Aggregate Exports Response to Trade Openness: Bounds Testing Approach for Pakistan

Transcription

1 World Applied Sciences Journal 17 (1): , 01 ISSN IDOSI Publicaions, 01 Aggregae Expors Response o Trade Openness: Bounds Tesing Approach for Pakisan Rao Muhammad Aif, Iqidar Ali Shah and Khalid Zaman Deparmen of Managemen Sciences, COMSATS Insiue of Informaion Technology, Abboabad, Pakisan Absrac: Trade liberalizaion has he poenial o promoe expors which plays dynamic role in he growh and developmen process of a naion s economy. This paper examines he impacs of rade openness measures on aggregae expors in Pakisan for he period of by using Auo Regressive Disribued Lag (ARDL) approach. The empirical finding reveals ha here exiss a unique long-run relaionship among real aggregae expor, rade openness policies, producion capaciy and world income. As, expor duy shows a saisically significan impac on expors in he shor run, however, in he long run expor duies aesed o be insignifican. World income and producion capaciy boh appears highly elasic in boh he shor run and in he long run. The relaive price is significan in he shor run bu insignifican in he long run which reflecs ha any change in relaive price due o exchange rae flucuaion will increase expor in he shor run. The rade openness dummy which refleced he impac of rade openness polices adoped in differed ime period prove o significan in he long run and insignifican in he shor which reflec he fac ha expors responds o hese policies in he longer ime period hen he shor run period. The shor-erm dynamic behavior of Pakisan s expor supply has been invesigaed by esimaing an error correcion model in which he error correcion erm has been found o be correcly signed and saisically significan. The resuls are robus no only in erms of saisical powers, bu also in erms of economic insinc. Jel Classificaion Code: E31 F41 Key words: Expors Trade openness Relaive prices Coinegraion Pakisan INTRODUCTION inpus for expor producion eliminaes he inpu ax source of bias and lowering ariffs on impors of raw Trade liberalizaion has he poenial o promoe maerials and capial good inpus for expors improves he expors which plays dynamic role in he growh and compeiiveness of he expor secor vis-à-vis he res of developmen process of a naion s economy [1]. he world. According o UNCTAD [] repor also emphasis he role Therefore, almos all he developing counries of of rade in deermining economic and social performance Asia, Lain America and Africa have adoped rade of counries, paricularly developing counries. openness policies over he las four decade. Inernaional rade encourages growh by enabling Similarly, Pakisan has also adoped rade openness counries o specialize in goods and services, increasing policies since mid 1980s under he series of Srucural compeiion and encouraging echnological change Adjusmen Programme (SAP) developed by Inernaional based on compeiive and comparaive advanage [3]. Moniory Fund and World Bank. The objecive of he As a resul, he world would be able o consume more and paper is: beer qualiy producs a lower prices which will increase oal world welfare. According o Milner [4], rade To examine he poenial impac of rade openness liberalizaion promoes expors by eliminaing he srong policies i.e. expors duies and oher non ariff ani-expor biases creaed by quaniaive resricions and barriers on expor growh boh in he shor run and ariffs. For example, liberalizing ariffs on inermediae long run. Correspondence Auhor: Khalid Zaman, Deparmen of Managemen Sciences, COMSATS Insiue of Informaion Technology, Abboabad, Pakisan. 91

2 World Appl. Sci. J., 17 (1): , 01 To examine how changes in price impac of rade openness on expor growh for compeiiveness, world income and producive developing counries of Asia, Lain America and Africa, capaciy affec he expor performance in he shor including Pakisan for ime periods. He run and long run. applied he dynamic panel daa models based on Fixed Effecs and Generalized Mehods of Momens The sudy is arranged in he following manner: esimaors. He concluded ha expor duies have negaive lieraure review is carried ou in secion while impac on expor growh and rade openness has daa source and variables are menioned in secion 3. significan posiive impac on expor growh and caused Research mehodology and economeric model is an increase in expor growh by percenage poins. explained in secion 4. Resuls and discussion are Moreover, he also found ha real exchange rae saed in secion 5, while