TRADE IMPLICATIONS OF EXTENDING THE TURKEY-EU CUSTOMS UNION TO AGRICULTURAL PRODUCTS *

Transcription

1 Prelmnary draft please, do not quote TRADE IMPLICATIONS OF EXTENDING THE TURKEY-EU CUSTOMS UNION TO AGRICULTURAL PRODUCTS * H. Ozan ERUYGUR Mddle East Techncal Unversty, Department of Economcs, Ankara 6531 Turkey Erol H. CAKMAK Mddle East Techncal Unversty, Department of Economcs, Ankara 6531 Turkey ABSTRACT Turkey s membershp of EU wll lead to the enlargement of already establshed customs unon between EU and Turkey to the agro-food products. Ths nvolves not only a full lberalzaton of agrcultural trade wthn the EU but also the mplementaton of a common external tarff. Trade dverson and creaton effects for agro-food trade wll emerge. Accordng to the artcle XXIV of GATT, the possble results of these counteractng effects are mportant. In ths paper, the trade dverson and creaton effects of the membershp of Turkey to the EU for the agro-food trade wll be calculated and analyzed usng the Armngton assumpton. Keywords: Elastctes of Substtuton, Armngton Elastctes, Fxed and Random Effect Panels, Trade Creaton, Trade Dverson, EU Membershp of Turkey. JEL Codes: C5, F15, F17, Q17 * The authors are grateful to the EU-MED AGPOL (Impacts of Agrcultural Trade Lberalzaton between the EU and Medterranean Countrest) Project funded by the European Commsson, 6 th Framework Programme on Research, Technologcal Development and Demonstraton.

2 I. INTRODUCTION Extendng Turkey-EU customs unon agreement to agrcultural products wll mply the adopton of a common external tarff (CET) and abolton of trade barrers between Turkey and EU. The blateral full lberalzaton of trade n agro-food products and the establshment of common external tarff wll have trade mplcatons not only for Turkey and EU, but also for the non-eu tradng partners of Turkey. In ths new stuaton, trade dverson and creaton effects for agro-food trade wll emerge. Accordng to the artcle XXIV of GATT, the possble results of these counteractng effects are mportant. The estmaton of the Armngton elastctes s descrbed n the followng secton. The thrd secton reports the smulaton results about trade dverson and creaton due to the applcaton of CET. The fnal secton s reserved for concludng remarks. II. ARMINGTON MODEL AND ESTIMATION OF ELASTICITIES The Armngton model assumes mperfect substtuton among goods from dfferent geographcal areas. The model uses a CES aggregaton functon whch mples that the substtuton of mports between any two pars of mportng partners are dentcal. Accordng to the choce of the CES functonal form, two dfferent specfcatons can be consdered. The non-nested specfcaton (Shells C. R. and Renert K. A., 1993, p.33) assumes that mports from regons or countres, as well as competng domestc producton all enter n the subutlty functon for a sector: 1 ρ ρ U = bm k k k (Eq.1) 2

3 where b k = 1, ρ s a constant greater than -1, and k 1 σ ρ =. Note that, ρ s the CES σ exponent and σ s the elastcty of substtuton where < σ <. 1 In ths CES functonal form, M k ncludes the quantty of domestc producton for good, as well. Tradtonally, CGE modelers assume that domestc producton substtutes wth an aggregate of mports from all sources. The second alternatve that Shells et al (1993) called nested specfcaton assumes that mports from dfferent sources are dfferentated products. In other worlds, n ths alternatve formulaton, M k does not nclude the quantty of domestc producton for good. Ths second form s generally used n order to analyze the preferental trade arrangements and/or customs unons. Ths nested specfcaton s exactly what has been used n ths study. Hence our model has the utlty functon of: U σ 1 σ 1 σ σ bm k k (Eq.2) k = Notce that n Eq.2, k represents the tradng partner, M k s the quantty of mports of product orgnatng from k, b k s a constant representng the level of preference for mports orgnatng from k. Armngton model mposes a two-step budgetng procedure. In the frst stage, the mporter decdes how much of a partcular commodty to mport. In ths stage the decson s determned accordng to the mport demand functon, M, of the mporter country, n other words, by the prce elastcty for total mport demand for product ; η. In the second stage gven the total amount mported, the mporter decdes how much to mport from each suppler. Ths decson s based on the elastcty of substtuton, σ. Solvng the consumer utlty functon gven n Eq.2 produces the followng equaton whch 1 If σ =, then the products are perfect complements, f σ = then the products are perfect substtutes. 3

