THE IMPACT OF DOLLAR-RAND VOLATILITY ON U.S. EXPORTS TO SOUTH AFRICA E. M. Ekanayake, Bethune-Cookman University Ranjini Thaver, Stetson University

Transcription

1 The Inernaional Journal of Business and Finance Research Volume 5 Number 3 THE IMPACT OF DOLLAR-RAND VOLATILITY ON U.S. EXPORTS TO SOUTH AFRICA E. M. Ekanayake, Behune-Cookman Universiy Ranjini Thaver, Seson Universiy ABSTRACT This sudy invesigaes he effecs of exchange rae volailiy on he op en caegories of expors by he Unied Saes o Souh Africa over a -year period from January 99 o December 9. The paper uses several measures of volailiy o generae a measure of exchange rae volailiy, which is hen esed in a model of U.S. expors o Souh Africa. We employ secoral rade daa a he -digi HS level o evaluae hese effecs on he op en individual commodiies raded. Uilizing bounds esing coinegraion, we esimae he shor- and long-run impac of exchange-rae volailiy on he US expors o Souh Africa. Our resuls sugges ha while he effecs of exchange rae volailiy on expors is mixed in he shor-run, in he long-run, exchange rae volailiy exers a negaive effec on he U.S. expors o Souh Africa. JEL : F4, F3 KEYWORDS: exchange raes, volailiy, expors, ARDL bounds esing, Souh Africa. INTRODUCTION The breakdown of he Breon Woods sysem and he adopion of he flexible exchange rae regime in 973 has led o a proliferaion of research on he impac of exchange rae volailiy (ERV) on real expors. The ineres in his research was promped by hree main developmens: (a) boh he real and nominal exchange raes have undergone periods of subsanial volailiy since 973; (b) during he same period, inernaional rade declined significanly among indusrialized counries; and (c) macroeconomic insabiliy in erms of oupu, inflaion, ineres raes, and employmen began o surface. Despie he sizeable number of sudies conduced, no real consensus abou he impac of exchange rae volailiy on expors has emerged. While a large number of sudies find ha ERV ends o reduce he level of rade, ohers find eiher weak or insignifican or posiive relaionships. For example, Onafowara and Owoye (8), Byrne, Darby, and MacDonald (8), Choudhry (5), Bahmanee-Oskooee (), Arize, e al. (), Arize (995), Chowdhury (993), Pozo (99), and Bahmani-Oskooee and Laifa (99), find evidence for negaive effecs. According o hese scholars, ERV may affec expors direcly hrough uncerainy and adjusmen coss for risk-averse exporing invesors. Furher, i may have an indirec effec hrough is impac on he srucure of oupu, invesmen and governmen policy. On he oher hand, Doyle (), Chou (), McKenzie and Brooks (997), Qian and Varangis (994), Kroner and Lasrapes (993), and Asseery and Peel (99) find evidence for a posiive effec for volailiy on expor volumes of some developed counries because exchange rae volailiy makes exporing more aracive o risk-oleran exporing firms. However, oher scholars such as Arisoelous (), Bahmani- Oskooee and Payesch (993), Bahmani-Oskooee (99), and Hooper and Kohlhagen (978) have repored no significan relaionship beween ERV and expors. Reasons for conradicory resuls by differen sudies may be due o a variey of facors, among hem: differen mehods used o measure ERV; he use of differen price deflaors; he differenial use of sample daa, for example, he use of aggregae expor daa versus secoral expor daa; differen ime-frame periods; ignoring impor dependency on inermediae and capial goods of he receiving counry, as is he 73

2 E. M. Ekanayake, R. Thaver IJBFR Vol. 5 No. 3 case wih many developing counries; and he absence of complex economeric mehods for sudying hese variaions. As a resul scholars sopped invesigaing he ERV-expor nexus by he lae 99 s. However, wih beer access o secoral daa and he developmen of more sophisicaed economeric models, recen sudies have begun evaluaing he ERV-expor connecion from a secoral perspecive. The raionale behind his is ha differen rade secors would be impaced differenially by ERV, and herefore may be more revealing han aggregae sudies. This sudy focuses on secoral expor rade from he Unied Saes o Souh Africa, a developing counry, using hree differen measures of volailiy ha may help o uncover he naure and sensiiviy of he relaionship beween ERV and secoral expors. We use he bounds esing approach o coinegraion o esablish a long-run relaionship among he explanaory variables. We also employ error-correcions models (ECM) o esablish he shor-run dynamics of he relaionship. In addiion, we use he generalized auoregressive condiional heeroskedasiciy (GARCH) model o generae one of he hree measures of ERV. Using his approach we invesigae he effecs of exchange rae volailiy on he op en caegories of expors by he U.S. o Souh Africa over a period of years using monhly daa from January 99 o December 9. Alhough Souh Africa accouns only for a very small share of U.S. oal rade, i is he larges African rading parner of he Unied Saes. On he oher hand, he Unied Saes is he hird larges marke for Souh African expors. To his end we provide a brief review of he lieraure in he nex secion. Thereafer, we