出口進口和經(jīng)濟增長的因果關系基于轉型經(jīng)濟的實證分析外文原文

1、出口、進口和經(jīng)濟增長的因果關系：基于轉型經(jīng)濟的實證分析外文原文 economics letters 94 2007 389?395/0>./locate/econbasecausality between exports, imports, and economic growth:evidence from transition economiestitus o. awokusedepartment of food and resource economics, 213 townsend hall, university of delaware, newark, de 19717, us

2、areceived 7 september 2005; received in revised form 7 august 2006; accepted 30 august 2006available online 28 november 2006abstracttheeconomicgrowtheffectsofceecmarketliberalizationandincreasingtradeaccesstothelargereumarketis an important empirical question worthy of investigation. this paper exam

3、ines the impact of export and importexpansion on growth in three transition economies. the empirical results suggest that trade stimulates economicgrowth 2006 elsevier b.v. all rights reserved.keywords: ceec nations; cointegration; economic growth; granger causality; tradejel classification: f43; c3

4、21. introductiondisagreements persist in the empirical literature regarding the causal direction of the effects of tradeopennessoneconomicgrowth.someanalystsarguethatcausalityflowsfromexportstoeconomicgrowthand denotes this as the export-led growth elg hypothesis balassa, 1978; bhagwati, 1978; edwar

5、ds,1998. export expansion and openness to foreign markets is viewed as a key determinant of economicgrowthbecauseofthepositiveexternalitiesitprovides.forexample,firmsinathrivingexportsectorcanenjoy the following benefits: efficient resource allocation, greater capacity utilization, exploitation ofec

6、onomies of scale, and increased technological innovation stimulated by foreign market competitiontel.: +1 302 831 1323; fax: +1 302 831 6243.e-mail address: .0165-1765/$ - see front matter2006 elsevier b.v. all rights reserved.doi:10.1016/j.econlet.2006.08.025390 t.o. awokuse / economics

7、 letters 94 2007 389?395helpmanandkrugman,1985.furthermore,exportscanprovideforeignexchangethatallowsformoreimports of intermediate goods which in turn raises capital formation and thus stimulate output growth.several studies have also shown that it is possible to have growth-led exports gle which h

8、as thereversecausalflowfromeconomicgrowthtoexportsgrowth.intheglecase,exportexpansioncouldbestimulated by productivity gains caused by increases in domestic levels of skilled-labor and technologybhagwati,1988;krugman,1984.thethirdalternativeisthatofimport-ledgrowthilgwhichsuggeststhateconomicgrowthc

9、ouldbedrivenprimarilybygrowthinimports.endogenousgrowthmodelsshowthatimportscanbeachannelforlong-runeconomicgrowthbecauseitprovidesdomesticfirmswithaccesstoneededintermediatefactorsand foreigntechnologycoeandhelpman,1995.growthinimportscanserve as a medium for the transfer of growth-enhancing foreig

10、n r&d knowledge from developed todeveloping countries lawrence and weinstein, 1999; mazumdar, 2000.after the collapse of central economic planning in the late 1980s, many former socialist central andeasterneuropeancountriesceecexperiencedmajoreconomiccrises.severalofthesecountrieshavesince taken

11、 steps to become members of the european union eu. as part of the conditions for euaccession, these nations signed the europe association agreements which required the adoption ofeconomic reforms and market liberalization policies which have led to significant expansion in theexport sector in severa

12、l countries. while many recent empirical studies have investigated the effect ofexports on economic growth, they largely focused on asian economies. due to the recent history of theceecs, very few studies have examined the trade-growth nexus for these economies dritsakis, 2004.rather, the majority o

13、f post-reform studies about the former socialist economies have focused on thenatureandchallengesoftheeconomictransitionprocess.themainfocusofpaststudieshasbeenontheissues of market failure and the development of free market institutions kaminski et al., 1996; worldbank,1996;hoekmananddjankov,1997.t

14、heeconomicgrowtheffectsofceecmarketliberalizationand increasing trade access to the larger eu market is an empirical question that needs furtherinvestigation.this paper examines the causal relationship between trade and economic growth for bulgaria,czech republic, and poland within an integrated fra

15、mework. in contrast to most previous elgstudies, this study specifies an augmented production function that explicitly tests for the effect ofboth exports and imports on economic growth. following toda and phillips 1993, this study adoptsrecent advances in time series modeling by specifying causal m

16、odels based on vector error correctionmodels. the empirical results indicate a bi-directional causal relationship between exports and growthin bulgaria and causality from imports to economic growth in the czech republic and poland. therest of this paper is organized as follows. section 2 discusses t

17、he analytical framework andmethodological issues. section 3 presents empirical analysis and results and section 4 summarizes thepaper's findings.2. analytical framework and methodological issuesearly empirical formulations tried to capture the causal link between exports and gdp growth byincorpo

