Volatility Spillovers between Stock Market Returns and Exchange Rate Changes: the New Zealand Case

Transcription

1 Volailiy Spillovers beween Sock Marke eurns and Exchange ae Changes: he New Zealand Case Choi, D.F.S., V. Fang and T.Y. Fu Deparmen of Finance, Waikao Managemen School, Universiy of Waikao, Hamilon, New Zealand Deparmen of Accouning and Finance, Faculy of Business and Economics, Monash Universiy, Ausralia Keywords: Volailiy Spillover, Sock eurns, Exchange rae, Exponenial GACH EXTENDED ABSTACT Kanas (000) firs examines volailiy spillovers beween sock marke reurns and exchange rae changes on six developed counries: he US, Canada, Japan, he UK, France, and Germany. He finds ha, for all counries, spillovers from exchange rae changes o sock marke reurns are insignifican. On he oher hand, spillovers from sock marke reurns o exchange rae changes are significan for all counries excep Germany. New Zealand (NZ) is a small marke which is characerized by down-markes and is impaced by inernaional movemens. In his paper, i is found ha, for NZ daa, when he exchange rae volailiy is higher, he sock marke volailiy is lower before he 997 sock marke crash. However, his volailiy spillover becomes significanly posiive afer he crash. On he oher hand, we find significan volailiy spillovers from sock marke reurns o NZ dollar movemens in he foreign exchange marke only before he997 crash bu no afer, i.e., volailiy spillovers beween exchange rae changes and sock marke reurns change over ime.. INTODUCTION Inernaional equiy invesmens are increasing rapidly since mid-980. Kanas (000) sudies volailiy spillovers beween exchange rae changes and sock marke reurns on six indusrialized counries, no including New Zealand and Ausralia. His empirical evidence shows ha, in general, sock reurn volailiy spills over o exchange rae changes while he reverse is no significan. The US and Ausralia are he op wo rading parners of NZ. Brailsford (996) firs invesigaes volailiy spillovers beween he NZ and Ausralian sock markes. He finds ha, afer conrolling for he effecs of he US volailiy, he Ausralian marke volailiy spills over o he New Zealand marke bu no he reverse. Brailsford has no ye sudied he volailiy spillover beween exchange rae changes and sock marke reurns wihin eiher NZ or Ausralia. Dungey (999) poins ou ha he NZ currency is mainly affeced by he inernaional facors as more han 50% of he decomposiion of NZ dollar volailiy in he foreign currency marke are due o overseas impacs. Moreover, he NZ sock marke is small, illiquid, limied o diversificaion, and exposed o exchange rae volailiy. Chen e al. (004) find ha, in he New Zealand sock marke, here is a lagged sock marke reacion o exchange rae flucuaions. Their resul indicaes ha he volailiy of he sock marke reurns migh no rise when he exchange rae uncerainy is higher. Unforunaely, Chen e al. have no furher repored he volailiy spillover of exchange rae changes o he sock marke reurns. This paper applies he same EGACH model o volailiy spillovers. I is found ha here is significan volailiy spillovers from exchanges rae changes o sock marke reurns; volailiy spillovers from sock marke reurns o exchange rae changes is marginally significan and changes from negaive before he 987 sock crash o posiive afer he crash. The res of he paper is organized as follows: Secion provides a lieraure review, Secion 3 deails he daa and he EGACH mehodology, Secion 4 discusses he empirical resuls, and Secion 5 concludes.. LITEATUE EVIEW Exchange rae changes are criical o a sock marke which is eiher impor or expor-oriened or boh. This is paricularly criical o he NZ marke. Empirical resuls show ha he exchange rae uncerainy has a significanly negaive effec on invesmen. Applying a rolling sandard deviaions mehod, Goldberg (993) finds evidence from he US indusry-level invesmen ha exchange rae uncerainy has significanly negaive long-run effecs on invesmen. Darby e al. (999), using a single-equaion esimaion in heir sudy, find a similar negaive exchange rae effec on aggregae invesmen based on he daa from five OECD counries. Carruh e al. (000) and Serven (003), 60

