Volatility spillovers between stock prices and exchange rates: empiral evidence from six APEC economies

Transcription

1 Dublin Insiue of Technology Conference papers School of Accouning and Finance Volailiy spillovers beween sock prices and exchange raes: empiral evidence from six APEC economies Lucia Morales Dublin Insiue of Technology, Mary O'Donnell Follow his and addiional works a: hps://arrow.di.ie/buschaccon Par of he Finance and Financial Managemen Commons Recommended Ciaion Morales, L., O'Donnell, M.: Volailiy spillovers beween sock prices and exchange raes: empiral evidence from six APEC economies. All China Economics Conference, Hong Kong, 18h.-0h. December, 006. This Conference Paper is brough o you for free and open access by he School of Accouning and Finance a ARROW@DIT. I has been acceped for inclusion in Conference papers by an auhorized adminisraor of ARROW@DIT. For more informaion, please conac yvonne.desmond@di.ie, arrow.admin@di.ie, brian.widdis@di.ie. This work is licensed under a Creaive Commons Aribuion- Noncommercial-Share Alike 3.0 License

2 VOLATILITY SPILLOVERS BETWEEN STOCK PRICES AND EXCHANGE RATES: EMPIRICAL EVIDENCE FROM APEC ECONOMIES Lucia Morales and Mary O Donnell Deparmen of Accouning and Finance Dublin Insiue of Technology Dublin Keywords: Sock reurns, Exchange rae changes, Volailiy spillovers, Bivariae EGARCH Topic Caegories: ACE Themes 3, 8, JEL Codes F, G Absrac This paper se ou o examine he volailiy linkages beween sock reurns and exchange raes in a number of Eas Asian markes. Overall, our main resuls indicae ha since he Asian financial crises, here exiss significan scope for invesors and porfolio managers o diversify heir asses beween socks and currencies in hese markes. In paricular, he lack of volailiy spillovers beween sock markes and exchange raes, and beween exchange raes and sock markes in all counries, excep Taiwan in he pos crises period indicaes ha here is scope for invesors o diversify heir invesmens and o benefi from poenial gains in he long run in his region. 1

3 1. Inroducion Several heoreical models have demonsraed heoreically he exchange rae beween wo currencies is affeced by sock price changes in he respecive counries; for example, Zapaero (1995 shows ha in fully inegraed financial markes, here is an explici linkage beween he volailiy of sock prices and he volailiy of he exchange rae. More recenly, Yang and Doong (00 noe ha given he rapid inegraion and deregulaion of inernaional financial markes in recen years, exchange raes have become more sensiive o sock marke innovaions. Exising empirical evidence on volailiy spillovers beween sock markes and exchange raes have ended o focus on he G-7 counries (see for example, Yang and Doong, 004; Kanas, 000, 00. Empirical evidence o dae ha has examined he ineracion beween sock markes and exchange raes for he Asian markes has focused on he effec of currency depreciaion on sock marke reurns and volailiy (Fang, 00. The only sudy ha has examined he relaionship beween he volailiy of sock prices and he volailiy of exchange raes in a number of Eas Asian markes is Wu (005 who examines volailiy spillovers beween he wo markes for he period of he Asian financial crises and he period afer he crises; however his sample only runs unil he end of 000. Thus here is no up o dae recen evidence on volailiy spillovers beween sock markes and exchange raes for Asian counries. We address he gap in he lieraure in his area by invesigaing his issue using daily daa for he period 1997 and for Hong Kong, Souh Korea, Singapore, Taiwan and Thailand. The layou of he paper is as follows. Secion 1 ses ou he heoreical and empirical evidence on he naure of linkages beween sock markes and exchange raes in he nex secion. Secion describes of he mehodology we use o assess he naure of volailiy spillovers beween he wo markes, and we discuss our daa. Secion 3 ses ou he resuls from our esimaion. We finish by summarising our main resuls and drawing some conclusions from our analysis. 1 Our sample runs from 1/1/1997 o 31/7/006 and he daa are sourced from Daasream.

