Asian Journal of Empirical Research

Transcription

1 Asian Journal of Empirical Research journal homepage: hp://aessweb.com/journal-deail.php?id=5004 ASSOCIATION BETWEEN ASIAN EQUITY MARKETS AND WESTERN MARKETS: EVIDENCE FROM THE INDEXES OF EQUITY MARKETS Jeffrey E. Jarre *1 Alina F. Klein 2 Eric Kyper 3 ABSTRACT This research examined he ime series characerisics of sock price indices for Hong Kong, Tokyo, New York (NYSE) and London (FTSE) equiy markes or sock exchanges during he period of 1991 o Specifically, we calculae he rae of reurn and he volailiy of reurn for all he markes and esimae he serial correlaion and co-movemen of he four markes. We find ha he average rae of reurn varies dramaically for he four equiy markes and across ime. Furher, we find ha sock prices are posiively serially correlaed in general. In he mulivariae regressions, we find ha here is lile evidence o show ha eiher he rae of reurn in cerain markes universally affecs he rae of reurn in oher equiy markes. I suggess ha he four markes are co-inegraed bu no universally across ime and wih each oher in pairwise dimensions. Lasly, we sudied and made conclusions concerning he mean and variaion in he volailiy of he raes of reurn in he four equiy markes sudied. Keywords: Volailiy, Correlaion, Equiy Markes. INTRODUCTION Our purpose is o sudy hree ses of weekly price indices: Tokyo Composie Index, NYSE Composie Index, London Composie Index and he Hang Seng (Hong Kong, hereofore, Hang Seng) Composie Index provided by common daa colleced during he period of Sudies of hese indices are imporan because of he rapid growh and influence of he Asian and especially he Chinese economies on he world, balance of rade and growh of Asian and oher economies hroughou he world (Chow e al. hereafer CMP, 1999; Jarre and Sun, 2009). 1 Universiy of Rhode Island Kingson, RI (USA). jejarre@mail.uri.edu 2 Lynchburg Universiy Lynchburg VA (USA) 3 Lynchburg Universiy Lynchburg VA (USA) 972

2 Previous sudies (Chen, 1991; Cheung and Ng, 1998; Liaw, ) described China as an economic power offering remendous opporuniies for invesmen and growing business reurns. Their financial markes for he earlier years in heir developmen were hough no o be fully developed when analyzed by he crieria developed by financial economiss using crieria for analyzing Wesern equiy markes (Fama, 1990, 1991; Wei and Wong, 1992; and Zhong e al. 1999). Chow and Lawler (2003, daa up and ill 2002), and laer, Jarre and Sun (2012), analyzed he price index for he Shanghai Sock Exchange in comparison wih he New York Sock Exchange Index in erms of is rae of reurn, volailiy and srucural changes in he movemen of he index. In his sudy, we propose o analyze he enire period from January 1991 o December 2012 dividing he period ino sub periods (sub samples) o analyze change associaed wih ime and especially significan economic evens. The comparisons have he purpose of revealing he behavior of sock movemens in an emerging marke in comparison wih an esablished Wesern marke. Previous sudies by Bailey e al. (2009), Jarre and Sun (2009A, B), Jarre and Kyper (2010) focused on oher issues in Chinese equiy markes. We will now focus on wo of he larges Asian equiy markes because hey are cenral areas of rade and economic aciviies no expressed only in he Shanghai equiy marke of China. Alhough previous sudies show ha Asian markes became and coninue o inegrae hemselves wih oher and small Asian equiy exchanges (Jarre and Sun, 2012). Anoher quesion relaes o wheher here is some inegraion beween he New York and Shanghai markes as seen by sudying he co-movemen of sock prices in hese exchanges. This will enable one o assess he degree of inegraion of he Chinese economy wih he res of he world as represened by he movemen of prices in he New York Sock Exchange (NYSE). We will also look a he correlaions among he Shanghai, NYSE and Hong Kong markes (Hang Seng Index) o examine heir inegraion as well. Lasly, one noes ha he Chinese financial markes are no open in he Wesern sense of he erm bu our sudy should yield some observaions abou he relaive openness of he Chinese financial markes. We examine boh he rae of reurn and he volailiy of he price indexes. The rae of reurn is he change in he naural logarihm of he price index for a given ime period. We follow Chow and Lawler (2003, hereafer CL) and Jarre and Sun (2012, hereafer, JS) in measuring he volailiy by he absolue value of he change raher han by is variance. The absolue value is less sensiive o exreme value as compared wih ARCH-ype models o sudy he residual variance of a ime series model. Saed differenly, we sudy he volailiy of he rae of reurn iself and no he residual in he ime series model of he rae of reurn. Following CL and JS, (1) he volailiy in he rae of reurn and no he ime series regression model residual is he subjec of ineres in financial research and (2), since log sock price behaves approximaely like a random walk, he rae of reurn iself and he residual of an auo regression of his rae are almos he same (CL, p. 18). The daa for his sudy include four ses of weekly price indices: Tokyo Composie Index, London Composie Index, NYSE Composie Index, and Hang Seng Composie Index provided during he period of The rae of reurn is 973

