Download a survey of trade theories - Wageningen UR E

S. van Berkum
H.vanMeijl
Onderzoekverslag 161
A SURVEY OFTRADETHEORIES
April 1998
3
j ^
«
SUI.J.ifnA
g
EX N O
MLV:
Agricultural Economics ResearchInstitute(LEI-DLO)
Afdeling Algemeen EconomischOnderzoek enStatistiek
£>
ABSTRACT
A SURVEYOFTRADE THEORIES
Berkum,S. vanand H.van Meijl
The Hague,Agricultural Economics ResearchInstitute (LEI-DLO),1998
ISBN90-5242-438-1
Onderzoekverslag 161
99p.,fig.
Thisreport isasurveyoftradetheoriesandreviewsinternational trade implicationsofthe newgrowththeories.Theoriesarereviewed byfocusing onquestions like
why countriestradewith eachother,whatcanbegainedbytradeandhowtrade patternscanbeexplained.Thesequestionshavebeenaddressedbymanyeconomistssince
Adam Smith and David Ricardo and stillthere ismuch controversy in explaining the
causesoftrade.Differences between countries,for instance in naturalfactor endowmentsandfactor prices,canbeamotive for trade betweentwo countries. Countries
trade in order to take advantage of these differences.Thisconcept of trade isbased
on(thetheory of) comparative advantage. However,other -more modern -theories
statethatcountriesmayalsotradebecausethereareinherentadvantagesinspecialization,arisingfrom the existence of economies of scale.Someother models in modern
tradetheoriesemphasize imperfectcompetition,productdifferentiation andtechnologygaps(innovation) acrossfirmsandcountriesasa majorsourceof explanations for
international trade. Finally,the trade implications of the 'new' growth theories will
alsobetaken into account becausethesetheoriesshedlight uponthe dynamicevolutionofcomparativeadvantage.Thisreviewresultsinasummaryofthemaincharacteristicsofthetheories,theirwayofexplaining internationaltrade,implications of trade
andthe influence of government intervention. Furthermore,the empirical evidence
of thetheories is discussed.
Tradetheories/Growththeories/Comparative advantage/Policy/Economies ofscale/
Productdifferentiation/Technology gaps/Knowledge
Thecontentsofthis report maybequoted orreproduced without further permission.
Dueacknowledgement isrequested.
CONTENTS
Page
FOREWORD
5
SUMMARY
7
1.
INTRODUCTION
19
2.
TRADITIONAL TRADETHEORIES
2.1 Classical theories
2.1.1 Absolute advantage (Smith)
2.1.2 Comparative advantage (Ricardo)
2.2 Neo-classical theory
2.2.1 Heckscher-Ohlin-Samuelson model
2.2.2 Specific factors model
2.3 Trade policy under perfect competition
2.4 Empirics
3.
MODERN TRADETHEORIES
43
3.1 Economies of scale and imperfect competition
43
3.1.1 External economies of scale
44
3.1.2 Internal economies of scale and imperfect competition 45
3.1.2.1 Monopolistic competition
45
3.1.2.2 Oligopolistic competition
49
3.1.3 Trade policy and imperfect competition
50
3.1.4 Empirics of trade theories related to economies
of scale and imperfect competition
54
3.2 Neo-technology trade theories
56
4.
TRADE IMPLICATIONS OFTHE NEW GROWTHTHEORIES
4.1 New Growth Theories
4.1.1 Knowledge asside product: external economies
of scale
4.1.2 Knowledge asinvestment: acombination of external
and internal economies of scale with monopolistic
competition
4.1.3 Empirics
4.2 Evolutionary Growth Theories
29
29
29
29
34
34
37
38
40
62
62
62
66
70
71
Page
5.
CONCLUSIONS
5.1 Evaluation of empirical tests of theories explaining
international trade
5.2 Government policy and international trade
REFERENCES
APPENDICES
1.
Traditional Trade Theories
2.
Modern Trade Theories
3.
Trade implications of the growth theories
75
75
76
79
89
90
92
95
FOREWORD
Most agricultural trade analysisfocuseson basicagricultural commodities
and restson traditional theoretical insights of comparative advantage, assuming perfect competitive markets on which goods are homogeneous and produced under atechnology of constant returns to scale. However, the observation of changing trade characteristics in agriculture and food products, private
businessconcentration and activegovernment policysuggestthat international
agricultural trade analysis implies investigating market structures other than
the competitive mode. Inorder to strengthen and buildup the theoretical and
empirical knowledge base inthe field of international trade inagricultural and
food products, the Agricultural Economics Research Institute (LEI-DLO)
launchedthe researchproject 'Policyand Patternsof International Trade' which
isfinanced bythe Institute's budget for Strategic Expertise Development (SEO
programme).
The objectives of this research project are to:
analyse international tradetheories,with the aimto answerthe question:
what determines international trade patterns and which role does government policy play in this?
assessthe usability of general trade theories in explaining agricultural
trade;
design aconcept for explaining world trade patterns in agricultural commodities.
This publication reports on the first objective of the project. Its main aim
is t o present an overview of the main theories that deal with international
trade, asthere arethe neo-classical, 'traditional' trade theories and the 'modern' trade theories. The trade implications of the 'new' growth theories are
also taken into account. This review of theories - written mainly during the
course of 1996 - acts as an input in the following stages of the project. The
secondstage,on the assessment of the general trade theories on their applicability for explaining agricultural trade, has been reported in Onderzoekverslag 162.
The director.
The Hague, April 1998
L.C.Zachariasse
SUMMARY
Thequestionwhycountriestradewitheachother hasbeenanswered in
many different ways.Differences between countries,for instance in natural
factor endowmentsandtechnology,canbeamotivefor trade betweencountries.Countriestradeinordertotakeadvantageofthesedifferences.Thisconcept of trade isbased on (the theory of) comparative advantage. However,
other recenttheoriesstatethatcountriesmayalsotrade becausethereareinherent advantages inspecialization,arisingfromthe existence of economies
of scale.Someother models in modern trade theories emphasize imperfect
competition,productdifferentiation andtechnology gaps(innovation) across
firmsandcountriesasamajorsourceof international trade.Finally,the 'new'
growth theories emphasize the endogenous generation of technological
changewhich hasimportant implicationsfor international trade.
Themainaimofthisstudyisto presentanoverviewofthe maintheories
that deal with international trade.The general structure of this overview is
illustratedinfigure 1.Inahistoricalsensewecanidentify three major streams
(depictedasthethree columns infigure 1).First,wecanidentify the classical
andneo-classicaltradetheorieswhichwecallthe TraditionalTradeTheories'.
Themostprominent modelofthisstreambecamethe neoclassical HeckscherOhlin model that dominated the field for almost eighty years.Despite their
theoretical dominance, some implications were not supported by empirical
evidence.Thisinducedeconomiststosearchfor newtradetheories.Thesenew
theories,whichwere mainlydeveloped inthe lateseventiesandearlyeighties,
arecollectedinoursecondmajorstream,the 'ModernTradeTheories'.Inthe
late eighties, ideasthat were generated bythese moderntrade theories inducedchangesinthegrowth literatureand ledtotheso-called 'new' growth
theories,which also shed light upon the dynamic evolution of comparative
advantage.Thetradeimplicationsofthesegrowththeoriesarealsotaken into
accountandcapturedinourthirdmainstream,'TradeImplicationsof Growth
Theories'.
Withineachmajorstreamafurtherclassificationinschoolsofthought is
made basedon crucial assumptions andthe main mechanism of trade.Each
school of thought ispresentedwith agrey ellipse infigure 1;the arrowsbetween two ellipses or schools indicate astrong relation between these two
schools.Arelationinthesensethatitbuildsforwardon(elementsof)aformer
onebutchangesor bringsinacrucial(new)assumption.
Animportant discriminatingaspectiswhether or nottechnology differs
betweencountries(thehorizontal lineinthe middleoffigure 1 representsthis
distinction).Whenaschoolofthought assumesthattechnology differsacross
countries it isdepicted inthe upper part of figure 1,and when aschoolassumesthattechnology isidenticalacrosscountries it isdepicted inthe lower
part. Because inthe new growth theorieswithin one schooltechnology can
varyacrosscountries,we identifiedtwo directionswithintheseschools.Each
direction isillustratedasawhite boxinfigure1.
Before we discussthe schoolsof thought in more detailwe discussthe
mainrelation(arrows)betweentheschools.Asmanyothertheories,thetrade
theories find someof their roots in The Wealth of Nations' of AdamSmith.
Smith showedthat trade ispossiblewhenonecountrycanproduceacertain
goodwith lesslabourthanthe other country andthe other country canproduceanothergoodmoreefficiently.Ricardoshowedthattradeisevenpossible
whenonecountrycanproduceallgoodsmoreefficientlythantheothercountry if the relativecostsof production oftwo goodsdiffer between countries.
This isknown asthe principle of comparative advantage which isstill one of
the most important concepts in trade theory. The Neo-classical HeckscherOhlin-Samuelson(H-O-S)modelelaboratedthetheory of Ricardoby introducinganother factor of production (capital),butassumingidentical production
techniques across countries.An implication of this change in assumptions is
thatfactor endowmentsbecamethemainexplanationfor trade.Thespecificfactormodelassumesthatonefactor isspecifictotheproductionofonegood;
becausethe income implications are different from the standard H-O-Sthis
theory istreated separately.
BecausesomeofthetradeimplicationsoftheH-O-Smodelwerenotsupportedbyempiricalevidence,themoderntradetheoriesreplacedtheassumption of constant returnsto scaleor identicalproductiontechnologies.Economiesofscalecanbeexternalor internalto afirm.With externaleconomies of
scaleperfectcompetitionremainswhyinternaleconomiesofscaleimplyimperfectcompetition.Thisdistinction isimportant becausetradeimplications differ
betweenthetwo approaches.Thethirdschoolwithinthe moderntradetheories,i.e.theneo-technologytradetheories,likethe classicaltheories,stressed
thecentral roleoftechnology. However, incontrastto thesetheoriestechnologicaldifferencesarenotstaticbetweencountriesbutaretemporarilycreated
byinnovations.Theevolutionarygrowththeoriesbuildforward onthistheory
andfocussedmoreindepthontheinnovationprocess; innovationsarecumulative,specificand irreversible.
Like the neo-technology trade theories, the more formalized 'new'
growththeoriesalsostressedtheroleofknowledgecreation.Thevariousways
in which knowledge creation can be modelled are taken from the modern
tradetheoriesthatstressedeconomiesofscale.Whenknowledge isaby-product of other activitiesor causedbylearning-by-doing effects,there isaclose
resemblancewith external economiesof scale.However,when knowledgeis
the intentionaloutcomeof economic behaviour firmshaveto investsomeresourcesinknowledgecreation.Thismeansthatfirmshavesomefixedcoststhat
leadto internaleconomiesof scaleand imperfect competition.Thisapproach
usesalsoelementsoftheneo-technologytradetheoriesbecauseaninnovation
leadsto (temporary) new products.Furthermore,it usessomeelements from
the external economiesof scaleapproach becausea part of the knowledge
createdbythefirmcanbeusedbyotherfirms.Importantwith regardto trade
and growth implications iswhether theseso-called knowledge spillovers are
nationalor internationalinnature.Inbothschoolsofthenewgrowththeories
the initialtrade pattern iscausedbydifferences infactor endowments or differencesininitial knowledgelevels.
Traditional theories
Withinthetraditionaltradetheories,classicaltheory(SmithandRicardo)
and neo-classical theory (Heckscher-Ohlin-Samuelson) are the schools of
thought to bedistinguished.Traditional trade theories focus on differences
amongcountriesthat arethe resultofdifferences intechnology (classicaltheory)ordifferencesinrelativefactorendowments(neo-classicaltheory).Oneof
thefirsttheoriesof international trade isthe classicaltheory of absolutecost
advantages.Accordingto Smith,tradeonlyappearswhenthere areabsolute
costdifferences betweencountries.DavidRicardoshowedtheshortcomings of
thistheorybecause,evenifonecountrycanproduceallgoodsmoreefficiently
thananothercountry,trade ispossibleandbeneficial.Apre-condition isthat
the relativeefficiencygapisnotthesamefor allgoods.Whenthisisthecase,
a country hasacomparative advantage in agood that hasthe highest efficiency gap. Inthe Ricardian model labour isthe only production factor and
differences in labour productivity arethe mainexplanationfor trade.Labour
productivitydiffersbetweencountriesbecausetheirtechnological knowledge
leveldiffers and/ortherearedifferences innatural circumstances(naturalresources,climate,soil,geographical position).
Theneo-classicaltheoryelaboratedthesetheoriesbyincludingmoreproductionfactors.However,contrarytotheclassicaltheoriesthistheoryassumed
identicalproductiontechniquesovercountries.Furthermore,thestandardneoclassicalHeckscher-Ohlin-Samuelson(H-O-S)modelassumesconstantreturnsto
scale,identicalconsumerpreferencesandperfectcompetition.Theseassumptionsimplythatdifferences infactorendowmentsarethe onlyexplanation of
trade inthe H-O-S model.Thelargerthedifference infactorendowments,the
moretradebetweencountriesandallthistradeisinter-industrytrade.Acountryexportsthegoodwhichmakesthe mostintensive useof it'sabundant factor of production.The relative abundant factor will gain from trade. In the
tradeequilibrium relativefactor pricesarethesameacrosscountries.
In the traditional theories gains from trade come from exchange and
specialization.Allcountrieswill benefitfromtradebecauseof amore rational
allocationof productive resourcesandlower relative pricesfor the importing
competing product.The lessbarriersto tradethere are,the more beneficial
tradewillbe.Therefore,freetradepolicyisseenasthebesttradepolicy,unless
countriescanimprovetheirtermsoftrade (only largecountries).Trade policy
to 'correct' domestic distortions or for political reasons(for instance incomedistributional effectsoftrade)aresecondbest.
Thespecific-factor modelisaspecialversionofthestandardneo-classical
H-O-S model.Inthe standard H-O-S model all production factors are mobile
betweensectors,while intheshort-termspecific-factor modelsomeareimmobile. This implies,for example that trade implications for production factor
rewardsaretotally different fromthestandard model.Tradeisbeneficial for
10
thespecificfactorthatisnecessarytoproducetheexportgoodanditreduces
theincomeofthespecificfactorusedintheimportgood.Thewelfareimplicationsoftradeforthemobilefactordependontheconsumptionpatternofa
workerandareambiguous.
Despitethetheoreticaldominanceoftheneo-classicalmodelforalong
periodthe implicationsofthismodelwerenotunambiguouslysupportedby
empiricalstudies.ThemostinfluentialstudywasdonebyLeontief(1953)who
foundthat importsoftheUS (acapitalabundantcountry)weremorecapital
intensivethanitsexports.However,theapplicationof Leontief'smethodhas
beencriticizedbymanyauthorsandthesuggestionthattheH-O-Stheoryperformsbadlyhasbeencounteracted.AsabyproductofLeontief'sresultsandthe
debatethatfollowed,theH-O-Stheoremhasbeenextendedtoallowforadditionalfactorsbeyondjustcapitalandlabourtoexplaintrade,ormakea clearer
distinctionbetweenskilledandunskilledlabourandhumanandphysical capitalasfactorsdetermining internationaltradeflows.
Modemtrade theories
Still,empiricalstudiesshowedthat-contrarytowhatwouldbeexpected
accordingtotheH-O-Stheory-mosttradeisbetweencountrieswiththesame
factorendowmentsthatamajorpartoftradebetween industrialcountries is
ofanintra-industrynatureandthat income-distributional effectsoftradeare
small.Thesecontradictionwithtraditionaltheoryinducedeconomiststosearch
for newtradetheories.The'new'tradetheorieselaboratedtheneo-classical
frameworkbyreplacingtheassumptionsofconstantreturnstoscaleandperfectcompetition.Asecond stream,the'neo-technologytheories',liketheclassicaltheories,stressedthe centralroleoftechnologyandproposedaradical
departurefromtheneo-classicalframework.
The'new'tradetheoriesassumed increasingreturnstoscaleandthisimpliesimperfectcompetitionunlesseconomiesofscaleareassumedtobetotaiy
externaltotheindividualfirms.Afirstapproachassumestheseso-called'externaleconomiesofscale'; anindustrystillcontainsmanysmallfirmsandperfect
competitionremains.Asecondapproachassumesinternaleconomiesofscale
whichleadtoimperfectcompetition.Withinthisapproachtwodirectionshave
been identified.Thefirstdirectionconcentrateson modellingeconomiesof
scaleandtreatsmarketimperfectionsassimplyaspossiblebyassuming monopolisticcompetition.Aseconddirectionconcentratesonimperfectcompetition
and useseconomiesof scaleto causethese market imperfections.Themain
marketstructuretheyuseis 'Cournot'or'Bertrand'oligopoly.Themaindifferencebetweenthese'new'tradetheoriesandtheneo-technologytradetheoriesisthattheformerassumeidenticalproductiontechnologiesacrosscountrieswhilethe latteremphasize (endogenous)technological innovationand
technologygapsacrossfirmsandcountriesasamajorreasonfor international
trade.
11
Economiesofscaleand(im)perfectcompetition
Accordingtothenewtradetheories,tradeis possiblebetweencountries
identicalinfactorendowments,technologiesandtastes.Inthesetheoriesthe
mainexplanationfortradeareeconomiesofscale.However, incontradiction
to the traditionaltheories,thedirectionofspecializationwith economiesof
scaleis often unknown.Consequently,thistheorygivesanimportantroleto
historyandaccidentindeterminingthepatternofinternationaltrade.
Economiesofscalecanbeexternalorinternal.Implicationsoftradediffer
wheneconomiesofscaleareexternalorinternal.Whenexternaleconomiesare
important,acountrystartingwithalargeindustrymayretainthatadvantage
evenifanothercountrycouldpotentiallyproducethesamegoodsmorecheaply.Gainsfromtradecomefromexchange,specializationandexploitingeconomiesofscale.However,thedivisionofwelfare betweencountriescanbevery
unequal,dependingonthespecializationpattern(whetherthecountryspecializesinthegoodproducedwithexternaleconomiesofscaleornot)andterms
oftrade(dependingonsupplyanddemandofgoodsinthetradeequilibrium).
Countriescanevenloosefromtrade.Inthatcasetrade(orindustrial)policies
canbebeneficial.
Withmonopolisticcompetition,anindustrycontainsasufficiently large
numberof'similar'firmsproducingdifferentiated'unique'productsandprofits
arecompetedawayintheequilibrium.Themainmechanismoftradeatamarketstructureofmonopolisticcompetitionare(internal)economiesofscaleand
productdifferentiationwhichcausetheproductionofeachvarietyto beconcentratedineachcountry.Eachcountryproducesadifferentsetofvarietiesof
acertainproduct.Becauseconsumersdisplaya'loveofvariety',theydemand
all varieties,which impliesthat acountry importseachof the varietiesproduced inother countriesandexportseachofthe varietiesdomesticallyproduced.So,therewillbeintra-industrytrade.However,itisunclearwhichcountryproduceswhichvariety.Againtheexactspecializationpatterndependson
historyandaccident.
Thegainsfromtradewithmonopolisticcompetitionarefromexchange,
specialization,exploitingeconomiesofscale, exitofredundantfirmsandmore
product variety. Income-distributional effectsof intra-industry trade areless
thanthoseofinter-industrytrade(H-O-Smodel)becausethereareadditional
gainsfromtrade.Becauseintra-industrytradewillbedominantbetweencountriesatasimilarlevelofeconomicdevelopment,tradewithoutseriousincomedistributionaleffectsismostlikelytohappenintradebetweencountriessimilar
intheirrelativefactorendowments.
Inanoligopolisticmarketstructure,thebehaviourofthefirmsinfluences
eachother.Trade occursbecauseof economiesofscale.However, if market
segmentation andpricediscriminationarepossible,therecanbetradeeven
withouteconomiesofscaleandcomparativeadvantage.Gainsfromtradeappearintheformofthepro-competitiveeffect(i.e.lower mark-up),theexitof
firmswhichareunabletocovertheirfixedcosts,andloweraveragecostsifthe
productionscaleofafirm increases. Welfare implicationsareunclear.
12
Theassumptionofconstantreturnstoscaleandperfectmarketsjustified
the domination of freetrade policyfor almostacentury.These 'new' trade
theoriesquestionthisfreetradepolicybyintroducingeconomiesofscale and
imperfect markets.Theyfindthatanactivetrade policycanbebeneficial in
certaincircumstances.Forinstance,if inan industrymakesexcessiveprofits,
such anindustryshouldbedesired.Undercertainconditions,theuseofexport
subsidiescanalsoshift profitsfromforeignto domesticfirms.Thisis thesocalled'strategic'tradepolicy.However,thenewargumentsforprotectionare
verydependentonspecificassumptions.Aslightchangeinoneoftheassumptionschangesorevenreversesthe implicationsofapolicy.Goodpolicythereforerequiresthatthegovernmenthasalotofinformationtohelpchoosethe
rightmodel.Thisinformationhastobeso detailedthat itis notreadilyavailable.Theempiricsalsoshowthatbenefitsfromdeviationsfromfreetradeare
small.Furthermore,empiricaltestsoftheories relatedto economiesofscale
and imperfect competition appear to berathersuggestive.Estimatesofthe
impactoftradepoliciesunderimperfectcompetitionlendnosupporttoastrategicrolefortradepolicy.Therefore,freetradeisstillconsideredtobeagood
ruleofthumb,althoughitis notoptimalunderimperfectcompetition.
Neo-technologytheories
Thecommonfeatureoftechnology-orientedtheoriesoftradeisan emphasisontechnological changeandthe resulting patternsoftrade.Inthese
theoriestradepatternsareexplainedintermsoftechnologicalprogress.Technological differences or gaps acrosscountriesare an endogenous outcome
throughfirmlevelproductandprocessinnovationthatreducescostsofproductionandgeneratesnewproducts.Theflowoftechnologicaldevelopmentsand
innovationisassumedtobenotfreeandinstantaneous,whichimpliesthata
firm/country hasat leastatemporary comparative advantage in production
and exports.Thedifference withthe Ricardiantrade modelsisthat inthose
models differences intechnology (productivity)for somegiven goodscause
trade,whereintheneo-technologytrademodelstradeis inducedbecausethe
innovatingcountrygeneratessomenewproductsthatothercountries,atleast
temporarily,areunabletoproduce.
EarlycontributionsinthisfieldhavebeenmadebyKravis,Posner,Vernon
andHirschinthe 1950sand1960s.Theseauthorsdescribeacontinualprocess
of innovative developments inwhichcountrieswherethe innovationsoccur
enjoytemporarilytechnologicaladvantageovertradingpartnersuntilthenew
technologyis imitatedinothercountries.Each ofthecontributorsstressesdifferent reasonsinexplainingwhycountrieswillproduceandexportnewproducts,liketheavailabilityoftechnologyto producenew products(asin Posner's'technologygap'model),theavailabilityofskilledlabour(Hirsch)orvicinityoftheirmarkets(asinVernon's'productcycle hypothesis').
Krugmanhasformalizedtheattemptstotrytoexplaintrade intermsof
technologyinhisNorth-Southmodel.InthismodeltheNorthinnovatesacontinuousstreamof newproductvarietiesonthemarket,whiletheSouthonly
imitatesafteratime-lag.Themainmechanismfortradearetechnologygaps
13
whichgrowwith innovationandclosewithimitation.Bothcountriesbenefit
fromtradebecauseofexchange,specializationandmoreproductvariety.However, while the world isbetter off asproduction isshifting to a lower-cost
country,theNorthmaylosefromimitationbytheSouth.Theinnovatingcountry maytry to increase the innovation rate (innovation policy) or consider
protectionisticmeasuresto reduceimitation,unlessthereisacorrectinternationallytechnologytransfersystem.
Most,ifnotallempiricaltestsofthetechnologytradetheoriestrytoexplainthepatternoftradeoftheUS, simplyassumingthattheUS istheinnovativeNortherncountrywithhighpercapitaincomesandrelativewagesandthe
restoftheworldtheimitatingSouth.Thesetestsareusuallyrathersuccessful.
SeveralauthorssoughtandfoundpositivecorrelationsbetweenUSexportperformanceacrossindustriesandvariousmeasuresofR&D.SinceR&Disrelated
to technological progress,whatever itscauseor effects,thisevidence lends
support to all technologytheories of trade.Severaltestsof the technology
theoryoftradehavealsointroducedadditionalexplanatoryvariables,includingthosethatareappropriatetothefactor proportionstheory,andsupport
the conclusionthatthere isastrongandpositivecorrelation betweentrade
performanceandtechnology-relatedvariables.Atthesametime,however, it
becomesclearthattechnology-relatedvariablesaremuchrelatedtotheability
of individuals,firms,orcountriesto developandexploittechnologywhichis
relatedtotheavailabilityof knowledgeandskills.It is,therefore,difficultto
distinguishevidencesupportingtechnologyfromevidencesupportinghuman
capitalorskillsas determinantsoftrade. Furthermore,morerecentobservationssuggestthattechnologicallevelsamongcountriesconvergerapidlyand
becauseofmultinationalsthespeedofdiffusionof innovationsaccelerate. So
itappearsthat,besidestheoldertestswithUSdata,empiricaltestsorapplicationoftheproductcycletheoryarelimitedinextent.Thesupportfromthese
studiesforthe idea ofthe 'technologygap' asdrivingforce behindtradeis
fragmentary at best,suggestingevidenceforthe ideathattransitoryadvantage resultingfrom innovationcanbeamajorfactor intradefor onlysome
industries.