4 determnes mport volume by sector and regon of orgn, M k, where P k s the partner specfc mport prce ncludng tarffs, P k = P (1 + t) where t s tarff rate. k σ o p k Mk = αkm P (Eq.3) where P = α P s the ndex of mport prces representng a prce for total mports from k k k all orgns, and α k s the quantty market share of country k n the base year. Note that Hckman and Lau (1973, p.351) showed that f we normalze our prces to unty n the base M k perod, then, one can show thatα k =. In ths case the Eq.3 can be rewrtten as M α k pk / p k o = αk P / p σ (Eq.4) where α = M / M. k k Armngton (1969, p.174) showed that takng the dfferental of both sdes of P = α P wll lead to: k k k dp P dp o k = Sk (Eq.5) k Pk where S = M P. Note that n ths study, t s assumed that the prce changes wll result o k. k k M P from tarff changes 2 new dp t t, so t s possble to wrte = P 1+ t old k k k old k k dfferental of Eq.3, Armngton (1969, p.174) showed also that dm k dm dp dpk = + σ M k M P Pk ( Effect1) ( Effect 2). In addton, takng the (Eq.6) 2 When Turkey enters to EU, a Common External Tarff (CET) of EU wll be appled by Turkey nstead of Turkey s current tarff rates. 4

5 dm dp where = η. The frst term represents the growth of the market for M k because of M P the prce change. Followng Unguru and Lozza (21, p.12), ths effect mples that the change n total mports wll be dstrbuted accordng to the ntal share of each partner. The second term represents the effect of relatve prce changes, or n other words the substtuton effect. The substtuton effect allows to estmate the trade dverson and to determne the wnners and losers of the CET across the tradng partners. Ths s the effect of substtutons between partner countres. Eq. 4 has been used for the estmaton. We wll get the followng equaton by takng the natural logarthm of Eq.4. α k pk / p k ln.ln o = σ (Eq.7) αk P / p In order to estmate ths equaton, we used the fxed and random effect models of panel data. We performed Hausman tests n order to choose the preferred model for each product,. Our approach s smlar to that of Unguru and Lozza (21). The man dfference s the fact that we performed Hausman tests n order to decde to fxed or random effect models of panel data, snce n some cases random effect model can be much more preferred to fxed model. Unguru and Lozza (21, p.26) used fxed effect model for all products. Notce that the estmatons are performed addng a trend term (trend) to (Eq.7) both n Fxed Effect and Random Effect specfcatons. For the panel data estmaton, the cross secton dmenson s regons, k, n other worlds country groups submtted to the same duty regme. The cross secton elements used n our study are k=eu15, EU1, USA, Chna, Latn Amerca, MENA and Rest of the World (ROW). The tme seres dmenson s t, that s years from 1992 untl 23. The model s estmated for each agro-food product group of our study, =1,2, 14 (for detals, see the Appendx). Followng Unguru and Lozza (21), n addton to the 14 dfferent agro-food products, we defned also the product groups such as Raw and Processed products. Ths extra classfcaton, n fact, doubles the number of products. 5

6 For the defnton of raw and processed products, we followed the defnton of EU. 3 The defntons are based on the Harmonzed System Combned Nomenclature snce the tarff data (Common External Tarf of EU) s based on ths codng system. 4 The ad-valorem equvalents of the CET (Common External Tarff) data are obtaned from UNCTAD database at 8 dgts of the Combned Nomenclature. The Turksh tarff data s obtaned from the Undersecretarat of Foregn Trade. The prce elastctes of mport, η, are estmated usng the smple specfcaton of: ln M = constant η ln P (Eq. 8) The estmated Armngton elastctes and elastctes of mports can be seen n Appendx. The values for these elastctes used n our smulaton study are provded n Table 1. All of the regressons are performed by Stata 8 and 23 s our base year. Table 1. Values for Elastctes of Substtuton and Prce Elastctes of Import. Elastcty of Substtuton, σ Prce Elastcty of Import, η Product Codes a Raw Processed Raw Processed Notes: a See Appendx Table A1 for the product names. The confdence nterval for all the elastctes n the table s.5. In the case that the estmated elastcty dd not fulfll ths requrement, we used the average elastcty obtaned from group of products. Source: Authors calculatons ex_en.htm 6

