EXCHANGE RATE VOLATILITY EFFECT ON TRADE VARIATIONS

Transcription

1 EXCHANGE RATE VOLATILITY EFFECT ON TRADE VARIATIONS By: MS.c Mmoza Agoll 1 Inroducory noes Exchange rae volaly can affec rade drecly, hrough uncerany and adjusmens coss, and ndrecly hrough s effec on he srucure of oupu, nvesmens as well as on governmen polces. Ths paper s focused on drec effec he exchange rae exer on rade, whle some characerscs of he ndrec mpacs are beng consdered as mporan background varables ha have defned he relaon and s frmness. The expeced effec s pared wh demand/supply elascy o relave prces; an nelasc expor demand/supply o relave prces makes unceran he heorecal presence of nomnal/real exchange rae effec on rade flows. If exchange rae dd or dd no maers, when radng, was also relaed o rade barrers (Bhagwa, 1996; pp13, hgh rade barrers would make he exchange rae neffecve n prce adjusmens o currency flucuaons. Facng dfferen elasces for mpors and expors demand o relave prces and radng counres specfcs, on one hand, and he sae of rade lberalzaon on he oher hand makes neresng he envronmen where exchange rae volaly plays s role on n rade conex. The paper res o emprcally nvesgae he effec exchange rae uncerany has played on rade volumes. Two dfferen measures of volaly ndex derved from nomnal and real exchange rae are esmaed, co negraon echnques s used o check counres shor/long erm evdences on he effec exchange rae volaly had played on blaeral rade and comparson s made from aggregaed resuls derved hrough pooled esmaon. In general here s evdence of a deerorang effec exchange rae volaly plays on rade volume, nn shor erm and consderng dsaggregaed counry daa expors have been sufferng mos, whle mpors do no reac sgnfcanly o exchange uncerany, manly due o he nelasc mpors demand. There are dfferen counry paern of hs effec and shor run adjusmens resuls mporan n boh n sgnfcance and magnude. The paper s organzed as follows: frs par offer a leraure revew followed by a summary overvew of he exchange rae regme n and emprcal evdences on he ssue. Secon and 3 presen he model, he economerc mehodology and daa ssue, whle secon 4 presens and dscuss dfferen esmaon resuls, fnally some concludng evdences are summarzed. 1 Trade Polcy analys ALBANIAN CENTER FOR INTERNATIONAL TRADE 1

2 I. Leraure revew The sandard macroeconomc argumen behnd rade flow adjusmens o nomnal exchange rae deprecaon usually mples real exchange rae deprecaon, a leas o he exen ha relave prces do no adjus by he same amoun as he nomnal exchange rae movemen. Relave prce adjusmens heory suggess ha such devaluaon of he domesc currency mples excess demand for expor orened goods (domesc goods, whch would resul n ncrease expors and decrease mpors. The effec exchange rae would exer on he rade flow s pared wh demand/supply elascy o relave prces. Facng an nelasc expors demand/supply o relave prces makes unceran he heorecal presence of nomnal/real exchange rae effec on rade flows. If exchange rae maers, or dd no maer when radng, was also closely relaed o rade barrers (Bhagwa, 1996; pp13 hgh rade barrers wll make he exchange rae neffecve n prce adjusmens o currency flucuaons. The macroeconomc debae on currency behavor and s relaon wh rade decson has been lad oward mcroeconomc analyses of opmzng frm behavor under uncerany and rsk. The consequences of exchange rae volaly (rsk measure on rade have long been of concerns, and hghly debaed among he economss. However, here s no consensus so far as o wheher he exchange rae volaly, maer, and f rade benefs or adversely reac o currency flucuaons remans unresolved. (McKenze, 1999 Tradonal models examned he exchange rae volaly effec on rade based on producer heory of he frm under uncerany, where frm profably s relaed o he movemens of he exchange rae. A rsk-averse frm, n a suaon of a dependency beween s prof and exchange behavor, would prefer o reduce rsk, reducng he level of rade. Baron (1976b suggesed ha when he exporer nvoces n foregn currency, he faces prce rsk. The quany demanded s known (conraced, snce prces may change durng conrac perod, he revenue sream and prof yeld uncerany. When a frm nvoces n local currency he frm faces quany rsk, quany demanded s no ceran because prces are unceran (mpor prces may become more compeve n currency apprecaon and consumer preferences may shf. Rskaverse frm would mnmze rsk, affecng prces, bu prce movemens are dfferen for hose frms nvocng n foregn currency (under demand cerany - prce would ncrease o mnmze rsk and hose nvocng n local currency (under demand uncerany prce would decrease. Hooper and Kohlhagen (1978 examned he effec of exchange rae volaly n a blaeral framework, where source of uncerany s he nomnal exchange rae. They showed ha here s a clear negave relaon beween, exchange rae volaly and he volume of rade, bu he effec on prces s ambguous dependng on he fac ha, are mporers or exporers who bear he rsk. De Grauwe (1988 assumng ha rsk averson s no consan, and s degree effec frm uly funcon convexy argued ha here are wo effecs of ncreased exchange rsk, a subsuon effec and an ncome effec, whch work n oppose drecon. The fnal effec of exchange rsk on rade would depend on he magnude of he subsuon and ncome effec. II. Some shor noes on he n exchange regme Examnng he effec of exchange rae and exchange rae volaly on rade remans an neresng oucome for he s foregn rade.trade lberalzaon and a flexble exchange rae regme, adoped by, n 199, have ncreased he exposure of rade flow o currency behavor rsk, whle beng affeced by specfc rade parners expor demand/supply characerscs. Regonal FTA s naves need careful moneary adjusmen o suppor rade and help he nassance of advanages, on he oher hand reducng currency rsk, wll mprove he rade and nvesmen envronmen. Afer aemps of he auhores o keep a fxed exchange rae, gven he fnancal condons of he counry hs moneary polcy regme couldn ress for long. In Augus 199, adoped a floang exchange rae regme, under a money growh arge (moneary sraegy, whch shfed he exchange rae deermnaon from sae monopoly o marke forces. The adopon of a flexble, raher han pegged

