Introduction. Business (International Economics Department), University of Zagreb, Croatia. 2 Such as CEFTA

Transcription

1 Gravty Model and Internatonal Trade: the Case of OECD Countres Mslav Jošć 1, Faculty of Economcs and Busness (Internatonal Economcs Department), Unversty of Zagreb, Croata KEY WORDS: Gravty model. Internatonal trade, Trade pattern, OECD, Transpost costs, GDP, Dstance UDC: ABSTRACT - Ths paper nvestgates relatonshp between trade varables such as exports and countryʹs macro varables whch s fully explaned usng gravty model of trade. Moreover, theoretcal background of gravty model on nternatonal trade s explaned n respect to change n transport costs. Gravty model approach s used n analysng trade pattern of the OECD countres n gravty equaton rangng from 1990 to 008. Emprcal evdence of ths work proves the expected sgns of varables n gravty equaton (GDP and Dstance) manly focusng on the role of transport costs n explanng varable Dstance. Introducton The purpose of ths work s to assess the degree of mpact of macro varables on the exports of OECD countres usng gravty model of nternatonal trade. In the past ffteen years European countres have experenced substantal trade ntegraton followed by the consttuton of the European Unon (Maastrcht Treaty, 1993) and varous PTA and RTA arrangements. The exstence of world trade polcy ncorporated n the GATT rules and the development of World Trade Organsaton provded sold ground for trade lberalsaton. Due to the avalablty of data, set of analysed countres s narrowed to European OECD member countres 3. European OECD member countres have ncreased ther exports wth the rest of the World from 137,4 bllons $US n 1990 to 468,4 bllons $US n (Table 1) whch s manly result of organc growth (new members) and lowerng tarrf rates amoung the members n the European Unon. Organc growth also boosted OECDʹs total economc strength measured by GDP (PPP) whch ncreased from 8343,1 bllons $US n 1990 to 103,9 bllons $US n (Table ). Ths relatonshp between exports and gross domestc product s analysed thoroughly n emprcal part of the study as well as other varable(s) n the model (e. g. Dstance). Blateral trade flow between European OECD member countres and the rest of the World s analysed on a set of data from 1990 to at a yearly bass. Trade flow n ths context s assocated wth exports of goods FOB. Exports serves as a dependent functon n a modelng gravty equaton usng economc sze of a country and partcular dstance between trade partners. 1 Mslav Jošć, unv. spec. oec. Currently workng as a Teachng and Research Assstant at Faculty of Economcs and Busness (Internatonal Economcs Department), Unversty of Zagreb, Croata. Such as CEFTA 3 As of October 008, OECD conssts of 30 members n total (3 from Europe), mostly hgh ncome countres (see OECD brochure) 4 Extracted from OECD Stats database (Internatonal Trade MEI) 5 Extracted from OECD Stats database (GDP) 6 Annually data not avalable for the year 008 at the tme of wrtng

2 48 CHALLENGES OF ECONOMIC SCIENCES IN THE 1 st CENTURY Theoretcal framework In order to determne the pattern of nternatonal trade t s wse to remember contrbuton of theores that precedeed the development of gravty model. Classcal theory of comparatve advantages defned by Davd Rcardo (1817) was a mlestone n researchng how trade can be mutually benefcal for all trade partners due to the labour costs and dfference n labour productvty. Introducton of Heckscher-Ohln trade theory (1919) based on relatve factor abundance and dfference n relatve factor prces has opened long-lastng academc debate about all causes and effects of trade. Refned verson of Heckscher-Ohln theory was made wth FPE theorem (Samuelson, 1948, 1949), Heckscher-Ohln-Vanek theorem and some emprcal testng that were mostly rejectng dea of real exstence of HO theory, notably Leontef for USA (1954) and Horvat (1968) for Yugoslav countres. Conclusons that arse from Heckscher-Ohln theory of nternatonal trade wll reman very mportant n the analyss of changes n output prces due to changes n transport costs. The frst formulaton of gravty model based on Newtonʹs Law of unversal gravtaton was made by Tnbergen (196). Analogy wth physcs served perfectly as a logcal explanaton of trade flow between countres. Other authors lke Armngton (1969), Anderson (1979), Bergstrand (1985), Helpman and Krugman (1985) and Brun et. al. (005) also gave ther contrbuton to the extenson of the model whch wll be dscussed later on. Sgnfcant fndngs n the same area but for South- East Europe were presented by Vujčć and Šošć (004) wth specal stress on trade potental of Croata. There are two major varables explanng blateral trade flow between trade partners: economc strength of a country (varable GDP), and geographcal proxmty (varable Dstance). The functonal form of the gravty model s gven n Equaton 1: E j = A GDP GDP D (1) β 1 β β3 j j where E = exports from country to country j A = constant GDP = gross domestc product D = dstance Better understandng of the gravty equaton n terms of elastcty s acheved usng loglnear structure (logarthmc values 7 of all varables) n Equaton : log Ej α j + β1 loggdp + β loggdpj + β3 = log D () j Where β, β β 1 and 3 parameters are nterpreted as coeffcents of elastcty of exports n respect to changes n ndependent varables (GDP and dstance). Ths change of dependent varable s not n absolute terms, t s rather relatve change due to nterpretaton of log structure and percentage changes that persst n β coeffcents. 7 Natural logarthm used for the log values

