THE EFFECTS OF EXCHANGE RATE VOLATILITY ON SOUTH AFRICA S TRADE WITH THE EUROPEAN UNION

Transcription

1 he Inernaonal Journal of Busness and Fnance Research Volume 6 Number HE EFFECS OF EXCHANGE RAE VOLAILIY ON SOUH AFRICA S RADE WIH HE EUROPEAN UNION E. M. Ekanayake, Behune-Cookman Unversy Ranjn L. haver, Seson Unversy Danel Plane, Seson Unversy ABSRAC In hs paper we analyze he effecs of he real exchange rae volaly on Souh Afrca s rade flows wh he European Unon over he perod 980 o Our sudy uses quarerly rade flows on Souh Afrca s expors and mpors and ulzes he bounds esng approach o conegraon, and error-correcon model. Our resuls reveal ha mpors depend posvely on he levels of domesc economc acvy and foregn exchange reserves bu negavely on relave prces and exchange rae volaly. In addon, expors depend posvely on he levels of foregn economc acvy bu negavely on relave prces and exchange rae volaly. Furhermore, he exchange volaly exers mxed effecs n he shor-run and n he long-run. JEL: F4, F3 KEYWORDS: Souh Afrca, mpors, expors, exchange rae volaly, panel conegraon INRODUCION Despe he szeable number of sudes conduced, no real consensus abou he mpac of exchange rae volaly on expors has emerged. Whle a large number of sudes fnd ha exchange rae volaly ends o reduce he level of rade, ohers fnd eher weak or nsgnfcan or posve relaonshps. For example, Onafowara and Owoye (2008), Byrne, Darby, and MacDonald (2008), Choudhry (2005), Bahmanee-Oskooee (2002), Arze, e al. (2000), Arze (995), Chowdhury (993), Pozo (992), and Bahman-Oskooee and Lafa (992), fnd evdence for negave effecs. Accordng o hese scholars, exchange rae volaly may affec expors drecly hrough uncerany and adjusmen coss for rsk-averse exporng nvesors. Furher, may have an ndrec effec hrough s mpac on he srucure of oupu, nvesmen and governmen polcy. On he oher hand, Doyle (200), Chou (2000), McKenze and Brooks (997), Qan and Varangs (994), Kroner and Lasrapes (993), and Asseery and Peel (99) fnd evdence for a posve effec for volaly on expor volumes of some developed counres because exchange rae volaly makes exporng more aracve o rsk-oleran exporng frms. However, oher scholars such as Arsoelous (200), Bahman-Oskooee and Payesch (993), Bahman-Oskooee (99), and Hooper and Kohlhagen (978) have repored no sgnfcan relaonshp beween exchange rae volaly and expors. Reasons for conradcory resuls by dfferen sudes may be due o a varey of facors, among hem: dfferen mehods used o measure exchange rae volaly; he use of dfferen prce deflaors; he dfferenal use of sample daa, for example, he use of aggregae expor daa versus secoral expor daa; dfferen me-frame perods; gnorng mpor dependency on nermedae and capal goods of he recevng counry, as s he case wh many developng counres; and he absence of complex economerc mehods for sudyng hese varaons. As a resul scholars sopped nvesgang he exchange rae volaly-expor nexus by he lae 990 s. However, wh beer access o secoral daa and he developmen of more sophscaed economerc models, recen sudes have begun evaluang he exchange rae volaly-expor connecon from a secoral perspecve. he raonale behnd hs s ha 3