3 exchange rae n found suppor n he leraure, especally n he conex of a small open economy n ranson, when he level of nernaonal reserves s below a mnmum hreshold (see Muço e al., Flexble exchange rae brough abou a rapd Lek devaluaon and an nflaonary suaon snce exchange rae shocks were ransmed o he prce level, manly due o he hgh wegh of mpors on GDP. Exchange rae regme was par of a full package of ranson reforms, ncludng he foregn rade secor whch was relefed from sae monopoly and prces were subsanally lberalzed. n Leke has been characerzed by deprecaon endences unl 1997, he 1997 crses was he reason of an abnormal LEKE deprecaon, hen an apprecaon rend s noced, whch s appled o almos all he currences, unl year 00, when Euro was nroduced and he nomnal exchange rae sared o behave dfferenly oward USD and EURO, LEKE was havng an apprecaon rend owards USD, whle followng a deprecaon pah agans EURO. The volaly ndex vares among currences; EURO s he one havng he more sable behavor, whle USD and Ialan Lra have been characerzed by a wder range of nomnal exchange rae he me (see sandard devaon of he nomnal exchange rae bu he volaly ndex shows ha Greek currency has had he hghs volaly (I refers o he value of he volaly ndex and he sandard devaon. The hgh sandard devaon shows ha volaly has been sharp; s value s hgher han he average volaly measure showng he presence of he repeaed hgh magnude of oscllaons. Table 1. Descrpve on nomnal exchange rae Currency daa USD DM DR LIT EURO Sd. Devaon Average Rao Sd.devaon/Averarage NER (% Volaly ndex Nomnal exchange rae - Average Nomnal exchange rae S.D Source: Bank of A number of papers have so far analyzed hese relaonshps and almos all of hem converge on he concluson of a modes mpac of he exchange rae on expor and a more pronounced and srong effec on mpor (see for example, Mançellar e al, 1999, Hadër e al, Kolas, IMF. Ths paper bulds on he oher conrbuons, esmang a measure of exchange rae uncerany buldng a volaly varable, n addon, a dsnc feaure of hs paper s ha looks on he currency effecs on rade, separaely for he hree man major radng parners,.e., Ialy, Greece and Germany, and comparave analyses are buld from aggregae resuls o counry specfcaons. II. Volaly Index and measuremen Two crucal ssues underpn emprcal nconssences n explorng exchange rae volaly relaon wh rade volume ( how an exporer perceves exchange rae rsk and ( how hs rsk s ncorporaed n he rade decson. When nvolvng volaly ndex, he use of nomnal versus real exchange rae s he frs pon o address. IMF suggess ha me dmenson should be consdered n economc decson when measurng exchange rae volaly. Shor run flucuaons n nomnal exchange raes are relevan for rader s decson as coss and prces are relavely rgd and herefore known. When he observaon horzon s exended, he relevan exchange rae becomes he one connecng domesc coss of producon and foregn prces convered no domesc currency; herefore real exchange rae volaly s he varable ha maers (IMF,