lay he empirical framework of our sudy by specifying our model. In he secion following ha we discuss variable definiions and ouline our daa sources. Empirical resuls from he bounds esing approach o coinegraion, and error-correcion model esimaes are presened in he penulimae secion. The final secion presens a summary and conclusion of he resuls obained in his sudy. LITERATURE REVIEW In his secion we presen a brief overview of sudies ha examine he ERV-rade nexus on U.S. rade flows using secoral daa. We begin by discussing he mos recen and sophisicaed sudies, employing coinegraion echniques using GARCH and ECM models, o older, less complex sudies. Bahmani-Oskooee and Hegery (9) invesigae he effecs of exchange rae flucuaions on rade flows beween he U.S. and Mexico using disaggregaed, indusry-level annual expor and impor daa for indusries from 96 o 4. They analyze boh he shor- and long-erm effecs of volailiy in he peso/dollar real exchange rae on Mexican-Unied Saes rade. They conclude ha in he shor-erm increased volailiy negaively affecs rade flows in mos indusries. Long-erm effecs however, are significan for only one-hird of he indusries sudied, and of his, only wo-hirds are negaive. They speculae ha increased Mexican inegraion and liberalizaion of economic policies allow for greaer adjusmens in he long-erm so ha volailiy is less of a problem in he long-erm han in he shor-erm. Byrne, Darby, and MacDonald (8) analyze he impac of ERV on he volume of bilaeral U.S. rade flows using homogenized and differeniaed secoral annual daa over he period 989- for a crosssecion of 6 EU counries and indusries. Their sudy finds ha clusering all indusries ogeher provides evidence of a negaive effec on rade from ERV, which confirms findings of oher sudies using aggregae daa. However, when invesigaing secoral rade differences, he effecs of ERV on rade is negaive and significan for differeniaed goods and insignifican for homogeneous goods, confirming recen sudies ha secoral differences are in fac crucial o explaining he differenial impac of volailiy on rade. They sugges ha a greaer degree of disaggregaion a he indusry level may provide more worhwhile resuls, which is wha we do in his sudy. 74

3 The Inernaional Journal of Business and Finance Research Volume 5 Number 3 Bahmani-Oskooee and Kovyryalova (8) invesigae he effec of exchange rae flucuaions on rade flows beween he U.S. and he Unied Kingdom using disaggregraed annual expor and impor daa for 77 commodiies indusries from 97 o 3. They analyze boh he shor- and long-erm effecs of real ERV on rade beween he U.S and he UK. Their resuls reveal ha he volailiy of he real dollar pound rae has a shor-erm significan effec on impors of 9 indusries and on expors of 99 indusries. In mos cases, such effecs are unfavorable. In he long run, however, he number of significan cases is somewha reduced: only 6 impor and 86 expor indusries are significanly and adversely affeced by ERV. The indusries affeced involve boh durable and non-durable goods, and include small as well as large indusries, supporing findings by aggregae sudies. In anoher sudy, Bahmani-Oskooee and Mira (8), invesigae he effecs of ERV on rade flows beween he U.S. and India, an emerging economy. Using annual daa from 4 indusries from 96 4, heir resuls demonsrae ha ERV has more shor-run han long-run effecs. In he shor-run, 7 indusries were affeced on he impor side and 5 on he expor side. The indusries affeced show India s increasing abiliy o produce impor subsiuable goods. However, in he long run, only a few indusries are affeced because he increasing dependence on rade beween India and he US cause indusries o respond inelasically o ERV. Using boh he nominal and he real exchange rae beween he Unied Saes dollar and he currencies of Canada and Japan, Choudhury (5) invesigaes he influence of exchange rae volailiy on U.S. real expors o Canada and Japan using aggregae monhly daa ranging from January 974 o December 998. The sudy uses condiional variance from he GARCH (, ) model as a measure of exchange rae volailiy, and finds significan and mosly negaive effecs of ERV on real expors. As in he above sudies, Sukar and Hassan () invesigae he relaionship beween U.S. rade volume and ERV using coinegraion and error-correcion models. Their sudy uses quarerly aggregae daa covering he period 975Q 993Q and a GARCH model o measure he exchange rae volailiy. Paralleling oher sudies, he auhors find evidence for a significanly negaive relaionship beween U.S. expor volume and ERV. However, unlike oher findings, hey reveal ha he shor-run dynamics of he ERV-rade relaionship is insignifican. They argue ha his resul may be due o he exisence of avenues for hedging agains exchange risks so as o neuralize he negaive impac of ERV. Oher scholars argue ha his shor-run insignifican relaionship may be because of he invesigaors use of aggregae daa, which ignores secoral differences. For example, while one secor may exhibi a negaive relaionship, anoher may exhibi an equal bu opposie effec so