18、rating exports into the aggregate production function balassa, 1978; sheehey, 1992. theaugmented production function including both exports and imports is expressed as:yf?k;l?; x ;m ? 1?t.o. awokuse / economics letters 94 2007 389?395 391where y represents real gdp growth, k, l, x, and m represent r

19、eal gross capital, labor, real exports, andreal imports, respectively. given the long-run nature of the trade and economic growth relationship, it isnecessary to test for cointegration, prior to granger causality analysis.since the cointegration methodology is fairly common and well-documented elsew

20、here engle andgranger, 1987; johansen, 1991; johansen and juselius, 1990, only a brief overview is provided.johansen1991modeledtimeseriesasareducedrankregressioninwhichtheycomputedtheimumlikelihoodestimatesinthemultivariateerrorcorrectionmodelecmwithgaussianerrors.themodelisbased on the error correc

21、tion representation given by:p 1xdz ? gdz ?pze ?2?t i t?i t?1 ti?1where z is an n×1 column vector of p variables, is an n×1 vector of constant terms, andtrepresent coefficient matrices, is a difference operator, k denotes the lag length, and ?n0,. thetcoefficient matrix is known as the imp

22、act matrix, and it contains information about the long-runrelationships. after pre-testing for the order of integration for each variable, johansen's methodologyrequires the estimation of eq. 2 and the residuals are then used to compute two likelihood ratio teststatistics: the trace and imal eig

23、envalue - tests. the asymptotic critical values for the tracelikelihood ratio tests are calculated via numerical simulations see osterwald-lenum, 1992.3. empirical analysis and resultsthe data set for each country consists of observations for real gdp growth gdp, real exportsexports, real imports im

24、ports, gross capital formation as proxy for capital capital, and labor forcelabor.thedataset,obtainedfromtheinternationalmonetaryfunddatabase,isquarterlyandcoverstheperiods 1994:1?2004:3 for bulgaria, 1993:1?2002:4 for czech republic, and 1995:1?2004:2 forpoland, respectively. all the variables are

25、in natural logarithms.first, the time series properties of the data are examined using two unit-root tests. the augmenteddickey and fuller 1979 adf test for the null hypothesis of non-stationarity while the kpss testproposedbykwiatkowskietal.,1992testforthenullhypothesisofstationarity.thecombination

26、oftheunit root tests results see table 1 suggeststhat the variables are integrated of order one i.e., i1. thisimplies the possibility of cointegrating relationships. table 2 provides the results for the johansencointegration testsbased onanecmusing anoptimallaglength oftwo.resultsfromboththetraceand

27、- tests indicate that the variables in the system are cointegrated. the existence of cointegratingrelationshipsimpliesthatanecmspecificationis appropriate.furthermore,theresiduals fromtheecmspecification are white noise.table 3 reports the results of granger causality tests based on the ecms. each c

28、olumn represents anequationforeachofthefivevariablesinthesystem.foreachvariable,atleastonechannelofcausalityisactive: short-run granger causality through the joint significance tests of the lagged-differencedcoefficients f-statistics or long-run causality through a statistically significant lagged e

29、rror-correctionterm t-statistics. a significant laggedectcoefficient implies that past equilibriumerrors affect currentoutcomes.inthecaseofbulgaria,boththeelgandglehypothesesaresupportedbythedataatthe5%levelofsignificance.thereisgrangercausalityfromexportstogdpgrowthp0.0118andthereverse392 t.o. awok

30、use / economics letters 94 2007 389?395table 1tests for unit rootvariable bulgaria czech republic polandadf kpss adf kpss adf kpsslevelsgdp ?2.699 0.164 ?1.899 0.150 ?2.421 0.334exports ?3.024 0.159 ?0.974 0.703 ?0.901 0.168capital ?4.683 0.212 ?3.277 0.167 ?2.194 0.153labor ?1.900 0.165 ?1.867 0.12

31、7 ?3.213 0.177imports ?2.163 0.176 ?1.581 0.718 ?1.685 0.1261st differencesgdp ?4.170 0.269 ?5.664 0.312 ?5.780 0.171exports ?6.101 0.087 ?4.873 0.401 ?5.515 0.300capital ?3.411 0.105 ?2.266 0.122 ?4.943 0.203labor ?6.914 0.341 ?6.321 0.142 ?7.384 0.329imports ?6.121 0.351 ?2.591 0.316 ?6.213 0.234a

32、nd denotesrejectionofthenullhypothesisofunitrootsfortheaugmenteddickey?fulleradftestsatthe5%and10%significance levelsand denotesrejectionofthenullhypothesisofstationarityforthekwiatkowski,phillips,schmidt,andshinkpsstestsatthe 5% and 10% significance levels.critical values at the 5% and 10% levels o

33、f significance for the adf with linear trend are?3.549 and?3.207, respectively.critical values at the 5% and 10% levels of significance for the kpss with linear trend are 0.146 and 0.119, respectively.table 2johansen cointegration test resultscointegrating bulgaria czech republic poland c5%rank rtra