4 . Lieraure Review Several heoreical models have analysed he link beween sock markes and currency markes. The asse marke approach o exchange rae deerminaion (Branson, 1983; Frankel, 1983 posis ha causaliy will run from sock prices o exchange rae changes as expecaions of financial asse price movemens affec he dynamics of exchange raes. Smih (199 derives an esimable equaion for he exchange rae where he sock price is included as an explanaory variable. The goods marke approach suggess causaliy runs in he opposie direcion, from exchange raes o sock prices (Mundell, 1963, 1964; Dornbusch and Fisher, In hese models, movemens in exchange raes affec he inernaional compeiiveness of firms which affecs real income and oupu and evenually sock prices. In Hekman s (1984 model he exchange rae is an explanaory variable for sock prices. Much of he available empirical evidence on he linkages beween sock markes and exchange raes has concenraed on he firs momens. Yang and Doong (004 noe ha here is a dearh of empirical evidence ha concenraes on he linkages beween he second momens of he disribuion of he variables. A number of sudies however have examined he exen o which volailiy from one sock marke spills over ino oher sock markes or beween differen asses 3. Kanas (000 was one of he firs sudies which analysed volailiy spillovers from sock reurns o exchange rae changes in he USA, he UK, Japan, Germany, France and Canada. He found evidence of spillovers from sock reurns o exchange rae changes for all counries excep Germany, suggesing ha he asse approach o exchange rae deerminaion is valid when formulaed in erms of he second momens of he exchange rae disribuion for he counries included in his analysis. Volailiy spillovers from exchange rae changes o sock reurns were insignifican for all counries. Yang and Doong (004 explored he naure of he mean and volailiy ransmission mechanism beween sock and foreign exchange markes for he G-7 counries. The resuls poin o significan volailiy spillovers and an asymmeric effec from he sock marke o he foreign exchange marke for France, Ialy, Japan and he US, suggesing inegraion beween sock and foreign exchange markes in hese counries. Verma, Jackson and Swisher (005, examined price and See for example Nieh and Lee (001, Yau and Nieh (006 for recen evidence on his opic. 3 See also for example, Nelson (1991, Koumos and Booh (1995, Laopodis (

5 volailiy spillovers from ineres raes and exchange raes o American Deposiory Receips (ADRs, originaing from Mexico, Brazil and Chile. In erms of volailiy spillovers, heir resuls indicaed ha boh ineres raes and exchange raes spillover o Brazilian and Chilean ADRs, whereas only exchange raes spillover o Mexican ADRs. In relaion o asymmery, he ineres raes of Mexico, Brazil and Chile as well as he exchange raes of Chile, indicae ha negaive innovaions increase volailiy more han posiive innovaions. The only exising evidence on his issue for Asian counries is Wu (005 who examines volailiy spillovers beween sock prices and exchange raes for Japan, Souh Korea, Indonesia, Philippines, Singapore, Thailand and Taiwan for he period , spliing he sample ino crises and recovery periods. He found a bi-direcional relaionship beween he volailiy of sock reurns and exchange rae changes during he recovery period in all counries excep Souh Korea, as well as significan conemporaneous relaionships beween he wo markes for mos of he counries. Furhermore, he found volailiy spillovers increased in he recovery period. 3. Mehodology and Daa The analysis will be conduced wih he purpose of invesigaing volailiy spillovers beween sock reurns and exchange rae changes for five Asian Markes, Hong Kong, Souh Korea, Singapore, Taiwan and Thailand, for he period 1 January 1997 o 7 July 006. The daa se consiss of daily closing values for he Hang Seng, Srai Times, Korea SE Composie, Taiwan Se Weighed and Thailand SE TISCO sock marke indices, and he Hong Kong$, Singapore$, Souh Korea Won, Taiwan New and Thai Bha foreign exchange raes agains he US$. Our sample has a oal of 485 observaions. All daa are aken from DaaSream. Following Kanas (000 we use coninuously compounded sock reurns and exchange rae changes calculaed as he firs differences of he naural log. Tha is, S= Sock s s e e Prices; = ln ( P ln ( P and E= Exchange Raes; ln ( P ln ( P S 1 E 1 =. As an iniial sep we perform a saionariy es on each of he relevan variables ha are included in our analysis o ensure ha he resuls from he analysis are no spurious. We apply he Dickey Fuller (DF es, or Augmened Dickey-Fuller es (ADF procedure if serial correlaion is presen. We also apply he Lagrange Muliplier (LMF es, o ensure ha a 4

