Disance, Inernaional Trade and Economic Growh Hossein Panahi, The Business School, Universiy of Hull, UK h.panahi@econ.hull.ac.uk Absrac Researchers have recenly become increasingly ineresed in a se of disance and growh conceps variously described as neighborhood effecs, borders effec, exernaliies or spillovers, access o regional and global marke and disance. In empirical research on economic growh and inernaional rade, localiy and geographical disance have all been revealed as of being of grea imporance. Oher hings being equal, i migh be expeced ha closer supplier of goods would be chosen raher han more disance ones because hey have lower ransacion coss as a resul of being cheaper or easier o access. From a heoreical poin of view, he effec of he disance on he rade is ambiguous because here are many differen mechanisms ha migh be a work. By aking a more direc approach and carrying ou a number of economerics ess ha examine he role of disance, his paper ries o fill he exising gap in his lieraure using he available evidences. We produced a graviy model for rade by using a sample conains 16 counries (including Iran and 15 major parners) and esed he possibiliy of his linkage. Wih respec of he naure of disance variable which is consan hrough he ime we used wo differen ype of regressions, pooled regression and cross-secion analysis o analyse he role of disance on inernaional rade. I is expeced ha GDP, populaion and openness posiively affec he volume of rade beween wo counries. This expecaion for exchange rae and disance were negaive correlaion. Our overall oucomes showed ha despie some sign differences in some counries, parners disance is significan in explaining he rade wih hem. Some excepions can be explained by oher erms like poliical reacions. In order o idenify he major deerminans of inernaional rade we also checked he robusness of our resul by dropping some of he variables from he regression. We found ha alhough in all specificaions, almos all he signs of parameer esimaes do no differ. However, some imporan changes may be observed beween specificaions. Firs, as he number of variables decreases, he value of he disance coefficien becomes smaller. This implies ha he impac of disance on rades will be significanly underesimaed when necessary variables such as openness are omied. Our finding from cross secion analysis highlighed ha he defined geographic variables appears o play an imporan role on he rade of Iran wih is parners and herefore on he long-run economic growh of Iran. Alhough exchange rae was no very significan in our samples bu our esimaes for disance were consisen wih he heory. Disance is an imporan deerminan of inernaional rade and herefore economic growh. Key words: Geographical Disance, Inernaional Trade, Economic Growh JEL classificaions: C51, F14, O47
Disance, Inernaional Trade and Economic Growh Hossein Panahi, Universiy of Hull, Hull, UK - Inroducion Researchers have recenly become increasingly ineresed in a se of disance and growh conceps variously described as: neighbourhood effecs, borders effec, exernaliies or spillovers, access o regional and global marke and locaion and disance. The main goal of his paper is o assess he relaionship beween Disance, inernaional rade and economic growh hrough regressions using daa for seleced counries over several decades. - Is Disance Imporan for Economic Growh? In empirical research on economic growh and inernaional rade localiy and geographical disance have all been revealed as of being of grea imporance. Oher hings being equal, i is expeced ha closer supplier of goods would be chosen raher han more disan ones because of having lower ransacion coss and herefore being cheaper or easier o access. In addiion, local raders are highly rusable because i is easy o inerac wih hem. On he oher hand, as a par of local neworks of relaionships hey have beer local knowledge. Despie hese expecaions here has been relaively lile previous aemp o invesigae he possible role of disance in influencing choice of rade parners. By aking a more direc approach and carrying ou a number of economerics ess ha examine he role of disance, his sudy seeks o fill he gap using he available evidences.