2 E. M. Ekanayake e al IJBFR Vol. 6 No dfferen rade secors would be mpaced dfferenally by exchange rae volaly, and herefore may be more revealng han aggregae sudes. hs sudy focuses on aggregae rade flows beween Souh Afrca, a developng counry, and he member counres of he European Unon (EU) o uncover he naure and sensvy of he relaonshp beween exchange rae volaly and rade flows. We use he recenly developed panel un roo and panel conegraon echnques for hs purpose. Usng hs approach we nvesgae he effecs of exchange rae volaly on Souh Afrca's expors o and mpors from he EU counres over a perod of 30 years usng quarerly daa from 980:Q o 2009:Q4. o hs end we provde a bref revew of he leraure n he nex secon. hereafer, we lay he emprcal framework of our sudy by specfyng our model. In he secon followng ha we dscuss varable defnons and oulne our daa sources. Emprcal resuls from he bounds esng approach o conegraon, and error-correcon model esmaes are presened n he penulmae secon. he fnal secon presens a summary and concluson of he resuls obaned n hs sudy. LIERAURE REVIEW In hs secon we presen a bref overvew of sudes ha examne he exchange rae volaly-rade nexus. We begn by dscussng he mos recen and sophscaed sudes, employng conegraon echnques and error-correcon models, o older, less complex sudes. Bahman-Oskooee and Hegery (2009) nvesgae he effecs of exchange rae flucuaons on rade flows beween he U.S. and Mexco usng dsaggregaed, ndusry-level annual expor and mpor daa for 02 ndusres from 962 o hey analyze boh he shor- and long-erm effecs of volaly n he peso/dollar real exchange rae on Mexcan-Uned Saes rade. hey conclude ha n he shor-erm ncreased volaly negavely affecs rade flows n mos ndusres. Long-erm effecs however, are sgnfcan for only one-hrd of he ndusres suded, and of hs, only wo-hrds are negave. hey speculae ha ncreased Mexcan negraon and lberalzaon of economc polces allow for greaer adjusmens n he long-erm so ha volaly s less of a problem n he long-erm han n he shor-erm. Byrne, Darby, and MacDonald (2008) analyze he mpac of exchange rae volaly on he volume of blaeral U.S. rade flows usng homogenzed and dfferenaed secoral annual daa over he perod for a cross-secon of 6 EU counres and 22 ndusres. her sudy fnds ha cluserng all ndusres ogeher provdes evdence of a negave effec on rade from exchange rae volaly, whch confrms fndngs of oher sudes usng aggregae daa. However, when nvesgang secoral rade dfferences, he effecs of exchange rae volaly on rade s negave and sgnfcan for dfferenaed goods and nsgnfcan for homogeneous goods, confrmng recen sudes ha secoral dfferences are n fac crucal o explanng he dfferenal mpac of volaly on rade. hey sugges ha a greaer degree of dsaggregaon a he ndusry level may provde more worhwhle resuls, whch s wha we do n hs sudy. Bahman-Oskooee and Kovyryalova (2008) nvesgae he effec of exchange rae flucuaons on rade flows beween he U.S. and he Uned Kngdom usng dsaggregraed annual expor and mpor daa for 77 commodes ndusres from 97 o hey analyze boh he shor- and long-erm effecs of real exchange rae volaly on rade beween he U.S. and he UK. her resuls reveal ha he volaly of he real dollar pound rae has a shor-erm sgnfcan effec on mpors of 09 ndusres and on expors of 99 ndusres. In mos cases, such effecs are unfavorable. In he long run, however, he number of sgnfcan cases s somewha reduced: only 62 mpor and 86 expor ndusres are sgnfcanly and adversely affeced by exchange rae volaly. he ndusres affeced nvolve boh durable and nondurable goods, and nclude small as well as large ndusres, supporng fndngs by aggregae sudes. 4

3 he Inernaonal Journal of Busness and Fnance Research Volume 6 Number In anoher sudy, Bahman-Oskooee and Mra (2008), nvesgae he effecs of exchange rae volaly on rade flows beween he U.S. and Inda, an emergng economy. Usng annual daa from 40 ndusres from , her resuls demonsrae ha exchange rae volaly has more shor-run han long-run effecs. In he shor-run, 7 ndusres were affeced on he mpor sde and 5 on he expor sde. he ndusres affeced show Inda s ncreasng ably o produce mpor subsuable goods. However, n he long run, only a few ndusres are affeced because he ncreasng dependence on rade beween Inda and he US cause ndusres o respond nelascally o exchange rae volaly. Usng boh he nomnal and he real exchange rae beween he Uned Saes dollar and he currences of Canada and Japan, Choudhury (2005) nvesgaes he nfluence of exchange rae volaly on U.S. real expors o Canada and Japan usng aggregae monhly daa rangng from January 974 o December 998. he sudy uses condonal varance from he GARCH (, ) model as a measure of exchange rae volaly, and fnds sgnfcan and mosly negave effecs of exchange rae volaly on real expors. As n he above sudes, Sukar and Hassan (200) nvesgae he relaonshp beween U.S. rade volume and exchange rae volaly usng conegraon and error-correcon models. her sudy uses quarerly aggregae daa coverng he perod 975Q 993Q2 and a GARCH model o measure he exchange rae volaly. Parallelng oher sudes, he auhors fnd evdence for a sgnfcanly negave relaonshp beween U.S. expor volume and exchange rae volaly. However, unlke oher fndngs, hey reveal ha he shor-run dynamcs of he exchange rae volaly -rade relaonshp s nsgnfcan. hey argue ha hs resul may be due o he exsence of avenues for hedgng agans exchange rsks so as o neuralze he negave mpac of exchange rae volaly. Oher scholars argue ha hs shor-run nsgnfcan relaonshp may be because of he nvesgaors use of aggregae daa, whch gnores secoral dfferences. For example, whle one secor may exhb a negave relaonshp, anoher may exhb an equal bu oppose effec so ha hey offse each oher. Arze (995), usng monhly seres from February 978 o June 986 analyzes he effecs of real exchange rae volaly on he proporons of blaeral expors of nne caegores of goods from he U.S. o seven major ndusral counres. he volaly measure employed s he sandard devaon of he monhly percenage change n he blaeral exchange rae beween he U.S. and he mporng counry from o - 2. he sudy reveals dfferenal effecs of exchange rae volaly across dfferen caegores of expors. he sudy also concludes ha exchange rae uncerany has a negave effec on U.S. real expors, and ha may have a major mpac on he allocaon of resources o dfferen ndusres dependng on rade elasces. Lasrapes and Koray (990) analyze he nerrelaonshps among exchange rae volaly, nernaonal rade, and macroeconomc varables usng he vecor auoregresson (VAR) model. he model esmaes U.S. mullaeral rade from 973 o 990 and ncludes a movng sandard devaon measure of real exchange rae volaly. Whle he resuls reveal some evdence of a sascally sgnfcan relaonshp beween volaly and rade, he movng average represenaon of he model mples a raher small quanave effec. he sudy concludes ha exchange rae volaly s nfluenced by he sae of he economy, a facor gnored n a varey of oher sudes. Klen (990) s one of he frs few scholars o analyze he effecs of exchange rae volaly on he proporon of dsaggregaed blaeral expors of nne caegores of goods from he U.S. o seven major ndusral counres usng fxed effecs framework. Usng monhly seres daa from February 978 o June 986, he sudy reveals ha n sx caegores of expors exchange rae volaly sgnfcanly affecs he volume of expors and n fve of hese caegores he effec s posve, suggesng ha real exchange rae volaly may n fac ncrease expors by rsk-akng frms. 5