3 calculaed as he change in he naural logarihm of he price index in a given period. The volailiy of reurns is calculaed as he absolue value of he change in he naural logarihm of he price index in a given period. We furher divide our sample ino hree subsamples: before 1997, afer 1997 and before, and afer. The enire sample period is from Ocober 1987 o Ocober Boh 1997 and are years in which he economic environmen changed. In urn his affeced he world s equiy markes. Hence, we separaed he daa ino hree sub-ime-periods. Lasly, we follow CL and JS in choosing he weekly daa as he bes choice among daily, weekly and monhly daa. To begin, we examine he characerisics of he equiy markes. We calculae he mean and variance of he rae of reurn and he mean and variance of he measure of volailiy. If boh hese measures reflec uncerainy, he volailiy in Asian sock prices should be more volaile han hose in New York. To sudy he co-movemens of he price in he wo markes, we calculae simple correlaions and muliple regressions. The muliple regressions include auo regressions as well as ordinary muliple regressions. There was no raional reason o examine curvilinear models since he sudy of residuals did no indicae nonlineariy in relaionships. The remainder of his paper is organized as follows: (1) he characerisics of he rae of reurn and he volailiy of reurn; (2) correlaion coefficiens; (3) regressions of he rae of reurn; (4) regressions of he volailiy of reurn; and (5) conclusions. Rae of Reurn and Volailiy of Shanghai and New York Price Indices Table-1 shows he informaion for he Hong Kong, Tokyo, London and New York sock price indices including he marke capializaion and he number of lised socks. The sizes of he four financial markes indicae ha he New York equiy marke is much larger han he oher hree equiy markes in erms of marke capializaion bu no he number of lisings. London is larges in erms of number of lisings indicaing ha he marke capializaion per lising is much smaller. Tokyo and Hong Kong are smaller han New York in erms of marke capializaion and especially marke capializaion per lising. There is a rich lieraure on he relaionship beween marke lisings and he size of he economy Levine and Zervos (1998). For a summary of curren discussion on his opic see Levine (2005). Table-1. Size and number of socks Sock Exchange Marke capializaion Number of Lisings (US$ rillion) Hang Seng (Hong Kong) London (FTSE) New York (NYSE) Tokyo Table-2 shows he means and sandard deviaions (variaion) of he raes of reurn for he four equiy markes, while Table-3 presens he means and sandard deviaions of he volailiy of 974

4 reurns in he same four markes. For Table-2, he means are highes for Hang Seng followed by New York, London and Tokyo which is negaive for he ime period analyzed. The sandard deviaion is highes for Hang Seng followed by Tokyo, wih New York and London being very close. As expeced, we observe ha he variaions in he Asian markes are greaer he Wesern markes. Observe he differences in he volailiy in he raes of reurn for he same four equiy markes. The Asian equiy markes have higher volailiy in he raes of reurn for he Wesern equiy markes. The variaion (sandard deviaion) in volailiy is again much higher in he Hang Seng marke han he ohers. The Tokyo marke s sandard deviaion is again larger han he wo Wesern markes. Table-2. Means and variaions (sandard deviaions) in raes of reurns Sock Exchange Hang Seng London New York Tokyo Mean Sandard deviaion Table-3. Means and variaions (sandard deviaions) in volailiy of reurns Sock Exchange Hang Seng London New York Tokyo Mean Sandard deviaion The mean rae of reurn for Hang Seng ( ) is abou weny-five percen larger han he mean raes of reurn for New York ( ), more han 50 percen larger han London ( ) and he mean rae of reurn for Tokyo is negaive ( ). Thus, mean rae of reurn for Hang Seng is growing a a rae much larger han for he oher equiy markes. All hree of hem represen developed economies whereas Hang Seng, while no underdeveloped, is grealy influenced by he dynamics of he Chinese equiy marke and economy (Pan e al. 2012) due o shares being crosslised wih he Shanghai sock exchanges. If we were o consider change in price levels for he wo naions by examining daa onhe consumer price indexes for he four naions (alhough no a perfec comparison), he changes in prices would no accoun for he major porion of he differences in he average raes of reurn. This leads o a conclusion ha he greaer mean rae of reurn for he Hang Seng Index is no aribuable o facors oher han he invesmen opporuniies in is marke and he influence of dual-lised equiies wih hose of China. Volailiy (as noed before) as measured by boh he sandard deviaion of he rae of reurn and by he mean volailiy in he rae of reurn is again larger for he Hang Seng sock marke han for he hree alernaive sock markes. Table-2 shows a much larger variaion for he Hang Seng sock index han for he ohers. Table-3 concurs by showing a much large mean volailiy of reurn for Hang Seng as well. This suggess a grea deal of uncerainy in he Hang Seng marke in comparison o he oher markes. Furhermore, he sandard deviaion of he measure in Table-3 is also much greaer for Hang Seng han i is for New York, London and Tokyo. These resuls are no a revelaion and are similar as hose of CL ad JS. This would lead one o observe ha he volailiy 975