TradeImplicationsofGrowth Theories
Whilealltradetheoriesaremainlystaticandfocusonallocationissues,
aninterestingdevelopmenthastakenplaceinthegrowthliterature.The'new'
growththeoriesbuildforwardonthestatic'new'trademodelsandputthem
inadynamiccontext.Thenewtradetheoriesprovidedthebuildingblockssuch
asthetreatmentofeconomiesofscaleandmarketimperfections.Liketheneotechnologytheoriesthe'new'growththeoriesstresstheroleoftechnological
change.Byputtingtheseelementsinadynamiccontextthese 'new' growth
theoriesdealwiththedynamicevolutionofcomparativeadvantageandthe
consequencesoftrade inaworldofglobaltechnologicalcompetition.
Thenewgrowththeoriesfoundseveralwaystoendogenizetechnological
changeinageneralequilibriummodel.Twoapproachescanbedistinguished.
The first approach assumesthat externalities or 'learning-by-doing' effects,
14
whichareby-productsofotheractivities,causegrowthandtheexternaleconomies of scale approach isusedto model theseeffects.Thesecond approach
assumesthat technological change isthe intentional outcome of economic
behaviour andfirms haveto 'invest' inknowledgecreationto obtaintechnologicalchange.Investments inknowledge canbeseenasakindof fixedcosts
andmonopolisticcompetitionmakesitpossibletocoverthesefixedcosts.Most
studiesthat usethesecondapproach,alsoassumethat knowledge generates
someexternalitiesandaretherefore amixture of both approaches.
In modelswhere external economiesof scaledetermine the dynamical
evolutionofthespecializationpattern(thefirstapproach),the centralmechanism isthat afirm -unintentionally -createsknowledge andthis knowledge
flowsdirectlyto allotherfirms,where it increasesthe productivity levelof the
productionfactorthatcanbeaccumulated.Inprincipletheinitialspecialization
andtrade pattern isdetermined bycomparative advantage (initialfactorendowments)orthe initial knowledge stock(technological capabilities).Thedynamic implications of theseexternaleconomiesof scalegrowth theories are
that acountry will built up knowledge or expertise inthe goods in which it
specializesandtherefore reinforcesitscomparativeadvantage inthesegoods.
Becausethetechnological opportunities differ between goodsthespecialization patterndeterminesthereforealsothewelfare levelandlongterm growth
of acountry.
Dependingonthespecificdemandconditions,trade or industrial policy
maybebeneficialifacountryspecializesinthelow-techgood;thismayreduce
welfareormayleadtolesswelfarethaninthecasethecountryhadspecialized
inthe high-tech good.Protection measuresor industrial policies mayreverse
thespecialization patternwhenspecialization inanothersector increaseswelfare. However,to deducethecorrectpolicyadviceisverydifficult becauseone
hasto knowtheexacttechnological opportunities ofdifferent goodsin different countries.
Modelsthatconcentrateontheinvestmentinknowledge(humancapital)
combine imperfect competition with externalities. Through investments in
R&D,afirm producesnewgoodsbyexpandingproductvarietyor quality.Furthermore,therearealsosomespilloversontheaggregatestockof knowledge.
Alargerstockof knowledge,inturn, reducesthecostsof producing blueprints
of newproducts.Thiscausesaconstantincentiveto investinR&D.Manufacturingwillthereforealsobegrowingataconstantratio.Importantforthe generation of endogenous growth isthat the incentive to invest in R&Ddoes not
decline.Inallthesekindof modelsthe growingstockof knowledge asaside
product of R&Dgeneratesthis constant incentive.The R&Dinvestments are
dependent on the specialization pattern which iscaused bythe principle of
comparativeadvantage(factorendowments),history,theinitialstockof knowledge,the scaleof acountry andthe demandstructure.Thesefactors determinethe numberof peoplethatareworking inthe R&Dsector,the high-tech
and the low-tech sector.Thewelfare andgrowth implications are therefore
alsodependentonthespecializationpattern.Importantiswhether knowledge
spilloversarenational or international inscope.
15
When there are international knowledge spillovers,all innovators will
have the same knowledge and national advantages in R&Darise only from
differences inrelativefactorprices(whicharedependent onresource endowments).Factorssuchasthe sizeof acountry andthe historyof its production
play no role inthe long-termtrade pattern;what only mattersarefactorendowments.
Withonlynationalknowledgespilloverstheinitialconditionsgovernlong
runoutcomes.Inmanysituationsthe countrywith the initial greater stock of
knowledgehasanadvantageinR&Dandaccumulatesknowledgemorequickly
thanitstradingpartner.Thissustainsandaddsto itsproductivity lead.History
alonedetermines long-runtrade patternsandgrowth rates(i.e.hysteresis).
Argumentsfor policyaretoobtainahigherwelfarelevelbyreversingthe
specializationpatternifthereareonlynationalknowledgespillovers.Industrial
policy (R&Dsubsidies) are considered first best andtrade policy measuresas
secondbest policy.
Evolutionary growth theoriesassumethattechnology playsthe fundamentalroleineconomiclife.Technologicalchangeandinnovationisacumulative,specificand irreversible process.Themaintrade mechanism isthatabsolutetechnologicaldifferences determinetheworld market positionof allsectors.Relativetechnologicalgapsplayaminor part.Theydeterminethespecialization pattern between sectorsaccordingto the mechanismof comparative
advantage.Futuregrowth andtechnological developments isdetermined by
thecurrentspecializationpattern.Thecurrentspecializationpatternofacountry hastherefore adynamic effect becausethis pattern determines in which
sectorstechnicalskillswillbeaccumulated,innovationswillbedone,economies
ofscalewill berealised,etcetera.Sectorsdiffer intheir growth opportunities
such that the present specialization pattern is extremely important for the
countries'futureeconomicperformance.Aspecializationpatternaccording to
the traditional mechanism of comparative advantage can lead acountry to
specializeinthoseindustries(sectors)andactivities inwhichthe opportunities
for growthandtechnological development areleast.Aspecialization pattern
which isstatic (Ricardian) efficient cantherefore bedynamic inefficient and
viceversa.Ifthistradeoff occurs,acountrycantrytochangethe specialization
pattern andfuture growth paththrough industrial ortrade policy.
Evaluationonpolicy
Traditional theories suggestthat trade isbeneficial for all countriesinvolvedandtherefore supportfreetrade.Onlyinsomecircumstances(alarge
countrymayimproveitstermsoftrade)there isanargumentfor trade policy.
Moderntradetheoriesextendthetraditional argumentsfortradepolicyand
addsomenewargumentsfor government policyinterventionsuchasthestrategictradeargument. However,thecircumstancesinwhichatrade policy may
bebeneficial appearto beveryspecific.Moreover,agovernment needsa lot
of very detailed information to make the right decision.And besides that,
empiricsshowthat benefitsfromdeviationsfromfreetradeareverysmall. So,
moderntradetheoriesstillsupportfreetrade,although it isrecognized that
16
freetradeisnotoptimal under imperfectcompetition.The'new'growththeoriesmostofthetimeassumefreeknowledgespilloversacrosscountries.Then,
tradepolicieswill notbebeneficial.However,theevolutionary growth theory
assumesthatacrucialpartofthegeneratedknowledge iscumulative,specific
andpathdependentandspilloversaretherefore localor national inscope.In
thatcasetradepolicymaybecomebeneficialbecausethegainsatstakecanbe
verylargeinsomecircumstances.Governmentpolicystillrequiresalotof informationaboutmanydifficulttomeasureeconomicvariablessuchastechnologicalopportunities,knowledgespillovers,andexternaleconomiesof scale.Still,
government policyseemsto bemoreworthwhilethaninthe caseofthe moderntradetheories becausethe gainsatstakearelarger.
17
1. INTRODUCTION
The survey of trade theories is part of the research project 'Policy and
Patterns of International Trade', executed by the Agricultural Economics Research Institute (LEI-DLO),financed bythe Institute's budget on Strategic Expertise Development (SEOprogramme).The objectives of this research project are
to:
1.
analyse international tradetheories,with the aimto answerthe question:
what determines international trade patterns and which role does government policy play in this?
2.
assessthe usability of general trade theories in explaining agricultural
trade;
3.
design aconcept for explaining world trade patterns in agricultural commodities.
The first activity in this research project, a review of literature, should
throw light uponthefactorsthat arecrucial inexplaining trade.A second activity ist o assessthe usability of thesetheories for researchon agricultural sectors.
We first describe the features of agricultural chains and markets (focusing on
allfactorsthat could beof importance) andthereafter confront the trade theorieswith these features and empirical data on the agricultural sector. This confrontation will result in depicting the factors of importance (and circumstances
under which they are of importance) in explaining agricultural trade. Then
these factors should become the building stones for adesign of a concept explaining world trade patterns in agricultural commodities. Of course, questions
of data-availability to operationalize the concept are addressed.
This report analyses international trade theories (objective 1of the project).An introduction to the trade theories ismade by classifying them accordingt o certain keyelements.Inthefollowing chapters,international trade theories are described more exhaustively, focusing on the explanation of international trade and the role of governments to play.
Why do countries trade with each other? What can be gained by trade
and how can trade patterns be explained? These questions have been addressed by many economists ever since Smith at the end of the eighteenth and
Ricardo in the early nineteenth century. Many different answers have been
given to the question what causestrade and still there ismuch controversy on
this issue.Differences between countries, for instance in natural factor endowments and factor prices, can be a motive for trade between t w o countries.
Countries trade in order to take advantage of these differences. This concept
of trade isbasedon (the theory of) comparative advantage. However, other more modern -theories state that countries may also trade because there are
inherent advantages inspecialization,arising from the existence of economies
19
of scale. Some other models in modern trade theories emphasize imperfect
competition, product differentiation and technology gaps (innovation) across
firms and countries asa major source of explanations for international trade.
Finally,the trade implications of the 'new' growth theories will also be taken
into account becausethesetheories shed light upon the dynamic evolution of
comparative advantage.
A General Classification of Theories of International Trade
Inthe economic literature, international trade theories can be classified
int w o major streams:the traditional and modern trade theories.Within both
schools, a further differentiation can be made, based on the major elements
the theory isfocused on. Inthe following, an outline of the theoretical principlesof each mainstream will begiven,plusthe thoughts of the most important
authors of each stream. International trade theories may be classified in the
following way:
1.
Traditional trade theories
- Mercantilism
- Classical theory
- Absolute advantage (Smith)
- Comparative advantage (Ricardo)
- Neo-classical theory (Heckscher-Ohlin-Samuelson)
- specific factors model (short term)
2.
Modern trade theories
- Economies of scale and imperfect competition
- External economies of scale
- Internal economies of scale and imperfect competition
- Monopolistic competition
- Oligopolistic competition
- Neo-technology trade theories
- Technology gap theory
3.
Trade Implications of the Growth Theories
- New Growth Theories
- Knowledge asside product: External economies of scale
- Knowledge as investment: A combination of external and internal
economies of scale with monopolistic competition
- Evolutionary Growth Theories
Theterm 'mercantilism'standsfor thetheory andsystemof political economy prevailing in Europe after the decline of feudalism (approximately 1,500
to 1,750).Thissystemwas based on national policies of accumulating gold bullion, establishing coloniesand amerchant marine,and developing industry and
mining to attain afavourable balance of trade. Mercantilism emphasized policies that encouraged exports of domestic products and discouraged imports.
20
Neo-mercantilism isa modern concept defined asatendency of a country to
accumulate large amounts of hard foreign exchange.
Traditional trade theories focus on differences among countries that are
the result of differences in technology (classicaltheory) or differences in relative factor endowments (neo-classicaltheory).One of the firsttheories of international trade isthe classicaltheory of absolute costadvantages. According to
Smith, the main representative of this school,trade only appears when there
are absolute cost differences between countries. David Ricardo showed the
shortcomings of thistheory because,even if one country can produce all goods
more efficiently than another country, trade ispossible and beneficial.A precondition isthat the relative efficiency gap isnot the samefor all goods. When
this isthe case,acountry hasacomparative advantage in agood that has the
highest efficiency gap.
The neo-classical theory, which dominated international trade theory for
a long time, elaborated these theories by including more production factors.
However, contrary t o the classicaltheories this theory assumed identical production techniques over countries. Furthermore, the standard neo-classical
Heckscher-Ohlin-Samuelson (H-O-S) model assumes constant returns to scale,
identical consumer preferences and perfect competition. These assumptions
imply that differences infactor endowments arethe only explanation of trade
in the H-O-Smodel.The larger the difference infactor endowments, the more
trade between countries and all this trade isinter-industry trade.
The specific-factor model isaspecialversion of the standard neo-classical
H-O-S model. Inthe standard model all production factors are mobile between
sectors,while in the specific-factor model one factor ismobile and one factor
isimmobile. We include this so-called short term version of the H-O-S model in
this review because the trade implication for production factor rewards are
totally different from the standard model.
Despite the theoretical dominance of the neo-classical model for a long
period the implications of this model were not supported by empirical studies.
The most influential study was done by Leontief (1953)who found that the US
(acapital abundant country) imports were more capital intensive than USexports. Furthermore, empirical studies showedthat mosttrade isbetween countries with the samefactor endowments andthat agreat part of trade between
industrial countries is intra-industry trade (Balassa 1967, Grubel and Lloyd
1975).
Thesecontradictionswithtraditional theory inducedeconomiststo search
for new trade theories. The 'new' trade theories elaborated the neo-classical
framework by replacing the most unrealistic assumptions of constant returns
to scale and perfect competition.A second stream,the so-called 'neo-technology theories', like the classicaltheories, stressedthe central role of technology
and proposed a more radical departure from the neo-classical framework.
The 'new' trade theories assumed increasing returnsto scale and this implies imperfect competition of scale unless economies of scale are assumed
total external to the individual firms. A first approach assumesthese so-called
'external economies of scale'; an industry still contains many small firms and
perfect competition remains. A second approach assumes internal economies
21
of scalewhich leadto imperfect competition. Within this approach t w o directions can be identified. The first direction concentrates on modelling economiesof scale and treat market imperfections assimple aspossible by assuming
monopolistic competition (Dixit and Norman 1980, Helpman and Krugman
1985).A second direction concentrates on imperfect competition and useseconomiesof scaleto causethese market imperfections.Themain market structure
they use is'Cournot' or 'Bertrand' oligopoly (Brander and Spencer 1985, Helpmanand Krugman 1989).Themaindifference betweenthese 'new' trade theories and the neo-technologytrade theories isthat the former assume identical
production technologies acrosscountries while the latter emphasize (endogenous) technological innovation andtechnology gapsacrossfirms and countries
asa major reason for international trade.
While all these trade theories are mainly static and focus on allocation
issues, an interesting development has taken place in the growth literature.
The 'new' growth theories build forward on the static 'new' trade models and
put them in adynamic context. The new trade theories provided the building
blocks suchasthe treatment of economies of scale and market imperfections.
Like the neo-technology theories the 'new' growth theories stressthe role of
technological change. By putting these elements in a dynamic context these
'new' growth theories deal therefore with the dynamic evolution of comparative advantage andthe consequences of trade in aworld of global technological competition (Romer 1990, Grossman and Helpman 1991b). This makes it
very interesting to include the trade implications of the growth theories in our
review of international trade theories.
Classification Criteria
1.
2.
3.
4.
5.
We usethe following criteria to survey the trade theories.
Assumptions: what are the main assumptions of a theory?
Central mechanism:what isthe central mechanism in atheory to explain
trade?
Implications:what arethe main implications of atheory (e.g.gains from
trade)?
Policy:what are the main effects of government policy in a theory?
Empirics:which kind of empirical testsaredone to test the theory and do
these tests support the theory?
Each criterion to survey trade theories consists of many elements or aspects. Next, the main key elements of each criteria to be distinguished are
shown. Most of the key elements are subdivided in a number of aspects which
need to be considered in reviewing trade theories.
22
Ad 1. Assumptions:
Supply:
Production factors
- number
- which? (land, labour, capital, human capital, knowledge)
- mobile across sectors
- immobile across sectors
- mobile across countries
- immobile across countries
- initial endowments differ between countries
- which factor can be accumulated (in caseof growth theories)?
Sectors/goods
- number
- homogeneous
- differentiated
Technology:
Economies of scale
- constant economies of scale
- increasing economies of scale
- external economies of scale
- internal economies of scale
Production technology
- identical between goods
- different between goods
- identical between countries
- different between countries
- no technological change
- technological change
- exogenous
- endogenous
- process innovations
- product innovations
- new varieties (horizontal product differentiation)
- consumer products
- producer products
- quality improvements (vertical product differentiation)
Knowledge spillovers
- knowledge spillovers do not exist
- knowledge spillovers do exist
23
national knowledge spillovers
international knowledge spillovers
knowledge spillovers across different kinds of goods
good specific knowledge spillovers
Demand:
Consumer preferences
- identical across income levels (i.e. homothetic)
- different between income levels
- identical between countries
- different between countries
Markets:
Good market
- perfect competition
- imperfect competition
- monopoly
- oligopoly
- monopolistic competition
Factor market
- perfect competition (markets clear)
- markets do not necessarily clear
Ad 2.
Main mechanism
differences in technology (technology gaps)
differences in factor endowments
differences in consumer preferences
economies of scale
imperfect competition
market segmentation and price discrimination
Extra for growth theories
Main growth mechanism
Ad 3.
Implications
Which country exports which goods?
- inter-industry trade
- intra-industry trade
Gains from trade
- specialization
- gains from exchange
- exploiting economies of scale
24
- more product variety
- pro-competitiveness effect
- higher rate of innovation
Do factor prices converge between countries?
Which production factors gain, which loose?
Extra for growth theories
Which factor can increase growth rate?
Welfare implications: Isthe growth rate optimal in the equilibrium?
- yes
- no
- which policy isneeded to obtain the optimal equilibrium?
Ad 4.
Policy
Arguments for government policy
- terms of trade
- infant industry argument
- externalities
- strategic trade argument
- political arguments
Which government policies can be implemented?
- trade policy (tariffs, quota)
- industrial policy (subsidies, taxes)
- innovation policies (subsidies)
- competition policy
Main effects of these policies on
- total welfare
- government budget, consumer, and producer surplus
- different sectors
- different factors of production
- neighbour countries
Ad 5.
Empirics
Isit possible to test the theory?
- no
- yes
- empirical studies support theory
- empirical studies contradict theory
Critical remarks
25
Elaboration
Inthe next chapters, each school of thought isdescribed by focusing on
the five specific issues mentioned above: assumptions, mechanisms, implications, government policy and empirical evidence of the theory. This review
should result inasummary of the main characteristicsof the theories,their way
of explaining international trade, implications of trade and the influence of
government intervention. Furthermore, the empirical evidence of the theories
are discussed.These issuesare summarized in a matrix.
The review of theories areto be usedasan input inthe following stages
of the project. Inthe second stage,the general trade theories are assessed on
their usability in explaining agricultural trade. Herewe take a broad definition
of agricultural products:these are not only agricultural products produced by
primary sectors, but all products produced in the agricultural chain, including,
for example, products produced by the food processing industries. In order t o
do thisevaluation, first the features, such asfor example production technology and market structure,ofthe food andagricultural chainsand markets have
to bedescribed.Thereafter, theseagricultural characteristicswill be confronted
with features of the trade theories reviewed,andthis results in an overview of
the major factors of importance in explaining agricultural trade.
We can illustrate this idea with the matrix 'Confronting characteristics of
product X with assumptions of the trade theories' presented below. The columns represent the theories that are studied in phase one and the rows depict
the characteristics of an (group of) agricultural product. When one characteristic of the agricultural product isembodied in acertain theory we put across in
the matrix. Inthisway we can identify which theory or theories explain a part
of the trade for acertain agricultural product.
Confronting characteristicsof product X with assumptionsof the trade theories
Theory 1
Characteristic 1
Characteristic 2
Theory n
X
X
Characteristic n
Theory2
X
X
X
X
X
X
X
The main objective of this study isto design aconcept to explain agricultural trade patterns. Most trade theories focus on one mechanism for trade
that isdependent on one or t w o assumptions. All the other assumptions are
treated assimpleaspossibleandassumedto be identical acrosscountries. However, in the real world countries differ in many aspects and many mechanisms
for trade work at the sametime (some mechanisms reinforce eachother, other
neutralise each other). In this study we try to design aframework or concept
for certain groups of agricultural productswith more or lesshomogenous characteristics (for each homogenous group adifferent framework). Such a frame26
work startswith the characteristics of the homogenous product group, suchas
homogenous or differentiated product, and incorporates all the trade mechanisms that are linked to these characteristics. This will be no framework w i t h
formulas and equations but aframework, containing the major factors of importance and circumstances under which these factors are of importance in
explaining agricultural trade.Suchaframework will beaqualitative model, in
which the importance of the factors distinguished are discussed in relation to
specific circumstances cq.features of agricultural product and markets.
27
2. TRADITIONAL TRADETHEORIES
2.1 Classical theories
Theclassicaltheories explaintrade bydifferences in production technologies between countries. First,we briefly discussthe absolute advantage theory
of Adam Smith and second we describe the comparative advantage theory
developed by David Ricardo in a more elaborated way.
2.1.1
Absolute advantage (Smith)
One of the first theories of international trade isthe classical theory of
absolute cost differences or absolute advantages. According to this theory
trade appears only when there are absolute cost differences between countries.We illustrate this principle with asimple example.Assume there are t w o
goods Xand Y,t w o countries Home (H)and Foreign (F),the only factor of production islabour, and the labour requirements to produce one good X and Y
are:
Labour requirements
X
Country H
Country F
20
10
Y
20
30
Country Fhas an absolute advantage in the production of good X and
country Hhasan absolute advantage in the production of good Y. According
to this theory the countries specialize in goods inwhich they have an absolute
advantage and consumers maximize their utility through international trade.
2.1.2
Comparative advantage (Ricardo)
David Ricardoshowedthe shortcomings of the theory of absolute advantage by demonstrating that trade is also possible and beneficial when one
country hasan absolute advantage in both goods,but when the efficiency gap
is not the same for both goods.
Basic assumptions
Production factors: one production factor, labour (domestically mobile
and internationally immobile)
Sector/goods: t w o homogeneous goods (X and Y)
29
Economies of scale:constant returns to scale
Technology: fixed labour requirements to produce each of both goods, these
requirements differ between goods and between countries. No technological
change.
Knowledge spillovers: no
Consumer preferences: identical across income and between countries
Market structure: perfect competition on both factor and good markets.
Main mechanism
Justasinthe absolute advantage theory,the main mechanism inthe classicaltheory of comparative advantage isthe difference in production technology between countries.
The technology of an economy issummarized by the productivity in labour in each industry and comparative advantage isthe result of international
differences in labour productivity. A country has acomparative advantage in
producing good Xwhen its ratio of labour requirements in good X t o that in
good Y islower than in another (foreign) country, i.e.the home country's relative labour productivity in good X ishigher than in the other country. We can
illustrate the principle of comparative advantage with the following example.
Labour requirements
X
Country H
Country F
5
10
Y
5
30
Country Hhasanabsolute advantage in both goods and according to the
theory of absolute advantage trade isnot possible. However, the relative efficiencydifferences between good XandYarenot equal between the countries.
In country Fit coststhe economy three times asmuch to produce one unit of
Y as it does t o produce one unit of X (pf=Px/ PY=1/3). In country Hthe labour
costof production are equal (ph= 1/1). These relative efficiency differences create possibilitiesof profitable exchange between the t w o countries. Ifthe world
price (p w) isequal to V2then it ispossibleto exchange t w o units of good X for
one unit of good Y.This world price isprofitable for country Hbecause it can
obtain 2 units of good X for one unit of good Y, while in production it can
obtain only one unit of X bygiving up one unit of Y.Therefore, despite country Hhaving an absolute advantage in the production of both goods, trade is
nevertheless profitable for this country; it isprofitable for country Hto specialize in commodity Yand import commodity X.Country F on the other hand can
obtain more units of Y for one unit of X by trading than by producing. It is
therefore profitable for country F to specialize in good X and import good Y.
Although country F does not have an absolute advantage inthe production of
either of the two commodities,shehasacomparative advantage inthe production of good X, inthe sensethat X can be produced relatively less expensively
than in country H. Therefore trade creates profitable possibilities for both
30
countries despite that one country ismore efficient in producing both goods.
The observation that trade depends only on comparative advantage and not
on absolute advantage isone of the major contributions of Ricardo.
Implications
Gains from trade
Inthe absence of trade,the relative pricesof both goods in each country
are determined bythe relative unit labour requirements.We can illustrate this
with figure 2.1 in which the production possibility curve (PPC)for country His
depicted (total amount of labour in country Hisequal to 50).Before trade the
autarky equilibrium price ratio istangent to the community indifference curve
(l 0 ) and the production possibility curve. Inthis casethe tangency is independent of the equilibrium point (A) andfully determined bythe slope of the PPC.
Therefore, only costconditions determine the relative domestic price level before trade.The pre-trade relative price (Px/PY)istherefore p f = 1/3 in country F
and ph=1in country H.The normal result of trade isthat the equilibrium world
price ( p j ends up somewhere between the pre-trade levels in the t w o countries (p f < pw<Ph)1).A possibleworld equilibrium price isincluded infigure 2.1.