This paper firs reviews previous researches which have sough o assess he role of disance and also localiy of counries in heir local and inernaional rades and herefore economic performances. I hen defines a heoreical approach used based on graviy models. The main body of he paper presens he mehodology of he survey and he chief dimensions of analysis: he role of counries disance from global economic aciviy and is effecs on economic developmen. - Discussion of Relaed Lieraure Whils here has been a rapidly expanding lieraure on he role of inernaional rade in economic growh of counries, here has been relaively less aenion paid o he role of disance as a possible influence on inernaional rade and herefore economic growh. Neverheless here have been some imporan earlier sudies ha evaluae he influences of disance in economic growh. On one hand, i is argued ha wih he decline in rade coss, including ranspor, informaion, ariff and communicaion coss, he imporance of disance has declined over ime. Therefore rade should have expanded geographically. 1 On he oher hand, mos graviy models find ha he imporance of disance has increased over ime. Leamer and Levinsohn (1995) poin ha:. he effec of disance on rade paerns is no diminishing over ime. Conrary o popular impression, he world is no geing dramaically smaller. 2 1 See for example: Cairncross F., (1997). 2 pp.1387-88
The imporance of disance is also confirmed in a sudy by Disdier and Head (2003). Using graviy models hey conclude ha he impac of disance on rade is significanly increasing over ime. On he relaionship beween FDI and disance, Freund and Hummels (2003) find ha FDI growh increases rade bu has had lile impac on he elasiciy of rade wih respec o disance. 3 Using differen empirical and heoreical models, he effecs of disance and localiy have been analysed by Afield e al. (2000) and also, Moreno, and Trehan (1997). The firs research has found no significan effec from disance on growh rae. In conrary, he second one has shown ha disance is a significan deerminan of rade. In a very similar research Cannon, Demery and Duck (2000), show ha whils boh longerm and cyclical movemens in regional oupu wihin he European Union are geographically clusered his clusering is linked no o geographic locaion per se bu o membership of a counry. By differen mehodology, Frankel (1997) repors ha disance increasing would have led o rade decreasing. Disdier and Head (2003) found ha here is an increasing impac of disance on rade since 1965. In similar work, Brun e al. (2002) also conclude ha here is an increasing impac of disance on rade by 11 percen over he las 35 years. Finally in a very recen paper Kuwamori (2006) presens an empirical invesigaion of he appropriaeness of disance as a deerminan of inernaional ranspor coss by using Philippine impor daa. He found ha 3 The coefficien of disance presens no clear rend when esimaed wih he sandard log-linear specificaion of he graviy model in cross-secion over several years by Coe e al. (2002) or in a panel over 35 years by Brun e al. (2002). When he model is esimaed non-linearly, Coe e al. (2002) find ha he coefficien for he disance variable shows a decline in 1975-2000. Wih an augmened ranspor cos funcion, Brun e al. (2002) find a decline in he coefficien of disance, hough he decline is largely confined o bilaeral rade among rich counries; for developing counries, he coefficien of disance does no decline over ime.
while disance is imporan in deermining ranspor coss, using disance alone as he proxy of inernaional ranspor coss is insufficien, and such use underesimaes he impac of disance on inernaional ranspor coss. 4 - Empirical Assessmen In he empirical lieraure i is argued ha disance as a mos popular proxy for ranspor coss is an imporan deerminan of inernaional rade. This can bes be addressed hrough his hough ha ranspor coss affec economic growh hrough reducing expor and herefore lowering aggregae saving and invesmen by exporers. They also reduce FDI because of higher price of impored capial goods. The assessmen of he influence of disance on economic growh hrough he inernaional rade is developed here using wo differen echniques. Because of he naure of disance variable using normal ime series analysis is irrelevan. Therefore firs a pooled regression for seleced counries is performed. Then I use a cross-secion echnique o run he defined graviy model. Here in he graviy model bilaeral rade is explained in erms of openness, he size of counries and he disance beween hem. Models which use disance as a proxy for ranspor coss, have recenly become he mos successful empirical models o explain he relaionship beween disance and rade and herefore o esimae he relaionship beween inernaional rade and economic growh. The sample conains 16 counries (including Iran and 15 major parners). 5 Daa on GDP, populaion, openness and exchange rae were exraced from WDT 2002 & 2006, Penn 4 Page 1. 5 These counries are 6 neighbourhood counries: Armenia, Azerbaijan, Iraq, Pakisan, Russia, Turkey, and 9 ohers rade parners including Ausralia, Canada, China, Germany, Ialy, Japan, Malaysia, UK and USA.