4 The paper wll nvolve boh nomnal and real exchange raes, n dervng volaly measures, bu consderng he shor pah me under observaon, as well as on heorecal and emprcal consderaons ha emphass he fac ha would be rrelevan wheher he volaly coeffcens are esmaed from real or nomnal exchange raes as he volaly s sourced solely from he nomnal exchange rae (McKenze and Brooks,1997 prory wll be gven o esmaon resuls derved from usng nomnal exchange rae. Anoher mporan ssue ha maers s how agens form her expecaons on uncerany exchange rae behavor may cause and over wha me horzon. There s almos nvarably a lag beween he me a whch an exporer conracs for sale and he me a whch delvery, paymen or boh are made. Mos of he emprcal leraure uses realzed exchange rae volaly, as proxed by measures such as he absolue percenage of change, lagged sandard devaon, or movng average varance around rend. These measures mpose an assumpon of adapve expecaons, wheren economc agens use only pas nformaon n predcng fuure exchange rae rsk. Movng average sandard devaon (MASD of he growh rae for he blaeral nomnal exchange rae s he frs a measure of volaly nvoked n he sudy (Kenen and Rodrk, ln mples he logarhmc form of he blaeral exchange rae seres. The opnon on he frs measures of volaly s ha does overesmaes he unceranes and expecaons, whle leraure, more broadly suppors he use of auoregressve specfcaon, as well as condonal varance ARCH/GARCH approach., whch s he oher measure used n he paper. We assume ha exporers/mporers form her expecaons followng GARCH (p, q process as follows: 1 m V = ( ln NER + 1 ln NER + - Where m s he order of movng average, whle m = 1 ln NER = α 0 + α1dummy + ε 1.1 h = α + βε 1 + γh 1 + δdummy + u Where ε ~N (0, h, and u are a whe process wh mean zero and varanceσ u.the condonal varance equaon descrbed above s a funcon of hree erms, he mean (α, he news abou volaly from he prevous perod measured as he lag of he square resdual from he mean equaon ε 1 - ARCH erm, and he las perod of forecas error varance h 1 (he GARCH erm. In addon we add he dummy varable o capure he nroducon of he Euro currency. Quarerly daa on nomnal blaeral exchange rae are aken from Cenral Bank of, for he perod 1993:003. Table 1: Esmaon of Volaly ndex Sample 1993:1 003:4 Ialy Greece Germany Varable Consan (-.3 ** ( ** (5.0 Dummy (.547 ** (66.10 ** (41.5 ** Varance Equaon Consan (4.373 (1.187 (1.48 ARCH (3.608 ** (.089 ** (3.003 ** GARCH -0.3 ( ** ( ( ** Process GARCH(1,1 GARCH(1,0 GARCH(1,1 4

5 The esmaons showed ha varably of nomnal exchange rae follows a GARCH (1, 1 process for he Ialan/German currency and a GARCH (1, 0 process for Greek currency over our sample perod. The expecaons erms derved from he general auo regressve condonal process (volaly daa are used n he rade equaons. III. Workng Model The emprcal leraure, acknowledges wo prmary deermnans of expor and mpor demand varaons derved hrough a paral equlbrum approach (Dornbusch, 1998 foregn ncome varable and relave prce varable (equaons.1 and.. Income sands for he economc acvy and he purchasng power of he radng parners, whle relave prce varable res o capure he prce power on shapng marke behavor (demand and supply. Exchange rae volaly s nvoked explcly n he funcon, consderng he currency movemen effec hrough uncerany on rade decson and also he effec on prce and compeveness level. LogM LogX = 1 + α Y + α 3 log( p + α 4V + α 5 α log( Dummy + ε (3.1 = 1 + α Y + α 3 log( p + α 4V + α 5 α log( Dummy + ε (3. M and X presen mpors and expors from/o, Y Y are gross domesc produc of counry and, P s he relave prce varable measured as a fracon of domesc and foregn expor un prce, V presens he varably ndex and a dummy varable ha dsngushes he perod where Euro was nroduced s also nvolved. An mproved economc acvy wll nduce rade, and we expec gross domesc produc o be posvely relaed o rade. The expeced relaon beween expor/mpors and relave prces wll depend on he dynamcs of he former, as well as on he elascy of demand and supply for expors. The nsably of exchange rae and s relaon wh rade has been esed o be ambguous. General gravy approach s also ncorporaed, n he hrough paper, ryng o enlarge he se of varable affecng rade, derved boh from paral and general equlbrum models of nernaonal rade (Lnneman, 1966 and Bergsrand Ths approach consders rade as an neracon process among dfferen economes where apar from he nfluence of prces n demand and supply allocaon process, oher mporan facor play an mporan role, facors such as dsance, culural dfferences, hsory, borders ec. IV. Daa and mehodology Quarerly seres of GDP and expor un prces are aken from Inernaonal Fnancal Sascs daabase. Quarerly Expor Un prce for was proxed by expor un prce seres for developng counres (IFS, whle blaeral expor/mpor daa n USD were drawn from ACIT daabase. Consderng he ambguy of he quesoned subjec n boh heorecal and emprcal arena, he analyss res o examne Expor Funcon Log(X j = α0 + α 1 log (YY j + α SIM j + α3 log Dj + α 4 log N j + α5p + α 6V + α7cmbr + α8dummy + u Impor Funcon Log(M j = α0 + α1 log (YY j + αsim j + α3 log Dj + α4 log N j + α5p + α 6V + α7cmbr + α8dummy + u j j 5

