., & Ghijsels, H. (1999). Addiive ouliers, GARCH and forecasing volailiy. Inernaional Journal of Forecasing, 15, 1-9. Goh, K.L., Wong, Y.C., & Kok, K.L. (2005). Financial crisis and ineremporal linkages across he ASEAN- 5 sock markes. Review of Quaniaive Finance and Accouning, 24, 359 377. Hammoudeh, S., & Li, H. (2008). Sudden changes in volailiy in emerging markes: he case of Gulf Arab sock markes. Inernaional Review of Financial Analysis, 17, 47-63. Hooy, C.H., Tan, H.B., & Nasir A.M. (2004). Risk sensiiviy of bank socks in Malaysia: empirical evidence across he Asian Financial Crisis. Asian Economic Journal, 18, 261-276. Inclan, C., & Tiao, G.C. (1994). Use of cumulaive sums of squares for rerospecive deecion of changes of variance. Journal of he American Saisical Associaion, 89, 913 923. Kang, S.H., Cho, H.G., & Yoon, S.M. (2009). Modelling sudden volailiy changes: evidence from Japanese and Korean sock markes. Physica, A388, 3543-3550. Kawakasu, H., & Morey, M.O. (1999). An empirical examinaion of financial liberalizaion and he efficiency of emerging marke sock prices. Journal of Financial Research, XXII, 385 411. Lasrapes, W.D. (1989). Exchange rae volailiy and U.S. moneary policy: an ARCH applicaion. Journal of Money, Credi and Banking, 21, 66-77. Law, S.H. (2006). Has sock marke volailiy in he Kuala Lumpur sock exchange reurned o pre-asian financial crisis levels? ASEAN Economic Bullein, 23, 212 219. Malik, F. (2003). Sudden changes in variance and Volailiy persisence in foreign exchange markes. Journal of Mulinaional Financial Managemen, 13, 217-30.
Marshall, B.R., & Cahan, R.H. (2005). Is echnical analysis profiable on a sock marke which has characerisics ha sugges i may be inefficien? Research in Inernaional Business and Finance, 19, 384-398. Nwogugu, M. (2006). Furher criique of GARCH/ARMA/VAR/EVT sochasic-volailiy models and relaed approaches. Applied Mahemaics and Compuaion, 182, 1735-1748. Nelson, D. (1991). Condiional Heeroskedasiciy in asse reurns: a new approach. Economerica, 59, 347 370. Ome, G., Khasawneh, M., & Khasawneh, J. (2002). Efficiency ess and volailiy effecs: evidence from Jordanian sock marke. Applied Economics Leers, 9, 817-821. Wang, P., & Moore, T. (2009). Sudden changes in volailiy: he case of five cenral European sock markes. Journal of Inernaional Financial Markes, Insiuions & Money, 19, 33-46. Wilson, B., Aggarwal, R., & Inclan, C. (1996). Deecing volailiy changes across he oil secor. Journal of Fuures Markes, 16, 313-30. Table 1. Summary saisics for he Malaysian sock marke weekly reurns KLCI Finance Indusrial Planaion Properies Mean 0.0012 0.0016 0.0013 0.0016 0.0019 Sd. Dev. 0.0129 0.0147 0.0126 0.0138 0.0156 Skewness 2.1998 2.3142 3.4857 1.1366 2.7368 Kurosis 30.4179 30.1676 45.8603 24.4249 31.9140 Jarque-Bera 34346.01 33829.48 83988.27 20676.11 38572.14 (P-value) 0.0000 0.0000 0.0000 0.0000 0.0000 Observaions 1069 1069 1069 1069 1069 Noes: Skewness measures he asymmery of he disribuion of he series around is mean. The skewness of a normal disribuion is zero. Kurosis measures he peakness or flaness of he disribuion of he series. The kurosis of he normal disribuion is 3. If he kurosis exceeds 3, he disribuion is lepokuric, and if less han 3 playkuric relaive o he normal disribuion. Table 2. Deeced changes in variance based on he ICSS Algorihm procedure Imporan change poins Evens KLCI 10/26/1990 Prior o he Gulf war 06/04/1993, 12/24/1993, Huge porfolio capial inflows 01/28/1994, 01/06/1995, Massive capial fligh 03/17/1995 08/29/1997, 09/05/1997 Asian financial and currency crisis 8/14/1998, 9/11/1998 Russian financial crisis 09/21/2001 Sepember 11 aacks 7/1/2004 The US-Iraq war and Severe Acue Respiraory Syndrome (SARS) oubreak 11/17/2006 Poliical crisis in Thailand and Eas Timorese crisis 7/6/2007 US sub-prime morgage crisis 8/8/2008, 4/24/2009 Global financial crisis and deleveraging aciviies in he Finance 10/29/1993 Huge porfolio capial inflows 01/14/1994, 01/06/1995, Massive capial fligh 03/17/1995 08/22/1997, 9/12/1997 Asian financial and currency crisis 08/14/1998, 09/11/1998 Russian financial defaul 10/15/1999, 3/30/2001 Inerne bubble 09/21/2001 Sepember 11 aacks 6/3/2004 The US-Iraq war and Severe Acue Respiraory Syndrome (SARS) oubreak