World Table version 6.1 and IMF Financial Saisics 2005, for 15 years (1990-2004). Using bilaeral rade daa here in graviy regressions is o capure he effec of geographical disance on bilaeral rade. The bilaeral rade daa beween counries is from Unied Naion Commodiy Trade Saisics Daabase (COMTRADE), 2004. The GDP daa are in inernaional prices (US Dollars) and herefore consruced o be comparable over ime and across economies. The disance variable in graviy models is a proxy for ranspor coss and oher rade coss ha vary wih disance. Using Gledisch Large daase, 6 and for he purpose of his model, a 16 16 disance marix ( d ij ) has been consruced ha conains he grea circle disance beween capial ciies in he kilomeres by means of he laiude and longiude of he capial in each counry. In ( d ij ) marix he diagonal elemens of he marix which is he disance of a counry by iself is 0. I is very small and differen from oher elemens. This causes ha he numerical accuracy is endangered. Therefore we rescale he daa by consrucing a weighing marix of disance ( w ). This is: ij w ij n 1/ d j 1 ij 1/ d ij (1) This marix links all he 16 counies wih each oher. The resuling weigh marix posulaes ha he influence of any variable beween wo counries decreases wih he inverse of he disance beween hem. 7 6 Gledisch daa se is available a: hp://weber.ucsd.edu/~kgledis/capdis.hml. 7 Each counry belongs o he neighbourhood of every oher counry. However, he relaive imporance of each counry in a paricular neighbourhood varies inversely wih is disance from he counry whose neighbourhood i is.
The graviy equaion for rade saes ha rade beween Iran and her 15 major parners a ime denoed by T, are proporional o he produc of counries GDPs a ime, denoed by Y, populaion POP, 8 degree of openness Open exchange rae ER and inversely proporional o heir disance, Dis ij. BD and LLD are Border and landlocked dummies. T 0 1Y 2Pop 3Open 4ER 5Disij 6BD 7 LLD (2) where 0, 1, 2, 3, 4, 5, 6and 7 are parameers o be esimaed and ij is an error erm showing all oher facors ha migh affec rade. I is expeced ha GDP, populaion and openness posiively affec he volume of rade beween wo counries. This expecaion for exchange rae and disance are negaive correlaion. - Pooled Regression To ge he overall effecs of defined variables, more specifically he role of disance on rade, using PcGive, a pooled regression for equaion (2) was performed. Resuls are presened in Table 1. The firs column in op of he Table 1 shows he esimaed coefficien. The signs of GDP and openness parameers are as expeced. Tha is, rade rises as GDP increases and counries become more open o inernaional rade. These signs for he oher measure of he counry size, populaion, and also exchange rae are opposie. The parameer esimae of he disance variable is also as expeced. I indicaes ha rade volume falls as disance 8 The resul is lile affeced when counry s working-age populaion is used. See: Frankel, Romer and Cyrus (1996).
increases. The Third column of able represens values which are generally significan. R² has a high value of 0.68 and iself indicaes a good esimaion. In order o idenify he major deerminans of inernaional rade I also checked he robusness of our resul by dropping some of he variables from he regression. In all specificaions, almos all he signs of parameer esimaes do no differ. However, some imporan changes may be observed beween specificaions. Firs, as he number of variables decreases, he value of he disance coefficien becomes smaller. This implies ha he impac of disance on rades will be significanly underesimaed when necessary variables such as openness are omied. Second, resuls in which openness variables and boh dummies are excluded from he regressions, have low explanaory power. This resul may imply ha he degree of openness and he characerisics associaed wih border and landlocked dummies are imporan in explaining variaions in disance. Finally he magniude of significancy is reducing somehow. - Border, and Landlocked Dummies The geographical daa has been used very rarely in growh economics argumens. Mos researchers ofen prefer o use he above mehod wih caegorical dummy variables o capure common characerisics of counries in a group. For example, he common border dummy can capure he economic imporance of counries nearby. In oher word, i can show his fac ha hrough free rade agreemens, neighbouring counries ofen give each oher preferenial marke access ha are ofen wide in scope.