ha hey offse each oher. Arize (995), using monhly series from February 978 o June 986 analyzes he effecs of real ERV on he proporions of bilaeral expors of nine caegories of goods from he U.S. o seven major indusrial counries. The volailiy measure employed is he sandard deviaion of he monhly percenage change in he bilaeral exchange rae beween he U.S. and he imporing counry from o -. The sudy reveals differenial effecs of ERV across differen caegories of expors. The sudy also concludes ha exchange rae uncerainy has a negaive effec on U.S. real expors, and ha i may have a major impac on he allocaion of resources o differen indusries depending on rade elasiciies. Lasrapes and Koray (99) analyze he inerrelaionships among exchange rae volailiy, inernaional rade, and macroeconomic variables using he vecor auoregression (VAR) model. The model esimaes U.S. mulilaeral rade from 973 o 99 and includes a moving sandard deviaion measure of real ERV. While he resuls reveal some evidence of a saisically significan relaionship beween volailiy and rade, he moving average represenaion of he model implies a raher small quaniaive effec. The sudy concludes ha ERV is influenced by he sae of he economy, a facor ignored in a variey of oher sudies. 75

4 E. M. Ekanayake, R. Thaver IJBFR Vol. 5 No. 3 Klein (99) is one of he firs few scholars o analyze he effecs of ERV on he proporion of disaggregaed bilaeral expors of nine caegories of goods from he U.S. o seven major indusrial counries using fixed effecs framework. Using monhly series daa from February 978 o June 986, he sudy reveals ha in six caegories of expors ERV significanly affecs he volume of expors and in five of hese caegories he effec is posiive, suggesing ha real ERV may in fac increase expors by riskaking firms. Koray and Lasrapes (989) examine he relaionship beween real ERV and bilaeral impors from five counries, namely, he UK, France, Germany, Japan, and Canada, employing a VAR model. The sudy uses aggregae monhly daa over a 7-year period from January 959 o December 985, and ess for differen effecs during boh he fixed and he flexible exchange rae regimes. Resuls sugges ha while he effecs of volailiy on impors is weak, permanen shocks o volailiy experience a negaive impac on impors. However, hose effecs are relaively more imporan during he flexible-rae han he fixedrae period. Finally, Cushman (988) ess for real exchange rae volailiy on U.S. bilaeral rade flows using annual daa from o sudy he effecs of he floaing exchange rae regime on ERV. The sudy finds evidence for significan negaive effecs in only wo of six U.S. expor flows wih one expor flow showing a significan posiive effec, confirming oher sudies of a weak risk-averse effec of ERV on exporing firms. One major problem wih mos of he sudies above is ha he sample period includes he period prior o he end of he fixed exchange regime, so resuls may include he lag effecs of fixed exchange raes on rade before 973 lingering on during he ransiion period afer he implemenaion of he floaing exchange rae regime. The curren sudy correcs for his poenial bias by using U.S. monhlydisaggregaed rade daa covering a -year period from January 99 o December 9. We focus on he op en expor producs in U.S.-Souh Africa rade o beer undersand how each indusry is affeced by ERV. The mehodology used in his sudy incorporaes many of he recen developmens in he lieraure, namely, bounds esing approach o coinegraion and error-correcion models, which may uncover he naure and sensiiviy of he ERV-rade nexus. In addiion, GARCH models are used o generae he ERV variable used in he sudy. MODEL SPECIFICATION The objecive of his sudy is o assess he effecs of exchange rae volailiy on he disaggregaed U.S. secoral expors o Souh Africa. Drawing on he exising empirical lieraure in his area, we specify ha a sandard long-run expor demand funcion for commodiy i may ake he following form (see, for example, Ozurk and Kalyonku, 9; Choudhry, 5; Arize, 998, 996, 995; and Asseery and Peel, 99): ln X = β + β lny + β ln P + β3 ln RER + β4 ln ERV + ε () i i Where X i is real expor volume of commodiy i in period, Y is he real income of Souh Africa in period, P i is he relaive price of expors of commodiy i in period, RER is he real exchange rae beween he U.S. dollar and he Souh African rand, ERV is a measure of exchange rae volailiy, and ε is a whie-noise disurbance erm. Economic heory posis ha he real income level of he domesic counry s rading parners would have a posiive effec on he demand for is expors. Therefore, a priori, we would expec ha β >. On he 76