6 sufficien number of lags have been added in he ADF es o ensure ha here is no serial correlaion presen and he resuls of he ADF es are valid. The LMF es is applied given ha i is valid in he presence of lagged dependen variables as well as having he advanage of esing for firs and higher orders of serial correlaion. If we found ha our variables are an I(0 process, meaning ha hey are saionary in levels, we will proceed o perform our EGARCH (p,q analysis, oherwise if we found ha our variables are no an I(0 process, wha i means ha hey are saionary in levels hen we will need o proceed and perform uni roo es in our variables, applying o hem firs differences, if we found ha our variables are an I(d process, i will means ha hey have o be inegraed a he same order hen we will be able o proceed wih he coinegraion es on our variables as is saed in he coinegraion es mehodology (Enders, 004. Using he Johansen Coinegraion es o invesigae he long-run relaionship beween Sock Prices and Exchange Raes, as Enders (004 noes given ha he resuls of he es can be quie sensiive o he lag lengh, he mos common procedure is o esimae a Vecor Auoregression (VAR model on he undifferenced daa in order o deermine he lag lengh for he Johansen es. We esimae he lag selecion ess up 0 lags. In erms of choosing beween he various lag lengh selecion crieria we follow Johansen e al. (000 who sugges ha when differen informaion crieria sugges differen lag lenghs, i is common pracice o prefer Hannan-Quinn (HQ crieria. Again, we ensure ha he lag lengh seleced for he VAR model is free from serial afer performing by applying he LMF es o es for serial correlaion up o he number of lags in he VAR model. There are five possible models o choose from for he Johansen es as follows. H (r : Π y -1 + B x = α B y -1 (1 H* 1 (r : Π y -1 + B x = α (B y -1 + p 0 ( H 1 (r : Π y -1 + B x = α (B y -1 + p 0 + α γ 0 (3 H* (r : Π y -1 + B x = α (B y -1 + p 0 + p 1 + α γ 0 (4 H (r : Π y -1 + B x = α (B y -1 + p 0 + p 1 + α (γ 0 +γ 1 (5 Equaion 1 has no deerminisic rends in he level daa and no inerceps in he coinegraing equaions. Equaion has no deerminisic rends in he level daa and he coinegraing equaions have inerceps. Equaion 3 has linear rends in he level daa bu he coinegraing 5

7 equaions only have inerceps. Equaion 4 has linear rends in boh he level daa and he coinegraing equaions, and equaion 5 has quadraic rends in he level daa and linear rends in he coinegraing equaions. Harris and Sollis (003 noe ha model 1 i.e. wih no deerminisic componens in he daa or coinegraion relaions, is unlikely o occur in pracice, as generally an inercep is needed o ake accoun of he unis of measuremen of he variables; hey also noe ha model 5 wih quadraic rends, is economically hard o jusify, as if he variables are enered in logs, as hey are in our model, as his would imply an every increasing or decreasing rae of change. This leaves a choice beween models -4. Johansen (199 suggess choosing he appropriae model according o he Panula principle; all hree models are esimaed; he Panula principle involves moving hrough each model for he null hypohesis of r=0, hen r=1 ec., and picking he model where he null hypohesis is rejeced for he firs ime. Chang and Caudill (005 noe ha he λ race es saisic is more robusness o boh skewness and excess kurosis han he λ max es saisic; for comparaive purposes, we show boh he resuls of he λ race and he λ max es saisics. We hen proceed wih our volailiy analysis and apply a bivariae exension of he EGARCH (p,q model in order o examine wheher he volailiy of sock reurns affecs and is affeced by he volailiy of exchange rae changes wihin each economy. The EGARCH specificaion (Nelson, 1991 is used in order o es wheher he volailiy spillover effecs are asymmeric. For example, an asymmeric spillover from sock reurns o exchange rae changes would sugges ha he effec of bad sock marke news on he exchange rae change is greaer han he effec of good news. The model is specified as follows: S E = a = a r r s, 0 + as, is i + ae, i E i + sλs, 1 + i= 1 i= 1 r r 0 + a i E i + as, is i + Eλ 1 + i= 1 i= 1 β e (6 S, β e (7 e e / N(0, σ S, Ω 1 S, / Ω 1 N(0, σ 6

8 The condiional variances of sock reurns and exchange raes changes are specified as follows: σ (8 σ ps = exp cs,0 + b j= 1 log( σ + δ S, S, j S, j S, S S, 1 S, 1 S, Szs, T 1 ps = exp c 0 + b j= 1 log( σ + δ j j E 1 1 Ezs, T 1 (9 σ = S, T ρ S, Eσ S, σ [( z E z + θ + δ [( z E z + θ ] S, E 1 1 S, Ez 1 [( z E z + θ + δ [( z E z + θ ] S S, 1 S, 1 Sz 1 We summarise each of he relevan erms in equaions (6-9 in Table 1. Error correcion erms (lagged residuals from he coinegraing regression of Sochasic error erms Informaion se a ime -1 Table 1 Descripion of Parameers Equaions (6-(9 Sock Reurns S, E, 1 Condiional (ime varying variances σ S, Sandardised residuals assumed o be normally disribued wih 0 mean and variances of σ, σ S, ARCH effec where he parameers, = /σ z S es, S, Exchange Rae Reurns λ S λ 1 e S, e E, Ω 1 Ω 1 e S, / Ω 1 ~ N(0, σ S, σ z = e /σ e E, / Ω 1 ~ N(0, σ ps pe Persisence of Volailiy b S, j b E, j j= 1 j= 1 θ S, S,θ E [ z ] allow his effec o be asymmeric S, E zs, + θ S, Sz S z, E z + θe, Ez E, [ ] Volailiy Spillover δ [ z E z + θ ] δ [ z E z + θ ] Measures of spillovers Asymmery of Spillovers 4 Correlaion Coefficien for Sandardised Residuals S, E 1 1 S, Ez E, 1 S S, 1 S, 1 SzSS 1 δ S,E δ S θ S,E θ S ρ S, E ρ S 4 θ S,E <0, θ S, E posiive shocks <0, implies ha negaive exchange rae shocks increase he volailiy of sock reurns more han 7