6 evdences hrough dfferen seps. Frsly a smple error correcon mehod s appled on he oal volume of rade and wo man exchange raes - LEKE/USD, EURO/USD. Ths esmaon s a l b bared, consderng only exchange rae as an explanaory varable of he rade varaon, bu smply res o evdence he drecon of he relaon beween rade and he wo man currences. Ths overall ndcaon wll be confroned wh resuls from esmang expor/mpor demand funcon where a se of deermnans of expors and mpors are consdered (derved from paral/general equlbrum nernaonal rade models. Effec of exchange rae on rade volume wll be esed separaely for he hree man radng parners, whch s Ialy, Greece, Germany, usng co negraon echnques, whle panel daa esmaons wll ry o esmae he resulan effec varably of dfferen currences had on s rade volume. V. Some emprcal evdences on exchange rae volaly relaon wh rade V.1 Trade volume and USD/EURO nomnal exchange rae Trade volume seems o have been negavely correlaed wh USD nomnal exchange rae movemens (n long run prospecve, whle no sgnfcanly relaed wh LEKE/EURO nomnal exchange rae (Table 6. Appendxes, coeffcen sascal sgnfcance. Expors and mpors dsplay dfferen sensvy wh respec o boh currences; mpors are almos hree mes more sensve o dollar movemen, han o EURO. Ths s because of a more sable EURO compared o USD (able 1. Sd. devaon measure comparson as well as ncreased nvocng n Euro (75% of s rade volume orgnaes from EU counres. Expors reacon o currency movemen doesn reflec frmness, he esmaed coeffcen are no hghly sgnfcan (USD esmaed coeffcen s sgnfcan a 10% level of confdence whle reacon coeffcen o Euro movemen doesn appear sgnfcan, Table 6-appendxes. Coeffcen magnude gves sgns of a hgher sensvy when nvocng n Euro, and hs manly o pegged naure of our currency o USD, whch doubles he rsk exporers have o bear when radng. Expors and mpors vary largely due o shor run adjusmen oward equlbrum, (ERC-erm sgnfcance; he relaon of rade wh EURO seems o be more ressan o sudden shor run changes, whle dollar movemen has been mrrored n a deerorang rade volume. Wha s he mporance currency rsk has n a general envronmen where rade ake place where oher very mporan facor smulaneously play he game, wha s he role each radng parner currency behavor has on rade allocaon, does he effec of exchange rae rsky behavor remans sold or s spread among oher varables, especally on prces? The followng par of he paper focuses on hese quesons, wh he am of amplfyng he analyses dmenson. V.II Counry evdences long run co negraon vecor Vecor Error Correcon (VEC mehod s beng used o pon ou he effec currency behavor for he s hree man radng parners has had on rade varaon and rade dsrbuon among counres. The VEC mehod s based on he Engle and Granger (1987 saemen ha a lnear combnaon of wo or more non-saonary seres may be, and he co negraon equaon saonary and f such a lnear combnaon exss. The VEC specfcaon resrcs he long run behavor of he endogenous varables, o converge o her co negraon relaonshp, whle allowng a wde range of shor run dynamcs. The reasons behnd adopng hs economerc mehod of esmaon, was o observe shor run behavor, whch n case of an ransory economy s characerzed by dynamc hgh economc srucural adjusmens, whch makes nfrmness long run endences. Pror applyng VEC esmaon mehod, me seres saonary s esed hrough Augmened Dckey-Fuller es, and he order of negraon s defned. The summary able of ADF he appendxes, show ha varables are non-saonary, unl beng dfferenaed wh he frs order. 6

7 Table. Un Roo es - summary able log( Ialy Greece Germany Y I(1 * I(1 I(1 * log( Y I(1 * I(1 I(1 log( p I(1 * I(1 I(1 ** V (MSDA I(1 I(1 I(1 V (ARCH_GARCH I(1 I(1 I(1 Log(Expors I(1 I(1 I(1 Log(Impors I(1 I(1 I(1 * Sgnfcan a 5% - 3 and 5 lags ** Sgnfcan a 1% -lags To esablsh weaher here s a long run relaonshp among he varables Johansen mehod was used, (Johansen, 1991 and Johansen and Juselus, For each counry, employng and ARCH/GARCH volaly measures, Johansen Lkelhood Rao es (race sascs and maxmum egenvalue 3 sascs were used o denfy he presence of common sochasc rends. For all counres he null hypohess of r=0 s rejeced n favor of he alernave hypohess r 1, ndcang he presence of long run co negrang relaon beween among expors/mpors and economc acvy, relave prces and exchange rae volaly. Table 3 Co-negraon es TRACE STATISTICS MAXIMUM EIGENVALUES Counry Funcon H 0 : r=0 H 1 : r=1 r 1 r= r r=3 r 3 r=4 r=0 r 1 r 1 r r r 3 r 3 r 4 Germany Expors Impors Ialy Expors Impors Greece Expors Impors ** Co negraed a 5% level of sgnfcance s beng esed usng crcal value, whch are for maxmum egenvalues, and for race sascs a r=0, r 1 hypohess ** Greek Impor funcon, manfess a weaker co-negraon vecor a 10% level of sgnfcance crcal value s 6,98 r-denoes he number of co negraon vecors Normalzed co negrang vecors (Table 4, correspondng o he maxmal egenvalues, he domnan long-run relaonshp showed ha he long run effec exchange rae volaly has played on rade dffers among he se of counres. Long run perceved currency rsk ncrease, wll sgnfcanly shrnk rade naves for exporers o Germany and Greece (refers o esmaed coeffcen, of volaly ndex, able-4, and whle expors o Ialy resul o be nsgnfcanly relaed o exchange unceranes. The negave effec of exchange rae rsk on expors s n lne wh heory and oher emprcal fndngs (IMF counry repor, 000, whle he sensvy manfesed s characerscs of expors responsveness founded n small developng counres. (Das, 003 Rahmasyah e al, The maxmum egenvalues can be compued from he race sasc by he followng relaons Qmax = T log( 1 λ + 1 = Qr Qr+ 1 7