2 adoping a GACH framework, find a highly significanly negaive impac from he real exchange rae uncerainy on invesmen. Choudhry (005) finds ha he effecs of exchange rae variabiliy on real expors are, in general, significanly negaive. There are a few explanaions proposed by classical economiss on why equiy prices are affeced by exchange rae changes. The flow models (e.g., Dornbusch and Fisher 980) sugges ha he currency flucuaions impac on he inernaional compeiiveness of he firm as well as he balance of he rade posiion. The real oupu of a counry is hus affeced by exchange rae changes. Share prices of companies are evenually influenced due o he changes in he curren and fuure cash flows of he firm. The exchange rae changes herefore should bear effecs on sock reurns. A recen sudy of Alaganar and Bhar (007) indicaes ha he firs- and second-order effecs of exchange rae changes impac significanly on diversified porfolios in he US share marke. They use weekly reurns of 6 World Equiy Benchmark Series (WEBS) in he sudy while each WEBS series represens a diversified porfolio of invesing in foreign shares in a counry ouside of he Unied Saes. The diversificaion echnique is efficienly applied in WEBS while racking he performance of a foreign counry s Morgan Sanley Capial Inernaional index hough WEBS are raded only in U.S. Dollars. Alaganar and Bhar (007) employ he GJ and GACH-M models in esing he impacs of exchange rae volailiy on reurns of diversified counry index porfolios. They find ha he exchange rae uncerainy is priced as a facor in WEBS reurns. Moreover, exchange rae volailiy has lagged posiive impacs on he volailiy of WEBS. They poin ou ha he second momen exchange rae informaion is imporan o diversificaions in he sock marke. The exchange rae risk is priced in reurns and he exchange rae volailiy is imporan o invesors. The empirical findings provided in he above-menioned sudies are for large developed markes. Chen, Naylor, and Lu (004) argue ha NZ sock prices are closely ied o exchange rae movemens compared o socks in oher counries. Their empirical resul suggess all NZ firms cash flows follow exchange rae flucuaions. Company foreign exchange exposures can be relaed o one single currency. In paricular, heir empirical resuls show ha NZ firms are more sensiive o he and exchange rae changes. Using hese wo exchange raes plus he in heir sudy, he auhors successfully idenify significan effecs of exchange rae flucuaions on he sock reurns. However, such effec oally disappears if monhly daa is used. In oher words, heir evidence suggess only delayed, bu no conemporaneous, marke reacions o exchange rae movemens. This sudy employs he EGACH model o invesigae he volailiy behaviour in NZ sock reurns and currency movemens. The GACH family has been widely used o capure and es he sock marke volailiy in he search of evidence how quickly he sock marke reflecs new informaion. Black (976) and Chrisie (98) noice ha a sock price decline is followed by an increase in he subsequen sock volailiy. This observaion is ermed as he leverage effec, which is esed in he GJ model developed by Glosen e al. (993). Blair e al. (00) explain ha a decline in he sock price will increase he deb-equiy raio and hence he equiy risk. If he leverage effec exiss in he NZ sock marke, we need o consider wheher he exchange rae volailiy effec is in fac a reflecion of he leverage effec. Pinfold e al. (00) poin ou ha he NZ sock marke is characerized by down marke condiions. As a resul, filering away he leverage effec in volailiy spillover ess is essenial. The EGACH model handles he leverage effec very efficienly. Kanas (000) firs uses EGACH models in invesigaing he volailiy spillover effecs beween he equiy marke and he foreign exchange marke in six indusrialized counries. Based on he daily daa of US, Canada, Japan, UK, France, and Germany for he period beween 986 and 998, his sudy shows significan symmeric spillover effecs from sock marke