9 The lag runcaion lengh p in he EGARCH model is deermined using he Likelihood Raio (LR es on alernaive specificaions. Hamilon (1994 defines he LR es as follows: L( ˆ θ L( ~ θ χ ( m, where L(θˆ denoes he value of he log likelihood funcion a he [ ] ~ unresriced esimae and L(θ denoes he value of he log likelihood funcion of he resriced esimae. Bollerslev-Woolridge robus -saisics are derived o ake ino accoun possible non-normaliy of he residuals. Given ha our sample period includes he Asian financial crises, in addiion o examining volailiy spillovers beween sock reurns and exchange raes for he enire period, we also spli our sample in order o compare he effec of volailiy spillovers during and afer he crises. Wu (005 noes ha he financial crises were riggered by Thailand s reques for assisance from he IMF he nd of July 1997 and ha mos counries had recovered from he crises by lae Thus our spli samples comprise he crises period of July 1997 o31 December 1998 and he pos crises period of 1 January July All resuls are generaed using he EVIEWS saisical program. 4. Empirical Resuls We begin by providing descripive saisics for sock reurns and exchange raes, in order o summarise he saisical characerisics of our sample which are se ou in Table 3. For he enire period, he sample means of sock reurns are posiive for all counries excep Taiwan as well as being posiive for Hong Kong and Souh Korea during he crises period and negaive for he oher counries in he crises period. In he pos crises period, he mean of all sock reurns were posiive. The highes mean reurns were for Souh Korea a.78% followed by Thailand a 1% for he enire sample; for he crises period hey were highes for Hong Kong and Korea for he pos crises period hey were highes for Singapore, followed by Korea and Hong Kong, wih Taiwan having he lowes mean reurns in he pos crises period. 5 Wu (005 examines he exen of volailiy spillover before and afer he crises bu our resuls differ from his in ha he defines he crises period from July 1997 o 30 Sepember 1998 and his sample for he pos crises period runs from 1 Ocober 1998 o 31 December 000; hus our pos crises sample is considerably longer and more up o dae. 8

10 The sandard deviaions of he sock reurns range from 0.7% o.7% for he enire period and from 0.5% o 3.% in he crises period and 0.03% o 1.9% in he pos crises period, indicaing ha he volailiy of sock reurns in general were lower in he pos crises period han during he financial crises. Boh he skewness and kurosis coefficiens indicae ha sock reurns are lepokuric relaive o he normal disribuion, which Caporale e al. (00 noe is a common finding for sock reurns. The Jarque-Bera es also rejecs he hypohesis ha sock reurns are normally disribued in all counries. Sock Reurns Table 3 Descripive Saisics Mean (x10 3 SD Skewness Kurosis JB Hong Kong Toal sample Crises Pos-crises Souh Korea Toal sample Crises Pos-crises Singapore Toal sample Crises Pos-crises Taiwan Toal sample Crises Pos-crises Thailand Toal sample Crises Pos-crises Exchange Raes Mean (x10 3 SD Skewness Kurosis JB Hong Kong Toal sample Crises Pos-crises Souh Korea Toal sample Crises Pos-crises Singapore Toal sample Crises Pos-crises

11 Taiwan Toal sample Crises Pos-crises Thailand Toal sample Crises Pos-crises The descripive saisics for he exchange rae reurns show ha he sample means are posiive for all counries for he enire period excep for Taiwan and Korea; for he crises period he means were negaive only for Taiwan and Thailand, and for he pos crises period were negaive only for Korea and Singapore. The volailiy of exchange rae reurns ranged from 0.04% in Hong Kong o 1.06% in Souh Korea for he enire period; volailiy was higher during he crises period han he pos crises period for all counries. Again he skewness and kurosis saisics indicae ha he disribuion of exchange rae reurns are nonnormal and he Jarque-Bera es also rejecs he hypohesis of normally disribued reurns for all periods for all counries. Table Augmened Dickey Fuller Tes Resuls Variables Toal Sample Crises Pos crises Hong Kong E S Souh Korea E S Singapore E S Taiwan E S Thailand E % criical values S The resuls from he ADF ess are given in Table. The values of he es saisics indicae ha we can rejec he null hypohesis of he exisence of uni roo in levels for all variables in all periods indicaing ha all series are I(0. 6 Given ha all variables are inegraed of he same order, and also given ha hey are an I(0 process is no need o process 6 The LMF es resuls indicaed ha he ADF ess were free from serial correlaion; for breviy we do no show he es resuls here. 10