5 is subjec o a greaer degree of variaion, ha is he spread in he disribuion in Hang Seng is greaer han he disribuion for he oher hree markes. Again, his is consisen wih he previous findings by CL and JS. This is no o say ha volailiy does no exis or is even small in he oher markes, bu only o say ha a risk-averse invesor is beer served by he developed markes of New York and London, whereas Tokyo appears no o serve well o any ype of invesor. To es how he wo equiy markes behave during exreme evens, we sor sample period o hree economic sub-periods and examine he mean and volailiy of he rae of reurn during hese periods: (1) 1987 hrough he las week of 1996; (2) 1997 o he end of ; and (3) afer unil he end of he sample daa period. By sudying hese hree periods, one may deermine if severe economic changes occurring in 1997 and affec he wo markes and wheher he changes are differen. We find ha a minuscule change in he mean rae of reurn for Hang Seng from period 1 o period 2 and he negaive mean rae of reurn in period 3 repored in Table-4, Panel A. The mean rae of reurn for London remained similar in periods 1 and 2 bu became negaive in period 3. For New York, he mean rae increased from period 1 o period 2 bu became negaive slighly in period 3. The Tokyo index was posiive slighly in period 1 bu declined o a negaive value in period 2 and more negaive in period 3. Afer he declines are very eviden in all markes. In sudying variaion, we observe he sandard deviaions of he rae of reurn are larges in all hree periods for he Hang Seng Index. In all hree ime periods he Tokyo exchange had he second larges variaion in raes of reurn bu no nearly as large as ha for he Hang Seng marke. The wo Wesern markes provided very similar resuls as hose of he enire ime period. The sandard deviaions were significanly smaller han Hang Seng and Tokyo, bu he variaion did increase largely in period 3. In Table-4, Panel B, we observe he changes in he mean and variaion in he volailiy of he raes of reurn for he four equiy markes across he hree ime periods. The mean volailiy in he raes of reurn did no change significanly in he hree ime periods. The order in erms of size of mean did no change. Hang Seng has he larges mean, followed by Tokyo wih London and New York being small bu very similar. For volailiy variaion, he sandard deviaion across all ime periods indicaed he greaes variaion in Hang Seng followed by Tokyo wih he London and New York markes varying for hird and fourh place depending on he ime period. The resuls demonsrae ha mean rae fell in period 3 and variaion and volailiy increased in period 3 relaive o he oher periods. Tokyo appeared o have he smalles posiive gains and was general in he negaive for he mean rae of reurn. Alhough Hang Seng may have performed beer han he wo Wesern markes and Tokyo according o he mean raes, here is no doub ha he variaion is much larger across he ime periods in Hang Seng han in he more developed markes of he Wesern naions and Tokyo. Our resuls vary from CL and TS due o a lenghier ime periods and he noion ha we have hree more developed markes, i.e. New York, London and Tokyo in comparison o he lesser developed marke of Hoang Seng. 976