Thisworld price impliesthat country Hcompletely specializes inthe production
of good Y (production point ispoint C)and consumes where the indifference
curve istangent to the world price (point B).Theconsumption point after trade
isassociatedwith ahigher utility level (!,>!(,),socountry Hgainsfrom trade (the
same istrue for country F,however this country specializes in good X).
Both countries derive gains from trade from this specialization because
by producing the good inwhich it hasahigher relative labour productivity and
trading this good for an other good inwhich it hasa relative lower productivity.This isamore efficient method than direct production of the good with a
lower productivity and it increasesthe consumption possibilities in both countries.
The total gains from trade (movement from A to C)can be divided into
gainsfrom exchange andgainsfrom specialization.Gainsfrom exchange occur
if one canobtain a higher utility level by simply changing one commodity for
another. Infigure 2.2this 'gainsfrom exchange effect' isdepicted bythe movement from Ato D:the production staysin point A while the price level changes
from the pretrade to the world price level.The movement from Dto Brepresent the gains from specialization; country Hspecializes in the production of
good Y (movement from A to C)while the price level staysthe same.
The gains from trade for a country are dependent on the world price
level. If the world price level isequal to the autarky level (p w= ph)then there
are no gains from trade for this country because the consumption (and the
associated indifference curve) stays in point A (figure 2.1). If the world price
1)
Note that at any price ratio outside this range,both countrieswould want to
specialize inthe production ofthesamecommodity,andthiscould not possibly
leadto anequilibrium situation.
31
declines (the relative price of good Y,the export product of the home country,
increases) the terms of trade of country H improves (the relative price of its
export product increases) and country Hobtains a higher indifference curve
(for example movement from A to B).Therefore the world price determines
how much a country will gain from trade and how the gains from trade are
divided over the countries.
In contrast to the autarky price level,which isonly determined by supply
conditions, the equilibrium world price level is determinated by supply and
demand conditions. The offer curves for both countries reflect the excess demand (imports) or excess supply (exports) for both commodities. These offer
curves for country Hand F are depicted in figure 2.3.The world price is determined where both offer curves intersect; point E (import or excessdemand for
good Y by country Hisequal to export or excesssupply of good X by country
F,and vice versa for product X). Becausethe world price is important for the
division of the gainsfrom trade over countries,we consider afew determinants
of the equilibrium price:
demand preferences
If the demand preferences in country H shift towards good X (indifference
curvesshift to the X-axis) then the offer curvefor country Hshifts from OAF h to
OA'F' h(figure 2.4).The world price level increases from pto p' (relative price of
goodX increases) andthe terms of trade for country Hdeteriorates. Ingeneral,
shifts in demand toward; the import commodity will tend to deteriorate the
terms of trade and reduce the gains from trade;
size of countries
If country H gets bigger (increase in labour) its PPCshifts outwards and the
offer curve shift from OAF h to OA'F' h .Therefore an increase in labour deteriorates the terms of trade. If country H gets even bigger and bigger then the
offer curve of country Fintersects the offer curve of country Hin the flat part
of the offer curve and the world price level becomes equal t o the pre-trade
level of the large country. The large country does not gain from trade in this
situation. Therefore, both countries will specialize completely when the world
price isbetweenthe costratiosof the t w o countries,this isonly possible if both
countries have the capability of producing a large enough quantity of one of
the goods to satisfy world demand;
production technology
Changes inthe production technology in general change the slope of the production possibility curve.A small increase in labour productivity from the import good does not change the terms of trade. An increase in the labour productivity of the export goods shiftsthe offer curve again from OAFht o OA'F' h
and deteriorates the terms of trade.
32
Figure2.2 Gains from exchangeand
specialization in theRicardian
trade model
Figure2.1 Ricardiantrade model
x
o
Figure2.3 Determination of the world Figure2.4 Demandandsupplyfactors
price in the Ricardian model
influencetheworldpriceinthe
Ricardian model
33
Policy
The main policy message from this theory isthat acountry should avoid
anything that may restrict trade with other countries.The governmental policiesshould be asnon-distortive aspossible asnonintervention in trade is considered to bethe optimal way to benefit from the gains of trade.These gains
are obvious, according to the model and are best served by free trade.
Only when a country can influence itsterms of trade or when there are
domestic distortions (externalities),trade policy canbe beneficial for a country.
These arguments are the same for all the traditional theories and these are
therefore treated in section 2.3.
2.2 Neo-classical theory
2.2.1
Heckscher-Ohlin-Samuelson model
The Heckscher-Ohlin-Samuelson theory provides important insights into
the relationship between commodity trade and factor endowments.
Basic assumptions:
Two production factors: labour and capital (both factors are domestically
mobile and internationally immobile). Initial factor endowments (apparent in
capital/labour ratios) differ between countries.
Economies of scale:constant returnst o scale
Technology: production functions aredifferent forthe two commodities (factor
intensities reflected incapital/labour ratios differ between goods) but identical
across countries. Notechnological change.
Knowledge spillovers: no
Consumer preferences: homothetic 1)and identical between countries.
Market structure: perfect competition.
Main mechanism
The central explanation for trade in the standard Heckscher-Ohlin-Samuelson (H-O-S) model isthe difference in factor endowments between countries. Taking into account all the assumptions, these differences in factor endowments leadto different factor pricesanddifferent pricesof goods between
countries. Because pricesof goods differ there's a reason for trade. Let us illustrate this with the following example. Assume that factor endowments differ
between countries and between goods.Thenthere isarelatively labour-abundant and arelativelycapital-abundant country and arelatively labour-intensive
and a capital intensive good. In autarky, the relative factor price of labour t o
capital will be lower in the labour-abundant country. The labour-intensive
good will therefore be relatively cheaper in this country. As in the Ricardian
theory when relative good prices differ, there isan incentive for trade.
1)
34
Similartastesbetween income levels.
We canelaborate this principle with the following illustration.Assume a
t w o goods (Xand Y) andt w o countries (Home and Foreign) model 1). Country
H is relatively capital abundant and Country Fis labour abundant. Good Y is
capital intensive and good X islabour intensive.The product possibility curves
of country Hand Fand the indifference curves are depicted in figure 2.5.
In the autarky situation, the relative price of good X to Y for country H
and Fisgiven bythe price lines Ph en Pf(price lines are determined by the tangency of product possibility curves and indifference curves).Given these relative prices,country Hproduces and consumesYha and Xhawhile country Fproduces and consumes Yfaand Xfa.Therefore inthe autarky situation the relative
price of good Xto Y ishigher in country H(P„/Pyfor H>P„/Pyfor F).According
to the principle of comparative advantage, country Hwill export good Y and
country Fwill export good X. Because the demand for good Y (X) increases in
country H (F)the relative price ratio of X to Y will decrease (increase). When
trade expands,eachcountry's exporting sector grows and the import-competing sector contracts. Factors of production move inthe same direction and result in income-distribution effects. Because the export sector uses relatively
more of the abundant factor, the relativefactor priceof thisfactor will increase
and therefore the relative price of the export good increases.This process will
continue until the relative prices in both countries are the same (world price is
P w in figure 2.5).
Inthe trade equilibrium Hproduces Yhpand consumesYhcand will therefore export Yhp-Yhcto country F.Country Fproduces Y fp and consumes Yfc and
will import Yfc-Y fp=Yhp-Y hc Forgood X it isthe other way around. It is import a n t t o know that trade enables each country t o reach a higher indifference
curve (I,in stead of l0).
Implications
Fromthe above illustration of the H-O-Stheorem several conclusions can
be draw:
1.
The Heckscher-Ohlin theorem
Given the assumptions of the model a country exports the good which
uses most intensive it's abundant factor of production. In our example,
the labour (capital) abundant country F(H) will export the labour (capital) intensive good X (Y) and will import the capital (labour) intensive
good Y (X).
2.
The factor-price-equalization theorem
Relative factor priceswill bethe same across countries in the trade equilibrium. Asin our example, in the trade equilibrium relative good prices
are the same in both countries. With the same production technology
and constant returns to scalethis isonly possible when factor prices are
identical.
1)
Thismodelwith 2goods,2factors,2countries isconsideredto bethe basicH-Omodel.
35
H
\ \
\
i
i
i
i
i
1
u-
—
1
T.
\\
\
1^
i
1
1
1 __l
T_
1
1
1
1
1
_1
\
Pw
H = production possibility curve of country H
F= production possibility curve of countryF
l 0 = indifference curve in autarky
I, = indifference curve with trade
Ph= Px/Pyfor country H, pre-trade relative price of X to Y
Pf = Px/Pyfor country F,pre-trade relative price of X to Y
P w = Px/Py in the trade equilibrium
Figuur 2.5 Relation between factor intensity and trade
3.
5.
36
The Stolper-Samuelson theorem
An increase inthe relative price of agood increases the factor reward of
the factor which isusedintensively and decreasesthe reward of the other
factor.
A combination of the Heckscher-Ohlintheorem and the Stolper-Samuelson theorem implies that the scarce production factor in a country will
loose from trade and the abundant production factor will gain from
trade.
The Rybczynski theorem
An exogenous increase inthe supply of one production factor leadsto an
increase in the production of the good that usesthis production factor
intensively and to adecrease of the other good.
6.
Trade will increase the welfare of both countries
In our example both countries reach a higher indifference curve in the
trade equilibrium.This iscaused bya more rational allocation of productive resourcesand lower relative pricesforthe import competing product.
Policy
The traditional theories explain trade asa result of differences between
countries. By using these differences between countries trade isbeneficial to
all countries.The lessbarriers to trade there are,the more beneficial trade will
be. Freetrade policy istherefore the besttrade policy under normal conditions
(arguments for trade policy are treated in section 2.3).
2.2.2
Specific factors model
The Heckscher-Ohlin model assumesthat all production factors can move
freely between sectors,which isclearlyalong run feature. Inthe short run capital isnot perfectly mobile because capital used in agriculture isquite different
from capital used in making cars. Inthe short run specific-factor model, there
are t w o factors of production, labour ismobile between sectors, but capital is
assumed to be immobile and therefore good-specific.
Another interpretation isthat the specificfactors model represents Ricardian technological differences. Now,the resemblance with the Ricardo model
isstriking.The only difference isthat Ricardo assumesafixed marginal product
to the mobilefactor labour whereasthisso-called Ricardo-Viner model assumes
diminishing marginal products to labour.
A third interpretation of the specific-factor model isthat this model explainstrade flows when there are really specific factors from their own nature.
Trade based on natural resources isan example of this interpretation.
Basic assumptions
Three production factors:labour (domestically mobile),two kindsof capital (good specific);allfactors are internationally immobile. Initial factor endowments differ between countries.
Sector goods: 2 homogeneous goods
Economies of scale:constant returns to scale.
Technology: production functions are different for the t w o commodities but
identical acrosscountries. No technological change.
Knowledge Spillovers: no.
Consumer preferences: homothetic and identical between countries.
Market structure: perfect competition on goods and factor markets.
Main mechanism
Incontrast to the Heckscher-Ohlin model,the international trade pattern
can not be predicted from initial factor endowments alone.The specialization
pattern isalso dependent on the nature of the production functions and the
allocation of capital between the t w o industries.
37
Implications
The Heckscher Ohlin theorem does not always hold. A relative labourabundant country will not alwaysexportthe labour intensive commodity.
The implications of the specific-factor model alsodiffer from those of the
Heckscher-Ohlin model on the field of income-distribution.Trade raises
the relative price of the good in which the country has a comparative
advantage (say good X). The mobile factor labour moves into sector X
and this raises the income of the specific factor in this sector. For the
good Ythe results arejust the opposite.Therefore,trade isbeneficial for
the specific factor that is necessary to produce the export good and it
reducesthe income of the specific factor used in the import good.
The marginal product of labour (the real return to labour) hasfallen in
the export sector X and risen in the import sector Y. The welfare of a
worker will depend on the consumption pattern. Ifthe consumer prefers
the export good Xwhich price hasincreased,the welfare of the consumer
declines,when the consumers prefers the import product Y the welfare
increases.Therefore the welfare implications from trade on labour are
ambiguous.
An increase inthe endowment of one of the specific factors reduces the
real income of both specific factors and increases the real income of labour the mobile factor (even if we keep the relative price ratio unchanged). This is in contrast with the Heckscher-Ohlin model in which
endowment changes do not influence factor prices when the relative
price ratio remains unchanged.
An increase in the amount of labour reduces the return to labour and
increase the return to both the specific factors.
An increase in one of the specificfactors increasesthe output of the industry that usesthis factor and reducesthe output of the other industry.
An increase inthe supplyof labour increases both outputs.Thefirst effect
is conform the Rybczynski theorem and the second effect is in contrast
with this theorem.
The equalization of commodity prices by international trade does not
equalize factor prices.
Policy
Again free trade policy isthe best for acountry's welfare.
2.3 Trade policy under perfect competition
In the traditional theory with perfect competition free trade isthe best
policy from a world point of view. However, for an individual country trade
policy can be beneficial under the following conditions. First, if a country can
influence the world price (i.e.a'large country' case),it can improve itsterms of
trade (terms of trade argument) at itstrading partners' expense.Second,trade
policy can be used to correct for domestic distortions. Third, trade policy can
also be desired for political reasons.
38
1.
2.
3.
Tradepolicieswhenacountrycan improveitstermsof trade
- Atariff raisesthedomesticprice levelwhichgeneratestariff revenues
attheexpenseofconsumersurpluses.Theneteffectcontainstwo elements.Onthe hand,there isa(consumer)distortion lossthrough the
wedge between supply anddemandandonthe other hand,thereis
a terms of trade gain because importers lower their price. For small
tariffs,the secondeffect (first order effect) dominatesthe first effect
(secondordereffect)andatariff istherefore beneficial.Forlarger tariffs,theoppositeistrue.Therefore,thereisanoptimaltariff foralarge
country (e.g.Corden,1971).
- An import quota isequivalentto atariff only if allthe rents are collected bythe domestic economy. Ifthis isnot the case,aquota (ora
Voluntary Export Restraint (VER)) isworse than atariff because the
rentsaretaken byforeigners (Bhagwati,1965).
- An export subsidyisnever beneficial becauseit worsenstheterms of
trade and itcreates(production) distortion effects.Theright policyis
an export tax which cancompensate the distortion losswith an improvement ofthetermsoftrade.Again,the secondeffect dominates
the first andasmallexporttax isbeneficial.
Domesticdistortion and tradepolicy
Marketfailuresordomesticdistortionscausethatconsumerandproducer
surplusdonotaccurately measuresocialcostsand benefits.Trade policy
canbe usefulto correctfor these distortions.Therearetwo maintypes
ofmarketdistortions.First,thereareimperfections infactor markets(differencesinwagesamong industries,unemployment,etc.).Second,there
areexternaleconomies:technologicalspilloversbetweenfirms, learning
effectsandnegativeexternalitiessuchaspollution.
Trade policy canshift productionto activitieswith positive externalities
andoutofactivitieswithnegativeexternalities.Thiscreatesanadditional
welfare gain inaddition to allthe effects mentioned inthe caseof the
termsof trade argument.Thismeansthat alltrade policiescanbecome
beneficialevenwhenthere isnotermsoftrade effect (smallcountryassumption). However, trade policy will always be a second best policy,
becauseaproductionsubsidycanachievethesameresultwithout generatingthe consumerdistortion.
Tradepolicy forpoliticalreasons:
Sometimestrade policy isusedto achievepoliticalgoals:
- National defence:you needcertain industriesbecausetheir output is
vital intime ofwar whenforeign suppliesare unavailable.
- Income-distribution: Sometimes trade policy is desired to achieve a
moredesirableincome-distributionortoavoidundesirableincome-distribution effects of trade.Asinthe casewith distortions,trade policy
isonly asecondbest policy whereas income-distribution policy isthe
first bestpolicy.
39
2.4
Empirics
Testof the Ricardian theory of comparative advantage
There areanumber of ways inwhichthe Ricardian model makes misleading preconditions (for example:that an extreme degree of specialization will
occur, or that, ignoring income-distributional effects of international trade
within countries, countries asawhole always gain from trade) and, obviously,
the Ricardian one-factor model isfar too simple to be a complete analysis of
either the causes or the effects of international trade. But according to Krugman and Obstfeld (1994:29)the basic prediction of the model -that countries
shouldtend to exportthose goods inwhichtheir productivity isrelatively high has been strongly confirmed by anumber of studies (for evidence,the authors
refer to studies by MacDougall, 1951/52, MacDougall et al., 1962,and Balassa,
1963).Therefore, Krugman and Obstfeld conclude that, alsobecauseof itsrelative simplicity,afocus on relative labour productivities canbeavery useful tool
inthinking about international trade. Learner (1994),however, islessconvinced
by the evidence from these empirical studies.To his opinion, first, 'the Ricardian model isnot sensibly interpreted literally' in the empirical work and,second, 'the studies are done without referring adequately to the range of alternative hypotheses that might be considered.' (Learner, 1994:71-72). Therefore,
Learner concludes that what we may have learned from this empirical work on
the classical comparative costtheory is 'not too much' (1994:71).
Testing the Heckscher-Ohlin theorem
The H-0 theorem isathree-way relationship of factor abundance, factor
intensity and trade. A proper test must therefore include all three of them.
Few, if any investigators have managed this and therefore what we have isa
large body of evidence that issuggestive buthardly conclusive asto the empirically validation of the H-0 theorem (Dearpdorff, 1984). Although the study
does not include a measuring of factor endowments, the work of Leontief
(1953) isconsidered asafirst, comprehensive empirical application of the H-0
theorem. He used an input-output table for the USeconomy to measure the
capital and labour embodied in USexports and import substitutes. He found
that the ratio of capital to labour embodied in USexports was smaller than
that embodied in import substitutes, while USAwas supposed to be acapitalabundant and labour-scarce country.Thedebate that followed from this 'paradox' concentrated mainly on the method Leontief applied. Several authors
havereapplied Leontief'sbasicmethodology andthe resultshavetypically reaffirmed the paradox inthe earlyyears but found that it may have disappeared
by 1970(see Dear^dorff, 1984:480-485,alsofor references).Asabyproduct of
Leontief's results and the debate that followed, the H-O theorem has been
extended to allow for additional factors beyond just capital and labour to explain trade, or make a clearer distinction between skilled, unskilled labour,
human and physical capital asfactors determining international trade flows.
An important steptowards resolving the 'Leontief paradox' hasnot been
as much the acknowledgement of additional factors of production that may
explain Leontief's results, but the notion by Learner (1980)that the USin Leon40
tiefs data wasanet exporter of both capital and labour.A trade surplus makes
it possiblefor acountry to beanet exporter even of the servicesof factors with
which it isrelatively poorly endowed. Learner showsthat when this happens,
the ratios of factors embodied in exports and imports need bear no particular
relationship to relative factor endowments. Then acapital abundant country
need not embody a higher ratio of capital to labour in its exports than its imports and Leontief's results are not paradoxical at all. Instead Learner shows
that avalid test of the unbalanced trade must be stated in terms of the factor
ratios embodied in production versusconsumption, rather than exports versus
imports.Taking this approach to Leontief's data hefinds that the USwas a net
exporter of both labour and capital services and that the capital-labour ratio
embodied in production was indeed greater than that embodied in consumption (Learner illustrates his analysis with a three-factor numerical example,
which is corrected later by Heravi, 1986, see Learner, 1994). The presumed
abundance of UScapital relative to labour issupported after all and the paradox disappears.
More recently Bowen et al. (1987) have attempted to test the H-O-model
using data for 27 countries and 12factors of production.Their study is based
on the idea that trading goods are actually an indirect way of trading factors
of production. According to the H-O-theory, by calculating factors of production embodied in acountry's exports and imports, acountry isexpected to be
a net exporter of the factors of production with which it isrelatively abundant U
endowed, and a net importer of those with which it is relatively poorly endowed. Bowen et al.calculatedthe ratio of eachcountry's endowment of each
factor to the world supply. They then compared these ratios with each country'sshare of world income. Ifthe factor proportions theory were right, acountrywould alwaysexport factorsfor whichthe factor shareexceededthe income
share and import factors for which it was less. However, the outcome of one
of the keytests in Bowen et al. isthat for two-thirds of the factors of production, trade ran inthe predicted direction lessthan 70 percent of the time. This
results confirm the Leontief paradox on global scale:trade does not run in the
direction the Heckscher-Ohlin theorem predicts.
Three empirical observations that contradict the implications of the HeckscherOhlin theory
1.
The first empirical observation is that major trade flows occur among
countries with practically the samefactor proportions. However, according the H-O-S theory the volume of trade is expected to be larger depending on the differences between countries.
2.
Grubel and Lloyd (1975) discovered that intra-industry trade is empirically
significant and its proportion of total world trade has grown over time.
However, the H-O-Smodel canonly explain inter-industry trade and not
intra-industry trade.
3.
Income-distribution effects of trade aresmall.However, the H-O-S model
predicted that there are large income redistribution effects associated
with trade.
41
Testsof the specific factors model
Testsof this model do not differ muchfrom tests of the Ricardo model if
one assumesthe specific factors to be fixed over time (again the relation between labour productivity andtrade iscentral).There have been notestsof the
specific factors model, however, astudy to the relation between international
events and the allocation of new investments could be interesting (Learner,
1994).
Conclusion
Sofar the resultsof empirical tests of the pure H-O-theory have shown t o
be rather disappointing and there are no tests of the specific-factor model at
all.At the sametime,empirical evidence supportsthe Ricardian model's prediction that countries will export goods inwhich their labour isespecially productive. Although the limitations of the Ricardian model are widely acknowledged, Krugman and Obstfeld (1994) conclude from these tests that the trade
pattern islargely driven by international differences intechnology rather than
resources. However, it isstill important to askwhat factors are embodied in a
country's exports and imports, astrade affects income-distribution. The H-Omodel and the specific-factor model,therefore, retain but to a more limited
use,asaway of predicting the income-distribution effects of trade and trade
policy.
Woods (1994) counteractsthe proposition that H-0 theorem is inaccurate
to explainglobaltrade patterns.Hearguesthatthe traditional Heckscher-Ohlin
theory probably provides a much better description of reality than is usually
supposed byempirical testsof the H-0 theory, suggesting that the H-O theory
performs badly.Woods statesthat H-0 models are likelyto work better if capital -defined asfinance - isexcluded than if it is included in the usual, wrong,
way. Capital is usually considered to be an immobile production factor, like
landand labour. However,financial capital isinternationally mobile and,therefore, does not generally influence the pattern of trade. Skill availability matters, according to Wood, in explaining North-South trade in manufactures, not
capital. He suggests that a skill-only H-0 model can explain the commodity
composition of North-South trade in manufactures rather well. His hypothesis
isthat the North, because of its larger supply of skilled (relative t o unskilled)
labour, exports skilled-intensive manufactures from the South, and imports
(unskilled) labour-intensive manufactures.Woodsfindsevidencefor hishypothesisthat North-South trade isbased on differences in the abundance of skills.
His conclusion is that when the H-0 theory is correctly specified - excluding
capital - it provides 'an accurate and illuminating description of a large part of
the global pattern of trade' (Woods, 1994:20).
42
3. MODERNTRADETHEORIES
3.1 Economies of scale and imperfect competition
Trade need not be the result of comparative advantage. Instead, it can
result from increasing returns or economies of scale.An industry is characterized by economies of scale when doubling of inputs more than doubles the
industry's production. Economiesof scaleprovide an incentivefor international
trade byconcentrating production on a limited number of goods.This enables
a country to produce these goods more efficiently than if it tried to produce
everything for itself. These specialized economies then trade with each other
t o be able t o consume the full range of goods.
Economies of scaleand market structure
However, economiesof scaleasadeterminant for international trade was
known for a long time.The problem wasthat economies of scale are inconsistent with the perfect competition standard Neo-classical model because they
imply adifferent market structure, namely imperfect competition.
Inthe first and oldest departure from the standard model economies of
scalewere assumedto beexternal to the firm. These so-called 'external economiesof scale' occur when the cost per unit depends on the size of the industry
but not necessarily on the size of the firm. An industry where economies of
scale are wholly external (where there are no advantages t o large firms) will
typically consists of many small firms and be perfectly competitive. This is the
Marshallian approach to increasing returns (see Krugman, 1990:65-74). The
early literature on the Marshallian approach (the early postwar period), however, seemed discouraging in that even with the simplest assumptions it
seemedt o lead to welter of multiple equilibria. Only since the work of Ethier
(1979, 1982) it has become clear that under certain circumstances it is possible
to bring order to this complexity.The new version of the Marshallian approach
distinguishes from the older approach by the way that it works from resource
allocationto trade.Byassumingthat atrading world reproducesthe aggregate
outcomes of a hypothetical perfectly integrated economy it follows that both
factor proportions and scale economies contribute to international trade and
both are sources of gains from trade.
Inthe seconddeparture economies of scaleatthe firm level areassumed.
Internal economies of scale give large firms a cost advantage over small and
leadto animperfectly competitive market structure.Within thisapproach there
aret w o directions.The first direction concentrates on modelling economies of
scaleandtreats market imperfections assimpleaspossible byassuming monopolistic competition (Dixit and Norman, 1980; Helpman and Krugman, 1985). A
second direction concentrates on imperfect competition and uses economies
43
of scaleto cause these market imperfections. The main market structure they
use is'Cournot' or 'Bertrand' oligopoly (Brander and Spencer, 1985; Helpman
and Krugman, 1989).
3.1.1
External economies of scale
Basic assumptions
Production factors: labour (and capital):this factor isdomestically mobile
and internationally immobile. Initial endowments of countries may be identical.
Sector/goods: 2 homogeneous goods
Economies of scale: one good produced with constant returns to scale, the
other good with external economies of scale which are country specific.
Technology: different acrossgoods and identical production functions between
countries, no technological change;
Knowledge spillovers: no
Consumer preferences: identical across incomes and between countries
Market structure: perfect competition in good and factor markets
Main mechanism
External economies are economies of scale that occur at the level of the
industry instead of the firm. To keep it assimple aspossible, imagine a world
with t w o countries,two goods and one factor of production (labour). Good A
isproduced with external economies of scaleandgood Bisproduced with constant returns to scale at the firm and industry level.Assume further that both
countries have the same production techniques and that both countries produce both goods.The fact that both countries produce good Bimplies equal
wages. But this meansthat whichever country had the larger A industry would
have lower costs in that industry. Trade will cause the relative size of that industry to increase still further; economies of scale reinforce the relative cost
advantage of this industry, now trade expands. This process isgoing on until
at least one country specializes. So, increasing returns leads to specialization
and trade.
Implications
1.
Trade ispossible between identical countries (same factor endowments,
technologies and tastes).
2.
With external economies of scale the direction of specialization is often
unknown. Thistheory gives an important role to history and accident in
determining the pattern of international trade. When external economies are important, a country starting with a large industry may retain
that advantage even if another country could potentially produce the
same goods more cheaply.
3.
The implications for the wage level in both countries are ambiguous. If
inthe trade equilibrium both countries produce the good with constant
returns to scale, the wages in both countries will be equal and higher
than before trade. If one country specializes in the good produced with
44
4.
external economies of scaleandfor which demand ishigh, than the relative price of this product will increase,and wages of this country will rise
relative to the other country. Real wages in the other country may still
rise (through gains from specialization and gains from exchange), however, if they consume alot of the more expensive goods,their real wages
may even decline.
An implication of these real wage movements isthat the division of welfare between countries isvery unequal and very dependent on the specialization pattern. Countries can even lose from trade.
Policy
A country may specialize in goodsthat are lessdesirable in terms of welfare.Trade or industrial policy can reversethe initial cost advantage and therefore reverse the specialization pattern. Inthis casetrade and industrial policies
can be beneficial.
Trade policy caneven be desirable from aworld point of view. This happenswhen asmall country hasan initial advantage in the sector with external
economies of scale and demand for this product is large.The sector with the
external economies of scalewill besituated inthe small country, but the country istoo small to achieve the optimal scale from a world point of view and
production costswill therefore be higher than necessarily.
Empirics
Identification of external economies isvery difficult. Asa matter of fact,
externalities are inherently hard to measures asby definition they do not leave
anytrace in market transactions.Stillthere are some examples which give evidence of the importance of external economies.Suchexamples arethe concentration of semiconductor manufacturers in California's 'SiliconValley', and the
concentration of financial and banking firms in London and New York.
Someof the most important external economies probably arisesfrom the
accumulation and spillovers of knowledge: production costsof individual firms
fall asthe industry asawhole accumulates experience. Dynamic scale economies, like external economies at apoint in time, potentially justify protectionism. The argument for temporary protection of industries to enable them to
gain experience isknown asthe infant industry argument.
3.1.2
3.1.2.1
Internal economies of scale and imperfect competition
Monopolistic competition
With monopolistic competition, an industry contains asufficiently large
number of 'similar' firms producing differentiated products and profits are
competed away inthe equilibrium. Products canbe differentiated in two manners.Onthe one side,each consumer has ataste for many different varieties,
this isthe so-called 'love of variety approach' (Dixit and Stiglitz 1977, Spence
1976).Product differentiation inthis case isonly producing adifferent variety.
Onthe other side,eachconsumer has its own preferred mix of attributes, this
45
isthe so-called 'diversity of tastes approach' (Lancaster 1980). Inthis case product differentiation shows up by products that consist of different attributes.
Although the latter approach hasthe advantage of greater realism,the former
approach ismostly used because it can be modelled more easily.
With product differentiation eachproduct isunique,therefore somefixed
costs are assumed to produce this product (internal economies of scale). Because product differentiation is possible, firms have some monopoly power
(downward sloping demand curve).However, additional firms canenter a profitable industry until monopoly profits are competed away (profits are equal to
fixed costs).The number of firms in the equilibrium are therefore dependent
on the size of the market (a large market will support a larger number of
firms), the amount of fixed costs (larger fixed costs will decrease the number
of firms) and the degree of product differentiation (if differentiation is valued
very much by consumers, more products will be there in the equilibrium).
The number of firms in a monopolistically competitive market, and the
pricesthey charge, are determined byt w o relationships. On the one side, the
more firms there are,the more intensively they compete and hence the lower
the industry price is.Thisrelationship ispresented by PP (seefigure 3.1).On the
other side,the more firms there are, the lesseach firm sells (given the size of
the market) and therefore the higher its average cost.This relationship is presented by CC.If price exceeds average cost,the industry will be making profit
and additional firms will be entering the industry; if the price islessthan average cost,the industry will be incurring losses and firms will leave the industry.
The long-run equilibrium price and number of firms occurs when price equals
average cost, at the intersection of PPandCC.
Basic assumptions
Production factors: labour (and capital):this factor isdomestically mobile
and internationally immobile.Initial factor endowments of countriesare identical in backbone model.
Sector/goods: n differentiated products.
Economies of scale: increasing returns to scale at firm level.
Technology: identical across goods and countries, no technological change.
Knowledge spillovers: no.
Market structure: monopolistic competition in good markets and perfect competition in factor markets.
Consumer preferences: identical across incomes and between countries.
Main mechanism
To concentrate on the mechanism caused by product differentiation and
internal economies of scale we assume identical factor endowments between
countries.We take the Dixit-Stiglitz assumption that eachconsumer hasataste
for many different varieties of a product. Product differentiation of product A
causes each country to produce a different set of varieties of product A. Because each country isassumedto demand all varieties (consumers have a 'love
of variety') each country exports each of itsown varieties and imports each of
46
PriceP
PriceP
<
c
P,
AC 2 , P2
!
\v
E
>/^
'
p;
-
AC,
1
n,
n2
Number of firms, n
Figure3.1 Equilibrium in a monopolisticcompetitive market
n3
Number of firms, n
Figure3.2 Effectsof alarger market
the varieties of the other country 1)So,there will be 'intra-industry' trade.This
intra-industry trade isessentially caused by scale economies. If there were no
scaleeconomies eachcountry would be ableto produce all varieties of product
A itself. Butthe existence of scaleeconomies make surethat each variety must
be concentrated in one country. So,it isthe combination of scale effects and
the specification of the product differentiation processwhich causes intra-industry trade.
Besidesthe increased number of varieties available to consumers, asecond possible gain from trade could be an increased scale of production which
together with internal increasing returns to scale implies lower average costs.
Whether or not the scale of production will increase isdependent on the specificformulations of, for example,the product differentiation function (Helpman, 1981). With Dixit-Stiglitz preferences (constant elasticity of demand),
trade offers greater variety but not greater scale.With Lancaster preferences,
trade leadsoften to amore elastic demand: in this case,a larger market leads
t o greater diversity and lower average costs (Helpman, 1981).
Thecasewhenthe scaleof production increasesafter trade is represented
by a downward shift in CCin figure 3.2.The result isasimultaneous increase
in the number of firms (and hence inthe variety of goods available) and a fall
1)
Inthemonopolisticcompetitionmodeleachfirm makesauniqueproductbecauseit isnot profitableto enterthe marketwith aproductthat alreadyexists.
47
in the price of each good. The integrated market supports more firms, each
producing at a larger scale and selling at a lower price, than either national
market did on its o w n .
Implications:
1.
Even with identical resource endowments, trade ispossible.All trade in
this caseisintra-industry trade andthe precise pattern of trade is indeterminate (which country exports which variety).
2.
Gains from trade appear in the form of:
- Increased product variety.
- Potential higher scale of production which results in lower average
costs.
3.
In a model where both internal economies of scale and relative factor
endowments are present, intra-industry trade iscaused by monopolistic
competition in combination with scale effects and inter-industry trade is
causedbythetraditional H-O-Smodel.Intra-industrytrade ismore important between similar countries and inter-industry trade becomes more
important the greater the difference in relative factor endowments.
4.
Intra-industry trade does not generate the same strong effects on income-distribution as inter-industry trade. Inthe models describing comparative advantage, trade had all its effects through changes in relative
prices,which inturn havestrong income-distributional effects (scarcefactors losefrom trade). However, when 1)countries aresimilar intheir relative factor supplies so that there is not much inter-industry trade and
when 2) scale economies and product differentiation are important, so
t h a t possible gains from larger scale and an increased choice are large,
intra-industry trade isthe dominant source of gains from trade. In these
circumstances,the income-distribution effects of trade will be small and
there will besubstantial extra gainsfrom intra-industry trade.The result
may well bethat despite the effects of trade on income-distribution, everyone (the owner of the abundant and of the scarcefactor) gains from
trade. Because intra-industry trade will be dominant between countries
at a similar level of economic development, trade without serious income-distributional effects ismost likelyto happen in manufactures trade
between advanced industrial countries (for examplethe Common Market
of the EU).
5.
It is possible to treat multinationals and trade in technology in these
models. Investment in knowledge is hard to model except asa kind of
fixed cost.Inthe monopolistic competition models it ispossiblethat firms
in one country develop a product and sell the knowledge about this
product abroad. Inthe foreign country a new monopolistic competitor
entersonthe basisof this knowledge (Feenstra andJudd, 1982).Forsome
knowledge this is not possible and knowledge can be only transferred
within a firm. This results in the existence of multinational enterprises
(MNE) (Helpman, 1985).
6.
Transport costs:when there aretransport coststhe sizeof the market has
an important influence on trade patterns. Ceteris paribus, countries ex48
port productsfor which they havea large home market (Krugman, 1980;
Helpman en Krugman, 1985).
3.1.2.2
Oligopolistic competition
Basic assumptions
Production factors: no assumptions about production factors in most
models.
Sector/goods: t w o or afew products.
Economies of scale:increasing returnsto scaleatfirm level (but not in Brander/Krugman's (1983) model).
Technology: identical across goods and countries, no technological change.
Knowledge spillovers: no.
Market structure: oligopolistic (usually Cournot) competition in good market
and perfect competition infactor markets. In Brander/Krugman model a Cournot duopoly isassumed.
Consumer preferences: identical across incomes and between countries.
Main mechanism
In this approach it isnormally assumed that internal economies of scale
lead t o an oligopolistic market structure. In such markets the behaviour of a
firm has influence on the behaviour of other firms. Inthe international trade
theory most of the time it isassumed that firms behave in a Cournot fashion:
imperfectly competitive firms take each others' output asgiven.
Oneof the main elements inthistheory isthe relation betweentrade and
market power. If firms compete inthe Cournot fashion its price will be higher
than marginal costs by a mark-up that depends on the perceived elasticity of
demand per firm. When trade occurseachfirm becomes part of a larger more
competitive market and it will perceive a higher elasticity of demand, leading
itt o expand output andto setting alower price.Therefore,trade or 'potential'
trade reduces monopolistic distortions. This effect iscalled the 'pro-competitive' effect.
Dixit and Norman (1980) show that there are t w o effects of opening
trade in a Cournot market with increasing returns to scale. On the one hand
the number of firms in each market will be reduced because a lower mark-up
(higher perceived elasticity becausethe number of competitors increases when
trade isopened) implies that somefirms are not able to cover their fixed costs.
This isthe so-called 'exit of redundant firms' effect which leadsto an increase
in the productive efficiency. On the other side increases the total number of
firms inboth marketstogether sothat competition increasesandthe monopoly
distortion decreases.This isagain the 'pro-competitive' effect.
Price discrimination asa new determinant of trade
If market segmentation and price discrimination are possible,there can
betrade evenwithout economiesof scaleandcomparative advantage (Brander
1981, Brander and Krugman 1983).Assume t w o identical monopolists in t w o
identical countries. They produce with constant marginal costs and take the
49
production level of the other firm aswell asthe distribution of this production
between markets asgiven.Trade isnot costless;there aretransportation costs.
Ifthe difference between marginal costs and the foreign price level is greater
than the transportation costs it is beneficial for the domestic monopolist t o
export its good.
The transportation costsimply that the market share in the foreign market will be lower than in the domestic market. This means that the domestic
firm will face ahigher elasticity of demand in the foreign market and its price
in the foreign market will be lower than in the home market. So,one would
expect that, if afirm would like to expand sales,additional salesat the domestic market are more profitable than additional exports. However, the marginal
price decrease of one extra unit sold has asmaller negative effect in the foreign market, becausethe number of products sold bythe domestic monopolist
in that market islower. When afirm charges a lower price for exported goods
than it does for the same goods sold domestically, this behaviour is called
'dumping'. The reason why afirm may choose to dump isthe difference in the
responsiveness of salesto prices inthe foreign and domestic markets. Because
both firms are identical. Brander and Krugman talk intheir model about 'reciprocaldumping'.Theresult istwo-way trade inthe sameproduct (cross-hauling).
This model gives,therefore, another explanation of intra-industry trade than
the interaction of product differentiation and economies of scale argument.
Inthis model, intra-industry trade ispossible in identical products and the driving force isprice discrimination.
Implications
1.
Gains from trade appear in the form of the pro-competitive effect, the
exit of firms which are unable to cover their fixed costs,and lower averagecostsifthe production scaleof afirm increases(ifthere are increasing
returns to scale).Welfare implications are ambiguous. Ascan be identified w i t h the Brander/Krugman model, there are t w o effects. First, the
pro-competitive effect, imports from the foreign monopolist imply that
the total supply in the domestic market increases which lowers the domestic price level.Second,some imports replace domestic deliverancies
which leadsto welfare lossthrough transportation costs.The total effect
isambiguous. Iftransportation costsare low,the first effect islarger than
the second and trade isbeneficial.
2
The most important implications of this Cournot-approach are on the
field of trade policy (seesection 3.1.3).
3.1.3
Trade policy and imperfect competition
The assumption of constant returns to scaleand perfect markets justified
the domination of free trade policy for almost a century. These 'new' trade
theories question this free trade policy by introducing economies of scale and
imperfect markets.They found that an active trade policy can be beneficial in
the certain circumstances. However, arguments against protection remain valid
50
in other circumstances.An overview of the arguments for and against protection under imperfect competition isgiven below.
Arguments for protection
1.
In case of foreign market power, the terms of trade may improve.
Under avariety of circumstances, aforeign firm with market power will
absorb apart of the tariff by increasing itsprice by lessthan the full tariff
(it reduces its mark-up).Thisargument issimilar to the optimal tariff argument inthetraditional tradetheories (seesection2.3). However, in this
case the 'large country' assumption is not necessarily, the only requirements are afirm with foreign market power with price discrimination.
2.
Rent shifting.
If in an industry excessive profits are earned, such an industry should be
desired. Under certain conditions, the use of export subsidies can also
shift profits from foreign to domestic firms.This isthe so-called 'strategic'
trade policy (Brander and Spencer, 1983, 1985; Dixit and Kyle, 1985) 1).
An argument thatjustifies export subsidies istotally incontradiction with
the traditional theories and extremely usefulfor lobbyists.Therefore, this
argument gets a lot of attention and we will treat it in a more elaborate
way.
The argument islinked to the Cournot model of oligopolistic behaviour
of firms. Cournot's idea isthat equilibrium (or stability) will occur when
eachfirm ismaximizing its profits through the choice of its own level of
production, given the output level of its rival. If firms are on equal footing atthe market place,there isno reasonto expect that one firm should
be intimidated bythe threat of the other to increase output in trying to
induce contraction of the rival.So,if afirm expands itsoutput, the other
firm might well match the increase, inducing a price war. But now suppose, a firm discovers to produce more efficiently and isable to reduce
its costs of producing additional output. Than the firm will expand its
output and the rival will contract because it has no reason to believe the
increased competitiveness is only temporary. The new equilibrium will
involve a higher market share and output for the firm with lower costs
and a smaller market share and output for the other firm.
Brander and Spencer makethe point that for afirm an export subsidy (or
production subsidy) hasthe sameeffect aslowered costs.A subsidy makes
it in the firm's interest to expand output, it is a credible threat to the
rival's position, who can best respond by contracting its output, and in
effect the subsidy inducesthe subsidized firm to stake out a larger share
of the international market. The firm's profits will increase, but it isalso
beneficial for the nation asthe benefits exceeds the cost to taxpayers:
1)
Strategic becauseit isnot profitableviewed inisolation but it alterscompetition
inthe future. Forexample afirm invests inexcesscapacitythat it doesnot use
butwhich deters potential competitors.
51
first, there isthe cost saving effect of the subsidy, which isequal t o the
transfer from government budget, and second,there isthe contraction
by the rival which in itself raises the domestic firm's profits by an additional amount.
Strategic export subsidies which improve the welfare of a country are
very dependent on the specific assumptions of the model.Various trade
theorists show that the argument is not true: 1) if the Cournot assumption isreplaced bythe Bertrand assumption (Eaton and Grossman, 1986),
2) if retaliation of foreign government occurs (Dixit and Kyle, 1985),3)as
other sectorswill beadversely affected ('general equilibrium'-argument),
so government needs very detailed information (Dixit and Grossman,
1986), 4) when entry eliminates supernormal profits and then subsidies
will only worsen terms of trade in the long run (Horstmann and Markusen, 1986).Therefore, the strategic trade argument seems to make a
case for protection, however, so much information is needed that it is
unlikely that this amount of information isavailable bythe government.
Reducing marginal cost.
Krugman (1984) showed another beneficial trade policy in a variant of
the Brander-Krugman model with declining marginal cost insteadof constant marginal cost.Assume acountry which protects itsdomestic market
w i t h atariff. The immediate result isthat the foreign firm sells less and
the domestic firm sells more in the domestic market. The marginal cost
of the domestic firm declines and the marginal cost of the foreign firm
increases. The indirect effect isthat the domestic firm sells more in the
unprotected foreign market. Krugman calls this 'import protection as
export promotion'. If there is no retaliation, this protection policy could
beworthwhile. Declining marginal costsare not very common, but Krugman showsthat this analysis isalso usable when time ismade explicit and
one considers dynamic economies of scale,arising for example from R&D
or learning effects (see also 'new' growth theories).
Prompting entry: atariff can lower the domestic price.
Venables (1985) shows that a small tariff can raise welfare by inducing
entry. He assumes free entry in the Brander/Krugman model in which
there are constant marginal costs with fixed costs.A tariff raises profitability in the home market and reduces profitability in the foreign market, which leadsto entry inthe home market (reduces degree of market
power) andexit inthe foreign market.Therearetwo effects onthe home
market. On the one hand,the increased competition in the home leads
to a lower price and on the other hand the tariff raisesthis price. Fora
small tariff, the first effect dominates the second effect so that the domestic price level actually declines.Welfare increases because consumers
gain (lower price level) and there are additional tariff revenues.
External effects.
The 'New' trade theories have broadened the basis of this argument by
emphasizing the positive external effects of R&D activities in connection
w i t h imperfect markets (Brander and Spencer, 1983). Positive external
effects appear especially by R&D activities because these have positive
52
spillover effects on other industries and because individual firms cannot
internalise all the benefits of its own R&D activities. The investments in
knowledge development arecharacterised byaconstant cost component
(These markets aretherefore imperfect which iscaused by economies of
scale). When the scale of production increases the cost of these investments per unit of product decline (Krugman, 1987).Investments in knowledge development are the source of positive spillover effects and the
contribution of the 'New' trade theories isthat these effects can now be
explicitly modelled.Another contribution isthat industries with external
effects can be identified, because these theories suggests that positive
external effects especially can befound in industries where R&Dcost are
a substantial part of total cost.
Arguments against protection
1.
Market power.
Normally, protection reduces competition in the domestic market and
increases the market power of domestic firms.
- domestic monopoly faces competitive foreign suppliers (Bhagwati,
1965).
Indeed, protection raisesthe market power of domesticfirms and leads
to ahigher domestic price level.Interesting isalsothat the equivalence
of import tariffs and quota's disappears in this case.In case of a tariff,
a monopolist cannot raise its price above the tariff-inclusive import
price otherwise it will loose the total domestic market. In case of a
quota, it can use its domestic market power because the amount of
imports are limited.Therefore, a quota will lead to a higher domestic
price and lower domestic output than a tariff.
- domestic monopoly facesforeign monopolist: non-cooperative behaviour (Krishna, 1984).
Incaseof aquota and Bertrandcompetition,the domesticfirm has t w o
options. Onthe one hand, it can be aggressive and charge a low price
that limits imports to lessthan the quota or it can be timid and charging a high price behind the protection wall of the quota. The higher
the price level of the foreign firm the more aggressive the domestic
firm will be.Normally, the profits of the domestic firms rise and those
of the foreign firm decline, although aquota can also raise profits of
both firms.
- domestic monopoly facesforeign monopolist: collusion or cooperative
behaviour (Rotemberg and Saloner, 1984).
Collusion can reversethe intuitive implications of the Bhagwati-model,
because in this case protection can increase competition and lead t o
lower prices than in the case of free trade because the penalty for
cheating on acollusive agreement isreduced.
2.
Inefficient Entry.
Normally, protection raisesthe sale of domestic firms which, in the case
of increasing returns t o scale, lowers average costs and increases efficiency. However, what happens if entry is possible?
53
- Excessiveentry: increasing returns to scale and cooperative behaviour
(Eastman and Stykolt 1960).Protection in acollusive industry will lead
to higher pricesand higher profits.The long run resultwill bethat new
firms will enter these markets,which decreasesthe scale of production
and therefore increases average costs. Therefore, protection creates
many small inefficient firms.
- Excessiveentry: increasing returnst o scale,non-cooperative behaviour
(Dixit and Norman, 1980).Trade increasesthe sizeof the market which
with Cournot competition implies higher salesperfirm and lower average costs (the number of firms in the world declines). Protection fragments markets and leads to higher production costs (in comparison
with free trade equilibrium the number of firms increases).
The theories basedon economies of scaleand imperfect competition extend the arguments for trade policy. Rent shifting is a new argument. The
range of possibilities to improve the terms of trade is increased because even
small countries can improve their terms of trade. Market distortions are inherent in imperfect competition so for example the wedge between price and
marginal costscreates an opportunity for government action,if consumers buy
to few domestic products. However, trade policy isonly asecond best policy in
this situation and industrial policythat eliminates the distortion isthe first best
policy.
The question iswhether these theories give asystematic new reason for
trade policy. The answer appears to be no because the new arguments, such
asfor example the strategic trade policy (rent shifting) are very dependent on
specific assumptions. A slight change in one of the assumptions changes or
even reversesthe implications of apolicy.Good policy requiresthat the government hasa lot of information to choose the right model. However, this information must besodetailed that it isnot available.The empirics also show that
the benefitsfrom deviations from free trade aresmall and that the value of an
optimal tariff islow (seesection 3.1.4). The potential for making policy errors
through lack of information or the possibility of retaliation byforeign governments makesthe possible lossesgreater thanthe benefits.Therefore, Krugman
(1987) concludesthat free trade isalmost never optimal under imperfect competition but that it isagood rule of thumb. Feenstra (1995) confirms this conclusion: according to his analysis of methods used to estimate the impact of
trade policies under imperfect competition, empirical results lend no support
t o astrategic role for trade policy.
3.1.4
Empirics of trade theories related to economies of scale and imperfect
competition
The empirical study of Grubel and Lloyd (1975) was the first impressive
work focusing on the importance of intra-industry trade.The authors did not
test a formalized theoretical model, but pointed at the pattern of trade between countrieswith similar factor endowments.Assuch,the Grubel and Lloyd
study can be considered to be evidence against the H-0 theorem. Together
54
w i t h the work of Leontief, the study stimulated thinking of new ways to explain international trade,focusing on increasing economies of scaleand imperfect competition.Onlysincethe late 1970snew models of monopolistic competition were seento allow atheory of trade inthe presence of increasing returns
to trade. Butalthoughthere are now manytheoretical models,especially in the
intra-industry framework, sofar there have been very few empirical studies
done to test them. Some of those are referred to by Krugman (1990: 257-261)
and Learner (1994) and are briefly discussed here.
The proposition that the proportion of intra-industry trade as opposed
to inter-industry trade should be positively correlated with the degree of similarity between countries' capital-labour ratios isconfirmed byempirical studies
of Loertscher andWolter (1980)andt w o by Helpman (1985and 1987).Thefirst
authors use differences in per capita income asa proxy for differences in resourceendowments andconfirmthe correlation usingacrosssection regression
for a single year. Helpman's analysis of 1985 confirms the proposition over a
number of yearsand also showsthat asthe industrial countries become more
similar over time the relative importance of intra-industry trade grows,just as
the model would suggest. In his 1987study, Helpman finds that for agroup of
14of the most industrialized countries both GNPsimilarity and trade intensity
have increased more or lessconstantly from 1956to 1981, giving the appearance that the model issupported. A third study, by Havryslynshyn and Civan
(1984),givesevidencefor the implication of thetheory that intra-industry trade
islikely to be more prevalent between advanced countries than between LDCs
under the assumption that advanced countries produce more differentiated
products. Learner (1994) criticizesthe empirical studies of Loertscher and Wolter and of Helpman, 1987,however, byquestioning the theoretical base of the
work. According to Learner, the link of the empirical studies and the theory is
very fuzzy, creating difficulties in interpreting the results of the regressions
computed. Empirical studies likethese may suggest apositive relation between
intra-industry trade and GDP level and similarity, they leave the question on
the roleof economies of scale in international relationship unresolved (Learner,
1994:84-89).
There are some empirical studies done attempting to quantify models by
calibrating them to data from actual industries. Examples are Dixit's (1988)
model of the USautomobile industry, Baldwin-Krugman's (1990) model of the
semiconductor industry, focusing on the USan Japanese market, and Venables
and Smith (1986) studying the UK refrigerator and footwear industries. The
results of these studies all indicate that modest tariffs are welfare improving,
and protection hasstrong export-promoting effects. The work of Dixit and of
Baldwin and Krugman are a bit more discussed below.
InDixit'sempiricalwork onthe USautomobile industry,the USauto market isrepresented asanoncooperative oligopoly of aCournot-type (conjectural
variations approach),with foreign carsdifferentiated from domestic.After calibrating his model with data from the industry. Dixit examines the effect of
policy variables like tariffs on foreign rivalsand aproduction subsidy to domestic producers.A modest tariff on imports appearsto be beneficial to USindustry for reasons of accessto home market and scaleeconomies. However, these
55
gainsareshown to be small.If aproduction subsidy isgranted,the additional
role for atariff isgreatly reduced,with the gains from adding tariffs asan instrument extremely small.
Baldwin and Krugman elaborated a model on the semiconductor industry. Semiconductor manufacture isan extraordinary dynamic industry, where
technological changeshappenvery rapidly.Thistechnological change islargely
endogenous, the result of R&Dand learning bydoing. Until the late 1970s,US
firms had dominated the market of semiconductors but than Japanese firms
took over. Baldwin and Krugman developed asimulation model, of which parameters are partly drawn from other published studies and partly estimated
bycalibrating the model to actual data. As in Dixit's study, the authors adopt
aconjectural variations approach inorderto matchthe observed industry structure. Krugman and Baldwin provide an assessment of the importance of market access by focusing on the impact of Japanese policy to protect domestic
markets for their own semiconductor industry on the ability to sell not only in
the domestic market but also inworld markets.The analysissuggeststhat privileged accessto the domestic market was decisive in giving Japanese firms the
ability to compete in the world market. However, these protection policies
resulted in higher Japanese prices,hurting consumerswithout generating compensating producer gain, so inwelfare terms the policy impact hasbeen negative.
Harrisand Cox(1984)developed ageneral equilibrium model for Canada
with increasing returns and imperfect (monopolistic) competition build in and
adopt the assumptionthat firmsareableto cooperate well enough to raise the
domestic price to the foreign price plustariff (Eastman-Stykolt pricing assumption). The excessprofits inthe protected (import-competing) industries lead to
entry of new firms and result in an inefficiently small scale industry structure.
Furthermore,the authorsshowthatthe inefficient scaleinthe Canadian export
industries isdue to USprotection policies.Combining these effects,the authors
findthat the costst o Canadafrom itspartial isolation from the USmarket are
several times higher than those estimated by conventional CGE models. Their
conclusion isthat free trade between the USand Canada would be beneficial
to both.
3.2 Neo-technology trade theories
The common feature of technology-oriented theories of trade isan emphasis on technological change and the resulting patterns of trade. In these
theories trade patterns are explained in terms of technological progress.Technological differences or gaps across countries are an endogenous outcome
through firm level product and processinnovation that reducescostsof production and generates new or better products.Theflow of technological developments and innovation isassumedto be not free and instantaneous, which implies that a firm/country has at least a temporary comparative advantage in
production and exports.The major early contributions of this field are done by
Kravis (1956), Posner (1961), and Vernon (1966).
56
Please note the difference with the Ricardian trade models in which differences in technology (productivity) for some given goods causetrade. In the
Neo-technology trade modelstrade iscaused bythe simple fact that the innovating country generates some new productsthat other countries, at least temporary, are unable to produce.
According to Kravis (1956), international trade iscaused by differences in
the availability of certain products among countries. When unavailability at
home isdue to lack of natural resources,the comparative advantage explanation would be perfectly adequate, but Kravis'sanalysis points at differences in
availability arising out of technological and product innovation. Technological
progress causescomparative advantage in trade by reducing costsof production or by supplying new products.
Posner (1961) sketches what may be called atechnology gap model.Posner observedthat asnew products and processes are continually being developed, the country in which these innovations occur will temporarily enjoy the
technological advantage over itstrading partners in these particular products.
This advantage will last only until the new technology is imitated in other
countries. But before that happens, the innovating country may export the
good eventhough it has no obvious basisfor comparative advantage in terms
of factor intensities and endowments. Overtime, each innovation is eventually
diffused around the world and the initial advantage is lost. However, asprogress continues, new discoveries are constantly being made and there exists a
constantly changing list of new products in which the innovating country enjoys a comparative advantage.
Vernon (1966) analyses shifts in international trade in constructing a
'product-cycle' hypothesis. Healso invokesthe role of foreign direct investment
by(multinational) firms in international trade.Vernon arguesthat the developersof new products must stay incloseproximity to their markets,soasto benefit from customer feedback in modifying the product and also to provide service. In addition, he arguesthat discovery of the innovation itself ishelped by
the proximity to thosewhose needsthe innovation will satisfy.Thus,both innovation and production tend to be concentrated in countries where new needs
andwantsarefirst makingthemselves known.Vernon explicitly rejects not only
factor proportions but also comparative costs asdetermining the location of
production -and later export -of new products. Instead, he predicts that new
products will be first produced in,and later exported from, the country where
they are first demanded. Only later still, when the product matures and becomes standardized,does its production move to a location of lower cost. According to Vernon's characterization,the product cycle might last from five to
twenty years.
Unlike Vernon, Hirsch(1967) goes backto the concept factor proportions
in dealing with a 'product cycle' hypothesis. He argues that new products go
through a cycle of systemic changes in technology. New products at first require large amounts of skilled labour intheir production and development. As
larger quantities are demanded, however, more capital intensive production
techniques become appropriate. Finally, when products mature and become
57
standardized,the production process becomes routine, and lessskilled labour
can play a greater and greater role. Hirsch goes on t o explain the location of
production byessentially applying the multi-factor H-0 theorem to this pattern
of factor intensities among new, growing, and mature products.
The first comprehensive formalisation of these attempts to try to explain
trade in terms of technology was offered by Krugman (1979a) in his NorthSouth-model.We will usethis elementary model to represent the technology
gap models.
Basic assumptions
Production factors:only labour. Thisfactor isdomestically mobile and internationally immobile;
Goods: n-different commodities (differentiated products);
Technology:the North canonly produce innovative goods (new product varieties), identical production functions (labour productivity) between North and
South for imitative goods;
Economies of scale:constant returns to scale;
Market structure: monopolistic competition;
Consumer preferences: homogeneous and identical between countries, consumers possess 'love of variety' (Dixit-Stiglitz utility function).
Main mechanism
The North innovates and introduces a continuous exogenous stream of
new product varieties on the market. Only after atime-lag the non-innovative
South imitates these new products. Because consumers in both regions display
'loveof variety' theywill demand all product varieties.The North will therefore
export new products and imports imitated productsandthe opposite istrue for
the South.The innovation and imitation rate will therefore determine the pattern of trade andthe time-lag in the adoption of new technologies isthe factor that gives rise to trade.
Implications
No fixed pattern of trade: each good isfirst exported bythe innovative
North and after a time lag it will be exported by the South (Vernon's
product live cycle).
Wages inthe North are higher: The labour productivity in both countries
is the same, however, for the new product varieties the North obtains
monopoly power and the associated rents flow to the only factor of production, labour. The wages will therefore be higher in the North than in
the South.Becausethe labour productivity for imitated goods isassumed
equal in both regions, only the South will produce imitated products.
An imitation of a new product bythe South (technology transfer) moves
production from the North to the South.At initial pricesthis reduces production costsand increasesworld output which causesallocative gains for
both countries. However, imitation also alters world distribution of income becausethe wages inthe North will decline and those inthe South
will rise.The decline inthe wage differential improves the terms of trade
58
for the South.Theterms of trade effect istherefore asecondary positive
effect for the South but a negative effect for the North. So, imitation is
beneficial for the South and the North may be worse off.
An innovation in the North raises the number of product varieties and
because consumers display love of variety, this isbeneficial for both the
South and the North. Production and capturing the rents takes place in
the North, sotheir wages increase and the terms of trade improves for
the North (a secondary benefit for the North and a loss for the South,
however in the Krugman model this secondary effect isalways smaller
than the love of variety effect).
The implication that innovation improves the terms of trade is in contradiction with the traditional theories inwhich technological change in the
export sector generally worsens the terms of trade.
Income inthe North depends partly on the rentsfrom their monopoly of
newly produced products. Implications of the imitation and innovation
process imply that the North must continuously innovate in order to preserve itswages or, to put it in Krugman's words, 'Like Alice and the Red
Queen, the developed region must keep running to stay in the same
place'.
World output will raise both by innovation and imitation.
An increase in labour in one region makesthis region worse off because
real wages decline and the terms of trade deteriorates.
International movement of production factors: a two-factor version of
the standard model, in which labour isimmobile between countries and
capital ismobile internationally has some interesting implications. Innovation in the North will be associated with capital inflows because it
raisesthe marginal product of capital inthis region.The inflow of capital
is an additional (next to the effects that appeared in the one factor
model) benefit for Northern wages andthe outflow of capital isan additional negative effect that hurtswages inthe South.Through this effect
Southern workers caneven loosefrom innovation.The profits from innovation arecollected bythe immobile factors:while the incomesof mobile
factors will be equalized the inequality of incomes of immobile factors
increases.The effects of imitation arejust opposite and induce a capital
inflow in the South.
Dollar (1986) combines this North-South-model with the H-O-Smodel. In
this model,the rate of imitation ispositively affected bythe North-South
wagedifferential.JensenandThursby (1987)assumeendogenous innovation byaNorthern monopolist (innovation dependson resources devoted
to R&D).One resultthat differsfrom the Krugman model isthat imitation
may improve the terms of trade for the North. Segerstrom, Anant and
Dinoupulos (1990) developed a model in which R&D is both costly and
risky (an innovation racethat afirm may win or lose).
59
Policy
Free trade policy is beneficial from aworld point of view:
- the decline of industries in developed countries will be a recurrent
event and from apoint of view of world efficiency it isdesirable. However, some countries may loosefrom trade therefore it isessential that
there will be some income redistribution effects.
Imitation from a nationalistic point of view
From a Northern point of view:
- imitation has two effects, on the one side it erodes Northern profits
but on the other side it increases world output. The total effect on
Northern income is ambiguous. If Northern countries loose they can
react intwo manners. First,they introduce 'innovation policy' that may
increasethe innovation rate and soupsetthe negative effects of imitation, and whether this is impossible or the effects are too small they
can usetrade policy to protect their markets.The latter implies losses
from aworld point of view, but Northern countries will only choose for
free trade if there isacorrect international technology transfer system.
From a Southern point of view:
- imitation has not only allocative gains but it also improves itsterms of
trade. Policiesthat increasethe rate of imitation are beneficial for the
South. However, there must be a correct technology transfer system
otherwise it encourages protection bythe Norththat makesthem even
worse off.
Innovation from a nationalistic point of view
From Northern point of view:
- innovation policy that increases innovation rate isbeneficial.
From aSouthern point of view:
- innovation in the North may hurt the South when it induces a capital
outflow in the South. Protectionism may be worthwhile in this case.
Critical remarks
- innovation and imitation are costless and exogenous;
- innovation isonly possible by introducing new varieties (product innovation), process-oriented innovations are not possible;
- the elasticity of substitution between any pair of goods isconstant.This
assumption isquestionable with regard to new and imitated goods.
Empirics
Most, if not all empirical testsof the technology trade theories try to explain the pattern of trade of the US, simply assumingthat the USisthe innovative Northern country with high capita incomesand relative wages andthe rest
of the world the imitating South.According to Dearndorff (1984) the tests of
these theories using USdata have tended to be rather successful.Several authors sought and found positive correlations between USexport performance
across industries and various measures of R&D.Since R&D isrelated to technological progress,whatever itscauseor effects,this evidence lends support to all
technology theories of trade. The technology theories point at different rea-
60
sons for their prediction that the North (the US) will be an exporter of new
products. Dearndorff, however, identifies severalstudiesthat support both the
technology gap and the product cycle explanations of UStrade (Dearndorff,
1984:495-499).Thesestudiesfocus on the correlation of UStrade performance
andvarioustechnology-related variables,suggested bythe technology theories
of trade. Inall studies, research and development asan important determinant
of USexports appears to find strong support.
Severaltestsof the technology theory of trade have also introduced additional explanatory variables, including those that are appropriate to the factor
proportions theory, and support too the conclusion there isastrong and positive correlations between trade performance andtechnology-related variables.
At the sametime, however, it becomes clear that technology-related variables
are much related to the ability of individuals, firms, countries to develop and
exploit technology which isrelated to the availability of knowledge and skills.
Dearndorff concludes therefore that it isdifficult to distinguish evidence supporting technology from evidence supporting human capital or skills as determinants of trade (1984: 499).
Dosi et al., 1990 isa recent study which shows that countries tend to be
strong exporters in industries in which they invest heavily in R&D.The process
of creating comparative advantage could bedescribed bythe product cycle but
although some empirical studies provided evidence for some industries, recent
work suggests that this explanation isnot very far-reaching. Gagnon and Rose
(1992) analyse USand Japanese trade statistics between 1962 and 1988 and
show that both countries' trade patterns at the 4-digit level remained remarkably stable: very few goods moved from exports to imports or vice versa.Alt h o u g h this result need not be inconsistent with the product cycle theory,
Krugman (1995:354) states that the results of this study indicates that there is
no evidence for a rapidly and constantly changing industrial structure of each
nation, or at least 'operates much more slowly than the rethoric of product
cycleenthusiastswould suggest'. Bythe end of the 1970sVernon (1979) himself
already questioned some elements of his own 'production cycle' explanation
of trade. Hequestioned whether production of new products hasto start near
their markets, mainly because he observed that wages and incomes have
caught up to the US in several other countries. Furthermore, Vernon argued
that the growth and spread of multinationals have undermined the product
cycle theory. Technological levels among advanced countries showed to converge rapidly and,especially because of multinationals, the speed of diffusion
of innovations accelerated.
Toconclude, empirical testsor application of the product cycletheory are
limited in extent. The support from these studies for the idea of the 'technology gap' asdriving force behind trade isfragmentary at the best, suggesting
evidence for the idea that transitory advantage resulting from innovation isa
major factor in trade for only some industries.
61
4. TRADE IMPLICATIONS OFTHE NEW
GROWTHTHEORIES
4.1 N e w Growth theories
Growth-accounting studiesshowthattechnical change isthe most important factor contributing to economic growth (e.g.Solow, 1957, Denison, 1962).
However, the standard Neo-classical growth model treated this important factor asexogenousto the system.Hence,the need hasarisento make technology
endogenous. The 'new' growth theories found several ways to endogenize
technological change in ageneral equilibrium model 1).There are t w o main
approachest o modelendogenousgrowth and both built forward on the 'new'
trade theories.
Thefirst approach assumesthat externalities or learning-by-doing effects,
which are by-products of other activities,cause growth and the 'externaleconomiesof scaleapproach' isusedto modeltheseeffects (seesection 3.1.1, within
the 'new' trade theories). The second approach assumes that technological
change isthe intentional outcome of economic behaviour and firms have t o
'invest' in knowledge creation to obtain technological change. Resources have
to be invested in activitiesthat are not directly productive.The 'internal economies of scale approach' within the new trade theories showed a direction to
model this process (see section 3.1.2). Investments in knowledge can be seen
asakind of fixed costsand monopolistic competition makes it possibleto cover
these fixed costs.Most studiesthat usethe second approach, assume also that
knowledge generates some externalities and are therefore a mixture of both
approaches. Becausethis hasbecome the dominating approach we discuss the
first approach and a mixture of both approaches.
The reason to include the 'new' growth theories in this overview is that
the 'new' growth theories put the static trade models into adynamic setting.
It is important that these theories deal therefore with the dynamic evolution
of comparative advantage andthe consequences of trade in aworld of global
technological competition (Grossman and Helpman, 1991b). Inthis survey we
focus on the dynamic trade implications of these 'new' growth theories.
4.1.1
Knowledge asside-product: external economies of scale
The central mechanism inthis 'knowledge asside-product' approach consistsof three elements.Thefirst element isthat afirm creates knowledge asa
side product t o normal business activities, such as producing goods or doing
1)
62
For an overview of the 'new' growth theories, seeVerspagen (1990),Van de
Klundert enSmulders(1991),SchneiderandZiesemer(1995)orVan Meijl (1995,
chapter2).
investments. Learning-by-doing is an important theoretical explanation for
these externalities. According to Arrow (1962) the acquisition of knowledge
(learning) ispositively related to experience. He argues that agood measure
of experience isinvestment because 'each new machine produced and put into
use iscapable of changing the environment in which production takes place,
sothat learning takes placewith continuous new stimuli' (Arrow, 1962,p.157).
The second important element in this approach is that newly created
knowledge by one firm flows directly to all other firms where it can be used
without payments in return:knowledge spillover effects exist.Thethird important element isthat the newly created knowledge increases the productivity
level of the production factor that can be accumulated (e.g.capital).
To understand the role of the three elements in the growth mechanism
it is illustrative to describe the standard Neoclassical growth model where
growth vanisheswithout an assumption of external technological progress. In
this standard model it isassumedthat there arediminishing returns to the production factor that can be accumulated: i.e. when you add more and more
capital to a piece of land the marginal benefits are diminishing. Therefore,
incentivesto invest in capital in this model vanish and growth will end. In the
'Knowledge asside-product approach' growth may not vanish because linked
t o the accumulation of capital new knowledge will be created and this knowledge makescapital more productive in all firms.The knowledge spillovers may
exactly counterbalance the diminishing returns of the factor that can be accumulated by increasing its productivity level.As aconsequence the incentive to
invest in the factor that can be accumulated may not diminish and this may
result in a positive growth rate.
In modelling terms,this approach has usedthe methodology of the 'external economies of scale' approach to achieve endogenous growth (seesection 3.1.1, within the 'new' trade theories). More specifically, firms invest in
new equipment which also generates some knowledge that increases the
productivity of the capital stockof the investingfirm but alsoof the other firms
(i.e. aspillover effect).The latter effect isnot taken into account by the investing firm and therefore increasing returns arewholly external to the firm which
allows perfect competition to remain at the firm level. However, the existence
of externalities causes increasing returns to scale in the aggregate production
function. This implies that if one firms doubles its inputs, the inputs of other
firms will also increase,and hencethis will result in a more than proportionate
increase in aggregate production.The latter causesthe economy to grow w i t h
a positive endogenous growth rate.When the positive spillover effect is large
enough the economy will grow with a positive growth rate (Romer 1986). Because individual producers do nottake into account their knowledge spillovers
on other producers acompetitive economy gets a lower growth rate than socially optimal.
Thisapproach recognizes that knowledge hassome non-rival good characteristics. However, technological progress or the growth of the knowledge
stock appears asaside product to activities that are not specifically directed to
the creation of knowledge.
63
Dynamic trade implications
Thetrade implications of these 'new' growth models are illustrated w i t h
the Lucas-ll model (1988) in which initial factor endowments determine the
pattern of specialization and the Young (1991) model in which there isa cont i n u u m of goods and the specialization pattern is determined by the initial
technological capabilities (i.e knowledge stock).
Initial factor endowments govern trade patterns: the LucasII model
Lucas (1988) describes in the second part of his paper a model in which
all human capital accumulation islearning by doing (no resources have t o be
invested to increase the human capital level).Technology isassumed identical
acrosscountries and there existsendogenous technological change by increasing the efficiency (process-oriented) via unbounded learning by doing effects
that are good-specific. Lucasidentifies a hightechnology good in which learning isfaster than in the other good. Furthermore, there are t w o kinds of human capital which are specific for the production of one of the goods. In this
model the trade pattern is determined by differences in the countries initial
relative human capital endowments. Because all human capital accumulation
is learning by doing, countries accumulate skills in goods in which they have
already a comparative advantage, and reinforce in this manner their initial
comparative advantage.
The intertemporal welfare effectsthat determine the desirability of acertain specialization pattern are dependent on two effects. First, learning effects
are highest inthe high technology good which leadsto afaster growth in production of this good. However, secondly,this faster production growth implies
that the terms of trade will move against it. The deterioration in the terms of
trade isdependent on the degree of substitutability between the t w o goods:
if the degree of substitutability islow (high) the terms of trade declines strongly (less).This implies that the country which specializes in the high tech good
getsthe highest real growth rate only if the substitution elasticity between the
t w o goods iselastic.An important result of this model isthat real growth rates
are determined endogenously and they can differ across countries.
Initial technological capabilities govern trade patterns: the Young model
Young (1991) investigates the dynamic effects of international trade in
a world consisting of a lessdeveloped country (LDC) and adeveloped country
(DC),the latter distinguished byahigher initial levelof knowledge.There exists
acontinuum of technical more sophisticated goods (to produce more advanced
goods,more experience/knowledge isneeded).Technological change iscaused
by learning by doing effects that are bounded in each good (per unit output
labour requirements decline until acertain boundary ismet when production
increases).At each point intime there aretwo setsof goods, one in which the
learning bydoing effect hasstopped and one in which learning by doing continues. In a static sense this is comparable to the Lucas model, however in a
dynamic sense it differs because it endogenizes the movement of goods out of
the learning by doing (infant industry) sector into the other (mature) sector.
Old goods will be discarded for more advanced goods and this causes an in64
crease inthe labour productivity. Over time, growth causesthe production of
a changing basket of goods,with both the quantity and variety of goods consumed increasing (there will be gains from increasing variety w i t h o u t the assumption of monopolistic competition). The trade pattern will not become
static but remains evolving.With trade the LDC(DC) specializes in the mature
(infant industry) goods where learning by doing has(not) ended.The LDC(DC)
experiences a lessthan or equal (greater than or equal) rate of technical progress and GDPgrowth to these under autarky. If the DC population is greater
than or equal to that of the LDCor when the initial technical gap is large, the
technical gap between these countries will grow without bound. Only when
the initial technical gap isnot too large and the LDCislarger there isapossibility of catching up or leapfrogging.
Finally, with regard to intertemporal welfare effects the results of this
Young modelareambiguousanddependfor exampleonthe initial technology
gap and on the sizeof the countries. Ifthe DCmaintains itstechnical lead, DC
consumers gain from trade by dynamic (higher rate of technical progress) and
the usual static gains from trade. Although its own rate of technical progress
decreases the welfare of LDCconsumers may still improve through the usual
static gains from trade, which increases asthe DC experiences technological
progress. In general it can be said that the welfare of the LDC will improve
when it is small relative t o the DC, it will be reduced when the LDC is much
larger than the DCand it cannot catch up.Ifthe LDCcatches up itswelfare may
increase or decrease (the latter ispossiblewhen the terms of trade deteriorates
relatively fast for more sophisticated products).
Main trade mechanism
Inprinciplethe specialization andtrade pattern isdetermined by comparative advantage (Lucas; initial factor endowments) or the initial knowledge
stock (Young;technological capabilities andcountry size).The dynamic implications of these external economies of scale growth theories are that in general
a country will built up knowledge or expertise inthe goods in which it specializes and therefore reinforces its comparative advantage in these goods, although the possibility of leapfrogging and reversing comparative advantage
remains (see, Young model). Because the technological opportunities differ
between goods the specialization pattern determines also the welfare level
and long term growth of a country.
Po//cy
Industrial policy t o correct for positive externalities (learning effects or
knowledge spillover effects).
Inthe Lucasand Young models learning effects are assumedto be external,agentsdo nottakethem into account. Ifthey didtheywould allocate
labour more to the good with the high growth potential in exchange for
a lessdesirable mix of current consumption. An industrial policy focused
on 'picking winners' (subsidizing the production of goods with a high
learning potential) is the right policy. However, Lucas remarks 'In the
model, 'picking winners' iseasy. If only it were so in reality! (1988:31)'.
65
Industrial or trade policy (second best) to reverse the specialization pattern:
industrial or trade policy (second best) can also be used t o obtain the
specialization pattern that leadsto the most desirable specialization pattern inthe long run.Subsidiesor protection can reversethe specialization
pattern. However,to deduce the correct policy advice inthis model isvery
difficult because one hasto know the exact technological opportunities
of different goods in different countries.
4.1.2
Knowledge as investment: a combination of external and internal
economies of scale with monopolistic competition
An obvious drawback of the external economies of scaleapproach isthat
technological progress isaside-product of other activities. In the knowledge
as investment approach technological progress becomes the intentional outcome of economic behaviour. The general method chosen in this approach is
to identify aseparate R&Dsectorsalong other sectors inthe economy.The R&D
sector produces blueprints of new goods and asaside-product general technological knowledge. The blueprints are specific and provide guidelines to produce a certain 'unique' product. The R&D sector sells these blueprints to the
production sector.The production sector cansellthese unique products above
marginal costs (i.e. imperfect competition) and can earn back the fixed costs of
buying the blue-print.Thegeneraltechnological knowledge cannot be applied
inthe production of goods, but hasamore general nature. Itaddsto a general
knowledge pool which can be used in the production of blue-prints. A larger
general knowledge pool, inturn, reducesthe costsof producing blueprints and
istherefore astimulusfor the development of new blueprints byallfirms in the
research sector.Thiscontinuous incentive to develop new products causesendogenous growth. Important for the generation of endogenous growth isthat
the incentive to invest in R&Ddoes not decline. Inall these kind of models the
growing stock of knowledge asaside product of R&D generates this constant
incentive,whether these modelsfocuson R&Dfirms expanding product variety
(Grossman and Helpman, 1991b)or on increasing quality of aconstant number
of varieties (Aghion and Howitt, 1992,Grossman and Helpman, 1991b).
In models where R&Dexpands product variety it isassumed that a larger
number of product varieties isvalued positively by consumers (in case of consumer products) or it increases the productivity of the final production sector
(in case of intermediate products). When an individual firm determines the
amount of R&Dto invest, it does not take into account that the knowledge it
will produce also increasesthe knowledge level of other firms because it does
not get compensation for this. Ifthe innovator would get compensated by the
other firms for this increase in knowledge it would have had invested more:
this iscalled the intertemporal spillover effect which dominates in these models.A R&D subsidy can give the firms the right incentive (Grossman and Helpman, 1991b).Therefore,the market providesjust asinthe external economies
of scale approach insufficient incentives for industrial research in comparison
w i t h what issocially optimal.
66
In models where R&Denhancesthe quality of products, it isalso assumed
that R&D performed by afirm contributes to ageneral knowledge stock that
can be used by all firms.This isagain an intertemporal spillover effect which
may lead t o under-investment in R&D. However, in these increasing quality
models,the firm that innovates getsthe whole market and destroysthe profits
of the previous innovator. Because an individual firm does not take this profit
destruction effect into account, this may lead to over-investment in R&D in
comparison whit what issociallyoptimal.Therefore, the t w o main market distortions are the profit destruction effect (displaced leaders lose a stream of
monopoly profits) and the knowledge spillover effect. The incentive to invest
in R&Dcantherefore betoo low ortoo high. Inthis case,the optimal policy can
beatax or subsidy (Aghion and Howitt, 1989,Grossmanand Helpman, 1991b).
Dynamic trade implications
Assumptions
Usually,these theories divide the labour market in unskilled labour and
human capital.They assumethat advances intechnology are most often engineered byskilled peoplewho have invested heavily inthe development of their
technical skills. Unskilled workers generally substitute quite imperfectly for
skilled labour (Helpman and Grossman, 1991). Furthermore, it isassumed that
relative factor endowments differ between countries and are fixed.
A further assumption isthat the various manufacturing activities differ
both in the intensity w i t h which they employ various primary inputs and in
their potential for contributing to innovation and productivity growth. In generalthesetheories assume atraditional good that isunskilled labour intensive,
a differentiated high tech good that is human capital intensive, and a R&D
sector that produces blueprints for the differentiated high tech good that is
even more human capital intensive.The potential for technological change is
low for the traditional good and high for the high tech good.The traditional
good is produced with constant returns under perfect competition and the
high tech good isproduced with increasing returns to scale in a monopolistic
market structure. In most casesthe production technology is identical across
countries, in some casesone country has a larger knowledge stock and therefore a higher productivity of human capital in the R&D sector.
The dynamic trade and welfare implications of this approach are also
dependent on the assumption whether knowledge spillovers are national or
international in scope. Does or does not knowledge flow across borders? The
trade and welfare implications of this approach are illustrated w i t h t w o backbone models. Inthe first model knowledge spillovers are international in scope
and in the second model knowledge spillovers are national in scope.
A.
International knowledge spillovers
Assume a t w o factor (human capital and unskilled labour), t w o sector
(traditional goods and high-tech differentiated goods), two-country
model.Thetwo sectorsdiffer intheir factor intensities and intheir contribution to technical progress. Countries differ in their relative factor en67
dowments. Further assume that goods must be produced in the country
in which they have been developed.
The human capital-abundant country (net) exports the human capitalintensive differentiated product (Heckscher-Ohlin theorem). There are
net-exports because there is intra-industry trade in the differentiated
sector. Eachfirm ineachcountry exports hisunique developed brand.The
relative labour-abundant country exports the traditional labour intensive
good. So far nothing new in comparison w i t h the 'new' trade theories
(seesection 2.1.2, Krugman, 1981,Dixit and Norman, 1980, Ethier, 1979).
But these 'new' growth theories add one important prediction to the
static models: The long run growth rates of output and GDP are linked
to resource endowments.
Ifthere isimperfect specialization in the long run,the steady state rates
of innovation inthe high tech sectorsof the two countries are the same,
because both countries keep an R&Dsector. However, the R&D and hightech sectors represent alarger fraction of value added inthe human-capital rich country.The labour richcountry specializes relatively inthe manufacturing of traditional goods,where opportunities for technical progress
are fewer. Therefore the human capital rich country experiences a faster
growth rateof output. Despitethis both countries enjoythe same growth
of real consumption because long run interest ratesarethe same in both
countries and each has access via trade to the entire set of innovative
products.
International knowledge spillover is a crucial feature for these results,
because this makes that all innovators have the same knowledge and
that national advantages in R&D arise only from differences in relative
factor prices (which are dependent on resource endowments). Factors
suchasthe sizeof acountry andthe history of its production play no role
inthe longterm trade pattern,what only matters arefactor endowments
(Grossman and Helpman, 1991b).
National knowledge spillovers
Consider aworld with only one production factor (labour) and in which
countries only differ in size and in their prior research experience. With
only national knowledge spillovers the initial conditions govern long run
outcomes. In many situations the country with the initially greater stock
of knowledge has an advantage in R&D and accumulates knowledge
more quickly than its trading partner. This sustains and adds to its productivity lead. History alone determines long-run trade patterns and
growth rates (i.e.hysteresis). Only when the countries are very different
in size the large country can overcome a modest knowledge lag if the
share of consumer spending devoted to traditional goods is relatively
small.
Thewelfare implications for acountry arevery dependent on the specialization pattern and the related wage rates.When the country that conductsthe world's R&Dalsoenjoys higher wages inthe long run,it isagain
possible that a country which specializes in the sector without national
68
knowledge spillovers losesfrom trade or can improve itswelfare by specialising inthe other sector. Inthis case R&D policy for the lagging country isjustified 1).A sufficiently large subsidy to R&D can be used to overcomethe initial productivity disadvantage.Thespecialization pattern will
be reversed.This iscalled policy-hysteresis: atemporary policy can have
permanent effects (Grossman and Helpman, 1991b).
Main trade mechanism
When innovation leadsto the development of new varieties of horizontally differentiated firms,the pattern of trade isdetermined bythe number of
blueprints in the hands of each country's firms. Over time the trade pattern
evolves in accordance with the number of new discoveries made by entrepreneurs in each country. This in turn depends upon the R&D investments that
take place in each location.
With quality competition innovators tryto improvethe quality of existing
products. Ifthey succeed,they capturethe world market.Thedirection of trade
in a particular product maytherefore reverse over time. In aggregate the pattern of trade isdetermined bythe number of products inwhich acountry takes
the lead. This isagain dependent on the amount of R&D investments.
In both casesthe R&D investments are dependent on the specialization
pattern which iscaused bythe principle of comparative advantage (factor endowments), history,the initial stock of knowledge, the scale of acountry and
the demand structure.Thesefactors determine the number of people that are
working in the R&D sector, the high tech sector and the low-tech sector.
Thewelfare andgrowth implications aredependent onthe specialization
pattern and whether knowledge spillovers are national or international in
scope. With international knowledge spillovers and no full specialization all
countries will gainfrom trade and no specialization pattern ispreferable. However, with national knowledge spillovers the welfare implications of certain
specialization patterns are very different and may even be negative. In this
situation certain specialization patterns may be preferable to others.
Po//cy
Industrial policy (first best) andtrade policy (second best) may be beneficial in the following circumstances:
in models where R&D expands product variety there isan under investment in R&D(because intertemporal knowledge spillovers dominate). An
R&D subsidy isthe right policy;
1)
However, with equal wages an R&D subsidy often reduces welfare despite it
increasesthegrowth rateofthegrossdomestic product.Themain reasonisthat
trade ingoodsandassetsproducesthe mostefficient outcome:'residentsof the
countrythat specializesinthe productionoftraditional goodsstill benefit from
innovations that aremade abroad becausethey investtheir savings in foreign
assetsandimportthe novelgoodsthat emergefrom their labs' (Grossman and
Helpman, 1991:232).
69
in models where R&D increases product quality there can bean under- or
overinvestment in R&D.An R&D subsidy or tax isthe right policy;
in models where knowledge spillovers are national in scale and wages
linked to the high tech goods are higher, atemporary protection policy
may change the specialization and trade pattern (policy hysteresis).
4.1.3
Empirics
The new growth theories developed usefulconceptsto deal with technological change and growth. However,just asinthe caseof the new trade theories (imperfect competition and economies of scale),the problem with this approach isthat it isdifficult to testthese theories empirically. It isextremely difficult to quantify certain effects and estimate their importance and therefore it
ishard to decide whether or not it is beneficial to imply some policies. At the
moment there seemsto be no need for more complicated models but priority
should be given to the empirical implementation of these theories.
The empirical evidence on the assumptions and implications of the new
growth theories hasjust started.Jones (1995a) hastested the relationship between changes in R&D spending and changes in macroeconomic growth implied by these models. More specifically, all the models that treat knowledge
asan investment good (R&D-based models) haveassumedthat past knowledge
is never rendered obsolete by new innovation and that therefore there are
constant returns to R&D investments, which hasthe following implications for
per capita growth rates: 'if the level of resources devoted to R&D, measured,
saybythe number of scientistsengaged in R&D,isdoubled,then the per capita
growth rate of output should also double, at least in the steady state.' Jones
shows that empirically this prediction receives little support. The number of
scientist engaged in R&D in advanced countries has grown dramatically over
the last 40 years and growth rates either have exhibited at constant mean or
have even declined on average.
Furthermore, Jones (1995b) showsthat if one releasesthe assumption of
constant returnsto scaleto R&D,the influence of policy (e.g.R&Dsubsidies) on
the long-run growth rate vanishes. However, how long isthe long run? Policies
can still have an influence on the period on the growth along the transition
path to the new steady state.Whether policies are still worthwhile depends on
the length of the transition path and the magnitude in which policies can
shorten this path.
Several authors investigated the assumption of whether spillovers are
national or international in scope because in models with only national spilloversgovernment policycanhave long lasting effects.Coeand Helpman (1993)
and Bernstein and Mohnen (1994) find some evidence for international spillovers. However, Branstetter (1996), who investigated both intra-national and
international spillovers, concludes that intranational spillovers are stronger
than international spillovers. There is also some evidence that spillovers are
geographically localized (Jaffe 1986,Acset al. 1992).Overall,there is evidence
that some spillovers are international in nature, but local or intra-national
70
knowledge spillovers exist too. This gives some support to models in which
government policy may have long lasting effects.
4.2 Evolutionary Growth Theories
A recent direction within economics assumes that technology plays the
fundamental role in economic life. In contradiction to the 'new' trade and
'new' growth theories they seethe process of technical change asa dynamic
evolutionair process. The innovation process is characterised by cumulative,
firm-specific and irreversible processes (Dosi et all., 1988).Therefore, technology cannot be reduced to freely available information or to a set of 'blueprints'. Furthermore, they assume out of equilibrium dynamics and bounded
rational behaviour. These 'technology' theories aretherefore a radical department from the traditional trade models which took technology data asexogenous to the economic process. However, becausethe evolutionary growth theories reject many of the standard neoclassical assumptions they are less formal
and more heterogeneous. Because of this heterogeneous nature we concentrate on some central features of the evolutionary economics.
Assumptions
Important assumptions in the evolutionary growth theory are that the
innovation process ischaracterised by cumulative,firm-specific and irreversible
processes; sectors differ in their contribution to the absolute competitiveness
of acountry; and countries differ intechnology skills and innovation capacity.
From innovation process to technology trade mode
The development of technology iscumulative and firm specific. Technological changes are often caused by learning by doing and learning by using,
suchthat technology isoften firm or region specific, and isgenerally embodied
in people and organisations (Dosi, 1988, Dosi, Pavitt and Soete 1990).Thisassumption of firm specific knowledge is in contrast with the new growth theorieswhich assumethat knowledge directly flows to ageneral knowledge stock
which can be used by all firms (national and international).
Technological paradigms (Dosi,1988) and technological trajectories (NelsonandWinter, 1977)send resources inacertain direction andtake care of the
factthat technical progress happens along arelatively ordered,cumulative and
irreversible pattern. The technological opportunities of afirm are, therefore,
constrained by activities in the past. Bysearching for innovations it will build
further on and isrestricted by his 'specific' knowledge.
Besides these private, firm-specific aspects of knowledge, some public
good features of knowledge play a role in the process of technical change.
First, the freely available information, such as publications. A second public
good feature isrelated t o the 'untraded interdependencies' between sectors,
firms andtechnologies.These interdependencies appear inthe form of technological complementarities, synergy effects, spillover effects and incentives and
71
restrictions which are not fully reflected in trading goods.The occurrence of
these technological externalities differs very much between sectors.
The direction and rate of technical change in amarket economy is different in every sector and dependent on (Dosi, Pavitt and Soete 1990):
the sources and nature of technological opportunities;
the nature of actual or potential markets;
the possibilities for successful innovators to appropriate asufficient proportion of the benefits of their innovative activities to justify the research
effort invested in such activities.
Because sectors differ in these characteristics, their contribution to the
national innovation capacity isvery different.
A very important implication of the analysis sofar isthat there exist unequivocal a-symmetricaldifferences between firms and countries intechnological capability.There are absolute cost differences between firms and countries.
Thesetheories aretherefore against all kinds of theories which assume identical production technologies between countries.
Trade mechanism
Dosi, Pavitt and Soete (1990) developed atrade model inwhich the principal part is played by absolute technological advantages (technology gaps)
between countries. These absolute advantages/disadvantages determine the
competitiveness and world market position of 'all' sectors and have therefore
a great influence on the income and employment position of a country. The
minor part is played by relative technological gaps between sectors within a
country.These relativetechnological gapsdetermine the specialization pattern
between sectors according the mechanism of comparative advantage.
We can illustrate this model with asimple example: absolute advantages
determine why acountry hasan average market share of 0.5% or 10%.Comparative advantage determines why one sector inthe country with an average
market share of 10% (0.5%) hasa market share of 9% (0.4%) and another sector has a market share of 11% (0.6%).
Dominant technologies, which influence the production in almost all sectors and display often huge positive externalities, are therefore extremely important in their contribution to the absolute competitiveness of a country
(MacDonald and Markusen, 1985).
Implications
1.
A nation's future growth and technological development isaffected by
the current composition of its industries and activities, and by its paradigms of how to develop and exploit technology (Patel and Pavitt, 1990).
2.
A specialization pattern according to the traditional mechanism of comparative advantage can lead a country to specialize in those industries
(sectors) and activities inwhich the opportunities for growth and technological development are least. The current specialization pattern of a
country hastherefore adynamic effect becausethis determines in which
sectors technical skills will be accumulated, innovations will be done,
72
3.
4.
5.
economies of scale will be realized, etcetera. Sectors differ in these opportunities suchthat the present specialization pattern isextremely important for the economic performance of countries in the future. A specialization pattern which is static (Ricardian) efficient can therefore be
dynamic inefficient.
The possibility of this trade-off between static and dynamic efficiency is
greater the greater the distance from a country to the technological
leader (Dosi, Pavitt and Soete, 1990).
Staticeconomies of plant sizeare not so important anymore through the
continuously arising of new technologies.The scaleeconomies which are
of greatest significance now areof two sorts (Freeman,Sharp and Walker
1991):
- economies of scaleandeconomies of scope in research,design, production and world-wide marketing networks. The costs of designing, developing andtesting new generations of products and services are often such that commercial viability is only possible if they are spread
over very large markets;
- dynamic economies of scale,which are associated with the accumulation of knowledge within a firm from R&D, design, production and
marketing (Pavitt, 1984).
Competition between firms isavery important factor in the innovation
process. Strong competition forces firms to innovate. The strength of a
sector can therefore be reflected in the presence of a relatively large
number of nationally based largefirms. Relativelytechnological strengths
of countries are therefore correlated with rivalry among the large firms
and not with gigantism (Patel and Pavitt, 1990).
Policy
Industrial and trade policy isbeneficial if:
- there is a tradeoff between static Ricardian efficiency and dynamic
Schumpeterian efficiency;
- a sector has a high contribution to the absolute competitiveness of a
country (isit adominant sector) because it generatesfor example huge
externalities;
- asector hasahigh dynamic potential in terms of growth and technological opportunities.
Empirics
Dosi, Pavitt and Soete (1990)find the following empirical relations which
supported their technology trade model:
- avariety of science andtechnology measures (e.g.R&D, Patents) gives
a constant picture of the aggregate distribution of innovative activities
between countries;
- international differences in innovative activities are reflected in differences in shares of world exports in most sectors,and in manufacturing
asa whole;
73
- export performance ispositively associated with differences in per capita innovative products and differences in labour productivity;
- changes intrade performance are more strongly associated with changes in innovative activities than in relative labour costs;
- international differences inper capita income havebeenclosely related
t o international differences in per capita innovative activity;
- international differences in the rate of growth of per capita income
have been associated with similar differences inthe rate of investment
and in the rate of growth of innovative activities.
74
5. CONCLUSIONS
5.1 Evaluation of empirical tests of theories explaining international
trade
Testingtradetheories hasprovento beextremely difficult. Someof these
problems intestingtheories have beenattached upon already. First,the theory
should be interpreted correctly inthe empirical tests. However, theories are in
most cases formulated in so stylized terms that a translation of a theoretical
model into an empirical application isvery hard to do. Second, it may be difficultt o find testable alternative hypotheses from the theoretical model.Third,
it isoften difficult to find any variable that closely measures the hypothetical
construct stipulated bythe theory (for instance,what are the 'relative autarky
prices',howto measure 'imperfect competition',what isagood proxy for 'technological development', etc.). All in all,the problem in testing atheory is t o
find outwhat needsto betested andto find data that develops aproper correspondence between the theory and observable events. Despite all the efforts
done sofar there ismuch controversy on how well tests have been done and
what these testssuggest about the validity of the model they were based on.
More often than not empirical work is criticized for being misdirected or incomplete astests of international trade models.
According to Learner (1994) the effects of empirical testing of international trade theories on the way economist think about the determinants of
international trade have been small.That isnot surprising to him because as
'by definition a model is not literally true, so there is no reason t o test it'
(1994:66). Learner, therefore, recommends researchers to 'estimate, not test',
expressing that data analysis should not take theoretical foundation too literally.Learner statesthat an influential empirical studywill haveto havea proper
balance of issues,theory and data.The central issuein international economics
which needsto be addressed ishow, if at all,governments should intervene in
international commerce. Attention seems too much focussed on formulating
theories while the issuesto be addressed are lost in the theoretical discussions
and questionson mappingthe theory into observable phenomena are ignored.
The latter isalso recognized by Krugman who finds that especially with respect
to the so-called newtrade theories '...the sophistication of our models in general seemsto have outrun our ability to match them up with data or evidence'
(Krugman, 1990:261).
The problems concerning the possibilities of empirical validation of trade
models elicit Learner's statement that 'to make progress, economists ought t o
abandon the idea that models are either true or false in favour of the notion
that models are sometimes useful and sometimes misleading' (Learner, 1993:
439). Models are only tools, nothing more and nothing less. Learner stresses
75
that each of the theoretical trade models isappropriate insome circumstances
and inappropriate in others,and therefore, empirical studies should not try to
test the validity of the theories. Instead, he claims,empirical work 'might identify the circumstances under which each of the tools is most appropriate, or
measure the 'amount' of trade that isdue to each of the sources. Neither of
these tasks have been accomplished or often even attempted' (Learner 1994:
69). Inour later work when we focus on the explanation of agricultural trade
patterns,we will tryto identify the circumstances under which eachof the (theoretical) explanations of trade isappropriate for agricultural products.
The circumstances meant in the former sentence are much related to
what has been identified asthe main mechanisms of the trade theories. The
main mechanisms inthe trade theories to explain international trade flows are
factor proportions, economies of scaleand differences intechnology (technology gaps). Each of these three explanations of trade isrelated t o specific key
commodity characteristics and national attributes (seee.g. Hufbauer, 1970 and
Choudhri, 1979).For instance,the factor-proportions theory focuses on capital
intensity of acommodity andthe relative endowment of capital and labour of
a country, while economies of scale is related to the relative importance of
economies of scale in the production and the size of the country. Technology
theories are related to technical complexities of products (R&D intensity), and
a country's income per capita. Our further research will focus on the question
which of the three major explanations of trade isusable in explaining agriculturaltrade through identifying the keycommodity characteristics in agriculture
and the related national attributes. Thiswill give usthe building stones for a
design of aconcept explaining world trade patterns in agricultural commodities.A next stage of our research inthis field will try to estimate (not test!) our
concept, for instance byfocussing on the bilateral agricultural trade between
the European Union and Central and East European Countries (CEECs).
5.2 Government policy and international trade
Thetraditional tradetheories explaintrade bydifferences between countries. Exploiting these differences increases national and world welfare. Trade
policies create 'barriers'to exploit these differences and are therefore welfare
reducing. No intervention in international commerce isthe best policy unless
countries can improve their terms of trade (only large countries), there exist
domestic distortions orfor political reasons(income-distribution). A country can
only improve itsterms of trade at the expense of other countries and it risks
therefore retaliation that makes everyone worse off. With respect to domestic
distortions and income-distributional effects of trade, trade policy is only a
second best policy. Industrial policy that adjusts directly the distortion or income policythat achievesthe desired income-distribution isthefirst best policy.
Consideringtheseargumentsthetraditional theories becamestrong supporters
of free trade.
The modern trade theories based on economies of scale and imperfect
competition extend the traditional arguments for trade policy and add some
76
new arguments suchasthe strategic trade argument. The range of possibilities
t o improve the terms of trade isincreased because this ispossible in the case
of foreign market power and even small countries can improve their terms of
trade. Market distortions are inherent in imperfect competition so,for example, the wedge between price and marginal costs creates an opportunity for
government action. However, in the former situation retaliation stays a problem and in the latter casetrade policy is,again, only a second-best policy. So
the extensions of the old arguments do not provide convincing arguments
against free trade.What about the new arguments?
The main new argument isthe strategic trade argument. Unlike in the
survey we now take abroad definition of this argument. Strategic trade policy
isaimed at keeping some industries that are desirable from awelfare point of
view w i t h i n the country. Industries can be desirable because they generate
above-normal rents (rent shifting from foreign to domestic firms),they exhibit
external economies of scale or they generate high externalities. The question
iswhether this argument gives asystematic new reason for trade policy? The
answer appearsto be no becausethe rent shifting argument isvery dependent
on specific assumptions.A slight change in one of the assumptions changes or
even reversesthe implications of a policy. The benefits related to the external
economies of scaleandthe externalities argument aresubtle and hard to measure. In both casesgood policy requires that the government hasa lot of information to choose the right model. However, this information must be so detailedthat in mostcasesit isnot available.Theempirics alsoshowthat the benefitsfrom deviations from free trade aresmall andthat the value of an optimal
tariff islow (seesection 3.1.4).The potential for making policy errors through
lack of information or the possibility of retaliation by foreign governments
makesthe possible lossesgreater than the benefits.Therefore, Krugman (1987)
concludes that free trade isalmost never optimal under imperfect competition
but that it isa good rule of thumb. Feenstra (1995) confirms this conclusion:
according to his analysis of methods usedto estimate the impact of trade policiesunder imperfect competition, empirical results lend no support to astrategic role for trade policy.
The 'new' growth theories built forward on the modern trade theories
and put these theories in adynamic context. They deal therefore with the dynamic evolution of comparative advantage. Likethe technology gap trade theoriesthey stressthe role of technological change. Important isthat knowledge
will be generated in current activities.The current specialization pattern determines therefore the activities in which a country accumulates knowledge or
exploits economies of scale. Because the technological, scale and growth opportunities differ largely between industries the current specialization pattern
becomes extremely important for a countries' future welfare. The latter crucially depends on the assumptions whether knowledge spillovers are national
or international in scale.When knowledge spillovers are international in scope
the specialization pattern does not influence welfare. However, when knowledge spillovers are national in scope or when there exist important dynamic
external economies of scalethe welfare level inthe future isdependent on the
current specialization pattern.
77
The 'new' growth theoriesthat built forward onthe neoclassical tradition
assumemost of the time free knowledge spillovers and therefore trade policies
will not be beneficial. However, the evolutionary theory assumesthat a crucial
part of the generated knowledge iscumulative, specific and path dependent
and spillovers aretherefore local or national in scope. Inthis casetrade policy
may become beneficial because the gains at stake can be very large in some
circumstances.Government policystill requires alot of information about many
difficult to measure economic variables such astechnological opportunities,
knowledge spillover and external economies of scales. However, government
policy is more worthwhile than in the case of the modern trade theories becausethe gains at stake are larger.
78
REFERENCES
Acs,Z., D.Audretsch and M. Feldman (1992)
Real Effects of Academic Research: Comment; American Economie Review, Vol.82, No. 1,pp. 363-367
Aghion, P.and P.Howitt (1992)
A Model of Growth Through Creative Destruction;Econometrica, Vol.60,
pp. 323-351
Arrow, K.J. (1962)
The Economic Implications of Learning by Doing; Review of Economic
Studies, Vol.29, pp. 155-173
Balassa, B. (1967)
Trade Liberalization among Industrial Countries: Objectives and Alternatives, New York, MacGraw Hill
Bernstein, J.I.and P.Mohnen (1994)
International R&D Spillovers between U.S. and Japanese R&D Intensive
Sectors; NBERWorking Paper, No. 4682
Bhagwati, J.N.(1965)
On the Equivalents of Tariffs and Quotas; in R.E. Baldwin et al., Trade,
Growth and the Balance of Payments: Essays in the Honour of Gottfried
Harberler, Amsterdam, North-Holland, pp. 53-67
Baldin, R.E., and P.R. Krugman (1990)
Market Accessand International Competition:aSimulation Study of 16K
Random Access Memories; in P.R. Krugman, Rethinking
International
Trade, Cambridge MIT Press
Barro, R.J.and X. Sala-i-Martin (1995)
Economic Growth; McGraw-Hill, New York
Brander, J.A. (1981)
'Intra-lndustry Trade in Identical Commodities';Journal of
Economics, Vol. 11, pp. 114
International
Brander, J.A. and P.R. Krugman (1983)
'A "Reciprocal Dumping" Model of International Trade',Journal of International Economics, Vol. 15,pp. 313-321
79
Brander, J.A. and B.J.Spencer (1985)
Export Subsidies and International Market Share Rivalry;Journal of International Economics, Vol. 18,pp. 83-100
Branstetter, L. (1996)
Are Knowledge Spillovers International or Intranational in Scope? MicroEconomic Evidence from the USand Japan; paper presented at the 'International Conference on Economics and Econometrics of Innovation',
Strassbourg, 3-5 June
Bowen, H.P., E.E. Learner and L.Sveikauskas (1987)
'Multicountry, Multifactor Testsof the Factor Abundance Theory'; American Economic Review 77 (December 1987), pp. 791-809
Coudri, E.U. (1979)
The Pattern of Trade in Individual Products: a Test of Simple Theories;
Weltwirtschaftliches Archiv, Vol. 115, pp. 81-98
Coe, D.T. and E.Helpman (1993)
International R&D Spillovers; NBERWorking Paper, No. 4444
Corden, W.M. (1971)
The Theory of Protection; Oxford, Clarendon Press
Cordon, M.W. (1984)
'The Normative Theory of International Trade'; in R.W. Jones and D.B.
Kenen (eds.)
Deardorff, A.V. (1984)
Testing TradeTheories; in R.W.Jonesand P.B.Kenen,Handbook of International Economics, Volume 1,North Holland, Amsterdam
Denison, E.F. (1962)
The Sources of Economic Growth in the United States and the Alternatives Before US; Committee for Economic Development
Dixit, A.K. (1988)
Optimal trade and Industrial Policy for the USAutomobile industry; in
R.Feenstra, Empirical Methods for International Trade, Cambridge, MIT
Press
Dixit, A.K. and G.M. Grossman (1986)
'Targeted Export Promotion with several Oligopolistic Industries'; Journal
of International Economics, Vol. 21, pp. 233-250
80
DixitA.K.andA.S.Kyle1(985)
The Useof ProtectionorSubsidiesfor EntryPromotion and Deterrence';
AmericanEconomicsReview, Vol.75,pp.139-152
Dixit,A.K.andV.Norman(1980)
Theoryof International Trade; Cambridge
Dixit,A.K.andJ.E. Stiglitz(1977)
MonopolisticCompetitionandOptimumProductDiversity;AmericanEconomicReview, Vol.67,pp.297-308
Dollar,D.(1986)
Technological Innovation, Capital Mobility and the Product-cycle in
North-SouthTrade;TheAmericanEconomicReview, Vol.76,pp.177-90
Dornbush,R.,S.FischerandP.Samuelson(1977)
ComparativeAdvantage,TradeandPayments inaRicardian Model with
aContinuumof Goods;American EconomicReview, Vol.67,pp.823-839
Dosi,G. (1988)
Sources,ProceduresandMicroeconomic Effectsof Innovation;Journal of
EconomicLiterature,Vol.26,pp.1120-71
Dosi,G.,C.Freeman,R.Nelson,G.Silverberg and L.Soete(1990)
Technical ChangeandEconomic Theory; London/NewYork, PinterPublishers
Dosi,G.,K.Pavitt and L.Soete(1990)
TheEconomicsof TechnicalChangeandInternational Trade;Harvester
Wheatsheaf,NewYork
Eastman,H.andS. Stykolt(1960)
A Modelfor the Studyof ProtectedOligopolies;EconomicJournal,
Vol.70,pp.336-347
Eaton,J.andG.M.Grossman(1986)
'OptimalTradeandIndustrial Policyunder Oligopoly'; QuarterlyJournal
of Economics, Vol.2,pp.383-406
Ethier.W.(1979)
'Internationally Decreasing CostsandWorld Trade';Journal of International Economics, Vol.9,pp. 1-24
Ethier.W. (1982)
'Decreasing Costsin International Tradeand Frank Graham's Argument
for Protection';Econometrica, Vol.50,pp. 1243-1268
81
Feenstra, R.C. (1995)
Estimating the Effects of Trade Policy; NBER,Working Paper No. 5051
Feenstra, R.C.and K. Judd (1982)
Tariffs,Technology Transfer, and Welfare;Journal of Political Economy,
Vol.90, pp. 1142-1165
Freeman, C, M. Sharp and W. Walker (1991)
Technology and the Future of Europe: Global Competition and the Environment in the 1990s
Gagnon, J.and A. Rose (1992)
How Pursuasiveisthe Product Cycle?An Empirical Dynamics of American
and Japanese Trade Flows; NBERWorking Paper, no. 3946
Grossman, G.M. (1986)
'Strategic Export Promotion: A Critique'; in P.R. Krugman (1986)
Grossman, G.M. and E. Helpman (1991a)
Quality Ladders in the Theory of Growth; Review of Economic Studies,
Vol. 58, pp. 43-61
Grossman, G.M. and E. Helpman (1991b)
Innovation and Growth in the Global Economy; Cambridge Mass.:MIT
Press
Grubel, H.G. and P.J.Loyd (1975)
Intra-industry Trade; the Theory and Measurement of International Trade
in Differentiated Products; London, MacMillan Press
Harris and Cox (1984)
Trade, Industrial Policyand Canadian Manufacturing; Toronto, University
of Toronto Press
Havrylyshyn, D. and E.Divan (1984)
'Inter-Industry Trade and the State of Development'; in P.K.M. Tharakan
(ed.). TheEconomics of Intra-industry Trade, Amsterdam
Helpman, E. (1981)
International Trade inthe Presenceof Product Differentiation, Economies
of Scaleand Monopolistic Competition:a Chamberlain-Heckscher-Ohlin
Approach;Journal of International Economics, vol. 11
Helpman, E. (1984)
'Increasing Returns, Imperfect Markets, andTrade Theory'; in R.W. Jones
and P.B. Kenen (eds.).Handbook of International Economics, Amsterdam
82
Helpman, E.(1985)
'International Trade in Differentiated Middle Products'; in D. Hague and
K.G.Jungenfeldt (eds.).Structural Adjustment in Developed Open Economies, New York
Helpman, E.and P.R. Krugman (1985)
Market Structure and Foreign Trade: Increasing Returns, Imperfect Competition, and TheInternational Economy; Cambridge, MIT Press
Helpman, E.and P.R. Krugman (1989)
Trade Policy and Market Structure; MIT Press, Cambridge
Hirsch, S.(1967)
Location of Industry and International Competitiveness; Clarendon Press
Oxford
Horstmann, I.and J.Markusen (1986)
Up the Average Cost Curve: Inefficient Entry en the New Protectionism;
Journal of International Economics, Vol. 20, pp. 225-248
Hufbauer, G.C (1970)
The Impact of National CharacteristicsandTechnology onthe Commodity
Composition of Trade in Manufactured goods; in R.Vernon, The technology factor in international trade, Columbia University Press, New York
Jaffe, A.B. (1986)
Technological Opportunity and Spillovers of R&D: Evidence from Firms'
Patents, Profits and Market Value;American Economic Review, Vol. 76,
No. 5, pp. 984-1001
Jensen, R.and M. Thursby (1987)
A decision Making Model of Innovation,Technology Transfer and Trade;
The Review of Economic Studies, Vol. 55,pp. 631-48
Jones, R.W. and P.B. Kenen (eds.) (1984)
Handbook of International Economics; Amsterdam
Jones, C.I. (1995a)
Time Series Tests of Endogonous Growth Models; Quarterly Journal of
Economics, Vol.110
Jones, C.I. (1995b)
R&D-Based Models of Economic Growth;Journal of Political Economy,
Vol. 103, no.4, pp.759-784
Kierzkowski, H. (ed.) (1984)
Monopolistic Competition inInternational Trade;Oxford,ClarendonPress
83
Kravis,I.B.(1956)
'Availability' and other Influences on the Commodity Composition of
Trade;Journal of Political Economy, Vol LXIV, pp. 143-155
Krishna, K. (1989)
Trade Restrictions asFacilitating Practices;Journal of International Economics
Krugman, P.R. (1979a)
A Model of Innovation,Technology Transfer, and the World Distribution
of Income;Journal of Political Economy, Vol.87, pp. 253-266
Krugman, P.R. (1979b)
Increasing Returns, Monopolistic Competition, and International Trade;
Journal of International Economics, Vol.9, pp. 469-479
Krugman, P.R. (1980)
Scale Economies, Product Differentiation andthe Pattern of Trade; American Economic Review, Vol. 70, pp. 950-959
Krugman, P.R. (1981)
Intra-lndustrySpecialization andthe Gainsfrom Trade;Journal of Political
Economy, Vol.89, pp. 959-973
Krugman, P.R. (1984)
Import Protection asExport Promotion: International Competition in the
Presence of Oligopolies and Economies of Scale; in H. Kierzkowski (ed.).
Monopolistic Competition and International Trade, Oxford
Krugman, P.R. (1986)
Strategic Trade Policyand the New International Economics; Cambridge,
MIT Press
Krugman, P.R. (1987)
'IsFreeTrade Passé?';Economic Perspectives, Vol. 1,pp. 131-144
Krugman, P.R. (1990)
Rethinking International Trade; Cambridge, MIT Press
Krugman, P.R. and M. Obstfeld (1994)
International Economics, Theory and Policy; Third edition, HarperCollins
College Publishers
Krugman, P. (1995)
Technological Change in International Trade,in:P.Stoneman (ed.); Handbook of the Economics of Innovation and Technological Change, Basil
Blackwell, Oxford/Cambridge
84
Helpman, E. (1985)
'International Trade in Differentiated Middle Products'; in D. Hague and
K.G.Jungenfeldt (eds.),Structural Adjustment in Developed Open Economies, New York
Helpman, E.and P.R. Krugman (1985)
Market Structure and Foreign Trade: Increasing Returns, Imperfect Competition, and TheInternational Economy; Cambridge, MIT Press
Helpman, E.and P.R. Krugman (1989)
Trade Policy and Market Structure; MIT Press, Cambridge
Hirsch, S.(1967)
Location of Industry and International Competitiveness; Clarendon Press
Oxford
Horstmann, I.and J.Markusen (1986)
Up the Average Cost Curve: Inefficient Entry en the New Protectionism;
Journal of International Economics, Vol.20, pp. 225-248
Hufbauer, G.C.(1970)
The Impact of National Characteristics andTechnology onthe Commodity
Composition of Trade in Manufactured goods; in R.Vernon, The technology factor in international trade, Columbia University Press, New York
Jaffe, A.B. (1986)
Technological Opportunity and Spillovers of R&D: Evidence from Firms'
Patents, Profits and Market Value;American Economic Review, Vol.76,
No. 5, pp. 984-1001
Jensen, R.and M. Thursby (1987)
A decision Making Model of Innovation,Technology Transfer and Trade;
TheReview of Economic Studies, Vol. 55,pp. 631-48
Jones, R.W. and P.B. Kenen (eds.) (1984)
Handbook of International Economics; Amsterdam
Jones, CI. (1995a)
Time Series Tests of Endogonous Growth Models; Quarterly Journal of
Economics, Vol.110
Jones, C.I. (1995b)
R&D-Based Models of Economic Growth; Journal of Political Economy,
Vol. 103, no.4, pp.759-784
Kierzkowski, H. (ed.)(1984)
Monopolistic Competition inInternational Trade;Oxford,Clarendon Press
83
Kravis, I.B. (1956)
'Availability' and other Influences on the Commodity Composition of
Trade;Journal of Political Economy, Vol LXIV, pp. 143-155
Krishna, K. (1989)
Trade Restrictions as Facillitating Practices;Journal of International Economics
Krugman, P.R. (1979a)
A Model of Innovation,Technology Transfer, and the World Distribution
of Income;Journal of Political Economy, Vol.87, pp. 253-266
Krugman, P.R. (1979b)
Increasing Returns, Monopolistic Competition, and International Trade;
Journal of International Economics, Vol.9, pp. 469-479
Krugman, P.R. (1980)
Scale Economies,Product Differentiation andthe Pattern of Trade; American Economic Review, Vol.70, pp. 950-959
Krugman, P.R. (1981)
Intra-lndustrySpecialization andthe GainsfromTrade;Journal of Political
Economy, Vol.89, pp. 959-973
Krugman, P.R. (1984)
Import Protection asExport Promotion: International Competition in the
Presence of Oligopolies and Economies of Scale; in H. Kierzkowski (ed.),
Monopolistic Competition and International Trade, Oxford
Krugman, P.R. (1986)
Strategic Trade Policyand the New International Economics; Cambridge,
MIT Press
Krugman, P.R. (1987)
'Is FreeTrade Passé?';Economic Perspectives, Vol. 1,pp. 131-144
Krugman, P.R. (1990)
Rethinking International Trade; Cambridge, MIT Press
Krugman, P.R. and M. Obstfeld (1994)
International Economics, Theory and Policy; Third edition, HarperCollins
College Publishers
Krugman, P. (1995)
Technological Change inInternationalTrade,in:P.Stoneman (ed.); Handbook of the Economics of Innovation and Technological Change, Basil
Blackwell, Oxford/Cambridge
84
Lancaster, K. (1980)
'Intra-industryTrade under Perfect Monopolistic Competition';Journal of
International Economics, Vol. 10,pp. 151-175
Learner, E.E.(1980)
'The Leontief Paradox, Reconsidered;Journal of Political Economy, vol.
88, pp. 495-503
Learner, E.E. (1994)
Testing TradeTheories; in D. Greenaway and A.L.Winters, Surveysin International Trade, Basil Blackwell, Oxford/Cambridge
Leontief, W. (1953)
'Domestic Production and Foreign Trade: TheAmerican Capital Position
Re-examined'; Proceedings of the American Philosophical Society,vol.97,
nr. 4, pp. 332-349
Lucas, R.E.B. (1988)
Onthe Mechanics of Economic Development; Journal of Monetary Economics, Vol.22, No. 1,pp. 3-42
MacDonald, B.N.and J.R. Markusen (1985)
A Rehabilitation of Absolute Advantage; Journal of Political Economy,
Vol. 93, pp. 277-297
Nelson, R.and S.Winter (1977)
An Evolutionary Theory of Economic Change; Harvard University, Cambridge
Patel and K. Pavitt(1988)
The International distribution of technological activities; Oxford Review
of economic Policy, Vol. 16
Pavitt, K.(1984)
Sectoral Patterns of Technical Change:towards aTaxonomy and aTheory; Research Policy, Vol. 13,No.6
Posner, M.V. (1961)
International Trade and Technical Change; Oxford Economic Papers,
Vol. 13
Richardson,J.D. (1989)
Empirical Researchand TradeLiberalizations with Imperfect Competition:
A Survey; NBER,Working Paper, No. 2883
85
Romer, P.M. (1986)
Increasing Returns and Long Run Growth;Journal of Political Economy,
Vol.94, No. 5, pp. 1002-1037
Romer, P.M. (1990)
Endogenous Technological Change;Journal of Political Economy, Vol.98,
No. 2, pp.S71
Rotemberg, J.and G.Salomer (1986)
Quotas and the Stability of Implicit Collusion; mimeo, MIT
Schneider, J.and T.Ziesemer (1994)
What's New and What's Old in New Growth Theory: Endogenous Technology, Micro-foundation and Growth Rate Predictions -A Critical Overview; MERIT Research Memorandum, No.94-029, Maastricht
Segerstrom, P.S.,T.C.A. Anant and E.Elias (1990)
'A Schumpeterian Model of the Product Cycle';American Economic Review
Solow, R.M. (1957)
Technical Change and the Aggregate Production Function; Review of
Economics and Statistics, Vol.39, pp. 312-320
Spence,A.M. (1976)
Product Selection,FixedCosts,and Monopolistic Competition; Review of
Economic Studies, Vol.43, pp. 217-235
Stokey, N. (1991)
Human Capital, Product Quality and Growth; Quarterly Journal of Economics, Vol. 106, pp. 587-616
Venables, A.J.(1985)
'Trade andTrade Policywith Imperfect Competition:The Caseof Identical
Products and Free Entry'; Journal of International Economics, Vol. 19,
pp. 1-19
Venables, A.J.and A. Smith (1986)
'Trade and Industrial Policy under Imperfect Competition'; Economic Policy,Vol. 3, pp. 621-672
Venables, A.J.and A. Smith (1988)
'Completing the Internal Market inthe European Community'; European
Economic Review, Special issue,september
86
Vernon, R.(1966)
International Investment and International Trade inthe ProductCycle;
QuartelyJournalof Economics, Vol.80,pp.190-207
Vernon, R.(1979)
TheProductCycleHypothesisina NewInternational Environment;Oxford
Bulletin of EconomicsandStatistics, Vol.41, pp.255-267
Verspagen,B.(1992)
UnevenGrowth Between Interdependent Economies: an Evolutionary
Viewon Technology Gaps, Trade and Growth; PhDthesis, UPM,Maastricht
Wood,A.
Give Heckscher and Ohlin a Chance!'; Weltwirtschaftliches Archiv 130
(1994)1,pp.20-49
Young,A.(1991)
LearningbyDoingandthe Dynamic Effectsof InternationalTrade;
QuarterlyJournalof Economics, Vol.106,pp.369-405
87
APPENDICES
89
à*
'S'S
.s-B
c e
Q) O
+= S
01
o
o
•OÎÈ
o-o
3
O
"
ai
T> C
C «
01
ra E
o
E x
J >
18.
8
£ë,
-o
E
o
c
o
\J
O TJ
C 01 J=
TJ
C
flj
S 5S
S? -M f+J -Q —
o>
ai
. Q 01 <->
E
2
. y oi u
.c
•H
.e
Ç
.g-È o
£
o
c
l/l
ra c t
oi + :
oS
01
•H
»J C
V» 3
o O
ra
oi
'M
M
15
a
<ü
E
o
E-a
S-E
Sg
"1
§1
.a .-e
ra c
C
i/t
O
>,l
D
t; _o
(/) O
t/1
»/> c
o
ai
O JZ
•H»
i j
_>*-Q
C IA
i_>
O
o o»
O
u
^
**M-
w 'u
"K c ai
2
aï o.
C il
01
Dl
C
0)
u
i/t
ra
o
— +-•
ra c
. Ü <u
••-•
O
a>
i—
3
ro
c
•M
u
E
o
C
O
nj
N
Hvt
C "w
a>
"«ü Q.
D» i / t
|rö
E
g
o
TJ
c
vi
•a
o
o
Ol
l/l
3
o
<
cu
<u
Ol
o
E
o
SZ
r\i
ra
.-
i/i
O
c
TL
3
O
i/i
'E
o
c
o
ai
+-<
c
+-»
ra
i/i
c
o
o
ti
Ol
c
01 .« Tï
01 •S -c
c
01
CL
g o "
•1-»
a» D l u
.O
c "c
ft) a> j z
at u
> ai
^
ï o
c
oK
E 2
0i
-o
ft «-*
o o
X c
01 o
E
o
E
o
01
l/l
c
'ra
ai
ra
!s
'u
01
CL
Ol
o
c
o
01
01
TJ
O
O
ai
ra
c
ra
>
TJ
ra
01
+-»
l/l
•ï
c
o
01
-c
l~
•s
!"
fc
>
(0
CL
ra
E
o
u
l/l
o
e
3
O
J3
3
O
oi
c
oi
ai
o
E
o
SZ
ra
3
o2°
a>
E
o
c
o
l_»
oi
+••
c
ral/l
•H1
c
u Ol <u
c c ?
<" <" 5
" ü l/l
ai <u ™
c
oj
£ £ "s
.y
c
.Q .Q u
+J +J V
c
c
«
£
-§
<U 01 - g
o
u
oK
E 2
o ij
o! ïi %
M- H -
*^
si ra
•o TJ c
"S
c
o
Dl
C
ro
CL
Ol
a.
a
E C
»f
i/> 2
90
<£
13
Tl
O
3
O
u
X C
ai O
E V»
o ro
N
l/t
|rö
c 'vj
ai
"(D Q.
Dl i/t
E
ui
01
3
ai
TJ
01
•ft
c
a
5
o
c
^
z
zÄ «J
O oi
P n_ o;
3
01
<
X
u
UI
E
C
O
ra <
S
i|2
3
TJ
O
O
£•
+-*
01
ti
tl
TJ
ti
01
t
oi
o
c
sz
-o
c
01
E
o
l_
01
E
t/i
u
4-*
l/l
4-»
2
E
o So
C o45 t
o a»
k_
r
IQ
o
•o
>
O
Q- O
t H-
t
o
"
c
a.
3
o S
x t! 3
ï ™s
U
(U
o *- -g
ȣ u 45
o
5
S
•ff
u
't
S
"> S*a>
a.
"> —o
«î
Ui?
CC
01
O
+-• T3
T
10
Xo
F
ra
c
o
a>
k_
vj
a Q.
E
r
3
XI
o
a z
c
i .
•
V
T3
H •o
o
4-» E
i)
E
o
o
a
V
X
v
TJ
tt c
3 3
"O O o E
C <-" E "•
'T c
10 ai
<u
o
E
c
o
(O
Ol <u
10
TJ
C
3
XI
i*
xi 2:
o
a
ïl
tU V
ai
Ä ç
01
ai <v
rs j ï
i- c
x~ <u
O T3
o
u (O
TJ (O
O'S
t C
a ,/.
o o
— XI
o o
a. ö
Q.
CU.S!
2
<u o
"O *^
10 i_
if
E
E«
£ o
<o S
x
•O <U
C u
xi_* —
(O c
10 >,*y>s
o
ï.ï
X
u ra uO£>
X ai
^3 2
Ml"
T3
<U
O)
<0
ai
x
(O . >
i ; to
-Hl
+•»
c
10
>
ai
>
E
o
u
| | »i
i— *" (~ r"
2 ra 3 g E.E
4ÏÏ
2 E
It; "S
X
a
(O
•sll§
ra
Q.
Q.
f
cio <5
T3
ra
a> . t c +^
^
*~ r*
•c I
* .?" c •ai o •£ "5
><
ai °- rii~
o
«
.•e E
3 °
o
£
a —
Xw ß
" ü 3
.2* £ O
' ^
3 X -ä> <"
O S "= >
S > .2 ^
t • c - w« ra:
â^
l/l
~ c _ S
3
O
X
0) - o »fr O (O ra
i«
^
•*-• k_
« ï 2c
*^ o C O — O)
£ 5 E ü o j ;
«j=5 o-5
c
o
Is
x
C
O)
Q.
a
<
u
5-3
is il
S
a. <
E
UI
u_ <u
91
•e
çu x
S«
't! o
o
tl -§3-5
o si o>
(U -p
E «
o E
"D <u
°Oro
E
c
I?
. y -o
X
n *= ~c
o •-.
3
ra t :
u o cu o o
X! l/l
—
c
ra
-s
u ra 12
ra
Ol a i
ci
C
o o
j,
i_ <y xt
3 := O
ï
NO
~ Cl
+J
c
O-D
2
£ • cu
+^
o ra
° -9E
«" X 2
. ï t a
ai oZ cuÊ
S>
c
o
t/l
F
o
fc u
V
O
O
01
o
a.
o
r
Ol
eu o
3
01
•H» -3- c
t- 0) c
oi > S=
c —m Oi
•3
£ C *-
o .h
i- M_ -n
c
.=
*- .*:
°S
|l
"*
CL
e o
O u
C u
o x;
KJ l/t
X
ra
w
o
F -2 "S '+=
c ~ ra c
_; -g V dl
— O p "D
ra e .E •—
~ 1 " CU
Q. F t/1 k_
+J
o5
CU o
C Xi
a> ra
ai.-t;
ï *" .?o
.E CU
- Ol
+3 u
c _
£"S -§1
c
u C
c-5
cu 5
CU ü
5 o
ra +-»
_ c
ra 3
.ü O
ra O
u O J,
•x: Di oi
g "O ra
5 S "5
(D TO
"D
O
O
Dl
O
O
Dl
r
r
u
CU
<V o
c
ra •- . ï ra
O t oJ2 E
•»'
C O
Ol 3 m
•s
o o ••
ai :=
(O Q .
w
- ?
w
raraCU
o
E
• P C ?
T3
O
C
CU O C
E
a
-S.Ï
raF
-Q
_cu
3
o
X I T3
.. o
(U
fc
E !J5 ra
u
ra
TJ TJ c c
CU
c O
O 0) " O
Cl
cu
3
X
O O
3
L
XI
ra
X *»UJ O
v)
£
u
— o
S
£
C
O »-
§ra
k_
Ili
+•/
t/1
u o
IA .—
3
CU
CU T 3
™£
O
r
o
•" 2
>> c
Di ra
e
o cu
VI
a
<U
c <u
.l
'ra
2.E 5 S
I-o
n o t
ra £
o
o
^
>.
ra
"n
o
o
Cl
l/l
3
O
<u
CU
o
F
o
X
QJ
0)
Ç n,
0)-E
'
ro
Ç O H_
•o
-c ï»
O) ïrt
C *+_
o o
o tc
2.
Cl 3
c o
cu c
o,
oi
C
ra
u
5««
£: cu
««°
u ~
ra +-<
_ c
ra 3
.y o
+j u
c
CU X o
e r a t
<u .y -5
"D !H c
CU
ra
•8
o
IN
•s
c
<u
a
a
El
Se
< à:
92
-D
O
O
Cl
• ^
O
t!
CU
O!
a!
E
o
c
o
-n
|2
I r <u
S
2
CU
3 >
o
o
L. CL
ai
i/i
£
uo o.
a>
A3
Dl°
x "
01
H-
<u
o
o
cra
O.
E
Q. O
ClS
ai
?
ai o S
o c •=
c
o fe
E
o
Q.
O
E
5ô
*£ '^
O <ü
m o.
.ï E
E o
o u
c .y
8*
"ö
« S£ §.
Së
£"5
tO
ö .E .1
"5 .t! .-e
2SS
c
o
« o
u
x
Z 0)
^ "O
x: \z
C TJ _ ^
O <-• C
•£l O ra
•SEE
Q.
„
lof
V Tl
««•8
o g
C Ol
O 3
ai ra
ra $
tt-ï c
=
m
0 Tl .
0 1 O .£>
•— O o
o ai a
c"
u
CD —
i/i U
° É»
o u J2
at
2 °:c
+•* o
Ç
E
~°I
Q.' -o
E +=
o
2 o «
E ma
£
C
u
ra
,CU
_
CU
c t
o o
'+= z
? ra
CU >
ra
E o»
'- v
Vx S
£
ai
a» K
c
T3
^ i CL
t
c °
c • ra
F i ra ^
_> T I
V
O
C
o
ni
X c —
3
'+-<
CU
<u o E
TÎ
C -P o
5 ra c
2 ü o
o .-c
o. ai
00
—-
Lo . 2 <u
•I * i 1
5, o.--
22
°--o
cS
•.
c
CU
vi
a i .E
c .*;
ü o -o
y .2 £
M- £
t!
<U
™ c CU
O
U
a> —
— o
ra o .
E O
ai o«
'S .E - r:
eE
c
'•£ -H *« ül fcc
3
"i
"> -n
o s; o
c a t:
E TJ
(11
—'^
ra c
> ra <u
MOJ +•'
ra c
E
o
c
o
>. ra
«Z TJ
°>
ai
S
r
5
>
'+J
ra
>reproduct
scialization
imexchang
o
^
CU
ra o
Sä
>,
X
*QJ + J
vi i £
Qj <U 3
X
E c
o .2
£|?
O Tl T
c + j ra
n
b
<c y
<" - 2 T l
Êi
O
u 3 e
•-
a i ' u cu o
CU
Tl
o
E
•M \Tt
c
o +-• c
sis
O T3
C <U
o y, i=
CU
•" ra j ?
cu
ra 0
3
XI
'i_
T3
o
OJ
Q.
(TJ
E
o
CU
Ô
£
CU
ra ó ra
ra
o Î ra+-»
£• **-
l/l
3
1/1 • o
»-
Ó
Tl
TJ
1/1
o
c
nj
T l
CU
E
CU
"O
nj
CU
cü
o
•H1
+-»
t/1
CU
c
e- 3
+-< I3A
«/i T l
ra Dl
u
CU
"fa
"O
o
o
Dl
O
«/I
0J
V
'+-*
<D ns
Q.
E
E
o o
u
nj
aj H t;
UJ o
i» • a
(U
a.
E
o
"O
ro
_0J
c
ra
40)
"O
t/t
ra ro
o
_<u
u
i/i
*•O
in
CU
+••
>*
•M
l/l
O
>,
E
£
o
c
o
flj
C
• ~
OJ
c
c
3
o
ra
E
o
CD 3
MCU
TJ
CU
ai
a c
V
CU 'o
ai Q . JÜ
c XCU fD
ra
u
x:
u T3 m
X
c O
CU
c
c
o
ra
N
•c- -c
ai
01
wi
ïï x:
ai ai
ra xi
o
o
c
3 >,
^*
- Era
ui
E
o
c
o
o
'ra
ra ra ai ' v j <-»
u
O
XI
<u
<>i
s
l/l
-o
c
eu
a
a.
<
z
<
<
ui
SS
Z
O
<
u
3
Q.
cl
c
'raai
o
vi
CU
3
O
ai
TJ
ra
+-»
E
o
Si
a|
fD O "
Ll_ <U
a g,
ra
_c i y o
5J2
93
vi
o
'»- '-P
o o 3 .t;
3 >
o
Ol
.E ï;
c ïï
il
S &.
5
rara
>
.2 E -
O
4-» •— «A
c
o
-
oi 01 + j
•-
ra _
^
a> ai .2
i ?
'S ^
t __
o mo
E 'S «r
g .ai
• >§
<
u
D
_
-c ?
m 2 "O
~«ï
0) ^
> +J ra
T:
il
t! "o ~
z
-o
•BE
o>
22
** °- «
TJ C
1
£
o 2
E £
O'S
— O
O +=
oi
c
o
2 E
>
3 ->•
cE c« 3S
.2 .o 8
01 o ,
C
n
is 3 2
l/l
»-
•8.3 SS
2 E
2 oi oi
£ g ^ •^ —
»« 5o cc t f. . =O)u ;
« 5 . 2 o° T>ï ï
c t
« o
mo
° -c c
Ö-* o
c S +:
£ -o s
o c oi
g. ra x
3 2- o
./> ra o .
JS oi o.
c o oi o >>
E c u
ff-5 C
2 u3 =
c
E -^ .2 >A
oi ra £ 2
*" E
Ol
> c E OJ
o Cl o
u •i= 01
O o»
vi Q.
.ï E
E ou
o
c .y
8*
«ö
™ SE g,
E o
dj O
u
C
3
OJ
H -
g
ra
c
oi -P
o
• O M-
ra o
ra
c
^ <" ra
+•<
O
01
3 ra • -
>
"O O. E
Q.
o
E
ra (_) <y
£
CU T3
S E
c
o
V
F
o
c
o
CD
c
_0>
V
'x
LU
O
3
•n
01
4->
c
fc.2
T)
c
ra
o oi.S
'S «i
1c
•*-»
Wl
Q.M-
4= t
°^i .0);ïvi +s^
oi .+;
II
g Q. Ol
°Q..E
EE
i:
_0j ra
ra c
£rSi
'Ë g .!
çu ra
> *#2 «
E 5
S5
E-g £-
gis
0 ra c
v c, °
°-o§
—2 =
2 S - £ ra
oi ra £
oi
0
Q. ra
&•* E
3 oi
11
Hz i i
01 i
oi
a.
a.
<
94
>u
o
Q.
c
01
E
3
Ol
O
c o „
g-S.2
ra ra S
1
1/1
^
J! " H -
wi C -—
<l» — Q .
=0*
ra c
C
m
wï +-« .—
T3 !P
«01 ?c
JT
O •=
E
.. p ^
•- o>
•o
03
fe
"So
ils
c _*
o c
01
4-» -1— Ol
1
Ä >" E
oi o 2 S «
Q..2 o»
E t-S
,—-. k * - (L> — o ra
Ë~
oi " c
•*-*
.ü (-
•c 2
oi ra 3
C M-
<u ' S
+J o
c
-1 •'l
f
2-o
ra *Ü tt -Q S
E £ 13 « E
ra
u
F o o
i
u ij.
c
k_
_c — -^
E ^
~ o
§.E
V> ^ j
oi : =
Ol + j - °
SÄ™
01 „o 2
'v
«
t E
3
T3
—o
E o
E-o S
ö
Q.'01
VI
y
E
E
E
UI
3 a;
ai ai
ai . *
C Ol
Ol • = • - y
o i - 9 01 i t
Ol £
o
k_
Ol
•£•
•2«
' Ol
•" *>
o
E
ra o
!-i •*
c
"5 .S
ai
T3
2 «
ra
O _ro
c
o
c
3
O
E
^
ï â
fc un
<U O l
£Ç "O
_0J
^-,
Ol
•*
«i r
ra + j
3 . a "o +;
o . 2 >s 5
3 _
.
o ro "v>
-Q . t
t
ui - *
">rä
c
o
^
ra
o
oi
J2 C L °
_ ro X>
7} u
_a> c
£
* .5
s
Dl
01
•c
L." C
3 3
O O
oi
c
O oi
'a. 3
ro u
u uro
00
01 00
"O 0*1
II
s
*
a
c - i=
'k_ v)
o . ra
3 01 _0I
TJ
2 *•
S,
°
f UI
01
ro
b
01
'Ë 2
oi o
_>» OJ
^
i-
c
=
Ç .c
ai ro
o.ra
w 3 0» _tt)
ro «U t i TO
{IJ •—
Cl
o
ro o i
-o
•o .
o c
o ai
01
ro C £
c
•-= 5
ra-£
«
"S-fi
al u Ol 01
st.
O *"
(-» M-
ai §
.. .ï
O £ b rä
f3 "5
o oi E«S .E y
S"o
oo
u
C
El
«s s^ c c
^
u
3
3
C
E-°
z
o
ra + j
T3 ^
O Ol
=
E
ai - S . - - -
• fN
VI
£ <"
c
"O fü (U ^
c +-* — ^
2o
c
ro ai
3
!fc O
•5 oi
• o -•F.
b
S 43-2
tï c fl 3 ^
<D ' Z
?«
oi ro
- ° E ^ ai c
rarôc - Ü T J 8
JS
01 t i
a
~J> 3
.c o
°!l
oi-c
ro
5 ou
o
•o S
c o
C "S *~- 01
.ai
_ j*5- oiN oi
rara
•e1-3
Ol
I!
a
K ü
5 ° £c
c
"o
c
1||
JP 01
C 3 ij
3 -C ro
_ro >-3 ra>.
Si V ^
E
•" C L - g
û
'•" O)
oi o
u
•S ;£
-»
•S E § _ro
O)
•o
ra §
o
Ol
-c
o
«
3
vl
• = ra S
§ £
c
• TJ
ra
lg
«
CU .Q O l
oi .+;
•gS.E
8,-C
3
« 0.-8 o m
û
So
»Bio
J S o E S
c ai
Ol oi
01
Ol
"O
ju
o ja -2
ti c ra
1- «1 Kl
r-§s-g
ra «
C .2
ra
'Ë g
Dl (U
TJ -C
O Ol
• 8-S
o; ü ,_
3§
'
o - J ai
-C ,
=5 °>
E
£
Ei -o S
S-8 £
i TJ
ü O
!t O
ai
s §I
oi-*
•o —
ai ra
ai
ro
c > E
ro o ro
o .E
a 1
" - C H_
Q. O l - C
'"> 5
"
o
ai c
4^
o TJ ro o
*" 7. •" m
-~Ol/l ^vi^K,O l
c _1
c
O
- o ••:
•K
ra -^
c « £ K.o.
5
™ rä 5!UTJOl TJ
c
^
o.
o
_
-a E
o ai
o
ra^3
88
j3
S TÎ E ï
3 Ol C <-*
2 Q. O 01
E 5, TJ JD
Ol Ol
x: .c
°«! t3
c o
01 —
O l Tc3
?
E «i
o .c
01
ro
ro01
c
ai ra
>< O
01
ui
H-
TJ
O
O
Ol
l_
O
ti
<£ £
95
-a
= -P?.™
•= o <u +; o o
v ~ c £ „
ai c
! t oO > 8 -Ë
01
l/l , Q
•^
en
ra
c
o :
oS
o ai
l/l u u u
.J. - a cn.
aai o
a oi c
> 13
Ç JD
ra g
<u c
OJ*
XI —
<D ra
s o ira
3
O
t
l/l _Q
•. o
o ai
en en
- = "O
Ol
c
"S
ra
'^ o
ra c
c .*
o OJ . £
O W i j
O
il
ai — *. en o
+-• ra . t
o ai
_- C J2
— c
ai "6
ra .
—
~
8.1
lit
vT *" - 0
•E.I
TI
i/i
ra
<u ra
u
ai
0)
3
t;
g
3
•o
O
^
a
o
c
o S
C ro c
°r — "a
o -= ai
2 o«
8c«
2 c-o
U — OJ - - .
ra 3 -a pi
D. -p
Q. 01 -— Ol c
O -O
c
3 X So
-=
ra s-,
a; • - ô ö
. u 1i t
g^
11
.-g
^ •—
O (U
-S3 -i 3
•o ï
o 3
a» 3 cn u
If
i/i
c
o
•=
>
c
•M +•< •—
v> O
wi r-
«>
rt>
(71
01 - ^
• - ai . ~
Kî "ci en
c c
« °£
î t ai
— +* + J > X I
i/i o
XI —
O 0J •S "C cn
I
*
ra en
c "O
0 ü
.s* Is*
<u
„
- *>
c e t ;
0J fO 3
Ol-C -o
O ^ O
Ol
I'
»i
Ol o
"g>C
t .Si
ra.Ec
\J
ip
û
O 3 ' <u
C U Q.
C m
21
"et i/i
><
CL
CL
<
96
.1 ?
O u
Û_ +-i
«
^
_
^3
C
s-i.
o
-Ü c = ai
CL c n .
Q
oS
Ol £ t
en * ;
«
o
o
S
^2
^ t
O u
JZ ra
5
QJ
•M
OJ
•M
C
o
u
O
E
o
ti
0
•D
(D
f5>
vi
O
> c
O
— OJ
5
J
•a
^f
o o
o n C
ai
l/l ai
aj
(U JZ OJ . c
+-»
+-< cn
OJ
ai
t/1
OJ
§
O
C
Ol
(TJ
«/i
a
«3
oc
? ra
o
c
U
\j
ai
•H
i+-
•° >> .
g. $ ?
<= ° E
gl «
C
nî
Q . <u
i> *-
; cn-Q
u
1
^
ai —
jz ra
— 01 c
; x> 5 r • -
iE ~ 2
; S tt at
OJ !•
ai en
x "e1
C m M - (Ü
o o c!
3 .h O
O O >r
J3 XI ° -
D. E
ra ai
c > J3
o o u
'•P =
£
a> <ü C
"g^-S
c g-,
s - Dl
"û c
d) ' F
11
= >C
c
1
Sg ?ë
01 —
. c ra
•M u
|I
°3
S e "g a
2 ra Q . c
cz v , o
S m » 3
ai x i ra x i
l/l
"i§
o
=
5&
„
3 _Q — i/i
(U
en .
"D £
oi ai
wi
i •— ni x i
f'SfiJ
«*» fD
T 3 Û)
C
XI
c
S
XI
C
<v
£ m
ni <U
-5:
ai
;a
C
^^
3^g
Ol c
m 3. »
!fc oi
ai . ra
- ^ : T ] i_ SI—- r u/ «
+cj +/
XI .— +J C u Ol
"Î cn
£ ra
Ol vi
i ê-2
c
3
O
u
c
ai
« - o en
"° en c
c +:
o jo
S S '•
o o
5o
E S-S ra
2 g-cnj;
o "Ö >,
U (D
OJ
•M OJ
cn b .Q
~°
m 5"ï 2
a
' P !S ^
S'5 3
C T3
^
C
u
*
•g-S 2
S Sg
rä S 5-°
'enjü
— o
u ai
ai >
.u§ •O§ra£
çg
a. E
o ",_
cn o
s.«
ai fe >
o t
t*
£ "O (ri ai
i/f
"D
i/i O
_ c O
o ai en
a . *;
is
a
f * ?S <" =. °
II
5
'u
£ 2 £ S •/ a
» ï o « 3 .g" S
. y .D — S
8gis
c ?c
^ c°
OJ
^
ro OJ
"Ö C .OJ
2
ai cn
>
"O 'P
Çc ra
I
cn c
= o u g
° cg
m
^
sf
t i C fc 3
ai o o o
ai cn
c
_o ± v _ en
o '**- 3 §\£
; i s,
™ ra
O
c
t Sg
X 2
£ <I
Z S
3 O U
< £ ui
252
ra
o
o
S1
o. ai a
S.73 c
S
ra
ÄOP
o •Ù s
Qj
i/i . s
0
il- "
ra
c
.2 g
m« S ï
ra
ps
S S E «!• O "1
3 O»
v*
c c ra o;
_
O
QJ
1- X I
01 - ^
, o
o g1 ra
>,
oi P
01 E 01
o 01 2, 01
Dl
O
01 c
- E2 >
.2 E «ü JS
—
ScJ s
<u !ri
£.
o> +^
O)
D. +<"
m 01
r
W
J S £ g ~" S 5
!É
01
•ST J3 73
_L
E •/>
tt o>
u 01
? "O
ra
g
<u c
D l - *
•o —
01 ra
2+=
~E
<u
*^
Ol
c <u
01 a i
3
.vj
t i
1/1 t/ï
i1 P c ^
0
3
.2- -o
c c
'T
ra
o '~ -S*
• - ™ .y «
E 01 râ ï
c if E
ra
01 ' 0
xi
XI
1/1 v*
Q-
SB c
U
«I
D - . Ï
?»
c
E-S««
SS
M-
3
1/1
c o
o •- E ?
">
Sc e E
./, -*
"ra
f
0 c
01 o
TJ -J3
v
01 c
ra S .2"o
* .E
.2 t N ï
I2
•—
C
01
ra
ra
.£ T3 ü Q.
ra o.
U
•5
1/1
ra g
o
_ro XI
01
S
c
ra ^
.£
-r =
$•
«i
01
c
o> o
E
1/1
S S<"
VI
r0
Dig 1 .?
C
E
rn
01
_0J
c
I'S
ra ra tî
u
Ö5
3
0
ra 1/1
0)
tu
ra — x i
-O XI — Di
>> a i - h
>/i Ol
g
01 s oai—ra E
E
F
"râ
3
(U > v
l_ 4-"
U Q;
0
Dl
>
XJ
«
.2 E
E M2-rä
- ^
ra c
SE
u u
cc ra > . ï
>.«
ra
' 0101
°-"S
ra .c- ra
T;
N
C Si.
•D
• =
° oü
o *
E 01 ra 3
ra a» c "5
Q.
^
M
q M- o -c >,
1/1
• ~ Kl . . L
._
tf 'ra
X!
t ^ !?
t!tt
cIr x<" óc:
ai _
Q.XI
1. O
X
01
0 XI
1- 01
01 u
- É * ! -° -5
i5ra
E
o <u
2?
ra
o J:
•ö E v • -
12
O
Ol
3
ra o . D
'E « b
ra
>i
01 +-»
>> <ü
'« g» O
'to
Ol
(IJ
Oi
£ 01
c
a
U Ç
*+- ra .b
o _c wi
^ u ra
S x tu
•D 2
c *"
IT,?
C r a y *
01
ra 01
v T3
ra—
ra
01 u
0 QJ
«•s
•èrS
-o _
\j
çm
5S
01
Dl
So
CT
ra x i
^ sts
E^
ra TJ s O
C
O C ra
oo
Dl
C
3
c:
> H u 0
X O TO3
ai += Q-
o
c
Ol
Dl
00
01 0 0
- o <J1
J.
XI
ra nj
o. "
u rj
\ \
^
c
tu
a.
a.
<
Ol
• - 8w tt
ra
c N
E0
Q . Ol
Dl C
>
0
> TR c . E g o
•*= ¥ 01 O W O)
ra .E £ " O <u ö l
raE g
fc - M T 3
O <» c
o.
S
K Z
H- <
< Ul
ss
11
M-
U
!r! "i il
01
c
Q.
•? S, 01
E » ai „,
ra _oi c c
g.•£5
- öi XI
•&s-S'a
c ï; ra
Q
•o
.n
f i rs i3o«f •+:ra> .E0)c
s:c XI ^ ra ,_
X
ui
u
Dl
l/l
Z
C
0J O
o
H H- E
y s>fc
J U IA <u
t 3 E x
g O ra ra
S
1/1
Q l v
+J
Q.
N
O
ri
3
ra CT
01
u-
97
"a ra
a* ai
TJ
C
£
O ni C
•° E
-2
ü
ra ai
E 3 c o E « ï
§- > , o u 0 3 = 3 :
ü x i z ; ra >p o oi
-i°
o
2c
o
.£
o
%^
~ 2
fc*
a i "O + ; ai > , - § ra - > ;
c
O vt
.— 01
•M .—
t c •= ,
o> ra E f ,
CTÏ
ÎS * £
'«•g
C
S uc u
.8s s S
S ra £>
<" E
g 1 «
•5 « « gai — .— 3
3 .
•D '
C
o.-s.a x: o)
c
3
" S8
ra + : 01 5 2 - ^
ai
— o>
'5 S
O" »<U c
Ol Ol
S "O
Ç V
oi ra
o
c
c
—
Dl.*
T> —
ai ra
e-s
S§
o -r; ai
c m fi
)Z c
°
C 01
<U O l
ra
E ai
=
« •» - o
E o> ra
*/» 't_ H
•= 0)
3
r;
CT
01 ra
•K + j O
c
=
•ra
o.
o
c
>.° g *
ra
>
ra
*<
1-
O = 3 o O 01
^ c ra er c — ,/,
t ; . E E t ; -E ra .9!
3
n "O
3 iP; >"
E <cJ
"O iP; X
o .2 S ° — O 3
O - X I . £ Q . Ü Ji ö
gE
o <u
* .E
01
> - 2
01
•o
E
« g1
ra|
00
c .E -c
ra o.= fe
-g>
<u 00
en
01
ö
o _
„
? -C r
>
w
.ï ™
<D O
01
TJ
(U u
Ol 3
C3
Ol
3
T3
O
5
o
c
^
-o
c
o>
Q.
O.
<
98
•e
Si E
£ "-O-B
ra « = S . ^
L
O
o
E 0.
wi "O
X!(IJ (ü
-ï
t -w g" c g c E
•g ra 01 ro o 3 n
x x : f- . e *- o il
01 + ; . b * : a i u >P
i
c
(Ü
i/i
u
T3
C
raju
"5
g
c
o - Ï ^ E
^î u u ra 3
"' <"
-• c
c Su
o
O i/i
oi u
c
—
ra
c
0; rô
(5 ° (U S O * ;
i_ C w Q. u O
c fl»
•*-»
.—
(D ni
•M »-
•— ra (-1
o
a
-=
.2
o t •£
°-Z- o.
"râ ra >
3 ^ S'a
"g £ S s
3
•o
c
•o
c
u
r
•" o
c
01
ra
*4-
J j- N
0
t ^
W> y
ï o j
ra ro 01
E EC
01
01 0
ra
•*-»
3
fO
0
T3
C
a. o>
c
ai
£
Ol
TJ
rara*;
ra
Cfü 0
D
C vi
« o>
- ' ü ra
C 01 o2
ra 01
. h +j . r
fc-
C
5
<u
u
01 o
Ol 01
.c ra
4-
ai
3
.~
"2
.E "5
Q.
01 o ü-
01
t-i ai
C
0 ^
x:
0
U Q.
a.<
c S:
ra 01
o
o
01 ~
S ra
c KS
SM
-a» _raN 01
-. 0 1 + ;
. , »M
T:
v
c
+j
aj
T3
01 > ,
0; . 2 o
O
C)
x a. S
01 —
x : a.
O .E
•0
El
>.
ra
a
c
(U
a3
>> ™ o
ra o
^
o>
CL
" _
°f: a *
>,
I &ï ° s!-ë
ai o
« tl
S
g
E
'P
a.
o
•§ 2 S ra
CT 10) c
;>
c
o
•=
ra
_ -M
3 O
•M
?11 i i
: = o>
O"
£ .2!
E ra
V
o>
>.°
'•»
10 5
o
<u c
Ol
Oï
"O
U
2
TJ
c
3
•M
vi
tl
01
3=
"u
C
:=
LL
5
O
• c ai
ê ra
ra (U
^
•4O
01
O
.C
1
k-
O
H-
^
•»-*
C _i.
S ä
tt o>
«
Ol
C Ol
'Z, i
«o
dl
c
CT.*
T>
—
Knowl
nation
lovers
ai ra
C IV
V O)
E "°
£$ JU
3
S> o
c
ra
• =
ra
Oï o
"O • =
± ra
5£
o <u
* .E
0 S
.i
C
01 +J
c
—
1
jresis not lin earity slated
• ^ Ol
+^ .—
3 i0 Ol
c
ra
S
0 -C
u
u
versth atare intra- ational
pilloveTSthai : areii ernatio
m
c v)
O
c -g
ü
Ö *"
-M
= S
=0
Q. fc
•* O
oi «"
C
o> •"
Q. S
E o
* ra
£ i/» - , 01 LO — 01
CS ra Q . Û - c
Î
Jl » ?
O
-theree:
nature
nallyin
- levelof
withgr
•£•
UJ
_
E en
.t
o>
01
TJ
'5 Ol
ra §
Si.o
#>
|MOU
(166
mpoj
<u «
*
a o
co
ai oa
TD o
i/» ^ ,
«Û IA
ra (ç
Ol «
•H1.3
•o
<uc
11
* a
tri
X
'T}
C
«1
a
Q.
<
O
K
S
UI
99