8 The reacon of expors o exchange rae varaons s also nduced by specfc characerscs expor demand manfess, demand for s expors s elasc o prce changes n case of Greece and Germany (he esmaed coeffcen vares from and -.13, and hs elascy was a reason ha makes rade volumes more sensve o exchange rae volaly n long erm. Expors o Ialy characerzed by conraced re-expors reflec nsensveness o currency and prces behavor (varables are no sgnfcan due o he fac ha n large par hey are denomnaed by-expors and herefore here s no msmach o affec ransacons. Impors volumes seem o have benefed from currency behavor (all he coeffcen are posve, and sgnfcan, whle relave prce coeffcen s eher no sgnfcan, Greece or Ialy, or less han uny Germany, able 4 n long run, and manly due o he nelascy naure of mpors demand. Table 4. Co negraon Equaons Counry Normalzed Co negraon Vecors Tes H 0 β 3 =0 Germany Expor LogX = 0.31 log( Y 1.45* log( p 9.4* V = -.81 * Impors LogM = 1.06 log( Y 0.74 * log( p * V = 3.39 * Greece Expor LogX = log( Y.13* log( p 8.05 * V = * Impors 1 LogM =.5 log( Y * V = 3.56 * Ialy Expor LogX = 6.81 log( Y * log( p 1.90 * V = Impors LogX =.11 log( Y * log( p * V =.99 * * Sgnfcan a 1% level of confdence 1 Relave prces for Greece was no sgnfcan and hey caused V. Counry Shor Run dynamc adjusmens Based on he represenaon heorem developed n Eager and Granger (1987 can be shown ha for he error correcon model exss for a co negrang vecor shown below: Expors: LogM = α 0 + α 1 R 1 + β * LogM + γ log( Y + δ log( p + τ V + α Dummy + ε Impors: LogX = α + α R + LogX Y p V Dummy β * + γ δ τ α ε log( log( Where R erm s he lagged error correcon erm, generaed from he resduals of co negraon vecor, sandng for paral shor run adjusmens ha correcs devaon from long run equlbrum. In he appled analyses ECT accouns for he fac ha acual expors/mpors do no adjus nsananeously o her long erm deermnans and shows how he sysem converges o long run equlbrum mpled by co negrang regresson. As a general vew here are dynamc shor run adjusmens, sascally sgnfcan, and of dfferen speeds among he se of counres (refers o coeffcen magnude, of error correcon erm, Table -5, ha have played role on rade varaons. Whle no sascally sgnfcan race of shor run pressure on expors volume was found for Greece, n expors o Ialy and Germany do manfes sgnfcan varaon due o shor run adjusmen pressure oward equlbrum level, almos 40% of shor run varaon n expors volume owards Ialy and Germany s resul of such dsored equlbrum suaons. Exchange rae rsk measure resuls sgnfcanly relaed o expors volume, whle manfesng dfferences n behavor n boh magnude and drecons comparng ndvdual counres. 8

9 Expors o Germany are shrunk due o currency behavor (refers o coeffcen sgn, able 5 and s neresng he long sandng mpac of exchange uncerany on expors. Ialan expors n shor run are a posve funcon of he rsky behavor of he currency erm; heorecally and emprcal evdences suppor such behavor, under he assumpon ha frm uly funcon s an ncreasng funcon on rsk Bu nsead of gong back o heorecal hypohess he coness of expors wh Ialy where a consderable par of expors volume s jus re-exporng, gves he relaon exchange uncerany expors, behavoral characerscs of mpors, raher han expors, and he resul s ncreasng expors despe experenced currency apprecaon and ncreased rsk faced due o exchange uncerany. Table 5 Error Correcon Resuls 1 EXPORT FUNCTION IMPORT FUNCTION Varables Germany Ialy Greece Varables Germany Ialy Greece ECT ( (-.449 ECT -0.8 ( ( (1.75 X( ( (-.86 M( (-.504 X(- M( (.8 X(-3 M(-3 Y ( (-4.48 Y( (.083 Y (- 8.8 Y(- (-4.79 Y (-3.57 Y(- (1.757 P(-1 P(-1 P(-.019 ( (-1.76 P( (-.01 P( P(-3 (-.35 V( ( (.449 V( ( (.635 V( V(- (-4.36 (-1.88 V( (.17 Dummy Dummy 0.58 ( (-.35 Adjused R Adjused R Insgnfcan varables are excluded from he able of resuls Expors o Greece reflec sensvy o shor run relave prces movemens, sensvy ha has affeced he negave pressure exchange volaly has exered on expors (Table, 5. EURO nroducon n he se of counres observed doesn resul of help for n expors, and hasn nduced so far any reacon n he expors allocaon, or on he volumes. Ths manly because of he shor perod of he effecve EURO nroducon, as well as of s apprecaon compare o USD. 9