3 Model of Market Economy for Countres n Transton 49 Most often trade flow s calculated as arthmetc mean of exports and mports, but n ths work dependent varable wll be solely exports n order to emphasze the role of exports for analysed countres. Equaton can also be extended usng other varables that could possbly measure economc strength of a country better, such as Populaton (POP), GDP per capta (GDP/POP) or combnaton of all mentoned. Furthermore, the extended gravty equaton can also be wrten n the followng form: log E j = α + β loggdp + β loggdp + β log POP + β log POP + j 1 GDP GDP j + β5 log β6 log + β7 log D POP + POP j j 3 j 4 j (3) The dea of ths paper s to test expected sgns of β 1 and β coeffcents on GDP varables from Equaton usng regresson analyss whch should comply wth the sgns from Table 3: Table 3. Expected sgns of β coeffcents β Expected sgn of β β + 1 β + Postve sgns n Table 3 arse from the postve mpact of hgher ncome on mports of countryʹs trade partner under constant margnal propensty to mport. Greatest breakthrough n explanng GDP varable n gravty model was made by Armngton (1968) and later on Bergstrand (1985). Ther contrbuton nvolves ntroducton of prce changes n form of GDP deflator and real exchange rates. Some recent fndngs by Bussère, Fdrmuc and Schnatz (008) nclude FDI flows as an addtonal varable n the analyss. Results have shown postve sgn for FDI but wth lower elastcty (see Bussère et. al., 005). An mpact of geographcal proxmty on trade s negatve, and suggests that dstance affects Ej < 0 Dj trade n the opposte way ( ) due to transport costs. Hence, the expected sgn of parameter β β < 3 for varable Dstance s negatve ( 3 0 ). How to measure dstance s a key queston n settng gravty model properly. One way s to use great crcle 8 dstance between captal ctes of trade partners (country and country j). The other way s to use auxlary varables that represent changes n prces that occur n process of trade between countres and j. These can be measured ether usng real exchange rate or the prce of ol. The latter s chosen as a sutable representatve of transport costs n emprcal part of the study. Due to the mentoned transport costs 9 t s mpossble to hold Heckscher-Ohlnʹs assumpton that nternatonal trade wll equalze prces of all tradeable goods n countres that partcpate n trade. Transport costs nhbt trade and dscrmnate prces n both trade partners as well. 8 Usng haversne formula from sphercal trgonometry 9 Also known as border effects

4 50 CHALLENGES OF ECONOMIC SCIENCES IN THE 1 st CENTURY α j Parameter accounts for all other unobservable varables that are not explaned drectly through gravty equaton and ncludes cultural, hstorcal, poltcal and language dfferences among countres. Emprcal results The goal of emprcal part of ths paper s to mplement lnear regresson analyss on a set of data for 3 European OECD countres n a tme perod between year 1990 and 007. Dependent varable (response varable) n a model s Exports for OECD countres (EOECD) and ndependent varables (explanatory varables) are as follows: GDP OECD (GDPOECD), GDP World (GDPW) and varable Dstance (D). Data for GDPs of trade partners are gven n Appendx (see Table and 4). Varable Dstance uses OECDʹs crude ol mport prces, CIF (US$ per barrel) see Appendx (Table 5). All data are avalable annually, except for year 008 whch s excluded from an analyss. Regresson equaton 10 wth three predctors n log-lnear form s gven n Equaton 4: log EOECD = α, + β1 loggdpoecd + β loggdpw + β3 log D, (4) OECD W OECD W Logarthmc values of all (dependent and ndependent) varables wth 5 decmal ponts precson are gven n Table 6 and serve as nput for computer support n regresson calculaton: Table 6. Log values of all varables n regresson analyss t LGEOECD LGGDPOECD LGGDPW LGD ,9304 9,0919 9,9930 3, ,91 9, ,04170, ,9763 9, ,10780, , , ,188, , , ,19333, ,58 9, ,9780, , , , , ,9110 9, ,31539, ,3554 9, ,30737, ,3706 9, ,3457, , , , , , ,731 10, , , , ,4036 3, ,636 9, , , , , , , , , , , , , , , , , , ,3715 Source: World Bank, OECD Stats Database, authorʹs calculaton Lnk between response varable and explanatory varables s shown n Table 7 usng coeffcents of lnear correlaton: 10 Sngle country gravty model used