6 Table-4. Rae of reurn and volailiy in hree subsamples of ime Panel-A: Rae of reurn Before 1997 Hang Seng London New York Tokyo Mean Sandard deviaion Hang Seng London New York Tokyo Mean Sandard deviaion Afer Hang Seng London New York Tokyo Mean Sandard deviaion Panel-B: Volailiy of reurn Before 1997 Hang Seng London New York Tokyo Mean Sandard deviaion Hang Seng London New York Tokyo Mean Sandard deviaion Afer Hang Seng London New York Tokyo Mean Sandard deviaion The economic change occurring in he world during he hree ime periods kep he Hang Seng marke more volaile and more profiable han he ones in New York, London and Tokyo. Riskaverse invesors were beer off in New York and London afer because of he smaller level of volailiy bu hey are giving up he possibiliy of higher reurn. The Tokyo marke appeared o be he leas wise choice for all hree ime period and also ended o be more risky han markes of New York and London. The differences in he sample saisics for he hree ime periods sugges ha he raes of reurn and volailiy in sock prices in nominal erms for he enire ime period sudied were no covariae saionary ime series. This is he same conclusion for a much larger ime period han observed by CL and TS. Their conclusions a his poin are no dispued bu only enhance by he sudy of he new and expanded analysis of he four markes. The previous sudies only observe one Wesern and one Asian marke (Shanghai, which is less developed and no much larger han Hang Seng). The Correlaion in Price Movemens Following CL and JS, we arrive a a preliminary view of he level of inegraion among he Han Seng, London, New York and Tokyo Exchanges by examining he simple (Pearson) correlaion coefficiens. Table-5 conains he Person Produc Momen correlaion coefficiens of he four exchanges. All daa came from known public sources. Noe ha he New York and London markes for he rae of reurn (Panel A) have a correlaion of 0.710, indicaing ha almos 50% of variaion in he wo markes is associaed wih each oher. The same Pearson Produc Momen Correlaion coefficiens for all oher wo-by-wo combinaion have coefficiens beween and Alhough smaller, hese coefficiens are all significan a he.01 level of significance and indicae some relaionship beween he raes of reurn in each marke. For volailiy (Panel B, 977

7 Table-5) in he raes of reurn, again he London-New York combinaion indicaes far greaer correlaion han for he oher possible combinaion. However, he oher combinaions aken wo a a ime and Shanghai-Kong Kong are and Hence, he associaion beween volailiy in he raes of reurn are saisically differen (a 1% level of significance), herefore, he associaion beween he volailiy in he raes of reurn among hese four markes aken wo a a ime are differ in he saisical sense. Unlike TS and o some exen CL, we observe ha hese markes influence each oher. Table-5 indicaes ha a firs glance ha he four exchanges covariae or are relaed o each oher over he ime period sudied. Whereas he London and New York markes do show more co-variaion, herefore hey are ofen influenced by he same economic facors. The smaller bu non insignifican correlaion coefficiens for he rae of reurn and volailiy of reurn for he oher combinaions of exchanges sugges ha hese equiy markes operae wih each oher and pairwise independence does no exis. These differ from he earlier sudies of CL and JS and herefore lead us o differen conclusion. However, he CL and JS sudy followed he relaion of Shanghai and New York in a small ime environmen. These resuls would never be he same bu our conclusions sugges a differen paern of behavior. Table-5. Raes of reurn and volailiy in hree subsample ime periods Panel-A: Raes of reurn Hang Seng London New York Tokyo Hank Seng 1 London New York Tokyo Panel-B: Volailiy of reurn Hang Seng London New York Tokyo Hank Seng 1 London New York Tokyo Previous research on he relaions beween large and small sock reurns in six Asian (Pacific- Basin) naions and he associaion among he same six Asian financial markes is exemplified by Jarre and Sun (2009A). Their purpose was o provide evidence of he cross-auocorrelaion of sock reurns in a lenghy ime period. Evidence was brough o bear as o he heoreical explanaions for sock marke behavior of Pacific-Basin naions including hose wih large financial markes, i.e. Japan and Hong Kong, and hose wih small financial markes, i.e. Thailand and Malaysia. This sudy hough differen han ours indicaes clearly he relaionship of large and small equiy markes and give us furher desire o learn more abou he coinegraion of equiy markes hroughou he world. We can now sill learn more abou he level of inegraion in hese four markes by analyzing muliple regressions, and in doing so we exclude he influence of he delayed effecs of lagged explanaory variables. 978