4 E. M. Ekanayake e al IJBFR Vol. 6 No Koray and Lasrapes (989) examne he relaonshp beween real exchange rae volaly and blaeral mpors from fve counres, namely, he UK, France, Germany, Japan, and Canada, employng a VAR model. he sudy uses aggregae monhly daa over a 7-year perod from January 959 o December 985, and ess for dfferen effecs durng boh he fxed and he flexble exchange rae regmes. Resuls sugges ha whle he effecs of volaly on mpors s weak, permanen shocks o volaly experence a negave mpac on mpors. However, hose effecs are relavely more mporan durng he flexble-rae han he fxed-rae perod. Fnally, Cushman (988) ess for real exchange rae volaly on U.S. blaeral rade flows usng annual daa from o sudy he effecs of he floang exchange rae regme on exchange rae volaly. he sudy fnds evdence for sgnfcan negave effecs n only wo of sx U.S. expor flows wh one expor flow showng a sgnfcan posve effec, confrmng oher sudes of a weak rsk-averse effec of exchange rae volaly on exporng frms. One major problem wh mos of he sudes above s ha he sample perod ncludes he perod pror o he end of he fxed exchange regme, so resuls may nclude he lag effecs of fxed exchange raes on rade before 973 lngerng on durng he ranson perod afer he mplemenaon of he floang exchange rae regme. he curren sudy correcs for hs poenal bas by usng Souh Afrcan quarerly aggregaed rade daa coverng a 30-year perod from 980:Q o 2009:Q4. he mehodology used n hs sudy ncorporaes many of he recen developmens n he leraure, namely, panel un roos and panel conegraon and error-correcon models, whch may uncover he naure and sensvy of he exchange rae volaly-rade nexus. MEHODOLOGY Model Specfcaon he objecve of hs sudy s o assess he effecs of exchange rae volaly on he aggregaed expors and mpors flows beween Souh Afrca and he member counres of he European Unon. Drawng on he exsng emprcal leraure n hs area, we specfy ha a sandard long-run expor demand funcon ha may ake he followng form (see, for example, Ozurk and Kalyonku, 2009; Choudhry, 2005; Arze, 998, 996, 995; and Asseery and Peel, 99): ln X β D + ε () = 0 + β lny + β2 ln RP + β3 ln RER + β4 lnvol + β5 where X s real expor volume of Souh Afrca n perod, Y s he real ncome of foregn counry n perod, RP s he relave prce of expors n perod, RER s he real exchange rae beween he Souh Afrcan rand and he local currency of a parner counry n perod, VOL s a measure of exchange rae volaly beween he Souh Afrcan rand and he local currency of a parner counry n perod, D s a dummy varable represenng he aparhed era ( ) n Souh Afrca, and ε s he error erm bounded wh classcal sascal properes. Economc heory poss ha he real ncome level of he domesc counry s radng parners would have a posve effec on he demand for s expors. herefore, a pror, we would expec ha β > 0. On he oher hand, f he relave prce of expors rse (fall), domesc goods become less (more) compeve han foregn goods, causng he demand for expors o fall (rse). herefore, a pror, one would expec ha β 2, whch measures he compeveness of Souh Afrcan expors relave o he gven counry's domesc producon, s negave. Smlarly, f a real deprecaon of he Souh Afrcan rand, refleced by a decrease n he RER, s o ncrease expor earnngs of ndusry, we would expec an esmae of β 3 o be negave. 6

5 he Inernaonal Journal of Busness and Fnance Research Volume 6 Number he las explanaory varable s a measure of exchange rae volaly. Varous measures of real exchange rae volaly have been proposed n he leraure. Some of hese measures nclude () he averages of absolue changes, (2) he sandard devaons of he seres, (3) he devaons from he rend, (4) he squared resduals from he ARIMA or ARCH or GARCH processes, and (5) he movng sample sandard devaon of he growh rae of he real exchange rae. Snce he effecs of exchange rae volaly on expors have been found o be emprcally and heorecally ambguous (Bredn, e al. 2003), β 4 could be eher posve or negave. Souh Afrca s assumed o be a small developng open-economy, renderng a prce-aker wh respec o mpors, and herefore suppors he use of sngle-equaon echnques for esmang he aggregae mpor demand funcon. We furher assume ha only normal goods are mpored, and ha as a developng counry, real foregn exchange reserves comprses an mporan varable n he funcon. Furher, he aparhed srucure and nernaonal sancons have boh had a sgnfcan mpac on aggregae mpor demand and are herefore ncluded n he model. he long-run aggregae mpor demand funcon for Souh Afrca (n naural logs) s hus specfed as: ln M α D + ε (2) = 0 + α lny + α 2 ln RP + α3 ln FR + α 4 lnvol + α5 where M s real mpor volume of Souh Afrca n perod, Y s he real ncome of Souh Afrca n perod, RP s he relave prce of mpors n perod, FR s he real foregn exchange reserves of Souh Afrca n perod, VOL s a measure of exchange rae volaly beween he Souh Afrcan rand and he local currency of a parner counry n perod, D s a dummy varable represenng he aparhed era ( ) n Souh Afrca, and ε s a whe-nose dsurbance erm bounded wh classcal sascal properes. he frs explanaory varable, Y, n he specfed model measures real domesc economc acvy, or he real GDP of Souh Afrca. Economc heory suggess ha ncome n he mporng counry s a major deermnan of a counry s mpors and has a posve mpac. hus, a pror, s expeced ha α > 0. he second explanaory varable, he relave prce of mpors, s calculaed as he rao of mpor prce o domesc prce. Economc heory poss ha an ncrease n he relave prce of mpors dscourages mpors and herefore α 2 s expeced o be negave. he hrd explanaory varable s a measure of avalably of foregn exchange, whch can be used o represen he ably o mpor. Followng Hoque and Yusop (200), we have ncluded he real foregn exchange reserve varable o capure he mpac of expor earnngs on mpor demand, as expor earnng s one of he major sources of foregn reserves. hs varable does no appear n he radonal mpor demand funcon. However, s an mporan deermnan of mpors for developng counres. Snce hgher real foregn reserves end o encourage mpors, we would expec ha α 3 > 0. he expeced sgns of α, α 2, and α 3 are borne ou n emprcal resuls by Hoque and Yusup (200), Aknlo (2008), Agbola and Damoense (2005), Narayan and Narayan (2005), Razafmahefa and Hamor (2005), Dua and Ahmed (2004), sonas and Chrsopoulos (2004), ang (2004, 2002a), Masubayash and Hamor (2003), Gumede (2000), Senhadj (998), and Mwega (993), among ohers. Snce he effecs of exchange rae volaly on mpors have been found o be emprcally and heorecally ambguous, α 4 could be eher posve or negave. Equaons () and (2) shows he long-run relaonshps among he dependen and ndependen varables n our models. hey can be represened whn a panel seng by ncorporang a subscrp "" depcng each of he European Unon counres n he sample. hs can be represened as follows: 7

6 E. M. Ekanayake e al IJBFR Vol. 6 No ln X, β 0 + β lny, + β2 ln RP, + β3 ln RER, + β4 lnvol, + ε, = (3) ln M, α 0 + α lny, + α 2 ln RP, + α3 ln FR, + α 4 lnvol, + ε, = (4) Panel un roo ess Before proceedng o conegraon echnques, we need o verfy ha all of he varables are negraed o he same order. In dong so, we have used he panel un roos es proposed by Breung (2005) and Breung and Pesaran (2008). Panel Conegraon ess We nvesgae he exsence of conegrang relaonshp usng he sandard panel ess for no conegraon proposed by Pedron (999, 2004). hese ess allow for heerogeney n he nerceps and slopes of he conegrang equaon. Pedron s ess provde seven es sascs: Whn dmenson (panel ess): () Panel ν -sasc; (2) Panel Phllps Perron ype ρ-sascs; (3) Panel Phllps Perron ype - sasc; and (4) Panel augmened Dckey Fuller (ADF) ype -sasc. Beween dmenson (group ess): (5) Group Phllps Perron ype ρ-sascs; (6) Group Phllps Perron ype -sasc; and (7) Group ADF ype -sasc. hese sascs are based on averages of he ndvdual auoregressve coeffcens assocaed wh he un roo ess of he resduals for each counry n he panel. All seven ess are dsrbued asympocally as sandard normal. Followng Pedron (999, 2004), he heerogeneous panel and heerogeneous group mean panel of rho (ρ), paramerc (ADF), and nonparamerc (PP) sascs are calculaed as follows: Panel ν - sasc: 2 2 Zν = Lˆ ê (5a) = = Panel ρ - sasc: ê Lˆ (ê ê λˆ ) = = Zρ = Lˆ (5b) = = Panel ADF - sasc: 2 * * * Z = ~ sn Lˆ ê* Lˆ ê ê (5c) = = = = Panel PP - sasc: Z pp = σ~ N Lˆ ê Lˆ (ê ê λˆ ) (5d) = = = = Group ρ - sasc: 8

7 he Inernaonal Journal of Busness and Fnance Research Volume 6 Number (ê ê λˆ ) = Z ~ ρ = ê (5e) = = Group ADF - sasc: 2 *2 *2 * * Z ~ = s ê ê ê (5f) = = = Panel PP - sasc: Z ~ pp = σˆ ê (ê ê λˆ ) (5g) = = = where j ŝ μ ; σˆ = ŝ + 2λˆ ; σ~ N k ~ *2 *2 s = ŝ N ; and j Lˆ = ηˆ + ηˆ ηˆ, j. = = j = k + = j + k 2 λˆ = μˆ μˆ, j ; = j = k + = j + = 2 N N 2 2 *2 *2 = Lˆ σˆ ; ŝ = μˆ ; N = = * μˆ,, μˆ, k = ρˆ εˆ, + μˆ ; εˆ ρˆ εˆ, + ρˆ k εˆ, j + μˆ j = he error erms εˆ, and ηˆ, are respecvely derved from he followng auxlary regressons: = ; and y = γˆ m xm + ηˆ. Of he seven es sascs, excep for he panel ν - sasc, he oher sx Pedron es sascs are lefaled ess. In order o fnd evdence for long-run relaonshp beween he varables, he null hypohess of no conegraon for hese ess should be rejeced. If he null hypohess canno be rejeced, here s no long-run relaonshp beween he varables. DAA SOURCES AND VARIABLES Our expor and mpor daa me seres span a 30-year perod from 980:Q hrough 2009:Q4, leadng o 20 quarerly observaons. Of he 27 member counres of he European Unon, only 20 counres were used n he sudy due o he unavalably of all relevan daa. Bulgara, Cyprus, Esona, Lava, Lhuana, Mala, and Slovena were dropped from he sample, and he res of he followng counres were used n he sudy: Ausra, Belgum, Czech Republc, Denmark, Fnland, France, Germany, Greece, Hungary, Ireland, Ialy, Luxembourg, Neherlands, Poland, Porugal, Romana, Slovaka, Span, Sweden, and he Uned Kngdom. In addon, whle he majory of he counres had daa for he enre sudy perod, he followng fve counres had daa only for he perod specfed: Czech Republc, 993:Q- 2009:Q4; Hungary, 995:Q-2009:Q4; Ireland, 997:Q-2009:Q4; Poland, 99:Q-2009:Q4; and Slovaka, 993:Q-2009:Q4. Quarerly daa on expors and mpors are aken from he Inernaonal Moneary Fund, Drecon of rade Sascs Daabase. Quarerly nomnal daa on expors were convered no real expor seres by deflang hem usng he expor prce ndex wh 2005 servng as he M m= 9

8 E. M. Ekanayake e al IJBFR Vol. 6 No base (=00). he expor prce ndex seres are aken from he Inernaonal Moneary Fund, Inernaonal Fnancal Sascs Daabase. he real foregn ncome varable for EU counres s he real quarerly GDP. he underlyng seres s obaned from he Organzaon for Economc Cooperaon and Developmen s onlne daabase. he real ncome varable for Souh Afrca s he real quarerly GDP of Souh Afrca. he underlyng seres s obaned from he Inernaonal Moneary Fund s Inernaonal Fnancal Sascs daabase. he daa on nomnal mpors, he mpor prce ndex, foregn exchange reserves seres, and domesc prce ndex are aken from he Inernaonal Moneary Fund s Inernaonal Fnancal Sascs daabase. Nomnal quarerly mpors n local currency are deflaed by he mpor prce ndex (2005 = 00) o oban he real mpor varable for Souh Afrca. he real GDP varable s compued n mllons of 2005 consan Rand. he relave prce of mpors seres s consruced as he rao of he mpor prce ndex (2005=00) o domesc prce ndex, as measured by he consumer prce ndex (CPI) (2005=00). o oban he real foregn reserves seres, we deflaed he nomnal foregn exchange reserves seres by he CPI (2005=00). Followng Bahman-Oskooee and Wang (2008, 2009), and Sekka and Varoudaks (2000), he real exchange rae, RER, s consruced as: f ER P RER = (6) SA P where RER s he real exchange rae, and a gven OECD counry a me, counry a me, and ER s he blaeral nomnal exchange rae beween Souh Afrca P s he consumer prce ndex (2005=00) of a gven OECD f P SA s he consumer prce ndex (2005=00) of Souh Afrca a me. he quarerly daa on nomnal exchange raes are aken from he IMF, Inernaonal Fnancal Sascs daabase. Fnally, we use a commonly used measure of exchange rae volaly n hs sudy, hough here are several alernave measures are avalable. I should be noed a hs juncure ha here s no unque way o measure real exchange rae volaly. Our measure of volaly s consruced followng Bredn, Founas, and Murphy (2003), Welwa, Ekanayake, and suj (999), Chowdhury (993), Lasrapes and Koray (990), and Koray and Lasrapes (989). Followng hese auhors he real exchange rae volaly measure s consruced as: 2 m 2 VOL = (ln RER + ln RER + 2) (7) m = where VOL s he volaly of real exchange rae, RER s he real exchange rae and m = 4 s he order of he movng average. Accordng o Koray and Lasrapes (989), hs measure can capure general movemens n real exchange rae volaly and exchange rae rsk over me. 20

9 he Inernaonal Journal of Busness and Fnance Research Volume 6 Number EMPIRICAL RESULS Panel Un Roo ess able shows he summary sascs of he man varables for he full sample. he sarng pon of our economerc analyss s o check wheher he varables ncluded n Equaons (3) and (4) conan panel un roos. In oher words, n Equaon (3), we need o check wheher [X, Y, RP, RER, VOL] conans a un roo and also wheher [M, Y, RP, FR, VOL] conans a un roo n Equaon (4). Whle here are several panel un roo ess are avalable, hs sudy uses he un roo es developed by Breung (2000, 2005). hese resuls of he expors model and he mpors model are presened n able 2. able : Basc Summary Sascs Expors Model Impors Model X Y RP RER VOL M Y RP FR VOL Mean Medan Maxmum Mnmum Sd. Devaon Skewness Kuross Jarque-Bera Probably Observaons 2,400 2,400 2,400 2,400 2,400 2,400 2,400 2,400 2,400 2,400 Noe: hs able shows he summary sascs of he man varables for each of he sx regons. able 2: Breung Panel Un Roo es Resuls Expors Model (Model 3) Impors Model (Model 4) Varable Breung -sasc Varable Breung -sasc Level: Level: ln(exp) (0.337) ln(imp) -.74 (0.337) ln(y) 3.03 (0.997) ln(y) -0.5 (0.307) ln(rer) (0.85) ln(fr).22 (0.997) ln(rp) 0.57 (0.998) ln(rp) -.03 (0.52) ln(vol) -.68 (0.264) ln(vol) -.33 (0.254) Frs Dfferences: Frs Dfferences: ln(exp) (0.000)*** ln(imp) (0.000)*** ln(y) (0.000)*** ln(y) (0.000)*** ln(rer) (0.007)*** ln(fr) (0.000)*** ln(rp) (0.000)*** ln(rp) -3.7 (0.000)*** ln(vol) (0.000)*** ln(vol) -3.4 (0.000)*** Noes: Breung ndcaes he Breung and Pesaran (2008) Breung -es for panel un roo ess. hs es examnes he null hypohess of un roo (non-saonary). he fgures n parenheses are he p-values. *** ndcaes he sgnfcance a he percen level. he panel un roo ess ndcae all he varables n boh he expors model and he mpors model are negraed of order one. Panel Conegraon ess Wh he respecve varables negraed of order one, he heerogeneous panel conegraon es advanced by Pedron (999, 2004), whch allows for cross-secon nerdependence wh dfferen ndvdual effecs, s performed and he resuls are presened n ables 3. he resuls for boh whn and beween dmenson panel conegraon es sascs are gven n he able. All seven es sascs rejec he null hypohess 2

10 E. M. Ekanayake e al IJBFR Vol. 6 No of no conegraon a he % sgnfcance level for boh expors model and mpors model, ndcang ha he fve varable are conegraed n each model. able 3: Heerogeneous Panel Conegraon Resuls (a) Expors Model (Eq.3) Panel Conegraon Sascs (whn-dmenson) es Sasc Panel ν-sasc (0.000)*** Panel ρ-sasc (0.000)*** Panel -sasc (0.000)*** Panel -sasc (0.000)*** Panel Conegraon Sascs (whn-dmenson) Group PP ype ρ-sasc (0.000)*** Group PP ype -sasc (0.000)*** Group ADF ype -sasc (0.000)*** (b) Impors Model (Eq.3) Panel Conegraon Sascs (whn-dmenson) es Sasc Panel ν-sasc (0.000)*** Panel ρ-sasc (0.000)*** Panel -sasc (0.000)*** Panel -sasc (0.000)*** Panel Conegraon Sascs (whn-dmenson) Group PP ype ρ-sasc (0.000)*** Group PP ype -sasc (0.000)*** Group ADF ype -sasc (0.000)*** Noes: he number of lag lengh was seleced auomacally based on SIC wh a maxmum lag of 5. he fgures n he parenheses are p-values. *** ndcaes he sgnfcance a he percen level. Afer havng found conssen evdence of conegraon, he Dynamc Leas Squares (DOLS) echnque for heerogeneous conegraed panels s esmaed. he resuls of he DOLS are presened n able 4 for boh expors model and for mpors model. As he resuls for he expors model show, all he coeffcens have he expeced sgns and all he coeffcens are sascally sgnfcan a he % level of sgnfcance. Exchange rae volaly varables has a posve sgn ndcang ha exchange rae volaly does no have any adverse mpac on Souh Afrca's expors o he European Unon. Gven ha he varables are expressed n naural logarhms, he coeffcens can be nerpreed as elascy esmaes. he resuls ndcae ha, n he expors model, a % ncrease n real foregn ncome ncreases real expors by 4.62%; a % ncrease n real exchange rae decreases real expors by 0.9%; and a % ncrease n he relave prce decreases real expors by 2.28%. Smlarly, as he resuls for he mpors model show, all he coeffcens have he expeced sgns and all he coeffcens are sascally sgnfcan a he % level of sgnfcance. Exchange rae volaly varables has a posve sgn ndcang ha exchange rae volaly does no have any adverse mpac on Souh Afrca's mpors from he European Unon. However, n boh models, he dummy varable represenng he aparhed era has a negave sgn ndcang ha rade flows beween Souh Afrca and he European Unon declned durng he aparhed perod. able 4: Dynamc Leas Squares (DOLS) Long-Run Elasces Dependen Var: ln(y) ln(rer) ln(rp) ln(vol) D Ln(X) 4.622*** (9.84) -0.93*** (3.44) *** (6.5) 0.044*** (5.5) -.440*** (7.49) Dependen Var: ln(y) ln(fr) ln(rp) ln(vol) D Ln(M) 3.03*** (9.92) 0.26*** (8.88) -0.08*** (3.74) 0.3*** (9.5) -.205*** (8.67) Noes: he fgures n parenheses are absolue values of -sascs. *** ndcaes he sascal sgnfcance a he percen level. Fnally, n order o evaluae he mpac of exchange rae volaly on rade flows durng he shor-run, a panel vecor error correcon model (VECM) proposed by Pesaran e al. (999) s esmaed. he resuls 22

11 he Inernaonal Journal of Busness and Fnance Research Volume 6 Number of hese models are presened n able 5 for expors model and for mpors model. hese resuls show evdence ha exchange rae volaly has an adverse mpac on expors n he shor-run. able 5: Dynamc Leas Squares (DOLS) Shor-Run Elasces Dep. Var. ln(y) ln(rer) ln(rp) ln(vol) D ECM - ln(x) 3.07*** (9.29) *** (8.44) -2.00*** (4.68) -0.04** (.99) 0.0 (.7) *** (9.67) Dep. Var. ln(y) ln(fr) ln(rp) ln(vol) D ECM - ln(m).034*** (8.70) (.39) (.53) (0.90) 0.03 (2.87) *** (9.52) Noes: he fgures n parenheses are absolue values of -sascs. ** and *** ndcae he sascal sgnfcance a he percen and 5 percen level, respecvely. All he varables (excep he dummy varable) are expressed n naural logarhm and herefore he coeffcens can be nerpreed as elasces. SUMMARY AND CONCLUSIONS In hs paper we have examned he dynamc relaonshp beween rade flows and exchange rae volaly n Souh Afrca's rade wh he European Unon n he conex of a mulvarae error-correcon model. Esmaes of he long-run expor and mpor demand funcons were obaned by employng he panel conegraon usng quarerly daa for he perod 980:Q :Q4. he panel un roo ess ndcae all he varables n boh he expors model and he mpors model are negraed of order one. he heerogeneous panel conegraon es advanced by Pedron (999, 2004) was used o es for panel conegraon for boh expors and mpors models. he resuls for boh whn and beween dmenson panel conegraon es sascs rejec he null hypohess of no conegraon a he % sgnfcance level for boh expors model and mpors model, ndcang ha he fve varable are conegraed n each model. Afer havng found conssen evdence of conegraon, he Dynamc Leas Squares (DOLS) echnque for heerogeneous conegraed panels s esmaed. As he resuls for he expors model show, all he coeffcens have he expeced sgns and all he coeffcens are sascally sgnfcan a he % level of sgnfcance. Exchange rae volaly varables has a posve sgn ndcang ha exchange rae volaly does no have any adverse mpac on Souh Afrca's expors o he European Unon. However, he resuls show evdence ha exchange rae volaly has an adverse mpac on expors n he shor-run. Smlarly, as he resuls for he mpors model show, all he coeffcens have he expeced sgns and all he coeffcens are sascally sgnfcan a he % level of sgnfcance. Exchange rae volaly varables has a posve sgn ndcang ha exchange rae volaly does no have any adverse mpac on Souh Afrca's mpors from he European Unon. However, n boh models, he dummy varable represenng he aparhed era has a negave sgn ndcang ha rade flows beween Souh Afrca and he European Unon declned durng he aparhed perod. REFERENCES Arsoelous, K. (200). Exchange Rae Volaly, Exchange Rae Regme, and rade Volume: Evdence from UK-US Expor Funcon ( ). Economc Leers, 72(), Arze, A. C. (998). he Effecs of Exchange-Rae Volaly on U.S. Impors: An Emprcal Invesgaon. Inernaonal Economc Journal, 2(2), Arze, A. C. (996). he Impac of Exchange-Rae Uncerany on U.S. Expor Growh: Evdence from Korean Daa. Inernaonal Economc Journal, 0(3),

12 E. M. Ekanayake e al IJBFR Vol. 6 No Arze, A. C. (995). he Effecs of Exchange-Rae Volaly on U.S. Expors: An Emprcal Invesgaon. Souhern Economc Journal, 62(), Arze, A. C., Osang,. and D. J. Sloje (2000). he Exchange Rae Volaly and Foregn rade: Evdence from hreen LDCs. Journal of Busness and Economc Sudes, 8, 0-7. Asseery, A. and D. A. Peel (99). he Effecs of Exchange Rae Volaly on Expors: Some New Esmaes. Economc Leers, 37(3), Bahman-Oskooee, M. and S. W. Hegery (2009). he Effecs of Exchange-Rae Volaly on Commody rade beween he Uned Saes and Mexco. Souhern Economc Journal, 75(4), Bahman-Oskooee, M. and M. Kovyryalova (2008). Impac of Exchange Rae Uncerany on rade Flows: Evdence from Commody rade beween he Uned Saes and he Uned Kngdom. he World Economy, 3(8), Bahman-Oskooee, M. and R. Mra (2008). Exchange Rae Rsk and Commody rade beween he U.S. and Inda. Open Economes Revew, 9(), Bahman-Oskooee, M. and Y. Wang (2009). Exchange Rae Sensvy of Ausrala/s rade Flows: Evdence from Indusry Daa. he Mancheser School, 77(), -6. Bahman-Oskooee, M. and Y. Wang (2008). Impac of Exchange Rae Uncerany on Commody rade beween he U.S. and Ausrala. Ausralan Economes Papers, 47(3), Bahman-Oskooee, M., and Z. Ardalan (2006). Exchange Rae Sensvy of U.S. rade Flows: Evdence from Indusry Daa. Souhern Economc Journal, 72, Bahman-Oskooee, M. (2002). Does Black Marke Exchange Rae Volaly Deer he rade Flows? Iranan Experence. Appled Economcs, 34, Bahman-Oskooee, M. (99). Exchange Rae Uncerany and rade Flows of Developng Counres. he Journal of Developng Areas, 25, Bahman-Oskooee, M. and S. Payeseh (993). Does Exchange Rae Volaly Deer rade Volume n LDCs. Journal of Economc Developmen, 8(2), Bredn, D., Founas, S. and E. Murphy (2003). An Emprcal Analyss of Shor-Run and Long-Run Irsh Expor Funcons: Does Exchange Rae Volaly Maer?. Inernaonal Revew of Appled Economcs, 7(2), Byrne, J., Darby, J. and R. MacDonald (2008). US rade and Exchange Rae Volaly: A Real Secoral Blaeral Analyss. Journal of Macroeconomcs, 30, Choudhury,. (2005). Exchange Rae Volaly and he Uned Saes Expors: Evdence from Canada and Japan. Journal of Japanese and Inernaonal Economes, 9, 5-7. Chowdhury, A. R. (993). Condonal Exchange Rae Volaly Depress rade Flows? Evdence from Error-Correcon Models. Revew of Economc and Sascs, 75,