reurns o foreign exchange rae changes. Moreover, he effecs of good news in he sock marke are he same in absolue erms as hose of he bad news for exchange rae volailiies. Sub-period ess show ha spillover effecs from he equiy markes o he foreign exchange markes increase afer he 987 sock marke. These resuls are consisen wih he views ha financial markes become more inegraed afer sock marke crashes and he volailiy relaionship beween sock reurns and exchange rae movemens change over ime. The significan volailiy spillovers from sock reurns o exchange rae changes have imporan implicaions for he asse approach o exchange rae deerminaion (Branson, 983; and Frankel 983), paricularly when inernaional equiy invesmens have been rising since he mid-980s. The higher level of cross-border financial asses flows beween, for example, domesic and foreign share markes resuls in he changes of demand for and supply of currencies in which he inernaional equiies are priced in. Exchange raes have o be adjused according o he inernaional equiy 6

3 flows. Therefore, he asse approach o exchange rae deerminaion suggess significan volailiy ransmission from he sock marke o he foreign exchange marke. While Kanas (000) finds a significan volailiy spillover from sock reurn changes o foreign currency flucuaions, i migh no be he case for he NZ economy. Dungey (999) argues ha i is he inernaional facor ha mainly affecs he volailiy of NZ dollars. In oher words, volailiy shocks in he local NZ sock markes can have lile effecs on he movemen of he NZ currency. 3. DATA AND METHODOLOGY 3.. Daa and descripive saisics We collec daily NZ oal marke index prices from Daasream. The exchange raes and as well as he are colleced from he eserve Bank of New Zealand. The USA and Ausralia markes are he wo larges expor markes of NZ firms. Chen e al. (004) use hese exchange raes in heir sudy. Our sample period spans from January 990 o December 004. There are 3,866 observaions. Adoping NZ daa for idenifying spillover effecs across sock and foreign exchange markes in fac provides a unique environmen for examining he volailiy relaionship beween sock reurns and foreign currency movemens. Dungey (999) finds ha i is he inernaional facor ha mainly affecs he volailiy of NZ dollars. In paricular, his paper indicaes ha he oal impac of inernaional facors can accoun for more han 50% of he decomposiion of volailiy. The NZ currency sars floaing independenly on he inernaional exchange marke since 985. The NZ dollar akes price-aker posiions because of is relaively small capializaion in he foreign exchange marke. The eserve Bank of NZ has no inervening moneary policy or inensions o inervene for NZ dollar value wihin our esing period. Our research focuses on he period before ha ime. Furhermore, NZ is no a world leader in many indusrial and commercial areas. The GDP figures deermine ha NZ is small and easily affeced by inernaional economical condiions. Local NZ firms are less diversified and hey have expors or impors orienaions. Because of he close linkage among local NZ companies, none of he firms can be exemped from inernaional facors influences. Moreover, alhough he NZ sock marke is very small and illiquid, he NZ sock exchange is efficien wih high marke awareness, especially o he exchange rae uncerainy. All hese unique and imporan characerisics of NZ well developed economy arac us o use her daa for invesigae how he NZ sock marke volailiy responds o he exchange rae variabiliy. In his sudy, he sock marke reurn and all exchange rae changes are defined as log relaive reurns. Table shows some descripive saisics of sock and foreign exchange marke reurns. Jarque-Bera es is performed for sock marke index and exchange rae reurns in order o check he normaliy of each series. The resuls show ha he disribuion of sock index and exchange rae reurns are no normal. We herefore can no longer assume normal disribuion in our daase. Ordinary leas squared regression ess are no appropriae in esing he relaionship beween sock reurns and exchange rae changes. In