7 III. SIMULATION RESULTS The adopton of the CET ncreases the total agro-food mports of Turkey by 12.5 percent (Fgure 1). The surge n processed products are hgher wth 15.3 percent, compared to the raw products that remans around 12 percent. At ths pont, t may be nformatve to look at the recent dstrbuton (wth 23 data) of Turkey s total agro-food product mports by ther country of orgn. Fgure 2A, 2B, 2C and 2D are prepared for ths purpose. Fgure 2C represents that the two major mport partners of Turkey are USA and EU-15 wth smlar percentages (26-27 %). Although a smlar pattern can be seen n raw agro-food products (Fgure 2A), the pcture for processed agro-food products s hghly dfferent and reveals an mportant feature of Turkey s current agro-food mport structure (Fgure 2B). In the case of processed products, we see that EU15 s the leadng mporter wth 73 % whle the share of USA drops drastcally to 12 %. In Fgure 2D, one can see the evoluton of the mport structure of Turkey for processed agro-food products 7

8 from 1992 untl 23. After 1993 there s a drastc ncrease n the share of EU15 and from 1994 tll 23 we see a statonary fluctuaton around 7 percent. In fact, returnng back to Fgure 1, the total ncrease of % obtaned from our smulatons for the agro-food products used n ths study s dstrbuted unequally accordng to the orgns of mportaton. The change n Turkey s mports of agro-food products accordng to the country of orgn s presented n Fgure 3. It s clear from the fgure that the EU countres would gan the most. The results ndcate that the EU-15 countres would ncrease ther exports of agro-food products to Turkey by 22.3 percent compared to the year 23. Accordng to the product type (Raw or Processed), the Turkey s mports of processed products wll ncrease by 25 percent. Wthn the EU member countres, the new members (EU1) countres would ncrease ther exports the most. The overall ncrease of 27 percent reveals ths stuaton. However, the man dfference between the EU15 and EU1 countres performance s the fact that the man share of EU15 s ncrease would be due to the processed products (24.9 percent) whereas for EU1 8

9 countres ths ncrease wll result manly from raw agro-food products (27.7 percent). Ths result reveals the dfference of the structure of the agro-food sectors n EU15 and EU1 countres. EU15 s 24.9 percent ncrease n processed agro-food products s really strkng, consderng the fact that the share of the EU15 n processed products mports of Turkey has been 73 percent despte rather low base for the processed products mports. These results dsclose the fact that these sectors n Turkey should mprove ther compettveness n order to survve wth the ncreasng foregn competton whch would result from the enlargement of Turkey-EU customs unon agreement to agro-food products. From Fgure 3, t can be observed that the two least benefcares of a possble EU accesson of Turkey would be USA and Chna. Latn Amercan and MENA countres arse as the medum benefcares. Another nterestng fndng of the smulaton results s the fact that although Chna would not beneft too much n total the ncrease n ther exports of processed agro-food products s really hgh wth an ncrease of 16.3 percent compared to ts overall ncrease of 2.8 percent. 9

10 In Fgure 4, we see the allocaton of effects that are dscussed n (Eq.6) on total (raw and processed) mports of agro-food products of our study. Recall that the Effect 2 representng the substtuton effect because of the change n the relatve prces between partner countres whereas the Effect 1 represents the change (enlargement or shrnkage) n market because of the prce change. In Fgure 4, the most mportant pont s that, except EU1 and EU15, the substtutons between partner countres wll negatvely affect exports to Turkey. In other words, all countres except EU members would experence a really hgh negatve substtuton effects due to the change n Turkey s mport prces n favor of EU agro-food products n the case of a possble accesson of Turkey to EU. Effect 2 allows us to estmate the trade dverson and to determne the losers and wnners of trade substtuton (Unguru and Lozza, p.12). The smulaton results ndcate that the most mportant trade dverson would take place for Latn Amercan countres (wth 16.5 percent). If there s no change n relatve prces, the ncrease of Turkey s mports from Latn Amercan countres would be 35.1 percent, however, because of the change n relatve prces of partner countres n favor of EU the substtuton effect would moderate ths ncrease at a rate of percent. Ths rate s representng, n fact, the total net trade creaton. Smlar stuatons can be seen for all mporter regons except EU n Fgure 4. USA wll experence a trade dverson at a degree of 1