12 we he Coinegraion ess, we will proceed direcly o perform our volailiy analysis using EGARCH (p,q modelling. Table 3 Likelihood Raio Tes for EGARCH Model Selecion for Condiional Variance Equaions Sock Reurns Exchange Raes Counry Toal Sample Crises Pos Crises Toal Sample Crises Pos Crises Hong Kong 1.6* 1.0* 17.3* 895.8* 7.4* 3.7 Souh Korea * * Singapore * 18.9* Taiwan 8.0* * 84.* * Thailand 159.1* *.1* * Noe: H 0 : EGARCH (1,1, H 1 : EGARCH(,1 The 5% criical value for he LR es disribued as χ wih degrees of freedom is * indicaes rejecion of he null hypohesis a 5% significance. The resuls in Table 3 indicae ha for he enire period we selec he EGARCH (,1 in he case Hong Kong, Taiwan and Thailand for sock prices, and for Hong Kong, Souh Korea, Taiwan and Thailand in he case of exchange rae changes. The EGARCH (1,1 model is seleced for Souh Korea and Singapore for sock prices and for Singapore for exchange rae changes. For he crises period, we selec he EGARCH (,1 model for sock prices for Hong Kong and Singapore and he EGARCH (1,1 for Singapore, Taiwan and Thailand. For exchange raes for he crises period, we selec he EGARCH (,1 model for Korea and Thailand, and he EGARCH (1,1 for Hong Kong, Singapore and Taiwan. For he pos crises period, he EGARCH (,1 is seleced for sock prices for Hong Kong and Taiwan, and for exchange raes for Hong Kong, Singapore, Taiwan and Thailand; he EGARCH (1,1 is seleced for sock prices for he pos crises period for Korea, Singapore and Thailand, and for exchange raes for Korea. The esimaed parameers from he EGARCH esimaion are se ou in Tables 7-9, for he oal sample, he crises period and he pos crises period respecively. Firsly, in relaion o he coefficiens on he volailiy persisence erm, he resuls indicae ha here is significan persisence in sock reurns volailiy for all counries for he enire period and he pos crises period, wih he excepion of Hong Kong. For he crises period, he persisence of volailiy is significan for all counries excep Thailand. For he exchange rae equaion, for 11

13 he oal sample he coefficiens are significan for all counries; for he crises period volailiy persisence is significan for all counries excep Thailand and for he pos crises period for all counries excep Taiwan. Wu (005 noes ha a necessary condiion for he volailiy persisence erms o be sable is ha he value of he esimaed coefficiens should be less han one; for our resuls, his applies in all cases excep for Hong Kong for he oal sample, alhough we noe ha he coefficiens are less han one in each of he sub-samples. In erms of he coefficiens for he volailiy spillover effecs from sock reurns o exchange rae changes, we find ha here are some significan differences beween he resuls for he hree ime periods. All coefficiens are significan for he enire sample. For he crises period, coefficiens are significan for Korea, Singapore and Thailand while for he pos crises period, volailiy spillovers from sock markes o exchange raes were only significan for Hong Kong. The significan coefficiens indicae ha he volailiy of sock reurns was a deerminan of he volailiy of he exchange rae as well as indicaive of inegraion beween sock markes and exchange rae markes. Furhermore, where significan, he resuls indicae ha volailiy informaion conained in sock prices impaced on he behaviour of exchange raes in hese markes. The lack of significan spillovers in he pos crises period in all markes a 1% level indicaes ha here is poenial for diversificaion beween sock markes and currency markes in hese counries. In erms of volailiy spillovers from exchange raes o sock markes, we found ha again for he enire period, he esimaed coefficiens were significan for all counries; however for he crises period, coefficiens for Hong Kong, Singapore and Taiwan were significan, while in he pos crises period, significan volailiy spillovers from exchange raes o sock markes were only found for Taiwan. The lack of significan spillovers from exchange rae changes o sock reurns found here for some counries is consisen wih resuls from Jorion (1990 as well as wih Yang and Doong (005. Jorion (1990 explained he lack of spillovers as possibly due o posiive exchange rae volailiy on sock reurns for some firms offseing negaive exchange rae volailiy on sock reurns for oher firms o give an insignifican or weak effec overall. In addiion o his, he use of insrumens o hedge exchange rae risk, may reduce he impac of exchange rae volailiy on sock markes; Gran and Marshall, 1997, and Bodnar e al. (1995 boh noe ha he use of hedging insrumens o ameliorae exchange rae risk is pervasive amongs larger companies which are he main 1