his paper, he EGACH model is adoped o examine he dynamic volailiy relaionship beween NZ sock reurn and foreign exchange rae changes. We also check he saionariy of sock reurns and exchange rae reurns. Table conains he resuls of Augmened Dickey-Fuller (ADF) uni roo es. The -saisics are highly significanly a he % level. All sock and exchange reurn series are inegraed wih I(0), i.e., saionary. Table. Descripive Saisics. This able presens basic descripive saisics of daily sock and foreign exchange reurn of New Zealand marke. We repor number of observaions, mean, median, maximum and minimum value, sandard deviaion, skewness and kurosis for each series. esuls of Jarque-Bera and ADF es are provided as he follows. Toal Marke Index NZD/ AUD NZD/ USD TWI Index Observaions Mean E E-05.96E-05 Median 0 9.4E Maximum Minimum Sd. Dev Skewness Kurosis Jarque-Bera ADF Tes [0] [0] [0] [0] [0] [0.000] [0] [0] 3.. The EGACH model We examine he relaionships beween he movemens of he sock marke volailiy and exchange rae flucuaions wihin an EGACH 6

4 model developed in Nelson (99). The mulivariae EGACH models can effecively capure he asymmeric effecs of innovaion on volailiy in sock reurns while he condiional volailiy of currency movemens can be included in he model a he same ime. This imporan feaure of he model gives direc and explici measuremen of he role of exchange rae flucuaions in explaining he ime series behaviour of sock reurn volailiy. Similarly, EGACH models can es wheher he exchange rae volailiy effecs are asymmerical in he foreign exchange marke, ha is, if he appreciaions of NZ dollar in he pas have he same impac on he fuure exchange rae volailiy as he depreciaions have. Alhough leverage effecs are idenified in he sock markes, he asymmeric volailiy impacs in he foreign exchange marke are no clear. The models are specified by he following equaions: = β + h h = exp = exp,,,0 ) h = β,0 + α, j j + β, j j + ε j= j= { α,0 + α, ) + α, ) + γ ln( h )} { α,0 + α, ) + α, ) + γ ln( h )} = ( Z E( Z ) + δ Z ) = ( Z E( Z ) + δ Z ) ) = ρ p, j j= p, α j h h + p β, j j= p j + ε where and denoe sock reurn and exchange rae changes. We use he, and, for respecively, o examine he exchange rae effecs o sock reurns. The number of lags, represened by p in he equaion, is opimized by he AIC informaion crieria in he resuls. The analysis searches for dynamic relaionship beween sock reurns and exchange rae changes. The Granger causaliy es is a join es ha examines he significance of relevan coefficiens. The coefficiens and in β, j β, j equaion () and () show he effecs of exchange rae changes o sock reurns and he reverse effecs respecively. The sochasic error erms (ε ) are obained in order o be used in he EGACH models. In equaions (3) o (7), h and h denoe he condiional variance of sock reurns and foreign exchange raes when h and h denoe lagged condiional variance included in he exponenial funcion. Z and Z denoe he sandardized innovaions of NZ sock reurns and exchange rae changes respecively () () (3) (4) (5) (6) (7) ( Z = ε h and Z = ε h ). The coefficiens α and, α es wheher here exis volailiy, spillovers across he sock marke and he foreign exchange marke. In paricular, α (, α ) indicaes, volailiy spillover effecs sourced from he foreign exchange (sock) marke o he sock (foreign exchange) marke. Moreover, coefficiens α and, α es he volailiy clusering in he sock reurn, and exchange rae series. Alhough volailiy auocorrelaions are ofen idenified in inernaional sock markes, i is no clear if volailiies in he NZ currency are auocorrelaed. A saisically significan α can provide, evidence ha volailiy in he foreign exchange marke is significanly affeced by he hisorical volailiy changes in he exchange raes. Finally, coefficiens γ and γ indicae volailiy persisence in he sock and foreign exchange marke respecively, indicaing if an informaion shock in he marke ends o have persis effecs for he fuure volailiy. Using EGACH models also allows us o examine he asymmeric effecs in he volailiy impacs. As menioned above, he