8 Regressions of he Rae of Reurn We define he rae of reurn o be he change in he naural logarihm of he sock price from period -1. According o he efficien markes hypohesis (EMH), he rae of reurn is difficul o predic wih any reasonable level of accuracy. Hence, we wish o deermine if here is validiy in his hypohesis and wheher raes of reurn in he four markes are correlaed afer excluding he influence of heir own lagged values. We consruc a model o explain he Hang Seng rae of reurn by is own pas values. By consrucing a model wih many lagged values of he rae of reurn and calculae he Akaike Informaion Crierion (AIC), we find ha AIC is minimized a a lag of one. In urn, we find he firs-order auo regression which appears in column (2) of Table-6. The coefficien (H 1 ) is wih a -saisic of 1.48 for all daa which is no significan a he reasonable level of α 0.05 (he p-value being large). According o his resul he weak form of he efficien markes hypohesis does hold for his ime period. For London, L1 is (-saisic =3.34), for New York, N1 is (-saisic = -2.03), and for Tokyo, T1 is (-saisic = 2.85). The resuls are mixed bu all have -saisics ha are significan a α.05. Nex, we will furher invesigae his phenomenon by dividing he ime series daa ino he same hree sub-periods analyzed previously. For all four equiy markes, we subdivide he daa ino he hree ime periods noed previously. In all hree ime periods he lagged variable of one resuled in coefficiens of 0.143, and for Hang Seng, bu only he firs had a sufficienly large enough -saisic o rejec he null hypohesis of he parameer equaling zero. Thus, he resuls are mixed for he Hang Seng marke. For he auo regression of London, we find he L1 coefficiens o be , and However, only he coefficien for period (before 1997) is significan a α Again he London marke yields mixed resuls. For New York, N1 = , and Only he period had a significan -saisic. Lasly, for Tokyo, he coefficiens for T1 are , and Only he period afer has a significan -saisic. These resuls are no enirely he same as hose of CL or JS, bu do indicae ha change occurred among he hree sample sub-periods sudied. The resuls are mixed in all four equiy markes indicaing differen effecs during he enire ime period and differenial effec in each of he sub periods. CL and JS observed ha lack of srong correlaion beween he equiy markes of New York and Shanghai which corroborae an earlier and similar sudy of Bekare and Harvey (1995). Wih he coninued and dynamic developmen of he of he Asian equiy markes, we should now observe he more recen analysis of heir coinegraion exhibied in Table-7 (panels A and B). In column 2, Table-7, we noe he value of L2 and is relaed -saisic of and The values indicae ha he curren Hang Seng rae of reurn (he response variable) associaed wih he curren London rae of reurn (he explanaory variable) is boh small and significan. Furher, wih a -saisic of 2.17 (for N0), we find ha he coefficien of he conemporaneous New York rae has associaion wih he Hang Seng rae of reurn for he enire ime period sudied. For NO, N1 and T0, wih T2 in column 2 in Panel B (All Daa), one observes small and significan - saisics. This suppors he noion ha he wo markes are inegraed and have some associaion 979

9 wih each oher during he enire ime period sudied. By examining he hree sub periods (e.g. before 1997, 1997 and afer ) one coninues o observe he small and some significan -saisics among all he equiy markes paired wih each oher. In general, he laer wo sub periods provide -saisicswere no significan. This is he only noiceable change associaed wih change in ime period. Hence, only for he periods saring in 1997 did he relaionship change beween he equiy markes sudied for he four equiy markes. Addiional ess such as he Chow es based on he F-disribuion do no rejec he hypohesis ha he coefficiens of he hree sub sample ime periods are he same. Table-6. Auo regressions of he raes of reurn on equiies lised in he four exchanges Regressions of he Volailiy of Raes of Reurn We consruc a regression model wih he purpose of explaining he volailiy in he Hang Seng and he oher hree equiy markes. Firs, we accoun for he effecs of heir own volailiy associaed wih heir pas values. Following CL and JS he appropriae number of lagged explanaory variables o include in he respecive models is deermined by (1) he significance of individual parameer esimaes; (2) by minimizing he AIC value; and (3) he presence or absence of serial correlaion in he residual. By including one lagged response variable a a ime, we follow CL and JS and observe he hree crieria o consruc a model explaining he curren volailiy in he four sock exchanges. 980

10 Table-7. Regressions of rae of reurn in all four markes Table-7. Regressions of rae of reurn in all four markes 981

11 In Table-8, we find for Hang Seng (All Daa) ha all lagged variables have significan -saisics (12.23, 4.19, 1.98, and 3.59). Tess for serial correlaion applied o he model having four lagged values yield a small and significan -saisics. Column 2 (All Daa) and Column (5) of Table-8 Panel A, and columns (2) and (5) of Panel B show he resuls for all four sock exchanges. London (All Daa) has large significan coefficiens for lags 1, 2 and 3 bu small and non-significan marginal coefficiens for lag 4. For New York, significan lags were found a 1, 2 and 3 bu no 4. Table-8. Auo regressions of volailiy of equiy prices in four exchanges. 982