Alernaively, landlocked dummy enable us o capure geographical indicaors ha may affec a counry's rade. Researches show ha land-lockedness implies a high ranspor cos. For example, Raballand (2003) concludes ha being landlocked would reduce rade by more han 80% when measured by a dummy variable. I esed he effecs of a border dummy and a landlocked dummy in he form of BDij and LLD. Table 2 shows he effec of border and landlocked dummies on Iran s rade j since 1990. Boh dummies had mosly he expeced posiive and negaive signs, bu border dummy was no significan while landlocked dummy was significan. I means, because of he easier access and cheaper shipping and ranspor coss, access o he sea is a good advanage in inernaional rade. - Non-Linear Esimaion: Cross-Secion Analysis A second way o obain a sense of how imporan geographical variables are o he inernaional rade and herefore economic growh is o esimae he effec of hese variables using a cross-secion echnique. Of course, cross-secion models mus be carefully specified if hey are o prove useful. In his secion an applicaion of he crosssecion model o our sample of 15 Iran s main rade parners is described. A non linear form of Equaion (2) is defined as: T Y 0 1 2 3 4 5 6 7 Pop Open ER Dis BD LLD ij (3) or in logarihmic form: T Y Pop Dis BD LLD 5 0 1 6 7 2 Open 3 ER 4 (4)
Once again, Y is growh naional produc in ime, Pop, Open and ER show he populaion, degree of openness and exchange rae of counries in ime. Dis indicaes he disance beween seleced counries and Iran which is consan in dae. As before, BD and LLD are dummy variables o capure neighbourhood and land lock effecs and finally, is an error erm showing all oher facors ha migh affec rade. Using PcGive and a cross-secion bach file described in appendix 1, a Maximum Likelihood (ML) esimaion has run for equaion (4). The resuls of his analysis are repored in Table 3. As i can be seen, for breviy we only summarily repor and inerpre he resuling parameer esimaes, bu emphasise heir significance for undersanding aggregaive behaviour. These resuls are imporan suppor for he proposiion ha geography does maer for inernaional rade and economic growh. Each raw shows he esimaed coefficiens of defined variables for each period of ime across he seleced counry. Furhermore, he esimaed coefficiens are indeed he elasiciies. As able 3 shows, he overall resul is as expeced from he lieraure. There is no single facor which can be isolaed as an imporan maer for rade across seleced counries for all years. Size of he counry as measured by GDP and populaion are found o affec inernaional rade. Populaion ha includes human capial seems imporan in some counries bu no in ohers. GDP as an alernaive measure of size is found o be more significan for rade. I has a posiive effec on rade hroughou he esimaes. Openness facor is mosly found imporan in economies. The esimaes for disance are very consisen wih he heory. I is highly significan in mos of he years. Exchange rae is no very significan in our samples. A similar conclusion for dummy variables is ij
applied. Despie boh dummies had mosly he expeced signs, bu hey are no significan. - Conclusion In empirical research on economic growh and inernaional rade localiy and geographical disance have all been revealed as of being of grea imporance. Oher hings being equal, i migh be expeced ha closer supplier of goods would be chosen raher han more disance ones because hey have lower ransacion coss as a resul of being cheaper or easier o access. Despie hese expecaions here has been relaively lile previous aemp o invesigae he possible role of disance in influencing choice of rade parners. From a heoreical poin of view, he effec of he disance on he rade is ambiguous because here are many differen mechanisms ha migh be a work. I is jus showed ha geographic disance is an imporan deerminan of inernaional rade. This can bes be addressed hrough he following hough experimen. A counry is considering where o source is supplies from, or where o locae is own producion. We saw ha despie some sign differences in some counries, parners disance is significan in explaining he rade wih hem. Some excepions can be explained by oher erms like poliical reacions. In order o idenify he major deerminans of inernaional rade I also checked he robusness of he resuls by dropping some of he variables from he regression. I found ha alhough in all specificaions, almos all he signs of parameer esimaes do no differ. However, some imporan changes may be observed beween specificaions. Firs, as he number of variables decreases, he value of he
disance coefficien becomes smaller. This implies ha he impac of disance on rades will be significanly underesimaed when necessary variables such as openness are omied. The general picure of finding from cross secion analysis also highlighed ha he defined geographic variables appears o play an imporan role on he rade of Iran wih is parners and herefore on he long-run economic growh of Iran. Alhough exchange rae was no very significan in our samples bu our esimaes for disance were consisen wih he heory. I was significan in mos of he years. I means disance is an imporan deerminan of inernaional rade and herefore economic growh. Therefore, here is endency in Iran s economy o concenrae is bilaeral rade mosly wih nearby counries raher han counries which are geographically far away. Of course here are some excepions which are generally because of some poliical consideraion. 2 R, which measures he correlaion beween he acual and fied values of variables confirms such a srong relaionship. Much remains o be done. This sudy has shown how disance effecs could be derived in a simple growh model, wihou appealing o oher geographical variables which cerainly do affec he economic performance of counry. However, wheher hese oher human and physical geography variables affec he economic performances of a counry or wha is he role of he size of neighbourhood counries markes, or how neighbourhood counries growh rae affec he economic growh of a counry remains o be invesigaed.
- References 1- Afield C.L.F., Cannon E. S., Demery D. and Duck N. W., (2000). Economic Growh and Geographical Proximiy, Economic Leers 68(1): 109-112. 2- Brun J.F., Carrere C., Guillaumon P., and de Melo J., (2002). Has Disance Died? Evidence from a Panel Graviy Model, CEPR Discussion Paper no. 3500. 3- Brun, J. F., Combes J. L. and Combes P. M., (2000). Do Geographical Spilovers Effecs on Growh Depend on Commercial Openness., CERDI Working Paper 9904, France. 4- Cannon E.S., Demery D. and Duck N.W., (2000). Does Disance Maer for Economic Performance? Evidence from European Union, Discussion Paper No. 00/509, Page 11, Deparmen of Economics, Universiy of Brisol, UK. 5- Cairncross Frances, (1997). The Deah of Disance: How he Communicaions Revoluion is Changing Our Lives. Boson: Harvard Business School Press. 6- Coe D.T., Subramanian A., Tamirisa N.T. and Bhavnani R., (2002). The Missing Globalizaion Puzzle, IMF Working Paper No. WP/02/171. 7- Disdier A.C. and Head K., (2003). Exaggeraed Repors on he Deah of Disance: Lessons from a Mea-Analysis, Mimeo, TEAM, Universie de Paris I. 8- Frankel J.A., (1997). Regional Trading Blocs in he World Economic Sysem, Insiue for Inernaional Economics, Washingon D.C. USA. 9- Frankel J. A., Romer D. and Cyrus T., (1996). Trade and Growh in Eas Asian Counries: Cause and Effec?, Cambridge, Naional Bureau of Economic Research (NBER) Working Paper No. 5732. 10- Freund C. and Hummels D., (2003). Why Hasn Disanc Died?, Mimeo, World Bank. 11- Gledisch, Krisian S. and Michael D. Ward, (2001). Measuring Space: A Minimum Disance Daabase and Applicaions o Inernaional Sudies, Journal of Peace Research 38(6): 739-58. 12- Kuwamori Hiroshi, (2006). The Role of Disance in Deermining Inernaional Transpor Coss: Evidence from Philippine Impor Daa, Discussion Paper No. 60, Insiue of Developing Economies, Japan Exernal Trade Organizaion. 13- Leamer E.E. and Levinsohn (1995). Inernaional Trade: The Evidence, In Handbook of Inernaional Economics, ed. Grossman,G.M. and Rogoff, K., vol.3, Elsevier. 14-Moreno R. and Trehan B., (1997). Locaion and Economic Growh of Naions, Journal of Economic Growh 2(4): 399-418. 15- Raballand G., (2003). Deerminans of he Negaive Impac of Being Landlocked on Trade: An Empirical Invesigaion Through he Cenral Asian Case, Comparaive Economic Sudies 45(4): 520-536.