he paper concludes in he las depreciaion posiively effec expor growh alhough secion. he size of elasiciy is small. The sudy also observed a posiive impac of world income growh on he expor of Lieraure Review: The idea ha he rade openness developing counries. plays an imporan role because i increases he growh Similarly, using ime series daa, Jayanhakumaran of expors and impors has been focused in various [11] examined he impac of rade openness reforms on sudies. Waler [5] sudied he effecs of non ariff manufacuring expors during he ime period of barriers (NTB) on expor growh for developing counries for Ausralia. The sudy concluded ha here is srong and concludes ha NTB operae almos he same way as link beween falling proecions and increase expors do he ariffs. According o him, impor charges have growh of manufacuring secor. His findings also similar effecs o hose of quoas and he impac of showed ha he relaionship beween expor growh and selecive subsidies o impor compeing supplier is inra-indusry rade is significan among he branches likewise similar o ha of ariffs as regards he volume of where impors increased and his seems o highligh he rade and he naional economic srucure bu quie significance of inra-indusry rade in encouraging expor differen wih respec o he fiscal aspecs and domesic growh. The expor performance of Bangladesh afer he prices. The sudy also concludes ha in some developing rapid rade openness in he early 1990s was also counries oher supply side facors like domesic evaluaed by Mohiuddin [1], using he secondary daa. absorpion, qualiy conrol, markeing echniques and The sudy concluded ha he rade-gdp raio of heir own impor resricions were judged even more Bangladesh has increased significanly in 1990s reflecing serious barriers o increase expors han he ariff and non he greaer openness of he economy o he exernal world ariff barriers. because is expors showed increase in 1990s as compared Using Flavery-Gemmell Modified Model, Greenway o previous decade. Expor srucure also changed from & Sapsford [6] examined he impac of rade openness he primary goods and radiional iems o he secondary measures on expors and economic growh for 19 and non-radiional iems. The dominan expor iem, he developing counries. The sudy concluded ha he rade ready made garmens which covered almos he 80 percen openness measures lead o a significan increase in of oal expors may become challenging and difficul o aggregae expors. Chadha [7] also find he posiive mainain afer he abolishmen of Muli-Fiber Arrangemen impac of reducion in NTB on expors growh for India (MFA) quoas. Wu and Zeng [13] examined he impac of using CGE model. Ahmed [8] sudied he relaionship rade openness measures on impors, expors and he beween expor growh and rade openness measure in balance of rade for 39 developing counries of Asia, Lain Bangladesh for by applying co-inegraion and America and Africa. The resuls esimaed showed a error correcion model echniques and found shor run as srong and consisen indicaion ha rade openness in well as long run relaionship beween inernal and exernal developing counries promoes boh impors and expors. variables. Similarly, Sharma e al [9] sudied he impac of Adewuyi & Akpokodje [14] examined he impac of rade rade openness measures on expor growh for Nepal and openness on Nigeria s impor and expor flows. did no find any relaionship beween proecion measures Economeric analysis revealed ha he world income has and expor growh. By aking he expor duies as a significan augmening impac on he aggregae expors. measure of rade openness, Paulino [10] esimaed he However, rade openness reforms have no originaed an 9

3 impac ha is powerful enough o enhance Nigeria s rade (SAFTA) on Pakisan was analyzed by Shaikh & flows. Shamsadini e al. [15] examines he relaionship Rahpoo [0]. They applied CGE model and he resul beween openness and growh in he seleced MENA suggesed ha Pakisan would experience highes counries over a period of The resul reveals welfare gain under he combine policy reform adoped ha here is a long-run relaionship beween openness in SAFTA hrough larger economies of scales in and growh in Algeria, Jordan, Kuwai, Lebanon and producion, increase in compeiion, specializaion, Syria. expansion of aggregae expors and greaer employmen In case of Pakisan, various researchers have opporuniies. analyzed he impac of devaluaion and oher rade relaed The impac of rade openness on differen measures on expor growh using 3SLS (hree sage leas componens of balance