5 Model of Market Economy for Countres n Transton 51 Table 7. Correlaton matrx LGEOECD LGGDPOECD LGGDPW LGD LGEOECD 1, , , , LGGDPOECD 0, , , , LGGDPW 0, , , , LGD 0, , , , Source: Authorʹs calculaton From Table 7 t s clear there s a hgh postve lnear correlaton between Worldʹs GDP and ρ, LGEOECD = 0, OECDʹs exports of goods ( LGGDPW ). Demand for OECDʹs export products s almost perfectly correlated wth the changes n Worldʹs GDP whch arses from the smlarty of trade structures of both tradng partners accordng to Lnderʹs (1967) hypothess of nternatonal trade. Results of regresson analyss based on Equaton 4 s presented n Table 8: Table 8. Regresson analyss on LGEOECD varable Dependent Varable: LGEOECD Method: Least Squares Sample: Included observatons: 18 Varable Coeffcent Std. Error t-statstc Prob. C LGGDPOECD LGGDPW LGD R-squared Mean dependent var Adjusted R-squared S.D. dependent var S.E. of regresson Akake nfo crteron Sum squared resd Schwarz crteron Log lkelhood F-statstc Durbn-Watson stat Prob(F-statstc) Source: Authorʹs calculaton Popularty of gravty model n explanng trade pattern s justfed due to hgh R-squared (0,990601) whch expresses goodness of ft between observed and predcted values. Test statstcs shows that only varable LGGDPW s sgnfcant n explanng varable LGEOECD under 5% sgnfcance level (N=18) on ndvdual bass 11. The same result can be shown usng p-values. Ths s consstent wth the fact that Worldʹs demand (measured by GDP) for EOECD > 0 GDP OECDʹs goods s postvely correlated,.e. W. Group test (F statstcs) shows the followng results: wth k=3 and N=18 crtcal value for F test wth α = 0,05 equals F(3,14)=3,34. Emprcal value from Table 8 (F-statstcs=491,8663) s 11 Crtcal value n t-test s t, 101 0,05 =

6 5 CHALLENGES OF ECONOMIC SCIENCES IN THE 1 st CENTURY hgher than crtcal value (491,8663>3,34) whch leads to concluson that at least one varable s sgnfcant n explanng dependent varable LGEOECD n the model under α = 0,05 sgnfcance level. All expected sgns from Table 3 ( β 1 and β ) are proven n regresson output and, once agan, t can be sad that a rse n economc strength measured by GDP ncreases trade 1 between countres that partcpate n t. β Interpretaton of coeffcents n log-lnear form s n terms of elastcty due to log values used. Varable Dstance doesnʹt conform wth gravty model expectatons and parameterʹs negatve sgn. Ths can be explaned usng real varables, n ths case ol prces, nstead of physcal ds- β tance. 3 coeffcent nterpreted as percentage change of OECDʹs exports n respect to changes n dstance s closer to zero and doesnʹt affect trade sgnfcantly. Concluson An mportance of gravty model s undsputed n every serous analyss of nternatonal trade flows. The goal of ths paper was to seek and fnd relatonshp between countryʹs exports and economc power of ts tradng partners as well as dstance. For ths purpose dependent varable was presented through OECDʹs exports n tme perod between 1990 and 007. Predcted varables for economc strength were presented wth GDP (PPP) and dstance was analysed through the changes n mport prces of ol n OECD countres. Expected parameters defned n theoretcal gravty model assumed postve correlaton between economc strength and exports of tradng partners. Ths assumpton was emprcally tested and proven usng t-test and F-test wth postve sgns of regresson parameters assgned to GDP varables. Demand for OECDʹs export products s almost perfectly correlated wth the changes n Worldʹs GDP whch arses from the smlarty of trade structures of both tradng partners accordng to Lnderʹs hypothess of nternatonal trade. Low nelastcty of exports n respect to changes n transport costs (ol prces) was acheved accordng to regresson output. Varable Dstance ddnʹt gve expected parameterʹs value (negatve) although ts regresson coeffcent was close to zero. Results of an emprcal test conducted on a set of data for European OECD countres show that gravty model serves as a good explanaton for trade flows and patterns of nternatonal trade. 1 Defned as exports of one country to another