10 There s a dfferen behavor of mpors varably n shor erm; here are a few varables beng sascally mporan, and a small explanaon power of he model (Table 5, Adjused R. The man reason of such resuls may be excludng he leadng facor of mpors demand, whch s he gap beween he domesc demand and supply. The sgnfcan Error Correcon erm shows ha mpors varaon, s raher resul of shor run srucural changes and adjusmen process, as a whole, han resul of sgnfcan nfluence from ndvdual varable behavor (number of varables sgnfcance n shor run reduces. The dfferences among counres n he way mpors have neraced wh exchange volaly are presen, here s an ncreased mpor funcon for Germany and Greece mpors o, ha goes n lne wh EURO nroducon effec and an observed decreasng Ialan mpors due o relave prces dsadvanage (German and Greek mpors are correlaed posvely wh exchange rae volaly whle here s a negave coeffcen of correlaon beween Ialan Impors and relave prces, able 5. Relave prces nsgnfcance n he overall pcure of he resuls enforces he dea of an nelasc mpor demand o prces (Mancellar e al, V.4 Aggregaed evdences on rade exchange volaly ndex Counry evdences showed behavoral dfferences, among he se of seleced counres, where dependng on he observed counry specfc elascy of mpor/expors demand we had dfferen magnude and drecon of he exchange rae volaly ndex on rade volume. A pooled mehod of esmaon s he followng sep he emprcal analyss goes hrough n order o observe he resulan effec of counres specfcs. Two se of esmaons were run an augmened gravy equaon, wh exchange rae volaly measure explcly ncorporaed and an esmable expor/mpor demand funcon (see equaon 3.1 and 3.. A proxy 4 for real exchange rae was nroduced, n order o wegh he effec flucuaons of nomnal exchange rae by wholesales prce ndexes. Table 6. Expors/mpors demand - Panel Daa Esmaon Fxed effec log( Y log( Y log( p V Dummy AR-erm -1 EXPORT FUNCTION IMPORT FUNCTION RER NER RER NER (-1.0 ( (9.84 ** ( ** (-.90 ** (-.364 ** (-0.98 ( (-.141 ** (-1.37 * (-1.67 ( (-0.67 (-0.47 (1.67 * (3. ** (3.99 ( R Adjused R Chow-Tes DW - es Observaons RER WPI ALBANIA T = NER *, where WPI whole sales prce ndex WPI COUNTRY 10

11 Aggregaed Expors showed o be a posve funcon of he s gross domesc produc, wh almos a unary elascy (esmaed coeffcen vary from 0.84 o 0.89, Table 6. The relave prces and exchange uncerany rsk (NER and RER have he expeced sgns; hey boh have negavely affeced expors volume. Trade sensvy magnude o currency flucuaons appears vulnerable o he measures appled (MSDA/Arch-Garch approach, bu hey resuls o be conssen n erms of drecon. Counry specfcs effec appears srong; consder he fac ha fxed effec mehod of panel esmaon resuled superor. Expors appear o have suffered sgnfcanly from real exchange rae volaly ndex, whle mpors a he end despe he fac ha who won more or less among he parners, hey have suffered lle from nomnal exchange rae movemens as well as from ncreased prces. (Refers o no so hgh sgnfcan esmaed coeffcen of real exchange rae volaly measure and relave prce varable, Impor Funcon, Table 6. Table 7 Gravy approach panel esmaon resuls EXPORT FUNCTION IMPORT FUNCTION RER NER RER NER Consan erm (15. ** ( ** ( ** ( ** log( Y * Y (0.69 (0.84 (.3818 (.539 ** SIM 1.67 (7.35 ** ( ** (4.103 ** ( ** log( p ( ** (-.86 ** ( (-0.5 Log(POP 1.63 (.89 ** (3.15 ** ( ** (3.879 ** Log(Ds -.69 ( ** (-8.35 ** ( ** ( ** V ( (-0.17 ( (-0.7 Dummy (-0.41 ( * (1.103 (1.739 ** AR-erm 0.1 (.33 R Adjused R Chow es * 1.06 DW - es Observaons Crcal values for chow es run s F (,105 =1.87 Prevous experences or observed changes seem o be sgnfcan n mpors behavor consderng he auoregressve erm sgnfcance, bu hs s no exclusvely relaed o currency varaons. I s par of he overall mpors adjusmens, o he economc srucural changes, n-counry overall rsk and sably n boh economc and polcal erms. Real exchange rae proxy, whch ncreases compare o he magnude of he nomnal exchange rae (refers o esmaed coeffcen, Table 6 and 7 negave effec on rade showng, due o he nflaonary effec emboded, whch ncreases uncerany and rsks. s expors haven ye reaced posvely o EURO nroducon, whle mpors are sgnfcanly beng an ncreased funcon of he facly of a unfed sable currency. Acually EURO apprecaon agans USD could have had he conrary effec on mpors f facng an elasc mpor demand o prces, whch s no he case (magnude of he esmaed coeffcen s less han unary, and he pooled esmaons does confrm agan. Oher facors drawn from he augmened gravy equaons play an mporan role n he varably of 11