of paymens for various square) echnique. Hasan and Khan [16] analyzed he developing and developed counries has been focused impac of devaluaion on rade balance, expors, impors in various sudies. However, very lile work has been and price level of Pakisan for he period of done in he conex of Pakisan so far. Therefore, he The sudy concluded ha he policy of devaluaion presen sudy esimae he long run and shor run impac (exchange rae managemen) improve he rade balance. of rade openness measures on aggregaes expors, Moreover, he Marshall Learner condiion of devaluaion impors and rade balance in Pakisan by using ARDL for Pakisanis is also saisfied which furher proves he model. argumen ha he devaluaion is o be successful in improving rade balance. The impor and expor elasiciies Daa Source and Variables Consrucion: The sudy were also esimaed by various researchers for Pakisan. implies annual observaions for he period of Afab and Aurangzeb [17] re-examined he long-run The daa is obained from World Developmen elasiciies and exisence of Marshall Lerner condiion. Indicaors published by he World Bank [1] and They empirically esed he rade performance of Pakisan GoP []. The variables used In his sudy are: wih en major rading parners and reconfirmed he aggregae expors and impors (fob), rade balance, saisfacion of Marshall - Lerner condiion using quarerly Real Gross Domesic Produc of he world and domesic daa for he period of By using Johansen counry (Pakisan), nominal exchange rae, expors co-inegraion echnique, Afzal [18] re-invesigaing and impor duies, rade openness and he ineracion Marshal Learner condiion for Pakisan by esimaing he dummies. All he daa was firsly convered in o long run rade elasiciies for he period of 1960 o 003. billion rupees (i.e., local currency uni of Pakisan) and The sudy concluded ha alhough Marshal Learner hen heir real values were obained by keeping 005 as condiion holds rue bu despie his he rade balance base year. does no improve significanly. I may be due o ha he devaluaion ses oher forces in moion ha end o Consrucion of Variables and Their Jusificaions: neuralize he posiive effecs of devaluaion. However, The following variables have been used in his sudy i.e., his sudy did no indicae he impacs of rade openness on impors and expors by using any appropriae measure Producion Capaciy of he Economy: Producion Capaciy of openness or liberalizaions. By using Lucas s is he supply side deerminan of expors [9, 3, 4]. An endogenous growh model, Dua and Ahmed [19] increase in producion capaciy will cause an increase in empirically esed he impac of rade policies on indusrial expors growh because addiional oupu can earn growh in Pakisan during he period The error foreign exchange by selling i in world marke. Therefore, correcion modeling and co-inegraion echniques were he posiive impac of GDP on expors growh is expeced. applied. The empirical finding suggesed ha here a In empirical lieraure, Kumar [5], Ahmed [4], Thirlwall unique long-run relaionship among growh of value [6] and Lopez [3] confirmed he posiive impac of GDP added oupu and, impor duies raes and real on expors. expors. The shor erm dynamic behavior of Pakisan's value added oupu performance was also found o Price Compeiiveness: A fall in he relaive domesic be saisically significan wih convenional signs. prices due o exchange rae depreciaion makes expors The effec of Souh Asian Free Trade Agreemen cheaper in inernaional markes resuling in increased 93

4 demand for expors; herefore we expec he posiive impac of real exchange rae on expor growh. Thirlwall [5] and Lopez [3] confirmed he inverse relaionship 1 beween expors growh and price compeiiveness. Real World GDP: World income has posiive impac on expors growh. An increase in world income level will lead o an increase expor growh. Thirllwall [5] and Lopez 16] esimaed he posiive relaionship beween world income and expor growh. Expor Duies: Expor ariff inversely effec he expor growh. An increase in expor duies will end o decrease expor growh and vice versa. Lopez [3] verified he inverse relaionship beween expor duies and expor growh. The daa of expor duies was colleced from CBR. Expor duies are define as he levies colleced on goods a he ime of expor o oher counry. We find he expor duies simply by dividing expor duies wih heir respecive value of he expors. Trade Openness Dummy: Pakisan governmen adoped he mixure of reforms packages for rade liberalizaion policies in he mid 1980s, herefore, he presen sudy