11 7.24 % but end up wth a net postve trade creaton of 4.57 %. MENA countres would lkely experence a trade dverson at a rate of 8.32 % wth a total net trade creaton of %. If we look at the sum of mports from all countres to Turkey, for the agro-food products of our study, a trade dverson of 7.88 % wth a 2.38 % trade creaton wll lkely result n a total net trade creaton at a rate of %. The wnners of the substtuton effect, not surprsngly, would be the EU countres. EU1 countres would be the most wnners of ths substtuton effects wth a rate of percent whle EU15 countres would be the second wnners of the substtuton effect resultng from reachng Turksh agro-food market wthout any tarffs. Lastly, rest of the world (ROW) would also experence a trade dverson at a rate of percent wth a fnal total net trade creaton of 12.7 percent. IV. CONCLUSION Turkey s membershp of EU wll lead to the enlargement of already establshed customs unon between EU and Turkey for the agrcultural products. Ths nvolves not only a full lberalzaton of agrcultural trade wthn the EU but also the mplementaton of a Common external tarff. In ths new stuaton, trade dverson and creaton effects for agro-food trade wll emerge. In terms of artcle XXIV of GATT, the possble results of these counteractng effects are mportant. In the frst part of the paper, we estmated the Armngton elastctes for Turkey for agro-food products of our study. Then, usng the Armngton assumpton, the trade dverson and creaton effects of Turkey s membershp for the agrcultural trade are calculated and analyzed. Our smulaton fndngs show that the wnners of a possble enlargement of Turkey-EU customs unon to agrcultural products are, not surprsngly, EU countres. Turkey s mports for agro-food products of our study wll ncrease by percent n total. The other countres wll also ncrease ther exports to Turkey, however, except EU, all of these countres wll be subject to some degrees of substtuton effects mplyng trade dversons for these regons. 11

12 REFERENCES Armngton P. S (1969), A Theory of Demand for Products Dstngushed by Place of Producton, IMF Stuff Papers, 16:1. Hckman B. G., and Lau L. J., (1973), Elastctes of Substtuton and Export Demands n a World Trade Model, European Economc Revew, vol.4, pp Shells C. R, and Renert K. A., (1993), Armngton Models and terms of Trade Effects: Some Econometrcs Evdence for North Amerca, Canadan Journal of Economcs, 26, pp Unguru M., and Lozza E. C., (21), The EU enlargement to the CEECs: Impacts on agrfood Trade wth Thrd Countres, ECOMOD Conference Paper. 12

13 APPENDIX TABLES A1. Products used, and defntons Our Codes, Descrptons Harmonzed System Codes 1 Lve anmals, meat & edble meat offal Dary, eggs, honey, & ed. Products 4 3 Edble Vegetables 7 4 Ed. Fruts & Nuts, Peel Of Ctrus/Melons 8 5 Cereals and Mllng Industry Products Ol Seeds/Msc. Grans/Med. Plants/Straw 12 7 Anmal Or Vegetable Fats, Ols & Waxes 15 8 Sugars & Sugar Confectonery 17 9 Preps. Of Cereals, Flour, Starch Or Mlk 19 1 Preps Of Vegs, Fruts, Nuts, Etc 2 11 Tobacco & Manuf. Tobacco Substtutes Other Foodstuffs Raw Hdes & Skns & Leather Cotton, Inc. Yarns & Woven Fabrcs Thereof 52 13

14 A2. Estmaton Results for Elastctes of Substtuton (-σ) by Product Groups Raw Processed Fxed Effect Model Random Effect Model Haussman Test Fxed Effect Model Random Effect Model Haussman Test (-4.87) (-5.3) (-4.65) (-4.14) (-3.11) (-6.76) (-4.25) (-4.12) (-7.61) (-9.11) (-6.4) (-6.66) (-8.99) (-8.6) (-3.81) (-4.41) (-2.94) (-3.51) (-3.6) (-3.66) (-2.64) (-2.79) (-1.17) (-1.73) (-4.39) (-5.52) (-1.38) (1.2) (-1.52) (-.65) (-9.71) (-14.88) (-5.21) (-5.49) (-9.64) (-8.29) (-.6) (-.6).9734 *Values n parenthess are t values. The bold values are used n the smulatons of our study snce they are sgnfcant and theory consstent. In order to decde between fxed effect and random effect models, we performed Hausman tests. The values n the Hausman test column are the prob. values. Hence f these values are less than.5 (a sgnfcance level), then accordng to the test the fxed effect model s proffered. 14

15 A3. Estmaton Results for Prce Elastctes of Imports (-η) by Product Groups RAW PROCESSED (-2.63) (-.22) (-.17) (-4.28) (-2.9) (-2.61) (-1.84) (-.69) (1.11) (-2.53) (-1.49) (-.91) (1.2) (-.1 ) (-7.27) (-1.89 ) (-.18) (1.46) (-4.32 ) *Values n parenthess are t values. The bold values are used n the smulatons of our study snce they are sgnfcant and theory consstent. Note that, degrees of freedom s low for the estmaton of these elastcty values snce we can not use panel data models. 15