14 componens of naional sock marke indices. The lack of significan spillovers from exchange raes o sock markes in he pos crises period may hus be indicaive of wider use of hedging by firms lised on he sock markes in hese counries han during he crises; i is also likely ha he crises iself may have conribued o more exensive use of hedging agains foreign exchange rae risk. A posiive sign on he spillover coefficien indicaes ha an increase in volailiy in one marke is associaed wih increased volailiy in he oher marke while a negaive coefficien indicaes ha increase volailiy in one marke is associaed wih decreased volailiy in he oher marke. In our resuls, we found a mixure of posiive and negaive significan coefficiens for he wo spillover erms; his indicaes ha he impac of volailiy from exchange rae markes o sock markes or vice versa is no consan eiher over ime for he same counry, or across counries for he same period of ime. Differen significan resuls for he enire sample and subsamples were also found for some Asian markes by Caporale e al (00 when esing for causal relaionships beween sock prices and exchange rae volailiy. For he asymmeric spillover effecs from sock reurns o exchange raes, we find significan coefficiens for all counries for all ime periods, wih he excepion of Hong Kong during he crises period where he coefficien was insignifican. Similarly, he asymmeric spillover effecs from exchange raes o sock prices are significan for all counries for all ime periods, wih he excepion of Thailand during he crises period. The insignifican coefficiens indicae ha he spillover effecs in hese insances are symmeric, ha is ha posiive and negaive shocks have he same impac on volailiy, or ha a decrease in sock reurns has he same impac on exchange rae volailiy as an increase in sock reurns. The posiive sign on all significan coefficiens indicaes ha unexpeced good news has a greaer impac on volailiy han unexpeced bad news. One possible explanaion for his is ha good news on sock prices may have a greaer impac on demand for currency so increasing volailiy as foreign invesors wan o increase holdings of rising socks; good news on exchange raes may have a greaer impac on demand for socks as invesors swich beween holdings of socks and currency, so impacing on sock marke volailiy. The correlaion coefficiens for he sandardised residuals from he sock and exchange rae equaions are significan for Korea, Taiwan and Singapore for he enire period and he crises period 13

15 indicaing significan conemporaneous relaionships beween he wo markes in hese counries for hese periods. The only significan correlaion for he pos crises period is for Thailand. The diagnosic ess on he sandardised residuals are lised in par B of he respecive ables. The Jarque-Bera es indicaes ha we rejec he hypohesis ha hey are normally disribued, hence jusifying he use of he Bollerslev-Woolridge robus -saisics. The Ljung-Box saisics for all hree periods for all counries indicae ha here are no residual linear or non linear dependencies; he one excepion o his is for linear dependency in he exchange rae equaion for Hong Kong for he oal sample, alhough for he crises and pos crises periods separaely, he linear dependencies are absen 7. Table 7a Volailiy Spillovers Beween Sock Reurns and Exchange Rae Changes: Toal Sample Hong Kong Souh Korea Singapore Taiwan Thailand Esimaed Parameers Volailiy Persisence (Sock Reurns ( b S Spillover: from Sock Reurns o Exchange Raes ( δ S, E Asymmeric Spillover effec:from Sock Reurns o Exchange Raes ( θ S, E Volailiy Persisence (Exchange Raes ( b E Spillover: from Exchange Raes o Sock Reurns ( δ S, E Asymmeric Spillover effec:from: Exchange Raes o Sock Reurns ( θ S, E Correlaion Coefficien ( ρ S, E ( ( ( * * 0.999* Table 7b Diagnosics on Sandardised Residuals: Residuals: Toal Sample Hong Kong Souh Korea Singapore Taiwan Thailand Sock reurn equaion Jarque-Bera (0.45 (0.898 (0.86 (0.014 (0.178 LB(0 7 Kanas (000 found similar linear dependencies for he UK in he sock reurn equaion. 14

16 LB²(0 Exchange rae equaion LB²( ( ( ( ( (0.998 Jarque-Bera LB( (0.097 ( (0.018 ( (1.000 ( (1.000 (

17 Table 8a Volailiy Spillovers Beween Sock Reurns and Exchange Raes: Crises period Esimaed Parameers Hong Kong Souh Korea Singapore Taiwan Thailand Volailiy Persisence (Sock Reurns ( b S ( ( ( ( ( Spillover: from Sock Reurns o Exchange Raes ( δ S, E (0.139 (0.017 (0.400 (0.001 Asymmeric Spillover effec:from Sock Reurns o (0.671 Exchange Raes ( θ S, E Volailiy Persisence (Exchange Raes ( b E (0.004 (0.038 (0.511 Spillover: from Exchange Raes o Sock Reurns ( δ S, E (0.019 (0.817 (0.0 (0.005 (0.950 Asymmeric Spillover effec:from: Exchange Raes o (0.094 Sock Reurns ( θ S, E Correlaion Coefficien ( ρ * -0.11* 0.138* S, E Table 8b Diagnosics on Sandardised Residuals: Residuals Hong Kong Souh Korea Singapore Taiwan Thailand Sock reurn equaion Jarque-Bera LB( (0.038 (0.591 (0.669 (0.611 ( (0.586 (1.000 (0.641 (0.476 (0.958 LB²(0 Exchange rae equaion Jarque-Bera LB( (0.961 (0.775 (0.709 (0.318 ( (1.000 (0.949 (0.997 (0.913 (1.000 LB²(0 16