aken rade openness dummy variables for finding he impac of hese reforms in he conex of Pakisan. Dummy variable akes he value of zero prior o he rade openness episode and one aferwards. Similarly by following he mehodology of Paulino and Thirwall [7], he presen sudy also inroduces wo ineracion dummies o capure he impac of rade openness measures on relaive prices and world income. There is a lack of inernaionally defined and globally comparable measure of a counry s rade orienaion because here are imporan concepual difficulies in defining and measuring rade openness. Several measures for rade openness are available in he curren lieraure adoped by differen researchers. For example some researchers define rade openness as he raio of expor plus impors o GDP while oher used dummy variable. Here we do no use impor plus expors o GDP raio index as a measure of rade openness because our inenion is o invesigae he effec of rade openness on aggregae impors and expors and if we use he index of openness iself includes he aggregae expors and impors as a variable hen he relaionship may become specious. Therefore we use openness dummy as an indicaor for he year in which major rade openness ook place. In order o measure he rade openness, he main approach used is he idenificaion of he years where he iming of openness is assessed by a se of guideline such as ariff reforms, relaxaion of non ariff barriers, expor promoion and exchange rae aleraion [10]. Moreover, i is also a commonly used index in he exising lieraure which capures also he exen of impor openness ha migh have spill over effec on expors. Therefore, openness dummy is bes suied for our sudy. Mehodological Frameowrk: To analyze he impac of various facors on aggregae expors for Pakisan, he following aggregae expors funcion is developed: Ex = f (P, Yw, PGDP, EXD, OP, OP YW, OP P ) (1) Where, Ex represens he aggregae expors; P is he Price compeiiveness; Yw is he world income PGDP is he Producion Capaciy of he Economy; OP is he rade openness dummy; EXD is raio of expor duies o oal expors where as OP*YW and OP*P are ineracion dummy capure he impac rade openness measures on relaive prices and world income. Economeric Framework of he Sudy: In his sudy auoregressive disribued lag (ARDL) model or bounds esing approach [8] has been used o check he exisence of shor and long-run relaionships beween expors growh, world income, price compeiiveness, producion capaciy, expor duies and rade openness dummies in he specific conex of Pakisan. Economeric heory designae a se of variables is co-inegraed if here is a linear combinaion among hem wihou sochasic rend. In his case, a long-run relaionship subsiss beween hese variables. However, his implicaion is only valid if he obligaion of he same order of inegraion has been me. Assume an explanaory variable, which is saionary a level is regressed wih anoher variable, which is non-saionary a level bu is firs-difference saionary, hen his will capiulae a spurious regression and hereby give a decepive and erraic conclusion. The use of he bounds echnique is based on hree validaions. Firs, Pesaran e al. [8] advocaed he use of 1 I is he rae of change of compeiiveness. P = P f+ NER-P d, Where; P = is he real exchange rae; P f= is he foreign price (CPI and USA), P d= is he domesic price (CPI of Pakisan); NER = nominal exchange rae The daa for World GDP is aken by he difference beween world GDP and domesic counry GDP, i.e. Y w = (World GDP-GDP). 94

5 he ARDL model for he esimaion of level relaionships LEX = 0 + 1LEX 1+ LP 1+ 3LPGDP 1+ because he model suggess ha once he order of he 4LEXD 1+ 5LYW 1+ 6OP + 7( OP YW ) + ARDL has been recognized, he relaionship can be p q esimaed by OLS. Second, he bounds es allows a 8( OP P) + 9 ( LEX ) i + 10 ( LP) i + mixure of I(1) and I(0) variables as regressors, ha is, i= 1 i= 0 he order of inegraion of appropriae variables may no r s necessarily be he same. Therefore, he ARDL echnique 11 ( LPGDP) i + 1 ( EXD) i + i= 0 i= 0 has he advanage of no requiring a specific idenificaion of he order of he underlying daa. Third, his echnique 13 ( LYW ) i + u is suiable for small or finie sample size [8]. i= 0 () Following Pesaran e al. [8], we assemble he vecor auoregression (VAR) of order p, denoed VAR (p), for he Where is he firs-difference operaor and u is a whiefollowing growh funcion: noise disurbance erm where as res of he variables are explained earlier. Equaion () indicaes ha aggregae p expors end o be influenced and explained