12 expors/mpors facors such as dsance and ransporaon coss, dfferences n economc mass and srucure, whch n cases of nelasc expor/mpors demands do pu some shade on prce varable and exchange rae volaly has on rade, whch reman deerorang for expors and n relaon o counry mpor demand specfcs, a benefng ool for mporers. Some concludng remarks: Some concludng remarks: Exchange rae volaly effec on rade volume, does exss, bu s magnude, sgnfcance and he drecon are a funcon of he dsagregaon level ha he analyses consder. Effec exchange rae volaly has played resuled hghly relaed o counres specfcs, and hese specfcs manly mean demand/supply elasces o relave prces. Relave prces become he ransory facor of he volaly effec on rade and he deermnng facor of he relaon frmness and drecon. Nonunformy of he exchange rae volaly effec on rade volumes when consderng dfferen rade parners s a fac ha draws some aenon on he role hs facor may play n makng comforable rade advanages regonally or n EU markes. Exchange rae volaly effec on mpors have been frmer n long erm, whle expors become vulnerable o he currency behavor n shor erm sep as well as when he analyses are dsaggregaed n counry erms. Trade sensvy o he wo mporan currences (USD and EURO appears dfferen, ye no oo sensve o EURO movemens, n boh long and shor run. 1

13 References: Asseery, A. and D.A. Peel (1991 The Effecs of Exchange Rae Volaly on Expors Economc Leers, 37, Baley, M.J., G.S. Tavlas, and Ulan, M. (1987 The Impac of Exchange-Rae Volaly on Expor Growh: Some Theorecal Consderaons and Emprcal Resuls Journal of Polcy Modelng, 9, Bn-Smagh, L. (1991 Exchange Rae Varably and Trade: Why s so Dffcul o Fnd Any Relaonshp Appled Economcs, 3, Bank of Monhly exchange rae Sascs ( Broda C. and Romals J. (004 Idenfyng he relaonshp beween rade and exchange rae volaly NBER workng paper. Barkoulas, T. John, Baum,F. Chrsopher, and Caglayan, Musafa (00 Exchange rae Effecs on he Volume and Varably of Trade Flows. Journal of Inernaonal Money and Fnance, Vol. 1, Calvo, G. A. and C.M. Renhar (000a Fear of Floang NBER Workng Paper No Coe A. (1994 Exchange rae volaly and rade, a Survey, Inernaonal Deparmen, Bank of Canada- Workng paper Caporale, T. and K. Doroodan (1994 Exchange Rae Varably and he Flow of Inernaonal Trade Economcs Leer 46, Chowdhury, Abdul R. (1993 Does Exchange Rae Volaly Depress Trade Flows? Evdence from Error Correcon Models The Revew of Economcs and Sascs, 75, DeGrauwe, P. (1988 Exchange Rae Varably and he Slowdown n Growh of Inernaonal Trade. Inernaonal Moneary Fund Saff Papers, 35, Durbn, J. (1970 Tesng for Seral Correlaon n Leas Square Regresson when some of he Regressors are Lagged Dependen Varables Economerca, 38, DeNards S. and Vcarell C (003 The mpac of he Euro on Trade, The (early effec s no so large ENEPRI Workng Paper Nr. 17 Dell Arcca G. (1999 Exchange rae flucuaons and rade flows, evdences from he European Unon, IMF Saff Papers, Vol. 46, Nr.3 Engle, R. F. and C.W.J. Granger (000 Long-Run Economc Relaonshps Oxford Unversy Press. 13

14 Engle, R, and C.W.J. Granger (1987 Coneraon and Error Correcon: Represenaon, Esmaon and Tesng Economerca, Vol. 55, Franke, G. (1991. Exchange Rae Volaly and Inernaonal Trade Journal of Inernaonal Money and Fnance, Vol. 10 (June, Feensa C. R and Kendall D.J (1991 Exchange Rae Volaly and Inernaonal Prces NBER Workng Seres No = Hooper, P. and S. Kohlhagen (1978 The Effec of Exchange Rae Uncerany on The Prces and Volumes of Inernaonal Trade Journal of Inernaonal Trade, 8, Inernaonal Moneary Fund. (1984 Exchange Rae Volaly and World Trade Washngon D.C. Inernaonal Moneary Fund - Inernaonal Fnancal Sascs ( Inernaonal Moneary Fund Counry Repor 003 Josef C. Pckard (003 Exchange rae volaly and blaeral rade flows: an analyss of U.S demand for ceran seel producs from Canada and Mexco Vrgna Polyechnc Insue and Sae Unversy, Economcs Mckenze, M. D. (1999 The Impac of Exchange Rae Volaly on Inernaonal Trade Flows Journal of Economc Surveys, Vol. 13, pp Mançellar A. e al (1999 Kurse e këmbm dhe ranzcon ekonomk Rose, A and Engle, CH. (000 Currency Unons and Inernaonal negraon Naonal Bureau of Economc Research, Inc. workng paper nr. 787 Rose, A. (1990 Exchange Raes and Trade Balance: An Emprcal Evdence from Developng Counres Economcs Leers, Vol. 34, (3, Rose K. A and Flood P. R (1999 Undersandng Exchange rae volaly whou he conrvance of Macroeconomcs Haas School of Busness, Unversy of Calforna Rahmasyah, T e al, (00 Exchange rae volaly, Trade and Fxng for Lfe n Thaland CIES Dscusson paper No. 01 Suj K. Das (003 Effec of exchange rae volaly on nernaonal rade, an emprcal analyss Vanderbl Unversy, Deparmen of Economcs, TN 3735 Taglon D.(00 Exchange rae volaly as a barrer o Trade, New Mehodologes and Recen Evdence Econome Inernaonale 89-90(00 p Wang, K. and Barre, C.(00 A new look a he rade Volume effec of Real Exchange rae Rsk Deparmen of Appled Economcs and Managemen, Cornell Unversy, Ihaca, New York Workng Paper