34、ce statistics trace statistics trace statisticsr0 104.378 90.744 138.945 69.819r1 42.987 48.382 57.803 47.856r2 16.148 25.977 26.758 29.797r3 5.076 8.733 4.633 15.495r4 0.982 2.683 0.174 3.841- - -statistics statistics statistics c5%r0 61.392 42.363 81.142 33.877r1 26.839 22.405 31.045 27.584r2 11.0

35、72 17.243 22.162 21.132r3 4.094 6.051 4.459 14.265r4 0.982 2.683 0.174 3.841r denotes the number of cointegrating vectors for cointegration test with constant within and outside the cointegrating vectors.johansen'scointegrationtestfollowsasequentialprocessforthedeterminationofthenumberofcointegr

36、ationvectors.westopatthe first r where we fail to reject the null hypothesis. the critical values c5% for the tests are taken from osterwald-lenum.*denotes rejection of the null hypothesis of cointegration rank r at the 5% significance level.t.o. awokuse / economics letters 94 2007 389?395 393table

37、3granger causality test results based on error correction models ecmindependent gdp dependent variablesvariablesexports capital labor importsbulgariagdp ? 28.8011 2.1072 22.6407 38.27800.0000 0.5505 0.0000 0.0000exports 10.9804 ? 2.5741 13.6154 7.62370.0118 0.4620 0.0035 0.0545capital 4.5984 2.3130

38、? 5.7842 0.86930.2037 0.5100 0.1226 0.8328labor 3.2183 6.6568 1.6661 ? 6.00410.3592 0.0837 0.6445 0.1114imports 2.8560 17.2791 0.6330 11.7167 ?0.4144 0.0006 0.8888 0.0084lagged ect ?1.07403 2.70768 0.86239 ?3.48998 2.05485czech rep.gdp ? 0.5030 8.2041 1.8525 2.00950.7776 0.0165 0.3960 0.3661exports

39、18.3790 ? 1.3907 0.5258 3.4484300.0001 0.4989 0.7688 0.1783capital 1.7229 1.5973 ? 1.6546 1.73540.4226 0.4499 0.4372 0.4199labor 4.8737 1.4497 2.7035 ? 0.22950.0874 0.4844 0.2588 0.8916imports 9.8766 2.5598 1.4947 2.6561 ?0.0072 0.2781 0.4736 0.2650lagged ect ?5.01776 ?3.02683 ?2.38648 ?0.28896 0.20

40、188polandgdp ? 0.5082 0.4545 3.1318 0.80800.4759 0.5002 0.0768 0.3687exports 0.4386 ? 0.4197 0.1925 0.71680.5078 0.5171 0.6608 0.3972capital 0.5977 0.4646 ? 3.2853 0.09660.4395 0.4955 0.0699 0.7559labor 0.6540 0.6785 0.4926 ? 0.31140.4187 0.4101 0.4828 0.5768imports 0.0006 0.3118 0.0219 3.6146 ?0.98

41、11 0.5766 0.8823 0.0573lagged ect 0.36376 1.28903 3.40543 5.11779 0.60777valuesinbracketsareestimatedt-statisticsforeachcointegrationequation.allothervaluesareasymptoticgrangercausality f-tests, values in parentheses are p-values.short-run causation from gdp growth to exports is also supported p0.00

42、00. although causality alsoflows from gdp growth to imports p0.0000, the reverse case does not hold implying a lack ofempirical support for the ilg hypothesis p0.4144. for the czech republic, the empirical resultssuggest that both the elg and ilg hypotheses are supported at the 5% level of significa

43、nce. the394 t.o. awokuse / economics letters 94 2007 389?395statistically significant error correction term indicates empirical evidence in support of long-run causalrelationshipfromexportsandimportstogdpgrowth.inpoland,onlytheilghypothesisissupportedbythedatap0.0295.overall,grangercausalitytestresu

44、ltssuggestthatimportsplayasmuchofaroleas1exports in stimulating economic growth in these countries.4. concluding remarksinrecentyears,therehasbeenmuchattentiondevotedtotheroleofinternationaltradeasanengineofgrowth. however, scant empirical studies exist on the causal relationship between trade and e

45、conomicgrowth for central and eastern european countries. this paper contributes to this literature by using aneoclassical growth modeling framework and multivariate cointegrated var methods to investigate thecontributionofbothexportsandimportstoeconomicgrowthinbulgaria,czechrepublic,andpoland.inthe

46、caseofbulgaria,theresultssuggestthatempiricalevidenceexistsforbothelgandglehypotheses.in the czech republic, granger causality flows from both exports and imports to economic growth thusproviding empirical support for both elg and ilg hypotheses. in contrast, only the ilg hypothesis issupportedbythe

47、polishdata.insummary,thisstudy'sfindingsindicatethattheexclusionofimportsandthe singular focus of many past studies on just the role of exports as the engine of growth may bemisleading or at best incomplete.acknowledgementthe author thanks hong yin for research assistance. the usual disclaimer applies.referencesbalassa, b