13 he Inernaonal Journal of Busness and Fnance Research Volume 6 Number Chowdhury, A. R. and M. Wheeler (2008). Does Exchange Rae Volaly Affec Foregn Drec Invesmen? Evdence from Four Economes. he Inernaonal rade Journal, 22(2), Chou, W. L. (2000). Exchange Rae Varably and Chna s Expors. Journal of Comparave Economcs, 28(), De Va, G. and A. Abbo (2004). Real Exchange Rae Volaly and US Expors: An ARDL Bounds esng Approach. Economc Issues, 9(), Doyle, E. (200). Exchange Rae Volaly and Irsh-UK rade, Appled Economcs, 33, Gheong, C., Mehar,. and L. V. Wllams (2005). he Effecs of Exchange Rae Varably on Prce Compeveness and rade Volumes n he U.K.: A Dsaggregaed Approach. Journal of Polcy Modelng, 27(8), Global rade Informaon Sysems, World rade Alas Daabase. Hooper, P. and Kohlhagen, S. (978). Real Exchange Rae Uncerany on he Prces and Volume of Inernaonal rade. Journal of Inernaonal Economcs, 8, Inernaonal Moneary Fund, Inernaonal Fnancal Sascs Daabase. Klen, M. W. (990). Secoral Effecs of Exchange Rae Volaly on Uned Saes Expors. Journal of Inernaonal Money and Fnance, 9(3), Koray, F. and W. D. Lasrapes (989). Real Exchange Rae Volaly and he US. Blaeral rade: A VAR Approach. Revew of Economcs and Sascs, 7(), Kroner, K. F. and W. D. Lasrapes (993). he Impac of Exchange Rae Volaly on Inernaonal rade: Reduce form Esmaes usng he GARCH-n-Mean Model. Journal of Inernaonal Money and Fnance, 2(6), Lasrapes, W. D. and F. Koray (990). Exchange Rae Volaly and U.S. Mullaeral rade Flows. Journal of Macroeconomcs, 2(2), McKenze, M. D. and R. D. Brooks (997). he Impac of Exchange Rae Volaly on German-US rade Flows. Journal of Inernaonal Fnancal Markes, Insuons and Money, 7(), Onafowora, O. A. and O. Owoye (2008). Exchange Rae Volaly and Expor Growh n Ngera. Appled Economcs, 40(2), Ozurk, I. and H. Kalyonku (2009). Exchange Rae Volaly and rade: An Emprcal Invesgaon from Cross-Counry Comparsons. Afrcan Developmen Revew, 2(3), Pesaran, M., Shn, Y., and R. Smh (200). Bounds esng Approaches o he Analyss Of Level Relaonshps. Journal of Appled Economercs, 6(3), Qan, Y. and P. Varangs (994). Does Exchange-Rae Volaly Hnder Expor Growh?: Addonal Evdence. Emprcal Economcs, 98,

14 E. M. Ekanayake e al IJBFR Vol. 6 No Welwa, A., Ekanayake, E. M. and H. suj (999). Real Exchange Rae Volaly and Sr Lanka's Expors o he Developed Counres, Journal of Economc Developmen, 24(), BIOGRAPHY Dr. E. M. Ekanayake s an Assocae Professor of Economcs a Behune-Cookman Unversy, Dayona Beach, Florda, USA. He earned hs Ph.D. n Economcs a he Florda Inernaonal Unversy, Mam n 996. He has many publcaons o hs cred. Conac nformaon: School of Busness, Behune- Cookman Unversy, 640 Dr. Mary McLeod Behune Blvd., Dayona Beach, FL 324, USA. Phone: (386) E-mal: ekanayakee@cookman.edu. Dr. Ranjn haver s an Assocae Professor of Economcs a Seson Unversy, Deland, Florda. She earned her Ph.D. n Economcs n 995 and an M.A. degree n Economcs n 989 a he Nore Dame Unversy. She s currenly he Char of Economcs Deparmen a Seson Unversy. Dr. haver s also he Drecor of Cener for Holsc Mcrocred Inaves (CHOMI), and he Drecor of CHOMI anzana. Conac nformaon: Deparmen of Economcs, Seson Unversy, Box 8392, Deland, FL 32723, USA. Phone: (386) E-mal: rhaver@seson.edu. Dr Danel R. Plane s an Assocae Professor of Compuer Scence a Seson Unversy, Deparmen of Mahemacs and Compuer Scence, 42 N. Woodland Blvd., DeLand, Florda, USA. He receved a PhD n Physcs from he Unversy of Nore Dame n 995 and a Maser of Engneerng n Compuer Scence from Norh Carolna Sae Unversy n He can be conaced a dplane@seson.edu. 26