5 The Inernaional Journal of Business and Finance Research Volume 5 Number 3 oher hand, if he relaive price of expors rise (fall), domesic goods become less (more) compeiive han foreign goods, causing he demand for expors o fall (rise). Therefore, a priori, one would expec ha β, which measures he compeiiveness of U.S. expors relaive o Souh African domesic producion, is negaive. Similarly, if a real depreciaion of he U.S. dollar, refleced by a decrease in he RER, is o increase expor earnings of indusry i, we would expec an esimae of β 3 o be negaive. Of course, his will a he same ime imply ha he Souh African impor demand for commodiy i is elasic. If, however, he Souh African impor demand for commodiy i were inelasic, we would expec β 3 o be posiive. The las explanaory variable is a measure of exchange rae volailiy. Various measures of real ERV have been proposed in he lieraure. Some of hese measures include () he averages of absolue changes, () he sandard deviaions of he series, (3) he deviaions from he rend, (4) he squared residuals from he ARIMA or ARCH or GARCH processes, and (5) he moving sample sandard deviaion of he growh rae of he real exchange rae. Since he effecs of ERV on expors have been found o be empirically and heoreically ambiguous (Bredin, e al. 3), β 4 could be eiher posiive or negaive. Equaion () shows he long-run relaionships among he dependen and independen variables in our model. Given he recen advances in ime-series analysis, in esimaing he long-run model oulined by equaion (), i is now a common pracice o disinguish he shor-run effecs from he long-run effecs. For his purpose, equaion () should be specified in an error-correcion modeling (ECM) forma. This mehod had been used in many recen sudies including Bahmani-Oskooee and Hegery (9), Bahmani- Oskooee and Wang (8, 9), Bahmani-Oskooee and Mira (8), Bahmani-Oskooee and Kovyryalova (8), and Bahmani-Oskooee and Ardalani (6). According o Bahmani-Oskooee and Wang (8), such an approach is warraned given ha he measure of exchange rae volailiy is a saionary variable (see, for example, De Via and Abbo, 4; Bahmani-Oskooee & Payeseh, 993; and Doyle, ), whereas he oher variables in equaion () are non-saionary. Therefore, following Pesaran, Shin, and Smih () and heir mehod of bounds esing or he Auoregressive Disribued Lag (ARDL) approach o coinegraion analysis, we rewrie equaion () as an error-correcion model in equaion () below. ln X = α + + n i= n i= β ln X i i i ϕ ln ERV + i n i= γ lny i + λ ln X i + n i= + λ lny δ ln P i i + n i= + λ ln P η ln RER i i + λ ln RER 3 + λ ln ERV 4 + ω () Where is he difference operaor and he oher variables are as defined earlier. Pesaran, Shin, and Smih s () bounds esing approach o coinegraion is based on wo procedural seps. The firs sep involves using an F-es or Wald es o es for join significance of he no coinegraion hypohesis H : λ = λ = λ = λ3 = λ4 = agains an alernaive hypohesis of coinegraion, H : λ, λ, λ, λ, λ. This es is performed using equaion (). The advanage of his approach is ha here is no 3 4 need o es for uni roos, as is commonly done in coinegraion analysis. Pesaran, Shin, and Smih () provide wo ses of criical values for a given significance level wih and wihou ime rend. One assumes ha he variables are saionary a he levels or I(), and he oher assumes ha he variables are saionary a he firs difference or I(). If he compued F-values exceed he upper criical bounds value, hen H is rejeced signaling coinegraion among he independen variables. If he compued F-value is below he criical bounds values, we fail o rejec H. Finally, if he compued F-saisic falls wihin he boundary, he resul is inconclusive. Afer esablishing coinegraion, he second sep involves esimaion of he long-erm elasiciies and he error-correcion model. 77

6 E. M. Ekanayake, R. Thaver IJBFR Vol. 5 No. 3 DATA SOURCES AND VARIABLES Our expor daa ime series spans a -year period from January 99 hrough December 9, leading o 4 monhly observaions. Monhly daa on real expor volume and prices are aken from he Global Trade Informaion Services, World Trade Alas Daabase. Monhly daa on real expor volumes and prices have been convered ino expor volume indices and expor price indices wih 5 serving as he base (=). The sudy focuses on he op en expor commodiies defined a he -digi Harmonized Sysem (HS) codes level, and seleced based on heir average expor value beween 99 and 9. They are: Machinery (HS 84); Passenger Vehicles (HS 87); Aircraf and Spacecraf (HS 88); Elecrical Machinery (HS 85); Opical and Medical Insrumens (HS 9); Organic Chemicals (HS 9); Mineral Fuel and Oil (HS 7); Cereals (HS ); Plasic (HS 39); and Miscellaneous Chemical Producs (HS 38). The real income variable for Souh Africa is proxied by he indusrial producion index (5=) of Souh Africa. The underlying series is obained from he Inernaional Moneary Fund s Inernaional Financial Saisics daabase and from he Organizaion for Economic Cooperaion and Developmen s online daabase. The relaive price raio for U.S. expors is calculaed as he raio of he expor price index of each commodiy o he price level, proxied by he consumer price index (5=) of Souh Africa. The expor price index for each of he expor producs is compued using he uni prices aken from he Global Trade Informaion Services, World Trade Alas Daabase, while he consumer price index is also obained from he Inernaional Moneary Fund s Inernaional Financial Saisics daabase. Following Bahmani-Oskooee and Wang (8, 9), and Sekka and Varoudakis (), he real exchange rae, RER, is consruced as: US ER P RER = (3) SA P where RER is he real exchange rae, Saes and Souh Africa defined as number of rand per U.S. dollar a ime, ER is he bilaeral nominal exchange rae beween he Unied SA P is he consumer price US index (5=) of Souh Africa a ime, and P is he consumer price index (5=) of he U.S. a ime. The monhly daa on nominal exchange raes are aken from he IMF, Inernaional Financial Saisics daabase. Finally, we use hree alernaive measures of exchange rae volailiy in his sudy so we may es he sensiiviy of our resuls. I should be noed a his juncure ha here is no unique way o measure real exchange rae volailiy. The firs ERV measure is obained using he esimaed GARCH (,) model, which has also been used in recen sudies by, among ohers, Chowdhury and Wheeler (8), Choudhury (5), and Gheong, Mehari, and Williams (5). We make use of real as opposed o nominal exchange raes in our measuremen. As Choudhury (5) poins ou, unlike oher measures of ERV ha can poenially ignore informaion on he sochasic processes by which exchange raes are generaed, ARCHype models capure he ime-varying condiional variance as a parameer generaed from a ime-series model of he condiional mean and variance of he growh rae, and hus are very useful in describing volailiy clusering. The GARCH (,) model we esimae is based on an auoregressive model of order ( AR () difference of he real exchange rae and i akes he following form: ) of he firs 78

7 The Inernaional Journal of Business and Finance Research Volume 5 Number 3 ln RER = β o + β ln RER + β ln RER + e, where e ~ N(, u ) (4) u = + αe + αu α (5) The esimaed condiional variance ( u ) from Equaion (4) is used as our measure of ERV. Our second measure of volailiy is consruced following Bredin, Founas, and Murphy (3), Weliwia, Ekanayake, and Tsujii (999), Chowdhury (993), Lasrapes and Koray (99), and Koray and Lasrapes (989). Following hese auhors he real exchange rae volailiy measure is consruced as: m VOL = (ln RER + i ln RER + i ) (6) m i= where VOL is he volailiy of real exchange rae, RER is he real exchange rae and m = 4 is he order of he moving average. According o Koray and Lasrapes (989), his measure can capure general movemens in real exchange rae volailiy and exchange rae risk over ime. We also experimened wih a hird measure of volailiy. This alernaive measure of exchange rae volailiy is defined as he ime-varying welve-monh coefficien of variaion of he real exchange rae given by: m ( RER + i RER) m i CV = m = (7) + RER where R ER is he mean of he bilaeral real exchange rae beween monhs and +m-. EMPIRICAL RESULTS Applying he ARDL approach o coinegraion o monhly daa from January 99 o December 9, we assess he expors of he U.S. o Souh Africa for he op en expor producs. Firs, we esimae equaion (). Following Bahmani-Oskooee and Mira (8) we impose a maximum of four lags on each firs differenced variable and employ Akaike s Informaion Crierion (AIC) o selec he opimum lag lengh. Choosing a combinaion of lags ha minimizes he AIC, we hen es wheher he variables for each indusry are coinegraed. These resuls are shown in Table. Table reveals ha seven of he en indusries (HS84, HS87, HS88, HS9, HS9, HS, HS39) encompass an F-saisic above he upper bound of 3.79, implying ha hese indusries five variables are coinegraed. This resul is consisen across indusries for all hree volailiy measures. The oher hree indusries (HS7, HS38, HS85) reveal an F-saisic below he lower bound of.6, indicaing no coinegraion among variables. Therefore, only hose seven indusries ha exhibi coinegraing relaionships among variables are used o analyze he effecs of ERV on expors. We firs esimae Equaions (4) and (5) for his period, and he resuls are shown in Table. The coefficiens of α, α, and α are all posiive and α + α =.94 <. These resuls ensure ha condiional variance is sricly posiive, hus saisfying he necessary condiions of he ARCH model in Equaion (5). Our findings also demonsrae ha he esimaed coefficiens of e and u are saisically 79

8 E. M. Ekanayake, R. Thaver IJBFR Vol. 5 No. 3 significan a he % level, implying ha significan ARCH and GARCH effecs exis in he daa. The prediced value of Equaion (5) provides our firs measure of real exchange rae volailiy. Table : Coinegraion Tes Resuls of Top Ten Expor Commodiies from he U.S. o Souh Africa Indusry Volailiy Measure F ECM Coinegraed? HS 84: Machinery 3.83** -.47 (5.) Yes 4.47** (.74) Yes 3 3.8** -.54 (.5) Yes HS 87: Passenger Vehicles 6.8** -.4 (.85) Yes 5.65** -.48 (3.9) Yes 3 6.3** -.4 (3.8) Yes HS 88: Aircraf and Spacecraf 8.54** -.35 (4.5) Yes 5.9** -.84 (.66) Yes 3 7.8** -.49 (4.5) Yes HS 85: Elecrical Machinery (.97) No (.8) No (.59) No HS 9: Opical and Medical Insrumens 9.7** (3.7) Yes 9.** -.57 (3.79) Yes ** (3.75) Yes HS 9: Organic Chemicals 9.5** -.79 (4.) Yes 9.9** -.75 (4.7) Yes 3 8.** (4.) Yes HS 7: Mineral Fuel and Oil ec (5.33) No (5.93) No (5.) No HS : Cereals 4.4** -.5 (.8) Yes 6.56** -.69 (.87) Yes 3 5.5** -.56 (.43) Yes HS 39: Plasic 6.59** -.38 (5.) Yes 9.56** (.6) Yes 3 7.