pas volailiy shocks can have asymmeric effecs on fuure volailiy changes in he marke. δ and δ in equaions (5) and (6) examine asymmery effecs in he sock marke and foreign exchange marke respecively. A posiive (negaive) and saisically significan δ indicaes good (bad) news in he sock marke has greaer impac on he volailiy of sock index. Likewise, δ ess if he appreciaions in NZ dollar have he same impac o he exchange rae volailiy as he depreciaions do. EGACH models are esimaed by using a maximizing log likelihood funcion ha is specified by he following equaion: ' L( ϕ ) = 0.5( NT )ln(π ) 0.5 (ln H + ε H ε ) T = where ϕ is he parameer vecor o be esimaed. N is he number of equaions. As we analyse volailiy spillover beween sock and foreign exchange markes, N equals o wo. T is he number of observaions in he sample; H is a ime-varying condiional variance covariance marix wih diagonal elemens given by equaion ' (3) and (4). ε = [ ε ε ] is he vecor of innovaions from sock and foreign exchange marke a ime. 63

5 4. EMPIICAL ESULTS 4.. Full sample period resuls We find he condiional variances of he sock reurns and exchange rae changes are significanly impaced by heir own sandardized innovaion in he pas. Coefficiens α, and α appear o be, saisically significan a he % level in all cases no maer wha exchange rae measuremens we apply in he equaions. The significan coefficiens α and, α indicae volailiy, clusering in boh NZ sock exchange and he flucuaions of NZ dollar in he inernaional foreign exchange marke. There also exiss volailiy persisence in sock reurns and exchange rae movemens. Volailiy persisence coefficiens γ and γ are all saisically significan a he % level while four ou of six values are less han and close o one. Our resuls are consisen o he resuls of boh Najand and Yung (99) and Chen e al. (00). Looking a he volailiy spillover coefficiens, we noice significan volailiy spillover effecs from he sock marke o he foreign exchange marke. The coefficien α has high saisical, significance for all exchange rae measuremens. The value of α is he highes when he TWI, index is used in he equaion as he index represens he overall performance of he kiwi dollar. The volailiy spillover from he sock reurns o is greaer han ha o. This aribues o he fac ha Ausralia and New Zealand are closely ied in many economical aspecs. Our resuls are consisen o he findings of Kanas (000) ha volailiy changes in he sock reurns have impacs on he movemens of foreign currency. We herefore provide supporive evidence for he asse approach o exchange rae deerminaion (Branson 983 and Frankel 983). While he NZ socks arac inernaional invesmens, he NZ dollar has o be adjused according o he inernaional equiy flows. The foreign exchange rae volailiy changes, on he oher hand, can spill over o he sock volailiy. The coefficien α, is saisically significan a he % level when is applied as he foreign exchange rae in he model. This indicaes ha he volailiy changes in exchange rae have spillover effecs on he sock reurns. However, he spillover effecs are no significan when and are used in he model. We noice ha he asymmery coefficiens δ appear o be negaive and saisically significan a he % level all he ime. This suggess srongly significan leverage effecs in he NZ sock exchange. Bad news ha resuls in sock decline in he marke leads o a more volaile sock movemen. This finding is consisen o he general leverage effec evidence in he lieraure. A he same ime, he asymmery coefficienδ is only saisically significan when is used in he model. This implies ha currency depreciaions in NZ dollar agains he US currency resul in more volailiy in he inernaional foreign exchange marke. This case, however, is no applicable o oher exchange rae variables for he NZ dollar. Table Volailiy spillover beween sock reurns and exchange rae changes (full sample period). Panel A: Sock volailiy α,0 α, α, γ δ a 0.43 a a -0.4 a (-0.07) (8.4) (-0.0) (380.80) (-5.38) a 0.64 a 0.09 a a a (-4.6) (30.8) (4.74) (466.89) (-.63) -0.4 a a a a (-4.97) (5.0) (-0.) (30.6) (-6.0) Panel