by is pas Z = + iz i + values. The srucural lags are esablished by using i= 1 minimum Akaike s informaion crieria (AIC). From he Where z is he vecor of boh x and y, where y is he esimaion of UECMs, he long-run elasiciies are he dependen variable defined as Aggregae Expors, x is he coefficien of one lagged explanaory variable (muliplied vecor marix which represens a se of explanaory by a negaive sign) divided by he coefficien of one variables i.e., World income level ;Price compeiiveness, lagged dependen variable [9]. For example, in equaion Producion Capaciy, Expor duies and Trade openness (), he long-run inequaliy, invesmen and growh dummy and is a ime or rend variable. According o elasiciies are ( / 1), ( 3/ 1), 4/ 1and 5/ 1respecively. Pesaran e al. [8], y mus be I(1) variable, bu he The shor-run effecs are capured by he coefficiens of regressor x can be eiher I(0) or I(1). We furher he firs-differenced variables in equaion (). developed a vecor error correcion model (VECM) as Afer regression of Equaion (), he Wald es follows: (F-saisic) was compued o differeniae he long-run p i p 1 relaionship beween he concerned variables. The Wald z = + + z 1 + y i + x i + es can be carry ou by imposing resricions on he i= 1 i= 1 esimaed long-run coefficiens of aggregae expors and is deerminans. The null and alernaive hypoheses are Where is he firs-difference operaor. The long-run as follows: muliplier marix as: H: 0 1 = = 3 = 4 = 5 = 0. (no long-run relaionship) YY YX Agains he alernaive hypohesis = H A: (a long-run relaionship XY XX exiss) The diagonal elemens of he marix are unresriced, so he seleced series can be eiher I(0) or I(1). If YY = 0, The compued F-saisic value will be evaluaed wih hen Y is I(1). In conras, if YY < 0, hen Y is I(0). he criical values abulaed in Table CI (iii) of The VECM procedures described above are Pesaran e al. [8]. According o hese auhors, he lower imperaive in he esing of a mos one coinegraing bound criical values assumed ha he explanaory vecor beween dependen variable y and a se of variables x are inegraed of order zero, or I(0), while he regressors x. To derive model, we followed he upper bound criical values assumed ha x are inegraed posulaions made by Pesaran e al. [8] in Case III, ha of order one, or I(1). Therefore, if he compued F-saisic is, unresriced inerceps and no rends. Afer imposing is smaller han he lower bound value, hen he null he resricions YY = 0 µ 0 and = 0, he funcion can be hypohesis is no rejeced and we conclude ha here is saed as he following unresriced error correcion model no long-run relaionship beween aggregae expors and (UECM): is deerminans. Conversely, if he compued F-saisic 95

6 is greaer han he upper bound value, hen aggregae he ADF and PP saisics do no exceed he Mackinnon expors and is deerminans share a long-run level criical values). However, when we ake he firs difference relaionship. On he oher hand, if he compued of hese variables, hen he ADF and PP saisics are F-saisic falls beween he lower and upper bound higher han heir respecive criical values (in absolue values, hen he resuls are inconclusive. erms). Bu he log of world income (LYW ) is saionary a level, here saisic is lower han he criical value RESULT AND DISCUSSION wihou aking he firs difference, herefore, we considered as I(0) variable. The Augmened Dickey-Fuller (ADF) and Phillips- This provides a good raionale for using he bounds Parron (PP) uni roo ess were used o check he order of es approach, or ARDL model, proposed by Pesaran e al inegraion of ime series variables. The resuls are [8]. The wo sep ARDL co inegraion procedure is repored in Table 1. The es resuls shown ha he log of aken for Pakisan by using annual observaion over he he aggregae expors (LEX ), log of price compeiiveness periods of 197 o 010. (LP ), log of he raio of expor duies and oal volume of The regression resuls of equaion are accouned in expors (LEXD ) and log of producion capaciy of he Table. In order o regress he auoregressive disribuive economy (LPGDP ) are non saionary series (i.e. a level, lag (ARDL) economeric model. We follow he general o Table 1: Augmened Dickey-Fuller (ADF) Tes on he levels and on he Firs Difference of he Variables ( ) Variables Consan/ consan & rend Level 1s Difference Order of Inegraion ADF Uni roo es LEX Consan (0) (0)*** I(1) LP Consan (0) (1)*** I(1)) LEXD Consan -.080(0) -6.4 (1) *** I(1) LYW Consan and rend -.041(0) * -4.71(0)*** I(0) LPGDP Consan 0.17 (0) -3.69(0)** I(1) Phillips-Perron Uni Roo Tes LEX Consan -0.63(3) -6.69(0) *** I(1) LP Consan 0.145(3) (1)*** I(1)) LEXD Consan (3) -6.