15 APPENDIXES: Table 1. Counry OLS Esmaons - EXPORTS Varables Ialy Greece Germany MSDA ARCH/GARCH MSDA ARCH/GARCH MSDA ARCH/GARCH Consan ( log( Y (5.446 log( p ( V ( ( ( ( ( ( ( ( ( ( ( ( ( ( ** -7.5 (-8.14 ** (8.04 ** (9.315 ** ( ( (- ( Dummy ( ( ( ( ( (-.43 AR-erm ( R Adjused R F-sasc Durbn- Wason Table. Counry OLS Esmaons Impor Funcon Varables Ialy Greece Germany MSDA ARCH/GARCH MSDA ARCH/GARCH MSDA ARCH/GARCH Consan (-1.67 * (.088 ** (6.058 ** (4.546 ** log(y (.133 ** (.0885 ** (.744 ** (.07 ** (-1.4 (-1.38 log( p (- ( ** (1.64 * (1.65 (.138 ** (.48 ** ** V (0.138 ( (-.18 ** (0.847 ( * ( Dummy (0.89 (0.904 (0.55 ( (3.76 ** (3.4 ** AR-erm ( (.568 R Adjused R F-sasc Durbn- Wason * Sgnfcance a 5% level of sgnfcance, ** Sgnfcance a 1% level of sgnfcance 15

16 Table 3. ADF-Tess resuls Ialy I(0 I(1 Number of Lags - varable MacKnnon Crcal values Dfference Order Log(Expors I(1 Log(Impors I(1 log( Y ** I(1 log( Y * I(1 log( p I(1 V (MSDA I(1 V (ARCH_GARCH I(1 RER(MSDA I(1 Greece Log(Expors I(1 Log(Impors I(1 log( Y ** I(1 log( Y I(1 log( p I(0 V (MSDA I(1 V (ARCH_GARCH I(1 RER(MSDA I(1 Germany Log(Expors I(1 Log(Impors I(1 log( Y ** I(1 log( Y 3 I(1 log( p I(1 V (MSDA I(1 V (ARCH_GARCH I(1 RER(MSDA I(1 * Sgnfcance a 10 % level of confdence ** Shows sgnfcance a 5% level of confdence 16

17 Nomnal exchange rae - n me Jan-93 Jul-93 Jan-94 Jan-95 Jul-94 Jul-95 Jan-96 Jul-96 Jul-97 Jan-97 Jul-98 Jan-98 Jan-99 Jul-99 Jan-00 Jul-00 Jul-01 Jan-01 Jan-0 Jul-0 Jan-03 Jul-03 USD DM Dr L* Table 4. Expors- Impors reacon oward USD/EURO Nomnal exchange rae EURO/LEKE Nomnal exchange rae USD/LEKE Varables Impors Expors Impors Expors Log(NER (-0.68 ( (-.90 ** (-1.66 * C ECT ( (-1.49 log(x (-4.13 log(x (-.768 log(m ( (-.798 ** log(m ( ** ( ** (-3.15 ** (-1.6 (-0.94 log(ner T (1.51 (-0.4 (-.04 ** (-3.36 ** log(ner T ( ( (0.90 Adjused R Varables are monhly me seres sarng from January 1993, unl November 003 * - Denoes sascal sgnfcance a 10% level of confdence, ** - Denoes sascal sgnfcance a 5% level of confdence 17

18 Table 1. Descrpve on nomnal exchange rae Currency daa USD DM DR LIT EURO Sd. Devaon Average Rao Sd.devaon/Averarage NER (% 18% 14% 14% 16% 6% Sd. Devaon whou year Average Rao Sd.devaon/Averarage NER (% 19% 11% 11% 13% 6% Sd. Devaon from 1998 unl Average Rao Sd.devaon/Averarage NER (% 7% 1% 13% 13% 6% Volaly ndex Nomnal exchange rae - Average Nomnal exchange rae S.D Response of expors o One S.D. Innovaons n ncome varable, relave prces and exchange rae volaly Response of mpors o one S.D nnovaon n counry GDP, relave prces and exchange volaly LGDPAL LPT RERV LGDP LPT RERV1 18