** -.57 (.6) Yes HS 38: Miscellaneous Chemical Producs (3.) No (.89) No (3.37) No Noe: This able summarizes he resuls of he bounds esing approach o coinegraion. The figures in parenheses are absolue value of he - saisic. ECM represens he error-correcion erm. Volailiy measures,, and 3 are defined earlier in equaions (4) and (5), (6), and (7), respecively. The upper bound criical value for he F-saisic wih unresriced inercep and no rend a he 5% level of significance is The lower bound criical value is.6. These values are aken from Pesaran, Shin, and Smih (, Table CI(iii) Case III, p. 3). ** indicaes he significance a he 5 percen level. Table : Esimaion of Real Exchange Rae Variance as a GARCH (, ) Process ln RER = ln + RER.6874 ln RER (.94) (.9) * (.84) u = e (.69) ** (3.45) ** Log L = N = u (7.3) ** Noe: This able shows he resuls of he firs measure of volailiy as defined by Equaions (4) and (5). The figures in parenheses are -saisics. * and ** indicae he saisical significance a he 5% and % level, respecively. The esimaed coefficiens for he seven coinegraed indusries are presened in Table 3. Following he sudies by Bahmani-Oskooee and Hegery (9), Bahmani-Oskooee and Wang (8, 9), Bahmani- Oskooee and Mira (8), Bahmani-Oskooee and Kovyryalova (8), and Bahmani-Oskooee and Ardalani (6), we repor only he shor-run volailiy coefficiens and all he long-run coefficiens. Shor-Run Effecs of Exchange Rae Volailiy: The shor-run esimaed coefficiens on ERV presened on he lef panel in Table 3 reveal a mixure of negaive and posiive signs regardless of he volailiy 8

9 The Inernaional Journal of Business and Finance Research Volume 5 Number 3 measure employed. There is also a significance variaion of he exchange rae volailiy on expors among indusries in he shor-run. The firs indusry, machinery, has a negaive and saisically significan coefficien regardless of he volailiy measure. The reason for his may be because of Souh Africa s abiliy o impor machinery from oher rading parners. Passenger vehicles indusry has a negaive bu saisically insignifican impac in he shor-run in all hree cases. The nex wo indusries, aircraf and spacecraf, and medical and opical insrumens, have posiive signs under all hree measures of volailiy. Each of he coefficiens is also saisically significan in all cases. The organic chemicals indusry has a posiive and saisically significan coefficien under he firs and hird measure of volailiy bu has a negaive and insignifican effec under he second measure. The las wo indusries, cereals and plasic, have mixed resuls. In general, he impac of ERV on expors for hese seven indusries is mixed in he shor-run. The U.S. dominaes hese indusries globally, and Souh Africa is impor dependen on hese producs, so even hough exchange raes have increased in volailiy since he 99 s, demand for hese goods have coninued. Tha cereals and plasics render mixed resuls may be due o Souh Africa s srong domesic producion in hese indusries. Long-Run Effecs of Exchange Rae Volailiy: The long-run coefficien esimaes are shown in he righ panel of Table 3. As economic heory posulaes, he real income variable renders a posiive sign in all cases, regardless of he volailiy measure. This coefficien is saisically significan in he majoriy of indusries including HS84, HS87, HS88, HS9, HS9, HS, and HS39; he coefficien for cereals (HS ) is insignifican under volailiy measures () and (3), while he coefficiens for plasic (HS 39) and cereals are insignifican under volailiy measure (). The relaive price variable displays he expeced negaive sign and is saisically significan a he % level in 9of he cases, and a he 5% level for machinery (HS 85) under volailiy measures () and (3). This resul is similar o hose of Bahmani- Oskooee and Mira (8), Bahmani-Oskooee and Kovyryalova (8), and Bahmani-Oskooee and Ardalani (6). The real exchange rae coefficien has a negaive sign in all cases and is saisically significan in he majoriy of cases, excep for machinery and passenger vehicles. Finally, he esimaed coefficiens on ERV show a mixure of negaive signs for machinery, passenger vehicles, opical and medical insrumens, organic chemicals, and plasic indusries and posiive signs for aircraf and spacecraf, and cereal, regardless of he volailiy measure used. Under volailiy measure (), five of he seven coefficiens are negaive and only hree coefficiens are saisically significan. Under volailiy measures () and (3), five of he seven coefficiens are negaive and four coefficiens are saisically significan a eiher he 5% or % levels. Thus, ERV has a negaive effec in five of he seven indusries presened in Table 3. They include machinery, passenger vehicles, opical and medical insrumens, organic chemicals, and plasic. Our findings are somewha similar o hose of Bahmani-Oskooee and Hegery (9) and Bahmani-Oskooee and Wang (8, 9). In general, in he long-run, ERV appears o have a negaive effec on he U.S. expors o Souh Africa. SUMMARY AND CONCLUSIONS In his paper we have examined he dynamic relaionship beween expors and exchange rae volailiy in Unied Saes' expors o Souh Africa in he conex of a mulivariae error-correcion model. Esimaes of he long-run expor demand funcions were obained by employing he bounds esing approach o coinegraion using monhly daa for he period January 99 - December 9. The coinegraion resuls clearly show ha here exiss a long-run equilibrium relaionship beween real expors and real foreign economic aciviy, relaive prices, real exchange rae, and real exchange rae volailiy, in seven of he en commodiies seleced. All he specificaions yielded expeced signs for he coefficiens. All our coefficiens are saisically significan eiher a he % or 5% levels. Of he seven 8

10 E. M. Ekanayake, R. Thaver IJBFR Vol. 5 No. 3 producs analyzed in deail, five of hem, namely, machinery, passenger vehicles, opical and medical insrumens, organic chemicals, and plasic, have negaive signs for he ERV variable indicaing ha ERV ends o deer expors of hese producs in he long-run. Table 3: Shor-Run and Long-Run Coefficien Esimaes Panel A: Volailiy Measure : ARCH Volailiy Measure as Defined in Equaions (4) and (5) Shor-Run Coefficien Esimaes Long-Run Coefficien Esimaes Indusry lnv lnv lnv lnv Consan 3 ln Y ln P ln RER ln V Machinery -.86* (.) ** (3.9) -.88** (3.43) -.37* (.73) -.68 (.3) Passenger Vehicles -.7 (.9) ** (5.33) -.88** (3.43) -.37* (.73) -. (.6) Aircraf & Spacecraf.38* (.) * (.8) -.75** (4.) -.37* (.73).34 (.5) Opical & Med. Ins..68* (.9).99** (3.).56.* (.3) -.533** (9.78) -.88** (5.46) -.76* (.96) Organic Chemicals.** (.93) * (.) -.953** (4.5) -.43 (.6) -.33** (.9) Cereals -.469* (.7) (.9) -.8** (9.3) -.39** (3.).759* (.3) Plasic -.55* (.) ** (3.88) -.96** (3.) -.587* (.4) -.38 (.53) Pane B: Volailiy Measure : Volailiy Measure as Defined in Equaion (6) Shor-Run Coefficien Esimaes Long-Run Coefficien Esimaes Indusry lnv lnv lnv lnv Consan 3 ln Y ln P ln RER ln V Machinery -.39* (.54) -.* (.58) ** (3.6) -.48* (.9) -.3 (.56) -.83* (7.95) Passenger Vehicles -. (.7) ** (5.7) -.788** (9.43) (.93) -.38 (.6) Aircraf & Spacecraf.64* (.4) * (.) -.7** (4.55) -.94* (.9).6 (.8) Opical & Med. Ins..38* (.3).46** (.8) * (.97) -.65** (9.47) -.77** (4.59) -.98** (3.35) Organic Chemicals -.7 (.58) * (.) -.48** (.99) -.4** (.96) -.374** (5.45) Cereals -. (.6) (.43) ** (9.38) -.64* (.3).44 (.) Plasic.54* (.58).35 (.85).3* (.) (.97) -.49** (5.) -.943** (.5) -.37** (6.46) Panel C: Volailiy Measure 3: Volailiy Measure as Defined in Equaion (7) Shor-Run Coefficien Esimaes Long-Run Coefficien Esimaes Indusry lnv lnv lnv lnv Consan 3 ln Y ln P ln RER ln V Machinery -.38* (.54) -. (.73) ** (3.6) -.477* (.9) (.5) -.858** (6.48) Passenger Vehicles -.86 (.45) ** (5.7) -.788** (9.43) (.94) -.38 (.8) Aircraf & Spacecraf.55* (.8).3* (.44) * (.) -.7** (4.55) -.94* (.8).6 (.7) Opical & Med. Ins..** (3.39) * (.97) -.65** (9.47) -.77** (4.59) -.98** (3.35) Organic Chemicals.3** (.78).9.63* (.8) -.48** (.99) -.4** (.97) -.374** (5.45) Cereals.58 (.9) (.4) ** (9.38) -.64* (.3).44 (.) Plasic.67 (.86) * (.) -.49** (5.) -.943** (.5) -.37** (4.59) Noe: This able summarizes he resuls obained using he ARDL model defined in Equaion (). The figures in parenheses are absolue value of -saisic. ** and * indicae he saisical significance a he % and 5% levels, respecively. The shor-run dynamics also indicae ha, in general, he impac of ERV on expors in hese seven indusries is mixed in he shor-run. These resuls poin ou o he decreasing compeiiveness of U.S. expors in he global economy despie he depreciaing value of he dollar over ime. I underscores he 8

11 The Inernaional Journal of Business and Finance Research Volume 5 Number 3 degree o which a developing counry such as Souh Africa has succeeded in finding alernaive markes in Europe and especially in Asia in he las decade. REFERENCES Arisoelous, K. (). Exchange Rae Volailiy, Exchange Rae Regime, and Trade Volume: Evidence from UK-US Expor Funcion ( ). Economic Leers, 7(), Arize, A. C. (998). The Effecs of Exchange-Rae Volailiy on U.S. Impors: An Empirical Invesigaion. Inernaional Economic Journal, (), 3-4. Arize, A. C. (996). The Impac of Exchange-Rae Uncerainy on U.S. Expor Growh: Evidence from Korean Daa. Inernaional Economic Journal, (3), Arize, A. C. (995). The Effecs of Exchange-Rae Volailiy on U.S. Expors: An Empirical Invesigaion. Souhern Economic Journal, 6(), Arize, A. C., Osang, T. and D. J. Sloje (). The Exchange Rae Volailiy and Foreign Trade: Evidence from Thireen LDCs. Journal of Business and Economic Sudies, 8, -7. Asseery, A. and D. A. Peel (99). The Effecs of Exchange Rae Volailiy on Expors: Some New Esimaes. Economic Leers, 37(3), Bahmani-Oskooee, M. and S. W. Hegery (9). The Effecs of Exchange-Rae Volailiy on Commodiy Trade beween he Unied Saes and Mexico. Souhern Economic Journal, 75(4), Bahmani-Oskooee, M. and M. Kovyryalova (8). Impac of Exchange Rae Uncerainy on Trade Flows: Evidence from Commodiy Trade beween he Unied Saes and he Unied Kingdom. The World Economy, 3(8), Bahmani-Oskooee, M. and R. Mira (8). Exchange Rae Risk and