(1) *** I(1) LYW Consan and rend -.1(3) * -4.7(0)*** I(0) LPGDP Consan (4) -3.69(0) *** I(1) Noe: The null hypohesis is ha he series is non-saionary, or conains a uni roo. The rejecion of he null hypohesis is based on MacKinnon (1996) criical values i.e., a level (consan): , and , a (consan linear rend): , and are significan a 1%, 5% and 10% level respecively. While a firs difference (consan): , and -.610, a (consan linear rend): , and and are significan a 1%, 5% and 10% level respecively.the lag lengh for ADF are seleced based on SIC crieria, his ranges from lag zero o lag four. The lag lengh for PP es is seleced based on Barle Kernal crieria, his ranges from lag zero o lag seven. * and *** represens 0.09 percen and 0.01 percen significance level. Table : Esimaed Model Based on Equaion () Variable Coefficien -Saisic Prob. Consan *** LEX *** LP LPGDP *** LEXD LYW *** OP *** OP YW OP P LEX LP ** LPGDP.0769*** LPGDP * LEXD ** LYW ** LYW * R-squared Akaike info crierion Adjused R-squared Schwarz crierion Durbin-Wason sa F-saisics 11.85*** Noe: *, ** and *** represens 10%, 5% and 1% significance level. 96

7 Table 3: Diagnosic and Sabiliy Tes F-Saisics Probabiliy NORM WHITE RAMSEY ARCH Seral Corr Noe: For normaliy es, we repor Jarque-Bera saisics. NORM, WHITE, RAMSEY, ARCH, Serial Corr, are non-normal errors normaliy es, whie Heeroscedasiciy es, Ramsey Regression Specificaion Error Tes and Auo regressive Condiional Heeroscedasiciy (ARCH Tes), Serial correlaion Lagrange Muliplier Tes (LM-ype Breusch-Godfrey-Tes). These saisics are disribued as Chi-square values and capure degree of freedom in parenhesis. specific rule in order o remove he highly insignifican or be presen an evidence of serial auocorrelaion ha errors less significan variables from our regression analysis. are normally disribued. Ramsey es also indicaes ha This process may help us o follow he imporan propery he resul is in favor of sabiliy of parameers. Table 4 of auoregressive disribuive lag (ARDL) model and repors he shor and long-run elasiciies of real make our regression analysis mos appropriae or aggregae expor funcion. desirable. The above procedure in esimaing equaion () The long-run coefficiens are derived by may appear as resul of one or none lags wih he normalizaion process in which co-inegraion vecor erm explanaory variables. Ineresingly, i capures all lags are divided by all explanaory variables. This hibernaing ha may have desirable significance power in erm of procedure may help us o cach up long-run esimaes of saisical inference. Table shows all he variables are real exchange rae o real oupu model. The derived or significan in he long run excep he price calculaing resuls are repored in Table 4. The heory compeiiveness and expor duies. However, in he explains ha here is a posiive effec of real world income shor-run, all explanaory variables are significan in and rade openness policies on real aggregae expors he shor run. where as expor duies and exchange rae have inverse The goodness of fi, model selecion crieria and he relaionship wih expor. The regression resuls prove resuls of Durbin-Wason saisics ha capure he hese argumens as here is a saisically significan problem of serial auocorrelaion are also repored in posiive relaionship beween real world income and real Table. Since, our dependen variable is real aggregae aggregae expors boh in he long run and he shor run. expors and is reacion along wih se of independen The coefficien of 1.5 and.7 suggess ha 1% change in variable. The value of adjused R-square is 0.65 ha he world income will increase he aggregae expors o shows ha 65 percen variaion in he deerminaion 1.5% in he shor run and.7% in he long run. Similarly of real aggregae expors is explained by our model. he counry s producive capaciy is also saisically The value of Durbin-Wason saisics is 1.86 ha significan boh in he long run and in he shor saisfies o he desirable level and shows he rejecion run. Price compeiiveness (real exchange rae) is of null hypohesis of serial auocorrelaion of any order. saisically significan in he shor run bu in he The model selecion crieria are menioned by Akaike long run is impac is insignifican. Similarly, he Informaion Crieria (AIC) and Schwarz Crieria ha is impac of expor duies is saisically significan in he -.5 and , respecively. Table 3 show he saisics shor run alhough is co-efficien of depics of