18 Table 9a Volailiy Spillovers Beween Sock Reurns and Exchange Raes: Pos crises period Hong Kong Souh Korea Singapore* Taiwan Thailand Esimaed Parameers Volailiy Persisence (Sock Reurns ( b S Spillover: from Sock Reurns o Exchange Raes ( δ S, E Asymmeric Spillover effec:from Sock Reurns o Exchange Raes ( θ S, E Volailiy Persisence (Exchange Raes ( b E Spillover: from Exchange Raes o Sock Reurns ( δ S, E Asymmeric Spillover effec:from: Exchange Raes o Sock Reurns ( θ S, E Correlaion Coefficien ( ρ ( ( ( ( ( ( ( ( ( ( ( S, E * *Hong Kong Exchange Raes Reurns follows a -disribuion, sandardised residuals, EGARCH(1,1 is he specificaion ha is working for his model. Thailand Exchange Raes follows a -disribuion, sandardised residuals, EGARCH (1,1 and lags (,3. *Singapore Sock Reurns sandardised residuals follows a -disribuion, EGARCH (1,1 Table 9b Diagnosics on Sandardised Residuals: Residuals Hong Kong Souh Korea Singapore Taiwan Thailand Sock reurn equaion Jarque-Bera LB( (0.511 (0.565 (0.903 (1.000 ( (0.319 (0.975 (0.781 (1.000 (1.000 LB²(0 Exchange rae equaion Jarque-Bera LB( (1.000 (0.34 (0.05 (0.583 ( (1.000 (0.977 (0.043 (0.609 (0.03 LB²(0 17

19 5. Summary and Conclusions This paper se ou o examine he volailiy linkages beween sock reurns and exchange raes in a number of Eas Asian markes. While here is a significan body of evidence which invesigaes he relaionship beween he firs momens of exchange raes and sock reurns, he evidence on volailiy linkages beween he wo markes is scarce and has generally been confined o invesigaion of he relaionship for developed counry markes. Thus our analysis make a clear conribuion in providing more up o dae informaion of he volailiy linkages beween sock prices and exchange raes which serves as a basis for increasing our undersanding of he naure of inegraion of sock and exchange rae markes in he counries we have examined. We examined he period , as well as spliing our sample o compare and conras he volailiy linkages beween he wo markes during he Asian financial crises, as well as afer he crises. Our resuls indicaed ha for mos markes, here exiss significan persisence in he volailiy of boh exchange raes and sock reurns in all periods. In addiion o his, we found significan volailiy spillovers beween sock reurns and exchange raes during he crises period for Korea, Singapore and Thailand. While Wu (005 invesigaes his relaionship for a number of Eas Asian counries, his sample ends in 000; our resuls are broadly consisen wih hose for he crises period even hough our crises sample period is slighly differen han Wu s. However, here are some difference in our resuls for he pos crises period; in paricular in conras o Wu we found no volailiy spillovers beween he sock and currency markes in Korea and Taiwan, and no volailiy spillovers from currency markes o sock markes in Singapore and Thailand. The reduced volailiy ransmission from currency markes o sock reurns in mos markes since he Asian financial crises may be indicaive of increased use of hedging insrumens by firms; he lack of volailiy spillovers from sock o currency markes in he pos crises period may be indicaive of he changes in he financial srucure which have aken place since he crises. Our resuls overall indicae ha since he Asian financial crises, here exiss significan scope for invesors and porfolio managers o diversify heir asses beween socks and currencies in hese markes. In paricular, he lack of volailiy spillovers beween sock markes and exchange raes in all counries, and beween exchange raes and sock markes in 18

20 all counries excep Taiwan in he pos crises period indicaes ha here is scope for invesors o diversify heir invesmens in hese markes. From he poin of view of policy, linkages beween he wo markes in Taiwan indicaes ha policymakers should ake ino accoun he impac of any exchange rae policy on sock markes, and he impac of policies relaing o he sock marke in erms of heir impac on he exchange rae. 19

21 6. References Andersen, T., Benzoni, L. and Lund, J., (004, Sochasic Volailiy, Mean Drif, and Jumps in he Shor-Term Ineres Rae, Deparmen of Finance, Working paper 14, J.L Kellogg Graduae School of Managemen, Norh-wesern Universiy. Bodnar, G., Hay, S., Marson, R amd Smihson, C., (1995, Wharon Survey of Derivaives Usuage by US Non-financial Firms, Financial Managemen, Vol. 4, pp Bollerslev, T., (1986, Generalized auoregressive condiional heeroscedasiciy, Journal of Economerics, Vol. 31, pp Branson, W. (1983, Macroeconomic Deerminans of Real Exchange Rae Risk, in R. Herring (Ed., Managing Foreign Exchange Risk, MA: Cambridge Universiy Press. Caporale, M., Piis, N. and Spagnolo, N, Tesing for Causaliy-in-Variance: An applicaion o he Eas Asian Markes, Inernaional Journal of Finance and Economics, Vol. 7, pp/ Chang, T. and Caudill, S., (005, Financial Developmen and Economic Growh: The Case of Taiwan, Applied Economics, Vol. 37, pp Dornbusch, R., and Fischer, S., (1980, Exchange Raes and he Curren Accoun, American Economic Review, Vol. 1, pp.7-1. Enders, W., (004, Applied Economeric Time Series. Chicheser, UK: Wiley. Fang, W., (00, The effec of currency depreciaion on sock reurns: evidence from five Eas Asian economies, Applied Economics Leers, Vol.9, pp Frankel, J., (1983, Moneary and Porfolio-Balance Models of Exchange Rae Deerminaion, in J. Bhandari and B. Puman (Eds, Economic Inerdependence and Flexible Exchange Raes, MA, MIT Press. Hamilon, J., (1994, Time Series Analysis, Princeon: Princeon Universiy Press. Karolyu, G. and Sulz, R.,(1996 Why do markes move ogeher? An invesigaion of US, Japan Sock Reurn Comovemens, Journal of Finance. Vol.51, pp Gran, K. and Marshall, A., (1997, Large UK Companies and Derivaives European Financial Managemen, Vol. 3, pp Griffin, J., Nardari, F. and Sulz, R., (004 Do Invesors Trade More when Socks have Performed Well? Evidence from 46 Counries. EFA 004 Maasrich Meeings Paper No. 511; Dice Cener Working Paper No , Ohio Sae Universiy. Johansen, S., (199, Deerminaion of coinegraion rank in he presence of a linear rend, Oxford Bullein of Economics and Saisics, Vol 54, pp Johansen, S., Mosconi, R and Nielsen, B., (000, Coinegraion analysis in he presence of srucural breaks in he deerminisic rend, Economerics Journal, Vol. 3, pp Jorion, P. (1990, The Exchange Rae Exposure of US Mulinaionals, Journal of Business, Vol.63, pp Kanas, A. (000, Volailiy Spillovers Beween Sock Reurns and Exchange Rae Changes: Inernaional Evidence, Journal of Business Finance and Accouning, Vol. 7, pp Kanas, A. (00, Is Exchange Rae Volailiy Influenced by Sock Reurn Volailiy? Evidence from he US, he UK and Japan, Applied Economics Leers, Vol. 9, Koumos, G., and Booh, G., (1995, Asymmeric Volailiy Transmission in Inernaional Sock Markes, Journal of Inernaional Money and Finance, Vol. 14, pp Laopodis, N., (1998, Asymmeric Volailiy Spillovers in Deusche Mark Exchange Raes, Journal of Mulinaional Financial Managemen, Vol.8, pp Mundell, R., (1963, Capial Mobiliy and Sabilizaion Policy under Fixed and Flexible Exchange Rae, Canadian Journal of Economics and Poliical Science, Vol. 9, pp Mundell, R., (1964, A reply: Capial Mobiliy and Size, Canadian Journal of Economics and Poliical Science, Vol. 30, pp

22 Nelson, D., (1991, Condiional Heeroscedasiciy in Asse Reurns: A New Approach, Economerica, Vol. 59, pp Nieh, C.C and Lee, C.F, (001, Dynamic Relaionship beween Sock Prices and Exchange Raes for G-7 Counries. The Quarerly Review and Finance 41, pp Smih, C., (199, Sock Markes and he Exchange Rae: A Muli-Counry Approach, Journal of Macroeconomics, Vol. 14, pp Wu, R., (005, Inernaional Transmission Effec of Volailiy beween he Financial Markes during he Asian Financial Crises, Transiion Sudies Review, Vol. 15, pp Yang, S. and Doong, S., (004, Price and Volailiy Spillovers beween Sock Prices and Exchange Raes: Empirical Evidence from he G-7 Counries, Inernaional Journal of Business and Economics, Vol. 3, pp Yau, H. and Nieh, C.C., (006, Inerrelaions among Sock prices of Taiwan and Japan and NTD/Yen Exchange Rae, Journal of Asian Economics, Vol. 17, pp Zapaero, F., (1995, Equilibrium Asse Prices and Exchange Raes, Journal of Economic Dynamics and Conrol, Vol. 19. pp