7 Model of Market Economy for Countres n Transton 53 Appendx Table 1. OECD Europe - Total Trade (n bllons of $US), OECD Europe Exports n goods 137,4 137, 144,9 136,3 156, 191,1 00,7 198,6 05,5 05,8 16,0 18,1 3,1 76,9 39,0 357,1 404,8 468,4 Imports n goods 14,1 144,5 150,7 134,7 15,0 184,3 193,3 191,8 0, 03,8 18,7 14, 3,4 70,0 34,4 357,1 408,0 474,5 Source: OECD Stats Database (008) Internatonal Trade MEI [onlne]. Avalable at: [October, 1, 008] Table. OECD Europe GDP (expendture approach, n bllons of $US), constant prces, constant PPPs, OECD base year, OECD Europe 8.343, , ,7 8.68, , 9.078,6 9.68, ,9 9.87, , , , , , , , ,8 1.03,9 Source: OECD Stats Database (008) GDP (expendture approach) [onlne]. Avalable at: [October, 1, 008] Table 3. GDP World (current US$, n bllons), World GDP (current US$) 1.877,3.964, , ,4 6.74, 9.667, 30.93, ,7 9.95, , , 31.70, , , 41.73, , , ,0 Source: World Bank (008) World Development Indcators database (World GDP, current US$) [onlne]. Avalable at: [November, 3, 008] Table 5. OECD crude ol mport prce, CIF, US$ per barrel, OECD crude ol mport prce, CIF, USD per barrel,3 19,3 18,4 16,4 15,6 17, 0,5 19,1 1,6 17,3 8,0 3,6 4,1 8,4 36,3 50,6 61,8 69, Source: OECD Stats Database (008) OECD crude ol mport prce, CIF, US$ per barrel [onlne]. Avalable at: [October, 1, 008] References Anderson, J. E. (1979) A Theoretcal Foundaton for the Gravty Equaton, Amercan Economc Revew 63, pp Armngton, P. (1969) A Theory of Demand for Products Dstngushed by Place of Producton. IMF Staff Papers 16 (3): pp Bergstrand, J. H. (1985) The Gravty Equaton n Internatonal Trade: Some Mcroeconomc Foundatons and Emprcal Evdence. Revew of Economcs and Statstcs 67 (3): pp Brun, J., Carrčre, C., Gullaumont, P. and Melo, J. (005) Has Dstance Ded? Evdence from a Panel Gravty Model. World Bank Economc Revew 19 (1): pp Bussère, M., Fdrmuc, J. and Schnatz, B. (008) EU Enlargement and Trade Integraton: Lessons from a Gravty Model, Revew of Development Economcs, 1 (3), pp Heckscher, E. (1919) The Effect of Foregn Trade on the Dstrbuton of Income, Ekonomsk Tdskrft, pp Helpman, E. (1987) Imperfect Competton and Internatonal Trade: Evdence from Fourteen Industral Countres. Journal of Japanese and Internatonal Economes 1 (1): pp Horvat, B. (1968) An Integrated System of Socal Accounts for an Economy of the Yugoslav Type, Yugoslav Insttute of Economc Research [onlne]. Avalable at: [October,, 008]

8 54 CHALLENGES OF ECONOMIC SCIENCES IN THE 1 st CENTURY Leontef, W. (1954) Domestc Producton and Foregn Trade: The Amercan Captal Poston Reexamned, Economa Internatozonale 7, (February), pp. 3-3 Lnder S. B. (1967) Trade and Trade Polcy for Development. New York Preager. OECD brochure (008) [onlne]. Avalable at: [October, 1, 008] OECD Stats Database (008) GDP (expendture approach) [onlne]. Avalable at: [October, 1, 008] OECD Stats Database (008) Internatonal Trade MEI [onlne]. Avalable at: [October, 1, 008] OECD Stats Database (008) OECD crude ol mport prce, CIF, US$ per barrel [onlne]. Avalable at: [October, 1, 008] Rcardo, D. (1817) On the Prncples of Poltcal Economy and Taxaton, London: John Murray [onlne]. Avalable at: [October,, 008] Samuelson, P. A. (1948) Internatonal Trade and the Equalsaton of Factor Prces, Economc Journal 58, (June), pp Samuelson, P. A. (1949) Internatonal Factor-Prce Equalsaton Once Agan, Economc Journal 59, (June), pp Tnbergen, J. (196) Shapng the World Economy: Suggestons for an Internatonal Economc Polcy. New York: The Twenteth Century Fund. Vujčć, B. and Šošć, V. (004) Trade Integraton n South-East Europe and the Trade Potental of Croata, mmeo, Natonal Bank of Croata, Belgrade [onlne]. Avalable at: [October,, 008] World Bank (008) World Development Indcators database (World GDP, current US$) [onlne]. Avalable at: [November, 3, 008]