Table 1: Deerminans of Trade of Iran wih Her Major Parners, (Pooled Regression), 1990-2004. Coefficien Sd.Error -value Consan 3.46073 1.1254 3.08 Y 0.01035 0.0188 1.75 POP 0.79785 0.5018 1.69 Open 1.11950 4.6747 2.40 ER 508.495 103.6 4.91 Dis -205454 8.0367-2.56 BD -2.10347 7.7321-2.72 LLD -1.4493 1.3865-1.55 --------------------------------------------------------------------------------------- sigma 1.421901 sigma^2 2.021803 R^2 0.684466 RSS 4.084041 TSS 5.707681 no. of observaions 210 (from 16 counries) no. of parameers 8 Wald (join): Chi^2(7) = 5.782 [0.000] ** Wald (dummy): Chi^2(1) = 9.467 [0.002] ** AR(1) es: N(0,1) = 1.754 [0.079] * AR(2) es: N(0,1) = 1.793 [0.073] *
Table 2: Aggregae impac of Border and Landlocked Dummy Variables on Iran s Trade wih Her Parners Year Border Dummy Landlocked Dummy R² & F 1990 0.125043 8.46804 R² = 0.09 (1.07) (84.2) F = 0.57 1991 0.394533 8.72816 R² = 0.09 (24.8 ) (0.30) F =0.58 1992 0.307483 8.70113 R² = 0.10 (1.13) (37.2) F = 0.64 1993-0.00070 9.41329 R² =0001 (-0.004) (55.1) F = 0001 1994 0.0530683 9.64739 R² = 0.001 (0.126) (24.4) F =0.008 1995 0.0548759 10.3536 R² = 0.007 (0.295) (61.0) F = 0.045 1996-0.180872 8.61920 R² = 0.01 (-0.351) (18.5) F = 0.06 1997 0.335555 10.0522 R² = 0.21 (1.77) (56.3) F = 1.56 1998 0.700491 9.92395 R² = 0.30 (2.27) (34.2) F = 2.58 1999 0.00417255 8.85037 R² = 0.001 (0.0135) (29.6) F = 0.009 2000 0.0679424 8.92361 R² = 0.003 (0.204) (28.5) F = 0.02 2001 0.193689 9.45859 R² = 0.055 (0.838) (49.5) F = 0.35 2002 0.0596606 9.61989 R² = 0.004 (0.231) (43.4) F = 0.027 2003-0.0564410 9.79145 R² = 0.01 (-0.343) (63.3) F = 0.059 2004-0.953261 9.78352 R² = 0.02 (-0.463) (4.88) F = 0.11 Figures in bracke show values
Table 3: A Maximum Likelihood Regression of Trade of Iran wih Her 15 Major Parners (Cross-Secion) 90 T90 31.36 0.21 Y90 0.87 POP90 1.93 Open90 0.23 ER 0.0004 Dis 3.39BD 0. 57LLD (3.09) (0.72) (-3.44) (-2.23) (3.42) (-4.30) (-0.04) (0.61) R²= 0.91 RSS= 2.57 F=8.86 ** DW=2.00 91 T91 6.68 0.81Y91 0.40 POP91 0.43 Open91 0.12 ER 0.0003 Dis 0.57BD 0. 75LLD (0.72) (2.88) (-1.76) (-0.62) (1.66) (-2.87) (-0.48) (0.75) R²= 0.91 RSS= 2.67 F=9.01 ** DW=1.90 92 T92 15.25 0.49 Y92 0.38 POP92 1.40 Open92 0.04 ER 0.0004 Dis 1.64.57BD 0. 16LLD (2.16) (2.14) (-2.74) (-2.26) (0.48) (-5.40) (-1.72) (0.18) R²= 0.96 RSS= 1.20 F=21.33 ** DW=1.85 93 T93 14.10 0.46 Y93 0.29 POP93 0.85 Open93 0.13 ER 0.0002 Dis 1.75BD 0. 09LLD (1.59) (1.65) (-1.51) (-1.08) (-1.36) (-3.87) (-1.56) (0.10) R²= 0.93 RSS= 1.83 F=10.88 ** DW=2.16 94 T94 10.44 0.58 Y94 0.28 POP94 0.40 Open94 0.16 ER 0.0002 Dis 1.01BD 0. 29LLD (1.33) (2.21) (-1.62) (-0.63) (1.83) (-3.65) (-0.90) (-0.35) R²= 0.90 RSS= 1.94 F=7.95 * DW=1.95 95 T95 8.67 0.65Y95 0.27 POP95 0.13 Open95 0.11 ER 0.0003 Dis 060BD 0. 