B: Exchange rae volailiy α,0 α, α, γ δ a a a 0.58 a -0.0 (-.93) (4.6) (4.64) (6.73) (-0.53) a 0.09 a 0.94 a 0.99 a a (-5.5) (3.08) (.36) (694.04) (-4.94) a 0.77 a 0.38 a a -0.0 (-50.83) (5.54) (0.97) (33.7) (-0.0) -saisics in parenheses. a saisically significan a % level b saisically significan a 5% level c saisically significan a 0% level 4.. Sub-periods resuls The 997 sock marke crash occurs in he middle of our full sample period. To check he robusness of our empirical resuls, we perform he ess for wo sub-periods, i.e., before and afer he 997 sock marke crash. Table 3 conains he resuls before he July 997 and Table 4 presens he resuls afer July 997. Table 3 shows significan volailiy spillover effecs from he foreign exchange marke o he sock exchange when and are used. However, we noice ha he coefficiens 64

6 α appear o be negaive in value. The spillover, coefficiens α afer he 997 sock marke crash, become posiive in Table 4. This implies ha he 997 sock marke crash causes he volailiy ransmission from he foreign exchange marke o he sock marke in NZ o change over ime. Comparing o he resuls for he whole period, we noice ha he foreign currency spillover effecs o sock reurns are easier o be idenified in subperiods. In paricular, he volailiy impac caused by flucuaions o NZ sock reurns is significan in he sub-periods esing while i is no he case in he whole period. Noe ha he spillover coefficiens α, for TWI parameers have differen signs in sub-periods, indicaing ha he effecs have changed over ime and herefore make he volailiy ransmission hard o be idenified in he whole period. The volailiy ransmissions from he sock marke o he foreign currency marke, on he oher hand, become insignifican for and TWI index afer he 997 sock marke crash. Table 4 indicaes ha he volailiy spillover from sock reurns only has impacs on he NZ dollar agains he Ausralian dollar. This again aribues o he close economical ie beween wo counries. However, able 3 shows ha he volailiy spillover from he sock reurns o NZ dollar flucuaions is significan across all currency parameers. Empirical resuls sugges ha he spillover effecs from he sock marke o he foreign exchange marke seem o vanish afer he 997 crash. This is consisen o he view ha NZ currency is driven by inernaional facors insead of domesic facors. While he news in he local sock marke had shor erm volailiy effecs on NZ dollar flucuaions, he volailiy spillovers disappeared afer he sock marke crash in 997. The sub-periods empirical resuls provide imporan insigh of he volailiy spillovers beween he sock marke and foreign exchange marke in NZ. The volailiy ransmissions have changed over ime. Table 3 Volailiy spillover beween sock reurns and exchange rae changes before 997 sock crash. Panel A: Sock volailiy α,0 α, α, γ δ a 0.3 a a (-3.97) (8.54) (-.0) (395.56) (-0.46) a 0.58 a a a a (-8.0) (9.96) (-3.59) (466.88) (-6.68) a 0.4 a a a -0.0 b (-4.49) (8.39) (-.35) (304.96) (-.09) Panel B: Exchange rae volailiy α,0 α, α, γ δ a 0. a a 0. a 0.8 a (-9.4) (3.5) (.4) (.65) (3.8) -.0 a a 0.0 a 0.99 a a (-0.67) (3.47) (.50) (694.04) (-.0) a 0.09 a 0. a 0.78 a b (-9.99) (3.08) (.89) (35.76) (-.3) Table 4 Volailiy spillover beween sock reurns and exchange rae changes afer 997 sock crash Panel A: Sock volailiy α,0 α, α, γ δ a 0.80 a a a (-5.9) (8.63) (-.3) (06.3) (-8.3) a a a 0.88 a c (-.6) (5.4) (.53) (4.79) (-.85) -.05 a 0.34 a a a b (-9.4) (.5) (3.74) (30.6) (-.9) Panel B: Exchange rae volailiy α,0 α, α, γ δ a 0.08 a 0.0 a 0.9 a (-3.3) (3.58) (4.68) (38.4) (-.037) a a 0.49 a a (-3.80) (3.9) (3.67) (-3.67) a a a a (-.39) (-0.0) (5.9) (33.7) (-3.) -saisics in parenheses. a saisically significan a % level b saisically significan a 5% level c saisically significan a 0% level 5. CONCLUDING EMAKS In his paper, he EGACH model is adoped o examine volailiy spillovers beween exchange rae changes and sock marke reurns in New Zealand. We effecively conrol he ofen down- 65

7 marke effecs in he NZ sock marke by incorporaing he leverage effec in he EGACH framework. We hen examine wheher he effec of volailiy spillover changes over ime by performing he es in he sub-periods. Kanas (000) suggess ha he volailiy spillover effecs from he sock marke o he foreign exchange marke becomes significan afer he 987 crash as financial markes become more inegraed. The NZ environmen, however, is very differen from he markes examined in Kanas (000). Dungey (999) poins ou ha he NZ currency is mainly driven by inernaional facors. We find ha local sock marke reurns in NZ have marginal effecs on he volailiy of he NZ dollar. The volailiy spillovers from he sock marke o he foreign exchange marke in NZ change over ime. Before he 997 sock crash, he spillover effecs from sock reurns o exchange rae changes are significan. However, he spillovers vanish afer he crash. This paper provides empirical evidence ha he asse approach o exchange rae deerminaion does no fi o small markes like NZ, whose currency is driven by inernaional facors. On he oher hand, we find significan volailiy spillovers from he foreign currency movemens o sock reurns in NZ. Our resuls are consisen o he argumen of Chen e al. (004) ha all NZ firms are affeced by NZ dollar flucuaions. The spillovers from he exchange rae o sock reurns are significan before and afer he 997 crash when leverage effecs in he NZ sock marke are incorporaed by he EGACH framework in he analysis. 6. EFEENCES Alaganar, V.T. and. Bhar (007), Empirical properies of currency risk in counry index porfolios, The Quarerly eview of Economics and Finance, Vol 47, Black, F. (976). Sudies of sock price volailiy changes. Proceedings of he American Saisical Associaion, Business and Economics Saisics Secion, Blair, B., Poon, S.H., and Taylor, S.J. (00). Asymmeric and crash effecs in sock volailiy for he S&P 00 index and is consiuens, Applied Financial Economics,, Brailsford, T.J. (996). Volailiy spillover across he Tasman, Ausralian Journal of Managemen, (), 3-7. Branson, W.H. (983). Macroeconomic Deerminans of eal Exchange ae isk, in.j. Herring (ed.), Managing Foreign Exchange isk (Cambridge, Universiy Press, MA). Carruh, A., Dickerson, A., and Henley, A. (000). Wha do we know abou invesmen under uncerainy? Journal of Economic Surveys, 4, Chen, J., Naylor, M., and Lu, X. (004). Some insighs ino he foreign exchange pricing puzzle: Evidence from a small open economy, Pacific-Basin Finance Journal,, Choudhry, T. (005). Exchange rae volailiy and he Unied Saes expors: evidence from Canada and Japan, Journal of Japanese and Inernaional Economies, 9, 5 7. Chrisie, A. (98). The sochasic behaviour of common sock variances: values, leverage and ineres rae effecs, Journal of Financial Economics, 0, Darby, J., Hughes H.A., Ireland, J., and Piscaelli, L. (999). The impac of exchange rae uncerainy on he level of invesmen, Economic Journal, 09, C55 - C67. Dornbusch,. and Fisher, S. (980). Exchange raes and he curren accoun, American Economic eview, 70, Dungey, M. (999). Decomposing exchange rae volailiy around he Pacific im, Journal of Asian Economics, 0, Frankel, J.A. (983). Moneary and Porfolio- Balance Models of Exchange rae Deerminaion, in J.S. Bhandari and B.H. Puman (eds.), Economic Inerdependence and Flexible Exchange raes (MIT Press, Cambridge, MA). Goldberg, L.S. (993). Exchange raes and invesmen in Unied Saes indusry, eview of Economics and Saisics, 4, Kanas, A. (000). Volailiy spillover beween sock reurns and exchange rae changes: Inernaional evidence, Journal of Business Finance & Accouning, 7(3) & (4), Najand, M. and Yung, K. (99). A GACH examinaion of he relaionship beween volume and price variabiliy in fuures markes, Journal of Fuures Markes, (5), Nelson, D. (99). Condiional heeroskedasiciy in asse reruns: a new approach, Economerica, 59, Pinfold, J.F., Wilson, W.. and Li, Q. (00). Book-o-marke and size as deerminans of reurns in small illiquid markes: he New Zealand case, Financial Services eview, 0, Serven, L. (003). eal exchange rae uncerainy and privae invesmen in developing 66

8 Counries, eview of Economics and Saisics, 85,