11/17/2006 Poliical crisis in Thailand and Eas Timorese crisis 8/8/2008 US sub-prime morgage crisis 10/24/2008, 6/12/2009 Global financial crisis and deleveraging aciviies in he 5/21/2010, 1/21/2011 European deb crisis Indusrial 08/20/1993 Huge porfolio capial inflows 01/14/1994, 01/06/1995, Massive capial fligh 02/24/1995 08/22/1997, 9/5/1997 Asian financial and currency crisis 7/3/1998, 08/14/1998, Russian financial defaul 09/11/1998 06/16/2000 Inerne bubble 09/21/2001 Sepember 11 aacks 8/26/2004-6/16/2006 Poliical crisis in Thailand and Eas Timorese crisis 7/20/2007, 11/23/2007, US sub-prime morgage crisis 1/18/2008 10/24/2008, 7/31/2009 Global financial crisis and deleveraging aciviies in he Planaion 10/25/1991-11/13/1992, 10/29/1993 Huge porfolio capial inflows 01/14/1994, 06/10/1994, Massive capial fligh 01/27/1995, 02/10/1995, 05/26/1995 08/22/1997, 09/12/1997 Asian financial and currency crisis 07/24/1998, 10/02/1998 Russian financial crisis 03/10/2000, 06/22/2001-9/28/2001 Sepember 11 aacks 9/18/2003, 7/8/2004 The US-Iraq war and Severe Acue Respiraory Syndrome (SARS) oubreak 1/27/2006 Poliical crisis in Thailand 10/19/2007 US sub-prime morgage crisis 7/11/2008, 1/2/2009, Global financial crisis and deleveraging aciviies in he 8/14/2009, 10/29/2010 European deb crisis Properies 06/11/1993 Huge porfolio capial inflows 05/12/1995, 12/15/1995 Massive capial fligh 03/07/1997, 08/22/1997, Asian financial and currency crisis 9/5/1997, 2/06/1998 08/14/1998, 09/11/1998 Russian financial crisis 03/10/2000 Inerne bubble 09/07/2001, 09/21/2001, Sepember 11 aacks. The invasion of Iraq 6/27/2002, 7/4/2002 5/20/2004 The US-Iraq war and Severe Acue Respiraory Syndrome (SARS) oubreak 12/30/2005 Poliical crisis in Thailand and Eas Timorese crisis. The energy shock by rising oil prices 2/2/2007, 5/4/2007-10/8/2007, 8/24/2007, US sub-prime morgage crisis 9/12/2008 6/19/2009 Global financial crisis and deleveraging aciviies in he
Table 3. The esimaes of ICSS-EGARCH-M parameers KLCI INDUSTRIAL FINANCE PLANTATION PROPERTIES 0.9701(0.000) 0.9498(0.000) 0.9471 (0.000) 0.9440(0.000) -0.9816(0.000) -0.9559(0.000) -0.9445 (0.000) -0.9477(0.000) 0.8464(0.000) -0.8173(0.000) (S. dev.) 0.1287(0.000) 0.1009(0.000) 0.1217 (0.000) 0.0463(0.051) 0.0691(0.047) -4.1737(0.005) -5.2794(0.000) -2.2509(0.000) -5.7589(0.000) -9.7346(0.002) 0.1205(0.121) 0.2252(0.009) 0.0005(0.992) 0.2740(0.003) 0.0931(0.275) -0.0991(0.048) -0.1564(0.006) 0.0030(0.939) -0.0590(0.377) 0.0411(0.497) 0.5767(0.000) 0.4559(0.000) 0.7481(0.000) 0.4218(0.005) -0.0928(0.792) d 1 0.5794(0.004) 0.7120(0.011) 1.3119(0.009) d 2 0.5312(0.034) -0.7291(0.011) 0.6277(0.005) -1.1104(0.058) d 3 1.2993(0.085) 0.5602(0.023) -1.7980(0.018) d 4-1.4775(0.052) -0.6387(0.004) -0.6974(0.008) 1.1302(0.081) d 5 0.8702(0.001) 1.7729(0.000) 2.5019(0.018) d 6-0.3668(0.063) -1.2396(0.002) d 7 4.5799(0.001) 0.9124(0.010) -1.4089(0.079) d 8-3.5331(0.004) -1.0181(0.004) 4.1831(0.014) d 9-0.2329(0.042) -3.6452(0.014) d 10-0.5388(0.026) -0.5798(0.008) 0.2724(0.069) -1.3415(0.016) d 11-0.5470(0.013) -0.6264(0.003) -0.3991(0.021) 1.5287(0.023) d 12-0.3784(0.031) -0.2166(0.019) -1.7935(0.013) d 13 0.7119(0.011) 0.2407(0.014) d 14 0.6778(0.021) 0.8627(0.050) d 15 1.9356(0.000) -0.8021(0.009) -1.3111(0.009) -1.5419(0.009) d 16-1.3285(0.001) 0.6253(0.039) 0.7227(0.088) d 17-0.5949(0.013) -0.9731(0.011) 1.4386(0.017) d 18-0.7034(0.008) 1.1018(0.002) d 19 d 20 0.9737(0.024) d 21-1.1470(0.012) d 22-0.8487(0.034) -1.3375(0.009) AIC -6.5530-6.5760-6.2154-6.3126-6.1001 SIC -6.4646-6.5016-6.1083-6.2241-6.0022 Log-likelihood 3518.32 3527.62 3342.07 3389.94 3272.39 Q(20) 18.51(0.422) 11.30(0.881) 25.33(0.116) 13.86(0.738) 21.48(0.256) Q 2 (20) 15.12(0.653) 9.89(0.935) 10.70(0.906) 26.73(0.084) 9.76(0.939) ARCH LM es 0.9015(0.585) 0.5759(0.931) 0.4041(0.991) 1.2968(0.171) 0.5590(0.940) Noes: p-values were repored in he parenhesis. Q(20) and Q2(20) were he Ljung-Box Q saisics for residuals and sandardized squared residuals a lag 20. The opimal srucure of ARCH LM es was deeced a lag 20. The empy cells in he able denoed non-applicable of he respecive independen variables.