Commodiy Trade beween he U.S. and India. Open Economies Review, 9(), 7 8. Bahmani-Oskooee, M. and Y. Wang (9). Exchange Rae Sensiiviy of Ausralia/s Trade Flows: Evidence from Indusry Daa. The Mancheser School, 77(), -6. Bahmani-Oskooee, M. and Y. Wang (8). Impac of Exchange Rae Uncerainy on Commodiy Trade beween he U.S. and Ausralia. Ausralian Economies Papers, 47(3), Bahmani-Oskooee, M., and Z. Ardalani (6). Exchange Rae Sensiiviy of U.S. Trade Flows: Evidence from Indusry Daa. Souhern Economic Journal, 7, Bahmani-Oskooee, M. (). Does Black Marke Exchange Rae Volailiy Deer he Trade Flows? Iranian Experience. Applied Economics, 34, Bahmani-Oskooee, M. (99). Exchange Rae Uncerainy and Trade Flows of Developing Counries. The Journal of Developing Areas, 5, Bahmani-Oskooee, M. and S. Payeseh (993). Does Exchange Rae Volailiy Deer Trade Volume in LDCs. Journal of Economic Developmen, 8(),

12 E. M. Ekanayake, R. Thaver IJBFR Vol. 5 No. 3 Bredin, D., Founas, S. and E. Murphy (3). An Empirical Analysis of Shor-Run and Long-Run Irish Expor Funcions: Does Exchange Rae Volailiy Maer?. Inernaional Review of Applied Economics, 7(), Byrne, J., Darby, J. and R. MacDonald (8). US Trade and Exchange Rae Volailiy: A Real Secoral Bilaeral Analysis. Journal of Macroeconomics, 3, Choudhury, T. (5). Exchange Rae Volailiy and he Unied Saes Expors: Evidence from Canada and Japan. Journal of Japanese and Inernaional Economies, 9, 5-7. Chowdhury, A. R. (993). Condiional Exchange Rae Volailiy Depress Trade Flows? Evidence from Error-Correcion Models. Review of Economic and Saisics, 75, Chowdhury, A. R. and M. Wheeler (8). Does Exchange Rae Volailiy Affec Foreign Direc Invesmen? Evidence from Four Economies. The Inernaional Trade Journal, (), Chou, W. L. (). Exchange Rae Variabiliy and China s Expors. Journal of Comparaive Economics, 8(), De Via, G. and A. Abbo (4). Real Exchange Rae Volailiy and US Expors: An ARDL Bounds Tesing Approach. Economic Issues, 9(), Doyle, E. (). Exchange Rae Volailiy and Irish-UK Trade, Applied Economics, 33, Gheong, C., Mehari, T. and L. V. Williams (5). The Effecs of Exchange Rae Variabiliy on Price Compeiiveness and Trade Volumes in he U.K.: A Disaggregaed Approach. Journal of Policy Modeling, 7(8), Global Trade Informaion Sysems, World Trade Alas Daabase. Hooper, P. and Kohlhagen, S. (978). Real Exchange Rae Uncerainy on he Prices and Volume of Inernaional Trade. Journal of Inernaional Economics, 8, Inernaional Moneary Fund, Inernaional Financial Saisics Daabase. Klein, M. W. (99). Secoral Effecs of Exchange Rae Volailiy on Unied Saes Expors. Journal of Inernaional Money and Finance, 9(3), Koray, F. and W. D. Lasrapes (989). Real Exchange Rae Volailiy and he US. Bilaeral Trade: A VAR Approach. Review of Economics and Saisics, 7(), Kroner, K. F. and W. D. Lasrapes (993). The Impac of Exchange Rae Volailiy on Inernaional Trade: Reduce form Esimaes using he GARCH-in-Mean Model. Journal of Inernaional Money and Finance, (6), Lasrapes, W. D. and F. Koray (99). Exchange Rae Volailiy and U.S. Mulilaeral Trade Flows. Journal of Macroeconomics, (), McKenzie, M. D. and R. D. Brooks (997). The Impac of Exchange Rae Volailiy on German-US Trade Flows. Journal of Inernaional Financial Markes, Insiuions and Money, 7(),

13 The Inernaional Journal of Business and Finance Research Volume 5 Number 3 Onafowora, O. A. and O. Owoye (8). Exchange Rae Volailiy and Expor Growh in Nigeria. Applied Economics, 4(), Ozurk, I. and H. Kalyonku (9). Exchange Rae Volailiy and Trade: An Empirical Invesigaion from Cross-Counry Comparisons. African Developmen Review, (3), Pesaran, M., Shin, Y., and R. Smih (). Bounds Tesing Approaches To The Analysis Of Level Relaionships. Journal of Applied Economerics, 6(3), Qian, Y. and P. Varangis (994). Does Exchange-Rae Volailiy Hinder Expor Growh?: Addiional Evidence. Empirical Economics, 98, Weliwia, A., Ekanayake, E. M. and H. Tsujii (999). Real Exchange Rae Volailiy and Sri Lanka's Expors o he Developed Counries, Journal of Economic Developmen, 4(), BIOGRAPHY Dr. E. M. Ekanayake is an Associae Professor of Economics a Behune-Cookman Universiy, Dayona Beach, Florida, USA and an Adjunc Professor of Economics a Embry-Riddle Aeronauical Universiy, Dayona Beach. He earned his Ph.D. in Economics a he Florida Inernaional Universiy, Miami in 996. He has many publicaions o his credi. Conac informaion: School of Business, Behune- Cookman Universiy, 64 Dr. Mary McLeod Behune Blvd., Dayona Beach, FL 34, USA. Phone: (386) ekanayakee@cookman.edu. Dr. Ranjini Thaver is an Associae Professor of Economics a Seson Universiy, Deland, Florida. She earned her Ph.D. in Economics in 995 and an M.A. degree in Economics in 989 a he Nore Dame Universiy. She is currenly he Chair of Economics Deparmen a Seson Universiy. Dr. Thaver is also he Direcor of Cener for Holisic Microcredi Iniiaives (CHOMI), and he Direcor of CHOMI Tanzania. Conac informaion: Deparmen of Economics, Seson Universiy, Box 839, Deland, FL 373, USA. Phone: (386) rhaver@seson.edu. 85