12 Lasly, for Tokyo, significan coefficiens were found a lags 1, 2 and 3 bu no 4. Hence, in all four markes, here ended o be significan auocorrelaion up o lag 3. Similar resuls were found also when we observe he resuls of sudying he individual hree sub periods noing ha he sample sizes for sub periods are smaller han for he enire ime period. Table-8 shows ha significan auo regressive coefficiens in equiy markes indicae an associaion wih is own lagged variables. I is a well-known ha observaions from previous sudies including CL and JS resuled in a similar analysis and conclusion. As before, we es for srucural change in each equiy marke by dividing our ime period ino hree sub periods. Columns 3, 4, 5, 6, 7, and 8 of Table-8, Panels A and B. Alhough he resuls are universally similar, for he mos par, he auo regressions for all four equiy markes exhibi oucomes ha are very similar o each oher. Tess for equaliy among he hree sub periods for each marke would show he same resuls. In conclusion, all four markes sugges ha each marke has some parameer sabiliy during he lenghy period sudied and he effec of changes in ime as expressed by he hree sub periods indicae ha his is rue for each marke bu wih some dispariies. A his poin, we inroduce lagged values of he oher markes o ascerain wheher he volailiy in he former marke indicaes Granger causaliy (Granger, 1969). To deermine Granger causaliy in Hang Seng volailiy, we choose he number of lagged values of New York, London and Tokyo volailiy according o he crieria noed before (e.g. AIC and he absence of serial correlaion in he residuals). Our resuls for he whole sample period (All Daa), repored in column 2 of Table 9, we have only lag 1 for New York, lag 1 for London, and lag 1 for Tokyo. The -saisic for New York is , and herefore is significan a a very small probabiliy, (significance level of α 0.01). This indicaes ha he volailiy in he Hang Seng equiy marke is associaed wih he lagged value in New York. Since he -saisics are no significan for London and Tokyo, we canno draw he same conclusion. CL and he JS comparisons were no for Hang Seng bu for Shanghai, hus hese resuls differ. Hence, we conclude ha only he New York volailiy a one lag was associaed wih Hang Seng for he enire ime period. Thus, Granger causaliy exiss beween Hang Seng and New York bu no for he ohers. Hence, his indicaes ha he volailiy in he markes for he enire ime period were likely independen of each oher. Unlike CL, we did no observe negaive coefficiens for N1 and N2 in he enire ime period sudied. To be consisen wih he findings of CL, we observed only he H 1 (lag of 1) and his ime he coefficien is no significan (a α = 0.05 or less). The AIC value suggess no including any lagged values of he Hang Seng variables. In addiion, he Breusch-Godfrey es revealed he absence of serial correlaion in he auoregressive model. The model wih one Shanghai lagged variable (S1) is conained in column 7 of Table-9. The negaive coefficien corroboraes he resuls of CL bu in his sudy his coefficien is no significan. Hence, by Granger causaliy, he Shanghai volailiy and New York volailiy do no have a Granger cause relaionship. Lasly, when we compare he Roo MSE of he Shanghai and New York models, we no ha he small residual variaion in he 983

13 New York regression and hence, i is more predicable han Shanghai volailiy. An addiional quesion relaes o wheher or no here is significan co-variaion of volailiy in a mulivariae seing. To incorporae insananeous causaliy in explaining Hang Seng volailiy, one adds he curren value of he variable in he oher markes in he auo regression. One observes he resul for Hang Seng in column 2 of Table-9, Panel A and he resuls for New York in column 6 of Table-9, Panel A. The coefficiens for he New York variables (All Daa) show some posiive coefficiens bu only N1 is significan. This would indicae ha exended ime period in his sudy resuled in some Hang Seng volailiy being relaed o New York volailiy in a lagged ime period. A wholly differen inerpreaion resuls when we observe he -saisics for N1 in he sub periods. The coefficien is significan only before The laer wo ime periods yield no significan coefficien. If we go line by line and column by column hrough all he daa in Table-9, we do no find consisency from period o period in pairwise combinaions. Thus, we could conclude ha he relaionships among he markes change during he sub periods indicaing he dynamic aspecs of he four equiy markes sudied. Volailiy is presen and changes he relaionships of markes due o economic condiions, law affecing hese markes he growh of emerging markes versus more esablished markes. There is lile doub ha marke volailiy is ever changing and he predicion of his volailiy is no easily accomplished. Table-9, Panel B, conains he analysis for London and Tokyo. Noe, for London and Hang Seng only H 0 and H 1 have significan -saisics for all daa and pre The coefficiens hen do no have significan -saisics for he 97 o 07 and he pos 07 sub periods. The same is rue for he relaionship of Tokyo and Hang Seng. Tokyo, alhough a Pan-Pacific exchange is no an emerging marke and ofen reacs more like he Wesern markes sudied here han he Hang Seng marke. Wihou going hrough he analysis o compare individual coefficiens, we observe he differen effecs of change in ime and he pairwise relaionship of markes. As long as economic condiions change, he resuls include emporal insabiliies in markes. Our sudy is lenghy and exhausive bu much of is resuls are no unnerving since we already know ha markes vary in prices and volailiy, bu hese facors have componens ha are predicable when using modern ime series analysis. For example, see Ray, Chen and Jarre (1997) where he auhors show ha firms lised on he Tokyo exchange conain componens (permanen and emporary) which may in urn, lead o beer predicions. Lasly, he resuls for he models for he volailiy in equiy reurns for all he equiy markes, we find he effec of he Asian equiies leading o he same for of emporal insabiliy of he parameers of he Wesern equiy markes. Simply saed he inclusion of he Hang Seng or Tokyo variables do no resul in sable relaionships hroughou he hree sub periods sudied. There are observed srucural changes relaed o each ime period. Hence, we conclude ha he concep of emporal sabiliy is no presen. 984

14 Table-9. Regressions of Volailiy of Equiy Reurns (Panel-A) Hang Seng (Hong Kong) New York Lag all daa pre pos all daa pre pos consan H H H H H N N N N N L L L2 L3 L4 T T R-squared S.E. of reg Table-9. Regressions of Volailiy of Equiy Reurns Panel B. London and Tokyo London Tokyo Lag all daa pre pos all daa pre pos consan H

15 H H2 H3 H4 N N N2 N3 N4 L L L L L T T T T T R-squared S.E. of reg Table-9. Regressions of Volailiy of Equiy Reurns (Panel-B) London Tokyo Lag all daa pre pos all daa pre pos consan H H H2 H3 986

16 H4 N N N2 N3 N4 L L L L L T T T T T R-squared S.E. of reg CONCLUDING REMARKS We colleced, analyzed and inerpreed an exensive daabase of he sock marke indices for equiy markes of New York, London, Tokyo and Hong Kong (Hang Seng). Our purpose is o draw conclusions concerning he relaionships of he various equiy markes expressed by an analysis of he mean and volailiy of raes of reurn in he four sock exchanges over a lenghy period of ime and during hree disinc sub periods. We firs examined he ime series characerisics of sock price indices for all four exchanges during he period of 1987 o Specifically, we calculaed he rae of reurn and he volailiy of reurns for he four markes and esimae he serial correlaion and co-movemen of he equiy markes. We find ha he mean raes of reurn in vary for he four equiy markes and noing ha Hang Seng and Tokyo have differences in heir raes of reurn. Volailiy in he raes of reurn also differs among he four equiy exchanges. Across he hree sub periods defined by ime he relaionship are no sable. This, perhaps, is he mos crucial of he general findings of his analysis. Relaionships among he markes across ime periods change. 987

17 Invesigaions ino he influences of he economic environmen in which he markes operae would indicae wha some of he causes are or a leas associaions wih changes in mean and variabiliy of raes of reurn and volailiy in he raes of reurn would add o ones knowledge of explaining and predicing relaionship among he four marke specific exchanges. Furhermore, we find ha serial correlaion also differs in he four equiy markes. The use of mulivariae ime series analysis (see Kuvia, 2010; Chen, Minary and Lai, 2005; and Juselius, 2006) may provide furher evidence for he lack of co-inegraion in hese sock exchanges. One las suggesion is o examine each individual index o access where here are emporary and permanen componens in hese indices in he manner of Ray e al. (1997) for he Japanese marke. This would answer quesions concerning emporal sabiliy in he indices of Chinese equiy markes. In he fuure, we expec sudies o coninue in areas of he so-called emerging markes of Asia (and Souh America as well) o relae in a similar conex o show correlaions among emerging markes, i.e. India, China and so forh, and he developed naions of Europe, Norh America and Japan. These sudies may also use highly regarded mulivariae ime series analysis including Transformaion and Inervenion analyses. REFERENCES Bailey, W., Cai, J., Cheung and F. Wang, (2009) Sock reurns, order imbalances and commonaliy: Evidence on individual, insiuional and proprieary invesors in China, Journal of Finance, Vol. 45, pp Bekaer, Geer and Harvey, Campbell R. (1997) Emerging equiy marke volailiy, Journal of Financial Economics, Vol. 43, pp Cheung, Y-W.and L. Ng., (1998) Inernaional evidence on he sock marke and aggregae economic aciviy. Journal of Empirical Finance, Vol. 5, pp Chen, N., (1991) Financial invesmen opporuniies and he macro economy, The Journal of Finance, Vol. 46, No. 2, pp Chen, L., Finney, M. and K.S. Lai, (2005) A Threshold Coinegraion of Asymmeric Price Transmission from Crude Oil o Gasoline Prices, Economerica Leers, Vol. 89, pp Chow, G. C., Fan, Z. and J. Hu, (1999) Shanghai sock prices as deermined by he presen value model, Journal of Comparaive Economics, Vol. 27, pp Chow, G. C. and C. C. Lawler, (2003) A ime series analysis of he Shanghai and New York sock price indices, Annals of Economics and Finance, Vol. 4, pp Eun, C.S. and W. Huang, () Asse Pricing in China s Domesic Sock Markes: Is here Logic, Pacific-Basin Finance Journal, Vol. 15, pp Fama, E., (1990) Sock reurns, expeced reurns, and real aciviy, Journal of Finance, Vol. 45, pp Fama, E., (1991) Efficien capial markes: II. Journal of Finance, Vol. 46, pp

18 Granger, C.W.J., (1969) Invesigaing causal relaion by economeric and cross-secional mehod, Economerica, Vol. 37, pp Jarre, J.E., Pan, X and S. Chen, (2009) Do he Chinese bourses (sock markes) predic economic growh? Inernaional Journal of Business and Economics, Vol. 8, pp Jarre, J.E. and J. Schilling, (2008) Daily Variaion and Predicing Sock Marke Reurns For The Frankfurer Börse (Sock Marke), Journal of Business Economics and Managemen, Vol. 9, pp Jarre, J.E. and Sun, Z., (2009A) Evidence and Explanaions for he Associaion among Six Asian (Pacific-Basin) Financial Markes, Applied Economics, Vol. 41, pp. 25, Jarre, J.E. and Sun, Z., (2009B) Explaining Pacific-Basin Financial Marke Reurns by Size of Firm, Inernaional Review of Applied Financial Issues and Economics, Vol. 1, pp Jarre, J.E. and E. Kyper, (2010) Seasonaliy in Ouliers of Sock Reurns for he Chinese Equiies Markes, Inernaional Review of Applied Financial Issues and Economics, Vol. 2, pp Jarre, J.E. and T. Sun, (2012) Associaion beween he New York and Shanghai Markes: Evidence from he Sock Price Indices, Journal of Business Economics and Managemen, Vol.13, No. 1, pp Juselius, K., (2006) The Coinegraed VAR Model: Mehodology and Applicaions, Oxford Universiy Press. Kuvia, T., (2010) Time Series Analysis of Transalanic Marke Ineracions: Evidence from Crude Oil and Gasoline Prices, Inernaional Journal of Business and Economics, Vol. 9, pp Levine, R. and S. Zervos (1998) Sock Markes, Banks, and Economic Growh, The American Economic Review, Vol. 88, No. 3, pp Levine, R., (2005) Finance and Growh: Theory and Evidence in Handbook of Economic Growh, Volume 1A, Philippe Aghion and Seven N. Durlauf, Eds., Elsevier, B.V., pp Liaw, K. T., () Invesmen Banking and Invesmen Opporuniies in China: A Comprehensive Guide for Finance Professionals John Wiley & Co.: New York Pan, X. Li, K. and J.E.Jarre (2012) The relaionship of high frequency ineracions beween Chines A-shares and Hong Kong H-shares of dual-lised companies, Journal of Research in Economics and Inernaional Finance (JREIF)(ISSN: ), Vol. 1, No. 4, pp Ray, B., Chen S., and J.E. Jarre (1997) Idenifying permanen and emporary componens in Japanese sock prices, Financial Engineering and he Japanese Markes, Vol. 4, No. 3, pp Wei, K. and K. Wong, (1992) Tes of inflaion and indusry porfolio sock reurns. Journal of Economics & Business, Vol. 44, pp Zhong, R.S., L. Gu and C.B. Lui, (1999) The Empirical Saisical Analysis of Chinese Sock Markes, China Financial and Poliical Economics Press, Beijing. 989