differen diagnosic and sabiliy es o check our all ha a 1 % fall in expor duies increase he aggregae sabiliy of he model. expor by 0.034%, however, i is insignifican in he Diagnosic and sabiliy es such as long run referring ha apar from expor duies here heeroscedasiciy, normaliy, serial correlaion, are some oher facors ha effec growh rae of auoregressive condiional heeroscedasiciy (ARCH) aggregae expors, for example, non quaniaive es and funcional-form misspecificaion (Ramsey resricion and oher counry specific insiuional Regression Specificaion Error Tes) are repored in Table elemens. Therefore he impac of rade openness 3 above. The impac of rade openness measure on reforms on aggregae expors seems o more aggregae expor passes all diagnosic ess. For normaliy significan (as rade openness dummy coefficien is 0.17% es, we repor Jarque-Bera saisics ha akes non- which is greaer han expor duies coefficien of 0.014%). normal errors, is value is along wih probabiliy In order o find ou he shor run dynamic value The null hypohesis of heeroscedasiciy is relaionship for real aggregae expor funcion, we employ rejeced ha suppor he evidence of no error correcion model (ECM). Table 5 shows he heeroscedasiciy across he erms. There also does no esimaion of error correcion model. 97

8 Table 4: Long-run and Shor run Elasiciies of Aggregae Expor Funcion Dependen Variable = LEX Variables Shor Run Long Run LP ** LPGDP * *** LEXD ** LYW *.706*** OP *** Noe: *, ** and *** denoe he significance level a 10%, 5% and 1% level of significance, respecively. Table 5: Error Correcion Model Variables Coefficien -Saisic Prob. LP *** LPGDP.07690*** LPGDP *** LEXD *** LYW *** LYW ** Consan *** ECT(-1) *** R-squared Akaike info crierion Adjused R-squared Schwarz crierion Durbin-Wason sa F-saisics 18.89*** Noe: ** and *** denoe he significance level a 5% and 1% level of significance, respecively. The error correcion erm indicaes he speed of aggregae expors. In he shor run, he aggregae expors adjusmen which resored equilibrium in he dynamic are significanly influenced by world income (i.e., 1.50), model wih in one year and i is calculaed from he long- producion capaciy (i.e., 1.63), prices (i.e., 0.55) and he run co-inegraion vecor. The ECM -1 coefficien shows expor duy raes (i.e., 0.034). how quickly variables reurn o equilibrium and i should The expor duy as measure of rade openness shows have saisically significan coefficien wih negaive a saisically significan impac on expors in he shor sign. The feedback coefficien of error correcion model run which means an increase in rade openness hrough erm is equal o which implies ha deviaion form he reducion in expor duies will increase he expors in he long erm expors is correced by abou 63.9% over he shor run. However, in he long run expor duies prove o following year. be insignifican. World income appears highly elasic in boh he shor run and in he long run suggesing ha CONCLUSION aggregaes expors are highly income elasic and depends upon he developed counries demand. The producion Using he Auo Regressive Disribued Lag Model capaciy also highly elasic boh in he long run and he approach wih annual daa from 197 o 010, he sudy shor run reflecing he fac ha our producion consiss invesigaed he impacs of rade openness measures on on expor oriened indusries. The relaive price is he aggregae expors of Pakisan. I was observed significan in he shor run bu insignifican in he long ha he variables in he aggregae expors funcion are run which reflecs ha any change in relaive price due o co-inegraed. Price, income, producion capaciy, expor exchange rae flucuaion will increase expor in he shor duies rae and he rade openness dummy all significanly run. The rade openness dummy which refleced he influence aggregae expors in Pakisan wih consisen impac of rade openness polices adoped in differed ime signs. In he long run, he rade openness dummy shows period prove o significan in he long run and a significan bu low posiive (0.17) impac, while he insignifican in he shor which reflec he fac ha expors impor duy rae as a measure of rade liberalizaion shows responds o hese policies in he longer ime period hen insignifican impac on he aggregae expors. Producion he shor run period. The error correcion erm in he capaciy shows he highes posiive impac (i.e., 3.01). model is found o be saisically significan, confirming Similarly, World Income also exhibis a posiive and he validiy of he long-run equilibrium relaionship. saisically significan impac (i.e.,.70); while The coefficien esimae of he error correcion erm is 0.69 compeiive prices (i.e., 0.045) show low impac on he indicaing a high speed of adjusmen o equilibrium. 98