50LLD (1.32) (2.78) (-1.33) (-0.22) (1.17) (-3.57) (-0.53) (-0.56) R²= 0.87 RSS=2.54 F=5.83 * DW=2.10 96 T96 32.30 0.01 Y96 0.57 POP96 0.70 Open96 0.29 ER 0.0004 Dis 4.44BD 2. 66LLD (1.31) (0.008) (0.64) (-0.30) (0.79) (-1.73) (-1.03) (-0.69) R²= 0.42 RSS= 40.18 F=0.61 DW=1.74 97 T97 32.80 0.07Y97 0.76 POP97 1.37 Open97 0.28 ER 0.0003 Dis 4.87BD 1. 58LLD (2.95) (0.18) (-1.93) (-1.17) (1.74) (-2.12) (-2.34) (-0.91) R²= 0.69 RSS= 8.37 F=1.86 DW=1.45 98 T98 25.28 0.11Y98 0.41 POP98 0.96 Open98 0.18 ER 0.0003 Dis 3.75BD 0. 83LLD (2.83) (0.39) (1.34) (-1.04) (1.60) (-2.39) (-2.21) (0.60) R²= 0.70 RSS= 1.99 F=5.26 * DW=1.89 99 T99 22.41 0.14 Y99 0.30 POP99 0.58 Open99 0.14 ER 0.0002 Dis 2.95BD 0. 88LLD (2.40) (0.44) (-0.98) (-0.61) (1.19) (-1.89) (-1.64) (-0.61) R²= 0.63 RSS= 5.71 F=1.49 DW=2.12 00 T00 27.95 0.08 Y00 0.501 POP00 1.33 Open00 0.22 ER 0.0002 Dis 4.36BD 1. 29LLD (3.51) (0.28) (-1.82) (-1.92) (2.05) (-2.50) (-2.87) (-1.02) R²= 0.80 RSS= 2.39 F=3.42 * DW=1.84 01 T01 29.07 0.03 Y01 0.48 POP01 0.46 Open01 0.16 ER 0.0002 Dis 4.01BD 1. 49LLD (4.09) (0.11) (-1.86) (-2.36) (1.78) (-3.06) (-2.95) (-1.30) R²= 0.81 RSS= 3.61 F=3.66 * DW=2.83 02 T02 25.97 0.08Y02 0.37 POP02 1.31 Open02 0.15 ER 0.0003 Dis 3.83BD 1. 11LLD (2.96) (0.25) (-1.24) (-1.74) (1.45) (-3.01) (-2.33) (-0.82) R²= 0.78 RSS= 4.94 F=2.96 DW=1.89 03 T03 25.05 0.03 Y03 0.23 POP03 1.25 Open03 0.16 ER 0.0003 Dis 3.75BD 0. 97LLD (2.67) (0.09) (-0.77) (-1.54) (1.48) (-3.04) (-2.18) (0.71) R²= 0.77 RSS= 5.22 F=2.80 DW=1.95 04 T04 11.78 0.25Y04 0.21 POP04 0.16 Open04 0.23 ER 0.0003 Dis 3.24BD 0. 07LLD (1.17) (0.62) (0.68) (-0.20) (1.94) (-2.74) (-1.71) (0.05) R²= 0.81 RSS= 3.76 F=6.08 * DW=2.03 Figures in bracke show values
Appendix 1: A Maximum Likelihood (ML) Esimaion Bach File for he Model of Trade. (Cross-Secion) Y = T90; Z = Consan,Y90, POP90, Open90, ER90, Dis, BD, LLD; Y = T91; Z = Consan,Y91, POP91, Open91, ER91, Dis, BD, LLD; Y = T92; Z = Consan,Y92, POP92, Open92, ER92, Dis, BD, LLD; Y = T93; Z = Consan,Y93, POP93, Open93, ER93, Dis, BD, LLD; Y = T94; Z = Consan,Y94, POP94, Open94, ER94, Dis, BD, LLD; Y = T95; Z = Consan,Y95, POP95, Open95, ER95, Dis, BD, LLD;
Y = T96; Z = Consan,Y96, POP96, Open96, ER96, Dis, BD, LLD; Y = T97; Z = Consan,Y97, POP97, Open97, ER97, Dis, BD, LLD; Y = T98; Z = Consan,Y98, POP98, Open98, ER98, Dis, BD, LLD; Y = T99; Z = Consan,Y99, POP99, Open99, ER99, Dis, BD, LLD; Y = T00; Z = Consan,Y00, POP00, Open00, ER00, Dis, BD, LLD; Y = T01; Z = Consan,Y01, POP01, Open01, ER01, Dis, BD, LLD; Y = T02; Z = Consan,Y02, POP02, Open02, ER02, Dis, BD, LLD;
Y = T03; Z = Consan,Y03, POP03, Open03, ER03, Dis, BD, LLD; Y = T04; Z = Consan,Y04, POP04, Open04, ER04, Dis, BD, LLD;