9 The policy implicaions are sraighforward. 1. Mohiuddin, G.K.M., 005. Trade Liberalizaion and For rapid expansion of expors, rade liberalizaion policies need o be associaed wih promoion of public policies for greaer uilizaion of producive capaciy of he economy. This is because any rade openness measures which encourage expors mus be suppored by increased poenial of expors. However, for effecive Expor Performance of Bangladesh. The Cos and Managemen, pp: Wu, Y. and L. Zeng, 008. The Impac of Trade Liberalizaion on he Trade Balance in Developing Counries. IMF Saff Papers 08/14, 14. Adewuyi, A.O. and G. Akpokodje, 010. Impac of policy analysis sudies may be underaken using daa a he disaggregae level. Trade reforms on Nigeria s Trade Flow. The REFERENCES 1. Bakare, A.S. and F.O. Fawehinmi, 011. Trade Openness and Is Impac on Nigeria s Non-Oil Indusrial Secor: Economics and Finance Review, 1(5): UNCTAD Unied Naions Conference on Trade and Developmen, Trade and Developmen Repor 1999, Geneva. 3. López, P., 005. The Impac of Trade Liberalizaion on Expors, Impors, he Balance of Paymens and Growh: he Case of Mexico. J. Pos Keynesian Economics, 7(4): Milner, C., The role of impor liberalizaion in expor promoion. In Expor Promoion Sraegies: Theory and Evidence from Developing Counries, ed. C. Milner, pp Harveser/Wheasheaf, London. 5. Waler, I., Non-ariff Barriers and he Expor Performance of Developing Economies. The American Economic Review, 61(): Greenway, D. and D. Sapsford, Wha does liberalizaion do for expors and growh, Welwirschafliches, 130(1): Chadha, R., India s Expor Performance: Are There Lessons from Eas Asia?. Paper presened a NCAER, May 9, Ahmed, N., 000. Expor responses o rade liberalizaion in Bangladesh: a coinegraion analysis. Applied Economics, 3: Sharma, K., S. Jayasuriya and E. Oczkowski, 000. Liberalizaion, Expor Incenives and Trade Inensiy: New Evidence from Nepalese Manufacuring Indusries. J. Asian Economics, 1: Paulino, S.A., 00. Trade Liberalizaion and Expor Performance in Seleced Developing Counries. Journal of Developmen Sudies, 39(1): Jayanhakumaran, K., 004. The impac of rade liberalizaion on manufacuring secor performance in developing counries: a survey of he lieraure. Thammasa Economic J., (): Inernaional Trade J., 4(4): Shamsadini, S., R. Moghaddasi and M. Kheirandish, 010. Relaionship beween rade openness and GDP-a panel daa analysis. World. Appl. Sci. J., 8(7): Hasan, M.A. and A.H. Khan, Impac of Devaluaion on Pakisan s Exernal Trade: An Economeric Approach. The Pakisan Developmen Review, 33(4): Afab, Z. and M. Aurangzeb, 00. The Long-run and Shor-run Impac of Exchange Rae Devaluaion on Pakisan s Trade Performance. The Pakisan Developmen Review 41(3): Afzal, M. (004). Esimaing Long - Run Trade Elasiciies in Pakisan: Coinegraion Approach, The Pakisan Developmen Review, 43(4), Dua, D. and N. Ahmed, 004. Trade liberalizaion and indusrial growh in Pakisan: a Coinegraion analysis. Applied Economics, 36: Shaikh, F.M. and M.S. Rahpoo, 009. Impac of Trade Liberalizaions and SAFTA on Pakisan s Economy by Using CGE Model, Inernaional J. Business of Managemen, 4(4): World Bank, 010. World Developmen Indicaors (WDI). Washingon, DC. hp:// daa/wdi010. GoP Governmen of Pakisan, Economic Survey , Economic Advisor's Wing, Minisry of Finance, Islamabad. 3. Beril, O., Model Consrucion in Inernaional Trade Theory. Published in Inducion, Growh and Trade. Clarendon Press. Oxford., pp: Ahmed, N., 001. Trade liberalizaion in Bangladesh An invesigaion ino rends. Dhaka: The Universiy Press Limied (UPL). 5. Kumar, N., Mulinaional Enerprises, Regional Economic Inegraion and Expor-Plaform Producion in he Hos Counries: An Empirical Analysis for he US and Japanese Corporaions. Elwirschafliches 134: Thirlwall, A.P., 003. Growh and Developmen: wih Special Reference o Developing Economies, 7h ediion (London: Palgrave-Macmillan). 99

10 7. Paulino, S.A. and A.P. Thirlwall, 004. The Impac of 9. Bardsen, G., Esimaion of long-run coefficiens Trade Liberalizaion on Expors, Impors and he in error correcion models. Oxford Bullein of Balance of Paymens of Developing Counries. Economics and Saisics, 51: Economic J., 114(1): Pesaran, M.H., Y. Shin and R.J. Smih, 001. Bound esing approaches o he analysis of level relaionships. J. Applied Economerics, 16: