C. Fundamental economic factors affecting international trade

world trade report 2013

112

The previous section has shown that the future of trade and economic growth depends on a range of factors. Predictions may change depending on how each of these factors develops. This section discusses how the fundamental economic factors shaping the future of international trade – namely demography, investment, technology, energy and other natural resources, transportation costs and the institutional framework – are likely to evolve in the coming years.

C. Fundamental economic factors affecting international trade

II – Factors shapIng the Future oF world trade

113

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

Contents 1 Demographicchange 114

2 Investment 134

3 Technology 152

4 Energyandothernaturalresources 166

5 Transportationcosts 179

6 Institutions 192

7 Conclusions 207

Appendix 217

Some key facts and findings

• Demographic change affects trade through its impact on countries’ comparative advantage and on import demand. An ageing population, migration, educational improvements and women’s participation in the labour force will all play a role in years to come, as will the continuing emergence of a global middle class.

• Investment in physical infrastructure can facilitate the integration of new players into international supply chains. The accumulation of capital and the build-up of knowledge and technology associated with investment, particularly foreign direct investment, can also enable countries to move up the value chain by altering their comparative advantage.

• New players have emerged among the countries driving technological progress. Countries representing 20 per cent of the world’s total population accounted for about 70 per cent of research and development (R&D) expenditure in 1999, but only about 40 per cent in 2010. Technology spillovers are largely regional and stronger among countries connected by production networks. In addition to the traditionally R&D intensive manufacturing sectors, knowledge-intensive business services are emerging as key drivers of knowledge accumulation.

• The shale gas revolution portends dramatic shifts in the future pattern of energy production and trade as North America becomes energy sufficient. Increasing water scarcity in the future in large swathes of the developing world may mean that the long-term decline in the share of food and agricultural products in international trade might be arrested or even reversed.

• Ample opportunities exist for policy actions, at the national and multilateral level, to reduce transportation costs and offset the effect of higher fuel costs in the future – improving the quantity and quality of transportation infrastructure, successfully concluding the Doha Round negotiations on trade facilitation, introducing more competition on transport routes, and supporting innovation.

• Improvements in institutional quality, notably in relation to contract enforcement, can reduce the costs of trade. Institutions are also a source of comparative advantage, and trade and institutions strongly influence each other.


114

Various economic theories use fundamental economicfactors to explain why countries trade and how tradepatternsevolve. InDavidRicardo’stheory,for instance,technologicaldifferencesbetweencountriesdeterminecomparativeadvantage. In theHeckscher-Ohlinmodel,relative factorendowments (labour,capitalandnaturalresources)shapetradepatterns.Thenewtradetheorypredicts that countries with larger economies – as aresult of growth in endowments and incomes – willdevelop an export edge in those goods consumed inrelatively greater quantities in the home market. The“newnew” trade theory identifies tradecostsasakeyimpediment to entry into trade. Others argue that thequalityofacountry’spoliticalandeconomicinstitutionscan be a key source of comparative advantage. Thissectionalsocoversfeedbackeffectsfromtradewhich,in turn, have an impact on the fundamental economicfactors shaping trade. Trade can lead to technologicalspillovers, for example, allowing countries with lesstechnologicalexpertisetoacquiremuch-neededknow-how. Engaging in trade can also help to strengthenpoliticalandeconomicinstitutions.

This section shows how developments in demography,investment, technology, energy and other naturalresources, transport costs and institutional quality arecapableofchangingtheoverallnatureoftrade:therolethat individualcountriesplay in international trade,howthey trade and what is traded with whom and why. Itexplores possible future scenarios for each factor andconcludes by describing their potential impact oncurrently observed trade trends, as discussed inSectionB.ThediscussionforeshadowsissuesthatcouldbecomecriticalfortheWTOaswellasforinternationalcooperation in the future–asubject thatwillbe takenupingreaterdetailinSectionEofthisreport.

1. Demographicchange

Theworld’spopulationisexpectedtoreach8.3billionby2030and9.3billionby2050.Mostofthisincreasewilltakeplaceincertaindevelopingcountriesthatareintheearly stagesof their demographic transitionandwhichwill seesignificant increases in theyoungworking-agepopulationofbothsexes. Inotherdevelopingcountriesandinmostdevelopedones,thedemographictransitionisalreadyinitsmostadvancedstage.Fertilityratesarelow,resultinginanageingpopulationandinashrinkinglabourforce. Insomeofthesecountries, immigration islikelytobethemainsourceofpopulationgrowthinthefuture. Furthermore, education and urbanization areadvancingeverywhereintheworld.Theobjectiveofthissection is to show how these long-term demographictrends are likely to affect international trade patternsthrough their impactoncomparativeadvantageaswellasonthelevelandcompositionofimportdemand.

(a) Thedemographictransitionandageing

Theworldisexperiencingdramaticchangesinthesizeandcompositionofitspopulation.Thesearetheresult

of theso-called “demographic transition”–aprocesswhich involves first a decline in mortality rates andthena reduction in fertility.Countriesareatdifferentstages of their demographic transition. The datapresentedinthefirstpartofthissectionwillshowthatsome countries are ageing quickly while others arereaping a “demographic dividend” from a youngerpopulation. These trends are likely to have an impactontradepatternsthroughtwomainchannels:changesincomparativeadvantageandchangesinthelevelandcompositionofimportdemand.Thesecondpartofthesectiondiscussesthesetwochannelsinmoredetail.

As clarified by Lee (2003), a country’s demographictransition occurs in four stages. In the first stage,mortalitystartsdecliningwhilefertilityremainshigh.Inthis phase, mortality reductions mainly affect theinfantpopulationandaremostlyrelatedtodeclinesincontagious diseases spread by air or water, and toimprovementsinnutrition.Sincemortalitydeclines,thepopulationincreasesandbecomesrelativelyyounger.

Thesecondstageofthetransitionischaracterizedbyadecline in fertility andan increase in theworking-agepopulation,astheyoungerpeoplereachedadulthood.1Duringthisphase,agrowinglabourforceandincreasedsavings can potentially boost economic growth,generating a “demographic dividend”. Next, ageingleadstorapidincreasesintheelderlypopulation,whilelow fertility reduces the growth of the working agepopulation, thus increasing the young- and old-agedependency ratios.2 The demographic transition endswhen the total dependency ratio is back to the pre-transition level but where the young-age dependencyratioislowwhiletheold-ageratioishigh.

TheglobaldemographictransitionisapparentinFigureC.1,whichshowspastandprojected fertility ratesandlifeexpectancy.Thedecreaseintotalfertility isclearlynoticeable. The Economist (2012) reports that almosthalftheworld’spopulation–3.2billion–alreadylivesin

FigureC.1:World fertility rate and life expectancy, 1800-2050

0

2Chi

ldre

n pe

r wom

an

4

6

1

3

5

7

0

20

Year

s

40

60

10

30

50

80

70

1800 1900 1950 2000 2050

Total fertility rate (left axis)Life expectancy at birth (right axis)

Source:Lee(2003).


115

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

countrieswithafertilityrateof2.1or less.Conversely,life expectancy at birth has followed a clear upwardtrend.Thesedevelopmentsindicatethattheworldasawholereapedademographicdividendinthe40yearsto2010 (The Economist, 2012). In 1970, there were75dependantsforevery100adultsofworkingage.In2010, the number of dependants dropped to just 52.Huge improvementswere registered not only inChinabut also in South-East Asia and North Africa, wheredependency ratios fell by 40 points. Even Europe andNorthAmericaendedtheperiodwithfewerdependantsthanatthebeginning.

Since 2010, however, the world population hasinexorably started to become older (see Figure C.2).Its sizewill continue togrowbutat a rate lower thanthe historical growth rates of the 19th and early20thcentury,asshowninFigureC.3.

Countriesareatdifferentstagesoftheirdemographictransition (Eberstadt, 2012). Developed economiesbeganthedemographic transition in the19thcentury.Inmostdevelopingcountries, thetransition laggedbyalmost a century. However, it progressed much morerapidly, thus implying that fertility and populationgrowth rates are converging relatively quickly at theglobal level (see Figure C.4). Lee (2003) notes thattheprocessofglobaldemographicconvergenceofthepast50years is inmarkedcontrastwith thegrowingeconomicdisparitiesoverthesameperiod.

However, these general trends mask noticeabledifferenceswithineachgroupofcountries,especiallyin fertility rates. Within developed countries, mostEuropean countries have very low fertility rates(forexample,Germanyat1.36,Italyat1.38andSpainat 1.41 in 2010) but some others have higher rates

FigureC.3:Size and growth rate of the world population, 1800-2050

0

0.4

% g

row

th ra

te

1.0

1.4

0.2

0.8

1.2

2.0

0.6

1.6

1.8

0

3

Bill

ions

5

7

2

4

6

10

1

9

8

1800 20501950 20001900

Population growth rate (left axis)Population size (right axis)

Source:Lee(2003).

FigureC.2:Age structure of the world population, 1800-2050 (percentage)

0

15

% o

f tot

al p

opul

atio

n

25

35

10

20

30

40

5

1800 1900 1950 2000 2050

Population under 15 Population over 65

Source:Lee(2003).

FigureC.4:Total fertility rate (TFR) and life expectancy by country group, 1950-2050

0

2

4

Chi

ldre

n pe

r wom

an

6

7

1

3

5

0

40

60

Year

s

80

90

30

20

10

50

70

TFR, more developed (left axis) TFR, less developed (left axis)Life expectancy, more developed (right axis) Life expectancy, less developed (right axis)

1950 1960 1970 1980 1995 2005 2015 2025 2035 2045

Source:UnitedNationsPopulationDivision,World Population Prospects: The 2010 Revision database.


116

(for instance,theUnitedKingdomat1.83andFranceat1.93).While the fertility rate in Japan is extremelylowat1.32,therateintheUnitedStatesis2.07.Withindeveloping countries, most Sub-Saharan Africancountries have high fertility rates, with an average of4.8 in 2010; this is the fastest-growing region of theworld in termsofpopulation.The fertility rate in India(2.73)isalsorelativelyhigh.Otherpopulousdevelopingcountries,however,havefertilityratesbelow2.Theseinclude the Republic of Korea (1.29), the RussianFederation (1.44), Thailand (1.63), China (1.64), Iran(1.77)andBrazil(1.90).

One of the implications of different demographicdynamics across countries is that the distribution ofworld population will continue to shift towardsdeveloping and emerging economies. As shown inFigureC.5, theshareofworldpopulation that lives insucheconomieswill risefrom85percent in2010to88percent in2050.Chinawillceasetobethemostpopulous country in the world in 2050; its share ofworldpopulationdroppingfrom20to14percentandbeing surpassed by India, which will account for18percentoftheworldpopulationin2050.3

One of the most dramatic consequences of thedemographic transition is the shift in age distributionof thepopulationat the laterstagesof the transition.Two variables that are of particular interest are thedependency ratio and the median age; these areshown for somepopulouscountries (China, Indiaandthe United States) and a range of regions (Sub-Saharan Africa, Middle East, Latin America and theEuropean Union) in Figure C.6 in order to highlightcertain patterns. Some countries and regions areshowntohaveafast-ageingpopulationandincreasesinthedependencyratio.China,forinstance,isageingfast: themedianagewasas lowas22years in1980but will reach the level of the United States (around38 years) in 2020 and the level of Europe (around46years)in2040.Moreover,China’sdependencyratio

willstarttogrowfromthelowlevelof37.5in2015totherelativelyhighlevelof64by2050–thesharpestriseintheworld(seeFigureC.6).AccordingtoLietal.(2012), the decline in labour force as a share of thepopulation will cause labour shortages and thuscontribute to risingwages inChina (seeSectionD.1).ToputitmorebluntlyinthewordsofTheEconomist,it“will bring an abrupt end to its cheap-labourmanufacturing”(The Economist,2012).4

In countries with relatively generous welfare systems,risingdependencyratiosimplyformidablechallengesintheprovisionofpensionsandhealthcarethatreliesontax revenues from the working population. Countrieswith intermediate fertility rates, such as the UnitedStates,willfinditeasiertocopewiththesechallengesthan countries with low fertility rates and acceleratedageing,suchasJapan.Thereare,conversely,countrieswhere demographic trends represent hugeopportunities, especially for India, Sub-Saharan Africaand Middle Eastern countries. Figure C.6 shows thatthey will have low median ages and will experiencedecreasesindependencyratiosinthecomingdecades.AsarguedbyThe Economist(2012),iftheycanimprovetheir public institutions, keep their economic policiesoutward-looking and invest more in education, as wasthecaseforEastAsia,thenAfrica,theMiddleEastandIndia could become the fastest-growing parts of theworldeconomywithinadecadeorso.5

(i) Ageing and comparative advantage

International differences in population dynamics havebeen identified as a factor determining comparativeadvantage and the composition of trade. Sometheoretical studies show that a country with slowerpopulationgrowthbecomesrelativelycapital-abundant,whileacountrywithfasterpopulationgrowthbecomesrelatively labour-abundant over time, thus registeringlower capital-labour ratios (“capital shallowing”). Thisgives rise to differences in autarky relative prices,6

FigureC.5:Share of world population, by country group, 2010 and 2050 (percentage)

20502010

Developed economies,12%

Congo, 2%

Indonesia, 3%

Other developingeconomies, 38%

Developing and emergingeconomies, 88%

India, 18%

Philippines, 2%

Bangladesh, 2%

Other developed economies, 8%

United States, 4%

Nigeria, 4%

China, 14%

China, 20%

Pakistan, 3%

Brazil, 2%

Other developed economies, 8%

Brazil, 3%

Other developingeconomies, 33%

Russian Federation, 2%

United States, 4%

Japan, 2%

Indonesia, 3%

India, 18%

Pakistan, 3%Nigeria, 2%

Bangladesh, 2%Developed economies, 15%

Developing and emergingeconomies, 85%



117

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

creating grounds for Heckscher-Ohlin trade in whichthe former country specializes in capital-intensivegoods and the latter country specializes in labour-intensivegoods(Sayan,2005;NaitoandZhao,2009).7

However, Yakita (2012) shows that countries with anageingpopulationarenotnecessarilynetexportersofcapital-intensive goods. A longer retirement promptsindividuals to invest more in human capital and toreduce the number of children. Moreover, a longerretirementdepressesdemand forconsumptiongoods(assumedtobelabour-intensive)intheworkingperiod,reducing their autarky relative price. If this relativepriceisbelowthefreetraderelativeprice,theageingeconomy ends up exporting labour-intensive goodsandimportingcapital-intensiveones.

Demographic change also has significant effects oncapital flows and the trade balance.8 However, theliterature does not provide unambiguous conclusionson the direction of these effects. Some studiesunderlinethatcountrieswhichareinarelativelymore

advanced stage of their demographic transition arecharacterized by net capital outflows and tradesurpluses. These studies show that higher lifeexpectancy determines an increase of savings forretirement, exerting pressure on the economy toexport capital to “younger” economies. Likewise, alower fertility rate reduces the size of the workingpopulation and investment demand, again inducingcapitalexports.On theotherhand,countries thatareintheinitialstagesofthedemographictransitionandhave relatively higher population growth will have netcapitalinflowsandtradedeficit.9

However,othershaveshownthateconomieswithhighand rising elderly dependency ratios can register netcapitalinflowsandtradedeficits.Forinstance,Higgins(1998)considers theeffectofdemographicvariableson savings, investment and the current accountbalance.Large,youngdependentpopulationsdepresssavingssupplywhileaugmenting investmentdemand.Savings and investment, in turn, are negativelyaffected by ageing. Therefore, the current account

FigureC.6:Dependency ratio and median age, selected countries and regions, 1950-2050 (percentageandyears)

Sub-Saharan Africa Middle East

India Latin America

Sub-Saharan Africa Middle East

India Latin America

0

10

19

50

19

60

19

70

19

80

19

90

20

00

20

10

20

20

20

30

20

40

20

50

20

30

40

50

60

80

100

(a) Dependency ratio (b) Median age

70

90

EU27 China United States EU27 China United States

0

5

19

50

19

60

19

70

19

80

19

90

20

00

20

10

20

20

20

30

20

40

20

50

10

15

20

25

30

40

50

35

45

0

10

19

50

19

60

19

70

19

80

19

90

20

00

20

10

20

20

20

30

20

40

20

50

20

30

40

50

60

80

100

70

90

0

5

19

50

19

60

19

70

19

80

19

90

20

00

20

10

20

20

20

30

20

40

20

50

10

15

20

25

30

40

50

35

45



118

balanceisnegativelyaffectedbothbylargeyoung-ageandold-agedependencyratios.Lührmann(2003)alsofinds that a high relative share of those aged 65 ormore in the population is associated with capitalinflows. This can be explained by declines in savingsand the repatriation of capital for consumption in oldage.10

Overall, little can be said definitively about theprospective effects of ageing on comparativeadvantage.Ifassociatedwithadecreaseinthelabourforceasashareofpopulation,ageingcan lead toanerosion of comparative advantage in labour-intensivemanufactured goods, as is foreseen for China. As aconsequenceofageing,countrieswithacomparativeadvantage in capital-intensive sectors may see thiscomparative advantage become stronger, but this isnot a general result. Finally, in order to assess theoverall impact on trade, it is important to considerdemand-side effects, in particular how ageing willaffectthelevelandthecompositionofdemand.Thisisthesubjectofthenextsection.

(ii) Demographic changes and the composition of demand

Demographicchangesareaffectingboththelevelandthe composition of consumption, with subsequenteffects on trade flows. The theoretical and empiricalliteratureonconsumptionoverthelifecycleprovidesauseful framework to understand the likely impact ofdemographyonfutureconsumptionandtradepatterns.

The life-cycle hypothesis assumes that individualsprefer to smooth consumption over their lifetimes.11Hence, they save during their working age, whenincomeishigher,anddis-saveintheirretirementperiod,whenincomeislower.Dataonconsumptionandincome,however, contradict the consumption and savingpatterns predicted by the basic life-cycle model inseveral respects. First of all, there is evidence of ahump-shaped relationship between households’ totalconsumption and age. This is mainly explained byhousehold composition effects, according to whichhouseholds’ expenditure increases with the number ofchildren (Attanasioetal.,1999;BrowningandEjrnæs,2009).Moreover,empiricalevidenceshowsthatsavingsoftheelderlydonotdecreaseasmuchasthelife-cyclemodel (in its simplest formulation) would predict. Thismainly depends on bequest motives (Hurd, 1989), orprecautionary savings, which are accumulated toaccommodate unexpected health or economic shocks(Carroll,1994;1997).12Liquidityconstraintsmightalsogenerate a pattern of consumption which is similar tothat determined by precautionary savings, withindividuals accumulating resources in order to smoothconsumption when facing economic shocks andimpossibilitytoborrow(Deaton,1991).13

Household composition effects are relevant toassessing the impact of demographic change on

demandpatterns.Inparticular,sinceahighernumberof children accounts for higher householdconsumption expenditure, one may expect, otherthings being equal, increased consumption in high-fertility, high-population growth countries, such asthose in Latin America and Sub-Saharan Africa.However,theabilitytofinanceconsumptiongrowthinthesecountriescruciallydependson theireconomicgrowth, which, in turn largely hinges on job creation(see Section D). Moreover, domestic demand andimport trends also depend on other economic andinstitutionalfactors,suchasfinancialintegrationandsocialsecurity,whichare likelytoaffecthouseholds’expenditurecapacity.

For countries at the most advanced stage of thedemographic transition, older groups will account forthe largest share of consumption.14 The effects ofageingonaggregateconsumption(and,consequently,on importdemand)will likelydependontheextentofthe decline in consumption following retirement, alsoknown as the “retirement consumption puzzle”.15

However,compositionaleffectsaremorerelevantthanlevel effects. Expenditures on some categories ofgoods, such as food, furnishing, clothing andaccessories, are noticeably reduced upon retirement,while expenditures on other categories remainconstant or increase (Hurst, 2008). Studies thatprojectfutureconsumptionpatternsinmoreadvancedeconomies based on current demographic, economicandsocialtrendsconcludethatservicesandhigh-techsectors will gain most in the coming decades (CBI,2012; Desvaux et al., 2010; Deutsche Bank, 2007;Lührmann, 2005; Oliveira Martins et al., 2005). Inparticular, consumption will increase most incommunication, transport, health, financial services,tourism services as well as in entertainment andcommunity services that target the senior citizenmarket.Sincenotall thesesectorsare tradeable, theimpact on international trade will also depend on thechange in demand for tradeable services relative tonon-tradeableones.

Thegradual convergenceof per capita income levelsacross countries, documented in Section D, is givingrise to another important phenomenon, namely theexpansion of the global middle class. According tothe World Bank (2007), in the period 2000-2030,theglobalmiddleclassisprojectedtogrowfromabouthalfabilliontoabout1.2billion,orfrom7.6to16.1percent of the world population. However, its share ofworld incomewill remainstableatabout14percent,reflecting decreasing inequality across countries.16Because of uneven population growth acrosscountries, the geographical distribution of the middleclass will change remarkably in the coming decades.Regions with relatively higher projected populationgrowth rates, such as South Asia and Sub-SaharanAfrica, will see their share of the global middle classincrease while other regions will see a decrease(KharasandGertz,2010;WorldBank,2007).


119

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

Theexpansionofthemiddleclassis likelytoresult inanincreaseindemandforgoodsandservices,suchascars, mobile phones, recreational equipment andservices, as well as food. Some Western foodcompanieshavealreadymodifiedtheirproducts,eitherto cater to Asian consumers’ tastes (The Economist,2013)or tomake themmore sophisticated.AsAsianconsumersbecomericher,theyaredemandinghigher-quality and healthier products (Atsmon et al., 2012).A decrease in the import share of low-value-addedproducts,suchasagriculturalgoods,andan increaseintheshareofhigher-value-addedgoods,suchascarsand office and telecom equipment, is already takingplace in the BRIC (Brazil, Russian Federation, India,China)group(Yamakawaetal.,2009).17

Traderemainskeytosustainingeconomicgrowthandthus the expansion of the middle class. This isparticularlytrueofcountriessuchasChinawheretheshareofdomesticconsumptioninGDPisstillrelativelylow.18Policiesthataddressincomeinequalitycanalsobe important to expanding the middle class and thuseconomicgrowth(KharasandGertz,2010).19

Another important trend in developing and emergingeconomies is the rise of education levels. Increasingdemand for education, combined with technologicaladvances,isfuellingariseineducationservicestrade.AccordingtoLimandSaner(2011),educationservices’exports grew, on average, by 12 per cent between2002 and 2007. The United States, Australia, theUnited Kingdom and Canada were among the topexporters; the Republic of Korea, the United States,Germany and India were among the top importers.Educationmarketsarealsogrowing inLatinAmericaand theMiddleEast (LimandSaner,2011). Typically,international students’ mobility, which corresponds tomode 2 (consumption abroad) of the GeneralAgreementonTradeinServices(GATS),hasbeenthemainchannelforeducationalservicestrade.However,long-distance education (mode 1 – cross-bordersupply) and the establishment of foreign branches ofeducationalinstitutions(mode3–foreigncommercialpresence)arealsogrowing.20

In short, demographic changes will affect trade boththroughtheirimpactoncomparativeadvantageandonpatternsofdemand.Onemightexpectcountrieswithhigh and rising old-age dependency ratios to switchfrom being net exporters to net importers of capital-intensive goods or to experience an erosion of theircomparative advantage in labour-intensivemanufactured goods. Ageing is also likely to beassociatedwitha relative increase in thedemand forgoods and services that are disproportionatelyconsumed by older groups of the population. Theemergenceof aglobalmiddle classwill alsohaveanimpact on the composition of global demand. Thegrowing number of relatively wealthy consumers inemerginganddevelopingeconomieswillopenupnewbusinessopportunitiesandexpandtrade.

(b) Changesinlabourforcecomposition

Two other notable labour force developments, bothlinked to the demographic transition, are likely toaffecttradeflows:arisingshareofeducatedworkersand increased female labour force participation. Thefollowingsectionexaminesthesetrendsinmoredetail,and then explores the channels through which theycanaffectcomparativeadvantageandtradepatterns.

(i) Skills

Overthelast60years,educationlevelshaveincreasedsubstantially in most countries. Using data from146 countries, Barro and Lee (2010) show that overtheperiod1950-2010theaveragenumberofyearsofschoolingamongindividualsaged15oroverincreasedfrom2.1to7.1indevelopingcountriesandfrom6.2to11.0 in developed countries (see Figure C.7). Thehighest growth rates were registered in the MiddleEastandNorthAfrica,Sub-SaharanAfricaandSouthAsia.21

BasedonthedataprovidedbyBarroandLee(2010),Fouré et al. (2012) project future secondary andtertiaryeducationenrolmentratesfortheworkingagepopulation to 2050. Their projections show that theeducational attainment profile of the workingpopulation will continue to increase, especially indeveloping countries, producing a convergence ineducational levels between both developing anddeveloped countries (see Figure C.8). The sameconclusion is reached by KC et al. (2010), who alsoexplain the underlying causes of this convergence.22In countries where the old-age dependency ratio isprojectedto increase,suchasChina,progresswillbe

FigureC.7:Educational attainment of the total population over age 15 by country group, 1950-2010 (years)

0

4

19

50

19

90

(a) Developing countries (b) Developed countries

19

70

20

00

20

10

19

50

19

90

19

70

20

00

20

10

8

12

2

6

Ave

rage

yea

rs o

f sch

oolin

g

10

Primary Secondary Tertiary

Source:BarroandLee(2010).


120

FigureC.8:Projections of secondary and tertiary enrolment rates, 1990-2050 (percentage)

United States Russian Federation European Union ChinaMiddle East South America India Sub-Saharan Africa

0

10

19

90

19

95

20

00

20

05

20

10

20

15

20

20

20

25

20

30

20

35

20

40

20

45

20

50

20

30

40

50

60

80

100

Tertiary enrolment rateSecondary enrolment rate

70

90

0

10

19

90

19

95

20

00

20

05

20

10

20

15

20

20

20

25

20

30

20

35

20

40

20

45

20

50

20

30

40

50

60

80

100

70

90

Source:Fouréetal.(2012).

definedintermsofthecompositionoftheworking-agepopulation.Incountrieswheretheold-agedependencyratioisprojectedtodecrease,suchasIndia,progresswillmostlybeintermsofthegrowingnumberofhighlyqualifiedpeopleaddedtothepotentiallabourforce.

In addition to highlighting global educationalconvergencegenerally,thesestudiesalsorevealspecificregional patterns. According to KC et al. (2010), LatinAmericawillregisterthemostrelevantimprovementsineducational attainment, mainly because of theinteractionbetweeneducationandfertilitydynamics.Inseveral Latin American countries, increases in schoolenrolment preceded fertility reductions, with the resultthat theyoungestandmosteducatedsegmentsof thepopulation are also bigger. This expanding populationof educated young people is found in several Asiancountries,suchasNepal,PakistanandCambodia,andin the Middle Eastern countries, such as Jordan andtheKingdomofSaudiArabia.

In Sub-Saharan Africa, the picture is more complex.Although education rates among 20 to 64-year-oldsareexpectedto improvesignificantly,somecountries,such as Ethiopia, Mali, Niger and Burkina Faso, arestarting from such a low base that by 2050 largeshares of the working age population (for instance,40percentinthecaseofEthiopiaand35percentinthecaseofBurkinaFaso)will stillhavenoeducationdespitesignificant improvements innationalaverages(KCetal.,2010).Theimplicationisthatthesecountriesmayfallbehindsignificantlycomparedwiththerestofthe world in terms of educational attainment of theworkingpopulationby2050.

Improvinghighereducationenrolmentrateswillrequiresubstantial effort and resources, especially incountries starting from a low base and in countrieswherethesizeoftheyoungpopulationisprojectedtoincreasesignificantly(KCetal.,2010).Anothercrucial

educationalchallengeistomakeprogressinschoolingquality, which remains uneven, even among countrieswith a similar level of educational attainment (BarroandLee,2010;HanushekandWoessmann,2009).Toensure that there are sufficient jobs created in highpopulationgrowthcountries,itwillalsobeimportanttomatcheducationalsupplyanddemandby,forinstance,establishing effective public-private partnershipsbetweenbusinessandeducationinstitutions.

These educational developments are likely to affecttradepatternsbecauseoftheirimpactoncomparativeadvantage. According to the Heckscher-Ohlin model,countries have a comparative advantage in sectorsthat make more intensive use of their relativelyabundant factors (see Section B.2). Several recentstudies have shown that the endowment of humancapital(relativetolabour)isanimportantdeterminantof comparative advantage and trade patterns.23

Building on these observations, Costinot (2009)suggests that comparative advantage is affected byworkers’ endowment of efficiency units of labour.When workers are more educated, they spend asmaller fraction of their time learning. Since learningcosts are relatively more important in more complexsectors, a country with educated workers has acomparativeadvantageinmorecomplexsectors.24

Comparative advantage can also be shaped by thedistribution of human capital across workers. InGrossman and Maggi (2000), for instance, there canbe trade between countries with similar aggregatefactor endowments, provided human capital is morewidely dispersed in one country than the other. Thecountrywitha relativelysimilarpopulation in termsofeducational levelsexportsthegoodwithaproductiontechnology characterized by complementaritiesbetween workers. The country with a diversepopulation,inturn,exportsthegoodwhosetechnologyis characterized by substitutability between


121

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

employees.25 Grossman and Maggi (2000) providesome examples in support of their theory. CountrieslikeJapanandGermany,withapoolofrelativelysimilarworkers, have a comparative advantage in industries,suchasautomobiles,thatrequirecareandprecisioninalongseriesofproductiontasks.Conversely,countriessuchastheUnitedStatesorItaly,withamorediversepoolofworkers,tendtohaveacomparativeadvantagein industries where the input of a few very talentedindividuals(e.g.fashiondesigners inthecaseof Italy)mattersmost.26

Demographicchanges that increaseoverall levelsofeducation will affect the relative endowment ofproductivefactorsandcontributethroughthevariouschannels outlined above to shaping the evolutionof comparative advantage and trade patterns.Developing countries, such as China, are alreadyexporting sophisticated goods to OECD countries(Rodrik, 2006; Schott, 2008). This generates theincreased overlap in the structure and in the skillcontent of exports from China and the high-incomecountriesdocumentedinSectionB.Thisphenomenonis partly due to processing exports (contractingmanufacturing for goods that are designedelsewhere) in sectors that may be labelled as high-tech industries.27 Wang and Wei (2010), however,report evidence that improvements in human capital(togetherwithgovernmentpoliciesintheformoftax-favoured,high-techzones)appeartocontributemostto the growing sophistication of China’s exports.Exportsofskill-intensivegoodstorichcountriescanbeasourceofgrowthforpoorcountries(MattooandSubramanian,2009a).Integratingalargernumberofskilled workers into their labour force (and adoptingtechnologies that most improve the productivity ofskilled labour) is therefore a promising option fordevelopingcountries.

(ii) Female employment

The demographic transition is also associated withchangesin labourforceparticipationrates(LFPRs).28

These changes depend on country characteristics,such as labour market institutions and social norms,andindividualcharacteristics,suchasageandgender.Between 1980 and 2008, the global male LFPRsdecreasedfrom82to77.7percent,mainlyasaresultof decreasing participation of young males who arestaying longer in education. The global female LFPRgrew in the 1980s from a starting point of 50.2 percent,reached52.2percentin1990,butthendeclinedbetween 1990 and 2008 to settle at 51.7 per cent(ILO, 2010). The limited increase in female LFPRscouldbeexplained,amongother things,by increasedfemale education, which decreases the participationrateofyoungfemales.

TheabovedatashowtherelevanceofeducationasadeterminantoffemaleLFPRs.Otherdemographicandeconomic factorsalsoplaya role.For instance,Galor

andWeil(1996)showthattechnologicalprogressandtheaccumulationofphysicalcapitalmakelabourmoreproductiveandincreasetheopportunitycostofraisingchildren,withnegativeeffectsonfertilityandpositiveeffects on women’s participation in the labourmarket.29 Moreover, Soares and Falcão (2008)emphasizetheroleofincreasesinadultlifeexpectancyindeterminingfemaleLFPRs.Inparticular,reductionsinadultmortalityincreasethereturnstoeducationforwomenandreducethegainsfromlargerfamilies,thusreducingfertility ratesand increasingwomen’s labourmarketactivities.

Female LFPRs are also likely to depend on thecountry’s level of development. The relationshipbetween the two variables seems to be U-shaped(Goldin, 1995; Mammen and Paxson, 2000).Participation rates are higher in subsistenceeconomies. Then, at the initial stage of development,educationandwagesincreaserelativelymoreformenthan for women. As household income increases,women reduce their labour market participation (theincome effect prevails).30 At a later developmentstage, thereareeducationalgainsforwomenaswell,raising the opportunity cost of child caring andincreasingfemalelabourmarketparticipation.31

Besides demographic and economic factors, otherimportant determinants of female labour marketparticipation are access to education, religious,cultural and social norms, and the institutionalframework (ILO, 2010). The impact of demographicchangemaybereducedoroffsetbyculturalandsocialnorms. For instance, analysing the determinants offemale LFPR in a sample of 160 countries between1960 and 2008, Tsani et al. (2012) found that, allthingsbeingequal,SouthernMediterraneancountrieshave significantly lower female LFPRs than othercountries.Theauthorssuggestthattheseresultsmayreflect region-specific social or institutional factorsthat act as barriers to women’s participation in thelabourmarket.

Figure C.9 shows past and projected data (for 1990and2020, respectively)on femaleLFPRsforselectedcountries and regions. The data highlight someinteresting patterns, which can be explained by thedemographic, economic and cultural factors outlinedabove. In the European Union, China and India, therewill be considerable reduction in LFPRs of youngwomen which is mainly the result of increased schoolattendance.32 Moreover, in the European Union therewillbeanincreaseinLFPRinmorematuresegmentsofthe female labour force. This is mainly related toincreased life expectancy, higher retirement ages andthe introduction of age and gender anti-discriminationlaws(Jaumotte,2003).Conversely,LFPRsofthemorematuresegmentsofthefemalepopulationareprojectedto increase only slightly in the case of China. In India,femaleLFPRsareexpectedtodecreaseforvirtuallyallage groups. Several factors may explain these


122

projections.First,continuedeconomicdevelopmentwilllead to lower participation by women in low-incomehouseholds.33Secondly, the specific characteristics ofIndia’sprocessofeconomicgrowth in the lastdecadeimply that increases in labour productivity growth areassociatedwithreductionsinemploymentgrowth(ILO,2012). Thirdly, according to Kingdon and Unni (2001),specific cultural and social norms, according to whichwomen’s labour is less socially acceptable in highercaste,may reduceLFPRsofwomenwith intermediatelevelsofeducation.

Cultural and social norms may also explain the lowLFPRscurrentlyobserved–andprojectedtocontinuein the future – in the Middle East (ILO, 2012).34Conversely,SouthandCentralAmericancountrieswillexperience significant increases in LFPRs for all agegroups.Thisincreaseisassociatedwiththefavourabledemographic trends outlined above, especially lowerfertility rates. In Sub-Saharan Africa, participation isalso increasing, mainly driven by increases in theworking-age population. However, Figure C.9 showsthatfemaleLFPRswerealreadyhighin1990,reflecting

the fact that several countries in the region were at averylowlevelofeconomicdevelopment.

Women’sincreasinglabourforceparticipationcanbeasource of comparative advantage if women aredisproportionately employed in particular sectors. Inmost developing countries, female employment isconcentrated in labour-intensive exports. UNCTAD(2004) reports that women’s participation in exportindustries such as textiles, clothing, pharmaceuticals,food processing, electronics and toy productionaveragesbetween53percentand90percentof thelabour force in African, Asian and Latin Americandeveloping countries. In South-East Asia, key exportindustriessuchastextilesandelectronicsreliedheavilyon relatively unskilled, but generally literate, women(Korinek, 2005). Between 1970 and 1995, women’sshare in the labour force in Indonesia, Malaysia andSingapore grew from between 26-31 per cent to37-40percent. In theRepublicofKorea, theshareofworking women in regular paid work increased from65 per cent in 1965 to 81 per cent in 1992, and inmining and manufacturing the female to male

FigureC.9:Women’s labour force participation rates in selected economies, 1990 and 2020 (percentageandagegroup)

0

20

40

60

80

100

15

-19

20

-24

25

-29

30

-34

35

-39

40

-44

45

-49

50

-54

55

-59

60

-64

65

+

European Union

1990 2020

0

20

40

60

80

100

15

-19

20

-24

25

-29

30

-34

35

-39

40

-44

45

-49

50

-54

55

-59

60

-64

65

+

China

1990 2020

0

20

40

60

80

100

15

-19

20

-24

25

-29

30

-34

35

-39

40

-44

45

-49

50

-54

55

-59

60

-64

65

+

Middle East

1990 2020

0

20

40

60

80

100

15

-19

20

-24

25

-29

30

-34

35

-39

40

-44

45

-49

50

-54

55

-59

60

-64

65

+

India

1990 2020

0

20

40

60

80

100

15

-19

20

-24

25

-29

30

-34

35

-39

40

-44

45

-49

50

-54

55

-59

60

-64

65

+

Sub-Saharan Africa

1990 2020

0

20

40

60

80

100

15

-19

20

-24

25

-29

30

-34

35

-39

40

-44

45

-49

50

-54

55

-59

60

-64

65

+

South and Central America

1990 2020

Source:ILO,Economically Active Population, Estimates and Projections (EAPEP),2011.


123

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

employmentratiorosefrom0.37to0.68(WorldBank,2001).35

BusseandSpielmann(2006)istheonlyempiricalstudythatanalysestheeffectofvariousmeasuresofgenderinequalityoncomparativeadvantage.Usingpaneldatafrom29countriesoversixseparateyears(1975,1980,1985, 1990, 1995 and 2000), they show that areduction in inequality in labour forceparticipation (i.e.highersharesof female tomale labourmarketactivityorahigherfemaleparticipationrate)isassociatedwithanimprovementinthecomparativeadvantageoflabour-intensive sectors.36 The relationship, however, losesstatisticalsignificancewhenhigh-incomecountriesareexcluded from the sample. This is surprising since, asnoted above, it is especially in most developingcountries thatwomenaredisproportionatelyemployedinlabour-intensiveexports.

In many developing countries, women’s increasedlabour force participation is likely to be accompaniedby higher education. KC et al. (2010) report thatcountriessuchasChile,ChinaandSouthAfricaoftenreach near universal secondary school attainmentamong women aged 20-39 by 2050. In India andPakistan,secondaryschoolattainmentamongwomenaged 20-39 is projected to increase from around40percentin2010tomorethan80percentin2050.Fromatheoreticalperspective,areducedgenderbiasin educational attainment (a measure of decreasinggender inequality) may positively or negatively affectcomparativeadvantagein labour-intensivegoods.Theempirical results of Busse and Spielmann (2006)indicate that a reduction in inequality in access toeducation (i.e. higher female literacy rates relative tomaleorhigherfemaleschoolenrolment)isassociatedwith an improved comparative advantage in labour-intensivesectors.

However,thecausallinkcouldruninbothdirections.AsshownbyVijaya (2003), insomedevelopingcountries,trade-related employment can lessen women’sincentivesto invest inhighereducationcomparedwithmen.Therefore,existinggendergapsineducationmaybe reinforced and even widened by greater tradeopenness. The explanation for this finding is that thedemandforfemalelabourremainsconcentratedinlow-skilled jobs,possiblybecausediscriminationclosesoffother higher-skilled opportunities, thus reducing theincentive to invest in higher education.37 However, areduction in discrimination would give women betteraccesstomoreskill-intensiveoccupationswhichwouldin turn shift comparative advantage from labour-intensivetoskill-intensivesectors.

Inconclusion,boththerisingshareofeducatedworkersandincreasedfemalelabourforceparticipationhaveanimpactoncomparativeadvantage. Inparticular,amoreeducatedworkforceincreasestheskillcontentandthesophisticationofexports,whichhasbeenanimportantsourceofgrowthforanumberofdevelopingcountries,

especiallyinEastAsia.Itishopedthatotherdevelopingcountries,especially inAfrica,willalsobeabletoreapthetrade-relatedbenefitsofincreasededucationinthefuture.Labourforceparticipationofwomenisintimatelyconnected with falling fertility rates and rising lifeexpectancy, but also with increased educationalopportunities.Inclusivefemalelabourforceparticipationhas effects on comparative advantage, can positivelyaffect import demand and can be a source of welfaregains.

(c) Migration

International migration has an important impact ondemographic change. It can influence populationgrowth directly by adding to or subtracting from thepopulation (both for the source and host countries)and indirectly by affecting fertility rates (UnitedNations,2011a).Moreover,internationalmigrantstendto be a unique population group in terms of age andeducation. This section suggests that internationalmigration can affect patterns of comparativeadvantagebyshiftingtheeducationandageprofileofboth source and host countries. This section alsoreviewsthetheoreticalandempirical literatureontherelationship of substitutability or complementaritybetween trade and migration. Finally, it considers thetradeeffectsofurbanization,whichisaconsequence,amongotherthings,ofinternalmigration.

The global stock of international migrants grew by38percentfrom1990to2010.However,internationalmigrants still constitute a very small fraction of theworld population, just 3.1 per cent (213.9 million) in2010. Migrants are concentrated in a few receivingcountries:in2010,tencountrieshostedmorethanhalfof the global international migrants’ stock.38 Themajority of international migrants reside in Europe,Asia and Northern America. Oceania and NorthernAmerica had the highest percentage of migrantsrelativetototalpopulationin2010(seeTableC.1).39

Migration is overwhelmingly from less developed tomoredevelopedcountriesandregions.From1990to2010,themigrantstockresidingintheNorth(Europeand Northern America plus Australia, New ZealandandJapan)but born in theSouth (all other countriesand regions) increased by 85 per cent, more thantwiceasfastastheglobalmigrantstock(38percent)(UnitedNations,2012a).

In traditional destinations for immigration, such asAustralia,Canada,NewZealandandtheUnitedStates,migrant inflows increasedsignificantlybetween1980and2008.40However,thegrowthratewaserraticandhighlyinfluencedbychangesinimmigrationpolicies.41In the United States, the main host country for theworld’s migrants, about 1.1 million permanentresidence permits were issued between 2005 and2010 (United Nations, 2011a). Immigrants to theUnited States mainly originate from Asia and from


124

Latin America and the Caribbean (with both regionsaccounting for 40 per cent of the total immigrantinflowsin2010).MexicoandChinaaccountfor13and7percentofthe2010inflows,respectively.Asiaalsorepresents the main region of origin of migrants toAustralia (shareof60percentof the total immigrantinflowsin2008)andCanada(shareof58percentofthe total immigrant inflow in 2009). In Europe,

Germany represents the main destination for Centraland Eastern European migrants, especially after theenlargement of the European Union in 2004 and2007.42 The majority of immigrants to Europeancountries in the period 2000-08 came from othercountriesinEurope.However,forsomeEuropeanhostcountries, such as France, the United Kingdom andSpain, immigrants mainly came from developing

TableC.1:International migrants by region (stocks), 1990-2010 (millionsandpercentage)

Number of international migrants (millions)

International migrants as percentage of the population

1990 2000 2010 1990 2000 2010

World 155.5 178.5 213.9 2.9 2.9 3.1

Moredevelopedregions 82.4 104.4 127.7 7.2 8.7 10.3

Lessdevelopedregions 73.2 74.1 86.2 1.8 1.5 1.5

Africa 16.0 17.1 19.3 2.5 2.1 1.9

Asia 50.9 51.9 61.3 1.6 1.4 1.5

Europe 49.4 57.6 69.8 6.9 7.9 9.5

LatinAmericaandtheCaribbean 7.1 6.5 7.5 1.6 1.2 1.3

NorthernAmerica 27.8 40.4 50.0 9.8 12.7 14.2

Oceania 4.4 5.0 6.0 16.2 16.1 16.8

Source:UnitedNationsPopulationDivision,WorldMigrantStockdatabase.

Note:Forthedefinitionofregions,seehttp://esa.un.org/MigAge/index.asp?panel=3.

BoxC.1: Has migration become more regionalized?

Oneof the trendsdocumented inSectionBof this report is the increased regionalizationofmerchandisetradeflows.Doesasimilarpatternemergeformigration?Answeringthisquestionisnoteasyduetoseveredatalimitations.InFigureC.10,historicaldataonmigrants’stockscompiledbytheWorldBankareusedfortheyears1990and2000.43Someinterestingfactsemerge.

First,migrantsfromAfrican,AsianandCommonwealthofIndependentStates(CIS)countriespredominantlyresideintheirrespectiveregionoforigin.Conversely,themajorityofmigrantsresidinginNorthAmericaandin theMiddleEastcome fromcountriesoutside the region.Europe fallsbetween the two,withashareofabout60percentofmigrantscomingfromwithinEurope.

Secondly,between1990and2000,theshareofintra-regionalmigrantsincreasedsignificantlyinSouthandCentralAmerica(from55to64percent),andtoaminorextentinNorthAmerica(from28to31percent)and Africa (from 85 to 87 per cent). Conversely, this share remained stable in Europe, and it slightlydecreased inall other regions (from32 to31per cent in theMiddleEast; from95 to93per cent inCIScountries;from85to84percentinAsia).

Thehighsharesofintra-regionalmigrationinAfrica,AsiaandtheCIScanbeexplainedmainlybymovementsacrossthebordersofneighbouringstates.AccordingtoRathaandShaw(2007),thisgeographicallylimitedcross-bordermigrationaccounted for80per cent of theSouth-Southmigrants’ stock in2007. The samestudy also shows that migrants from Burkina Faso to neighbouring Côte d’Ivoire account for the highestshare of South-South migrants in Africa, while migrants from Bangladesh to India represent the highestshareofSouth-SouthmigrantsinSouthAsia.IntheCISregion,migrantsmainlymovebetweentheRussianFederation and Ukraine and between the Russian Federation and Kazakhstan. Other countries with highlevelsofcross-bordermigrationareSouthAfrica,whichisthemaindestinationformigrantsfromLesotho,MozambiqueandZimbabwe,andThailand,which is themaindestinationformigrants fromCambodia,LaoPDRandMyanmar(IOM,2008).

The relevance of cross-border migration among developing countries reflects low levels of wealth andeducationofthepopulationatorigin,whichlimit individuals’andhouseholds’abilitytoaffordlong-distancemigration. Since it is mainly short-distance and temporary, cross-border migration can be equated withinternalmigration.


125

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

However,sinceittakesplacebetweenareaswithrelativelysimilar incomelevels,cross-bordermigrationislikelytobedrivenmorebythedesiretoreduceriskanddiversifyincomeratherthanbygeographicalincomedifferences(RathaandShaw,2007).44

Institutionalfactors,suchasthepresenceofpreferentialtradeagreements(PTAs)orregionalconsultativeprocesses (RCPs) on migration, may also help to explain patterns of intra-regional versus extra-regionalmigration.45ArecentstudybyOrefice(2012)showsthatPTAshavebeenadeterminantofmigrationinflowsfor29OECDcountries in theperiod1998-2008. Inparticular, visa-and-asylumand labourmarket relatedprovisions,whenincludedinPTAs,stimulatebilateralmigrationflows.Inthisstudy,however,nodistinctionismade between intra- and cross-regional PTAs because of data limitations. In the future, more researchshouldbeconducted,with theaimofdiscerning theeffectsof institutional factorson intra- versusextra-regionalmigration.

FigureC.10:Intra-regional and extra-regional migrants (stocks), 1990 and 2000 (thousandsandpercentage)

North America

28%31%

72%

69%

1990 2000 Europe1990 2000

Middle East1990 2000

CIS1990 2000

Asia1990 2000

Africa1990 2000

South and CentralAmerica

1990 2000

28,484

5,213 5,509

41,299

55% 64%

85% 87%

60%

40%

68%

95%

85% 84%

93%

32%

69%

31%

33,658

60%

40%

40,551

27,36924,805

23,929 26,292

10,030 12,173

13,17416,438

Extra-regional Intra-regional

Source:WorldBankGlobal Bilateral Migration Database(GBMD).

Note:Graphsforregionsareshowntoscale,withtheexceptionofSouthandCentralAmericaandAfrica,whichshareadifferentscale.ColoursandboundariesdonotimplyanyjudgmentonthepartofWTOastothelegalstatusofanyfrontierorterritory.

regions.46 A more detailed analysis of migrationpatterns within regions (intra-regional) and acrossregions(extra-regional)ispresentedinBoxC.1.

As argued above, migration can directly influencepopulationgrowthbyaddingtoorsubtractingfromthepopulationofthecountriesconcerned.Fiftyyearsago,the impact of net migration on overall populationgrowth was negligible in virtually all countries andregions.Morerecently,netmigrationhasbecomemoreimportant to developed countries due to low fertilityrates. As shown in Figure C.11, by 1990-2000 netmigration was already the main driver of populationgrowth in developed countries.47 This trend willcontinue in the future. From 2010 to 2050, the netnumber of international migrants moving to moredevelopedregionsisprojectedtobe87million.Sinceit is projected that deaths will exceed births by

11 million, the overall population growth will be76 million. From 2050 to 2100, the net number ofinternational migrants moving to more developedregionsisprojectedtobe49million.Givenanexcessofdeathsoverbirthsof24million,thiswillresultinanoverallgrowthof25million(UnitedNations,2011b).

Migrationalsoimpactspopulationchangeindirectlybyinfluencingfertilityratesinthecountryoforiginandinthe host country. However, recent evidence suggeststhat migrants adapt over time to the host country’sfertility norms (Kulu, 2005).48 Thus, any positiveimpact on host-country fertility that internationalmigration from high- to low-fertility countries mighthave is likely tobetemporary.Migrants’adaptationtothehost country’s normsaffects fertility levels in thecountryoforiginaswellbecausethefertilitynormsofthehost countryare, toa certaindegree, transferred


126

backtothecountryoforigin.Forinstance,BertoliandMarchetta (2012) show that Egyptian couples have asignificantly higher number of children when thehusbandreturnstohishomecountryafterhavingbeena migrant in a high-fertility Arab country. Moreover,migration’s impact on fertility rates is not limited tomigrantsandtheirhouseholdsbutcanspillovertothewiderpopulationinthecountryoforigin.Usingmacro-leveldataforabout150hostcountriesin2000,Beineetal.(2012)estimatethata1percentdecreaseinthefertilitylevelinthehostcountryreducesfertilityratesinthecountryoforiginby0.3percent.

Migrants are generally younger than the nativepopulation.Forinstance,themedianageofimmigrantsinEUmemberstatesin2009rangedfrom24.9years(in Portugal) to 33.7 years (in Latvia), relative to amedianageof theEU-27populationof40.9years.49More importantly, individuals of working age areover-represented among international migrants, asFigureC.12showsforEUmemberstates.50

Accordingly, migration is projected to reducedependency ratios in a number of economies, asindicated by Table C.2.51 The impact of migration isvery noticeable in oil-exporting Middle Easterncountries,suchastheUnitedArabEmirates,Qatarandthe State of Kuwait but it is also noticeable in HongKong (China), Switzerland and southern Europeancountries.However,notwithstandingarelativelygreaterimpact in certain economies, the overall impact ofmigrationontheagestructureoftheworldpopulationislikelytobemodest,especiallyincountrieswheretheageingprocess ismostadvanced,suchasJapan.TheUnited Nations (2011a) concludes that migrationcannotreversethetrendofpopulationageing.

Theimpactofmigrationontheoriginandhostcountriescrucially depends on the skills distribution betweenmigrantsandthenativepopulation.TableC.3providesacomparison between the education structure of thenativepopulationand immigrants inOECDdestinations.Thelastrowofthetableshowsthat,between1990and2000, on average, immigrants are more educated thanthe native population. Thus, immigration is associatedwithanet“braingain” inhostcountries.However, therearesignificantdifferencesacrosscountries.Forinstance,immigrantsaremoreskilledthanthenativepopulationincountries where the nationals’ education level is low(such as Mexico and Turkey) or in countries where theimmigration policy favours the entry of highly educatedindividuals(suchasAustralia,CanadaandNewZealand).In contrast, immigrants are less skilled than the nativepopulation in countries where the nationals’ level ofeducationishigh,suchastheUnitedStatesandFrance.

Table C.3 also shows that during the period1990-2000 theoverall shareof high-skill immigrantstoOECDcountriesincreasedfrom30to35percent.Inthesameperiod,thenumberofhigh-skillimmigrantsincreased by 64 per cent (from 12.6 to 20.7 million),whilethenumberof low-skill immigrantsincreasedby22percent(from20.1to25.7million).However,mostimmigrants to OECD countries are medium- or low-skilled individuals (Docquier et al., 2009). Asunderlined by Widmaier and Dumont (2011), this islargelyexplainedbylabourneedsintheso-called“3Djob” sector (dirty, dangerous, difficult) and low-wagesectors,suchasagriculture,constructionanddomesticservices. Here, too, there is significant heterogeneityacrossOECDcountries. InsouthernEurope,migrantsaremainly low-skilled,while inCanada,Australia andNewZealand,migrantsaremostlyhighlyeducated.

FigureC.11:Contribution of natural increase and net migration to net population change in developed countries, 1950-55 to 2045-50 (percentage)

-2

0

2

4

6

8

10

12

14

19

50

-19

55

19

55

-19

60

19

60

-19

65

19

65

-19

70

19

70

-19

75

19

75

-19

80

19

80

-19

85

19

85

-19

90

19

90

-19

95

19

95

-20

00

20

00

-20

05

20

05

-20

10

20

10

-20

15

20

15

-20

20

20

20

-20

25

20

25

-20

30

20

30

-20

35

20

35

-20

40

20

40

-20

45

20

45

-20

50

Rate of net change Rate of natural increase Rate of net migration



127

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

Theemigrationofskilledindividuals(“braindrain”)haslongbeenapolicyconcernintheircountriesoforigin(see thediscussion inDocquierandRapoport,2012).Table C.4 shows data on the stock on high-skilledemigrantsandhigh-skillemigrationratesbyregionfortheyears1990and2000.Thetableshowsthat,unlikehigh-skillemigrationstocks,high-skillemigrationratesremainedfairlystableoverthisperiod.52Inbothyears,thereisconsiderablevariationacrosscountrieswithinregions.

Forinstance,withinEastAsiaandthePacific,therateis 3 per cent in Australia but rises to 15 per cent inSouth-eastern Asia and to about 47 per cent in thePacific Islands. Within South and Central America,theraterangesfrom18percentinSouthAmericato27percent inCentralAmerica to65percent in theCaribbean (in this sub-region, the countries with thehighestskilledemigrationratesareJamaicaandHaiti,withratesof85and83percent, respectively).SomeAfrican countries are also characterized by skilledemigration rates that are significantly higher thanthe regional average. This is the case for Gambia(68 per cent), Sierra Leone (49 per cent), Ghana(45percent)andKenya(40percent)amongothers.53Whethertheemigrationofskilledindividualsisharmfulor beneficial for the countries of origin is a questionthatwillbeanalysedinmoredetailbelow.

(i) Migration and trade

Labourmigrationcanhavedistinctshort-andlong-runeffects in the host country.54 The short-run effectscanbestbeunderstoodinaspecific-factorframework.Consider an economy with two sectors, agriculture

and manufacturing, and three factors of production:labour, land and capital. Labour is mobile acrosssectors, while land and capital are specific to theagricultural and to the manufacturing sector,respectively.Atconstantrelativeprices,anincreaseintheendowmentof labour (due to immigration) resultsin an increase in the output of both sectors becausemoreworkersareemployed.55Sincecapitaland landcannot move between sectors, labour intensity (theamountoflabourrelativetotheamountofthespecificfactor)inproductionincreasesinbothsectors,leadingto a fall in wage rates (under the assumption thatmarkets are perfectly competitive and workers arepaid their marginal productivity). Since the output ofboth sectors increases symmetrically, there is nochange in the overall composition of output and oncomparativeadvantage.

The long-run effects of immigration, however, aredifferent due to the inter-sectoral mobility ofproduction factors. Consider an economy with twosectors, shoes and computers, and two factors ofproduction:labourandcapital.Bothfactorscanfreelymoveacrosssectors,andtheshoesector isrelativelymore labour intensive than the computer sector. TheRybczynskitheorempredictsthat,atconstantrelativeprices,anincreaseintheendowmentoflabourduetoimmigration will lead to an increase in the output ofshoes and to a decrease in the output of computers.Thelogic isthefollowing: inthe longrun,thecapital-labour ratio will remain unchanged in both sectors.Therefore,notonlywilltheadditionallabourbeentirelyabsorbed by the shoe sector, but there will also besome reallocation of labour and capital from thecomputer to theshoesector.Therefore,production inthe shoe sector will expand while production in thecomputer sector will contract.56 If the host countryenjoyed a comparative advantage in the shoe sector,this comparative advantage will be strengthened. If,

FigureC.12:Age structure of the national and non-national populations, EU, 2010 (percentage)

0

25

50

75

3% 2% 1% 0% 1% 2% 3%

Nationals Foreign nationals

Male Female

Age

100+

Source:Eurostat,Migrationandmigrantpopulationstatistics.

Note:TheagedistributionisbasedontheaggregateofallEUmemberstates.Allmigrants,bothfromEUandnon-EUmemberstates,areconsideredasforeignnationals.

TableC.2:Countries with the greatest increase in dependency ratio under zero-migration scenario, 2050

Dependency ratio in 2010

Dependency ratio in 2050

Rank CountryMedium variant

Zero-migration scenario

1UnitedArabEmirates

25 37 104

2 Qatar 20 38 95

3HongKong,China

32 78 108

4Kuwait,theStateof

34 57 79

5 Switzerland 48 72 88

6 Spain 47 87 98

7 Canada 44 70 80

8 Greece 48 82 92

9 Austria 48 77 86

10 Italy 53 88 96

Source:UnitedNations(2011a).


128

TableC.3:Percentage of high-skill immigrants and nationals in OECD countries, 1990-20001990 2000

Percentage of high skill

among natives


among immigrants

Ratio immigrants/

natives


among natives


among immigrants

Ratio immigrants/

natives

Australia 31.1% 34.6% 1.11 34.0% 40.3% 1.19

Austria 11.2% 8.4% 0.75 14.4% 12.7% 0.88

Belgium 20.8% 12.7% 0.61 27.5% 19.8% 0.72

Canada 43.8% 50.7% 1.16 51.5% 58.8% 1.14

CzechRepublic 8.5% 5.6% 0.66 10.8% 11.5% 1.06

Denmark 19.6% 13.8% 0.71 21.6% 17.3% 0.80

Finland 20.2% 16.0% 0.79 26.3% 23.8% 0.91

France 21.9% 9.9% 0.45 21.9% 16.4% 0.75

Germany 21.8% 16.9% 0.78 25.5% 21.8% 0.85

Greece 10.9% 15.1% 1.39 15.2% 15.0% 0.99

Hungary 10.1% 7.6% 0.75 12.0% 11.6% 0.97

Iceland 11.0% 24.0% 2.17 15.5% 31.4% 2.02

Ireland 14.6% 26.5% 1.82 19.4% 41.1% 2.12

Italy 6.3% 15.4% 2.45 8.7% 15.4% 1.78

Japan 21.2% 22.5% 1.06 24.0% 28.1% 1.17

Korea,Republicof 13.4% 33.1% 2.48 25.8% 38.1% 1.48

Luxembourg 20.8% 17.1% 0.82 27.5% 21.7% 0.79

Mexico 9.1% 33.8% 3.70 11.2% 44.9% 3.99

Netherlands 16.2% 17.3% 1.07 22.0% 22.0% 1.00

NewZealand 23.3% 43.6% 1.87 25.9% 40.9% 1.58

Norway 15.7% 25.2% 1.60 21.8% 28.7% 1.32

Poland 7.9% 12.0% 1.53 11.1% 14.0% 1.26

Portugal 6.5% 20.1% 3.08 8.8% 18.6% 2.10

SlovakRepublic 9.5% 7.7% 0.81 11.6% 15.2% 1.31

SouthAfrica 3.8% 16.0% 4.27 10.3% 22.0% 2.13

Spain 9.5% 16.7% 1.76 12.2% 18.5% 1.51

Sweden 20.5% 17.7% 0.86 27.5% 25.7% 0.93

Switzerland 17.2% 15.1% 0.88 17.2% 18.6% 1.08

Turkey 5.0% 11.4% 2.30 8.5% 21.5% 2.54

UnitedKingdom 13.9% 20.3% 1.46 17.8% 34.9% 1.96

UnitedStates 39.2% 41.2% 1.05 51.3% 42.7% 0.83

OECD 21.6% 29.7% 1.37 27.1% 34.8% 1.29

Source:Docquieretal.(2009).

however, its comparative advantage was in thecomputer sector, this will be weakened and possiblyreversedbyimmigration.

The example can be slightly modified to understandtheeffectsofskill-biasedmigration.Ifthecompositionof migrants is relatively more skilled, in the short runthewage rateofskilled labourwilldecrease,while in

the long run the output of skilled labour-intensivesectors will increase at the expense of unskilledlabour-intensive sectors. The same logic holds whenimmigrants are unskilled. Empirical research onadjustmentatthequantitymarginislimitedbutthefewexisting studies confirm the theoretical predictions.HansonandSlaughter(2002),forinstance,documenttherapidgrowthinapparel,textiles,foodproductsand


129

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

TableC.4:High-skill emigrant stocks and emigration rates by region, 1990 and 2000

1990 2000

Stock of high-skill emigrants

(thousands)

High-skill emigration rate

Stock of high-skill emigrants

(thousands)

High-skill emigration rate

Africa 742 11.5% 1,407 10.6%

Asia 3,349 4.9% 6,304 5.7%

CommonwealthofIndependentStates(CIS)

226 1.0% 681 2.0%

Europe 4,843 9.2% 6,535 9.2%

MiddleEast 479 12.3% 769 9.8%

NorthAmerica 1,085 1.4% 1,900 1.7%

SouthandCentralAmerica 1,559 10.0% 2,735 10.1%

Source:Docquieretal.(2009).

Note:Foragivenregion,thehigh-skillemigrationrateisdefinedastheshareofhighlyeducatedemigrantsfromtheregioninthetotalofhighlyeducatedemigrantsandnativesoftheregion.

otherlabour-intensiveindustriesinCaliforniaafterthearrivalofrelativelylow-skilledMexicanmigrants.57

A closely related question is whether trade andmigrationaresubstitutesorcomplements.Thegeneralpresumption is that theyaresubstitutes,aspredictedby the standard Heckscher-Ohlin-Samuelson (HOS)trademodel.Consider thecaseof twocountries, twogoods and two factors. As shown by Mundell (1957),there is a one-to-one relationship between relativecommodity prices and relative factor prices. Thisrelationship is identical for both countries due to theassumption of equal technology. If, due to free trade,commoditypricesareequalized,thenfactorpricesarealsoequalized.Bythesametoken,if,duetofreefactormobility, factor prices are equalized, then commodityprices are also equalized. These factor prices andcommoditypricesmustbethesameasinthecaseoffree trade.58 Therefore, trade and immigration aresubstitutes.

AsonemovesawayfromtheassumptionsthatdefinetheHOSmodel,however,thenatureoftherelationshipeasily changes, and trade and factor mobility can becomplements. Gaston and Nelson (2013) introduce aslight modification of the example discussed above,where the host country has a superior technology inthe production of the labour-intensive good. Thistechnological superiority gives rise to a comparativeadvantage in the labour-intensive good (for a givenwage-rental, theautarkypriceofthisgoodis lower inthe host than in the foreign country). If, due to freetrade,commoditypricesareequalized,thewage-rentalin thehostcountrywillexceedthewagerental in theforeign country. This will provide an incentive tomigrate from the foreign to the host country. If suchmigration is allowed, labour will flow to the hostcountry, increasing its comparative advantage in the

labour-intensive good through Rybczynski effects.Migration is, therefore, complementary to trade.Suppose now that, due to free factor mobility, factorprices are equalized. The relative price of the labour-intensive good will be lower in the host country thanabroad. If trade isallowed,productionwill increase inthe comparative advantage good. Migration is,therefore,complementarytotrade.59

Ultimately,itisanempiricalquestionwhethertradeandmigrationaresubstitutesorcomplements.Mostof theempirical evidence points towards complementarity.Using data for the United States from 1948 to 1983,Wong(1988)findsthattradeisaquantitycomplementto immigration.UsingUKdatafortheperiod1975-96,HijzenandWright (2010)show thatskilled immigrantsarequantitycomplementswithtrade.Unskilledworkersare quantity substitutes but the result is statisticallyinsignificant.60 The large literature on the effects ofmigrant networks on trade (see Box C.2), while notproviding a rigorous test based on general equilibriummodels, also points towards complementarity betweenmigration and trade. The policy implication is thatrestrictive immigration policies may not only restrictmigrationflowsbutalsotradeflows.

Immigration is not only a labour supply shock; it alsoaffects total factor productivity and consequentlyinternational trade. Peri (2012) offers convincingevidence that immigration to theUnitedStateshasapositive effect on total factor productivity and anegative effect on the skill-bias of productiontechnologies(i.e.itpromotestheadoptionofunskilled-efficient technologies). These effects can be jointlyexplainedbytwomechanisms.

First, Acemoglu’s (2002) theory of directed technicalchange predicts that the availability of a production


130

factor induces firms to adopt technologies that aremoreefficientand intensive intheuseofthatfactor.64Secondly, Peri and Sparber (2009) show thatimmigration can drive specialization according tocomparative advantage at the task level. They assertthat native workers and immigrants are imperfectsubstitutes in production, even if they have similar(limited)educationalattainments.Sinceimmigrantsarelikely to have imperfect communication skills, but

manual skills similar to those of native workers,they have a comparative advantage in occupationsrequiring manual labour, while less educated nativeworkers have a comparative advantage in occupationsdemandingcommunicationskills.Immigration,therefore,encourages workers to specialize, with consequentproductivity gains. Peri and Sparber (2009) offerempirical support for this hypothesis, using US data.Their main conclusion is that, due to specialization in

BoxC.2: Migrant networks and trade

Thepresenceofmigrantnetworkscanpromotetradebetweentheiroriginandhostcountries inat leasttwoways.First,theymighthelpovercomeinformationalbarrierstointernationaltraderelatedtolanguage,cultureor institutions, facilitate thecreationofbusiness relationshipsandmakevaluable informationonforeign sales and sourcing opportunities more readily available. Secondly, migrants boost trade if theyderivehigherutilityfromgoodsproducedintheirhostcountries.FelbermayrandToubal(2012)refertothefirstchannelasthetrade-costchannelandtothesecondasthepreferencechannel.61

Since the seminal contribution of Gould (1994), several studies have tried to quantify the positiveassociation between immigration and trade.62 The “business and social network effect” of immigrantsreceived large empirical support (see, for instance, Rauch and Trindade, 2002). In a recent paper,AleksynskaandPeri(2012)examine,asameasureofthetradebusinessnetworkofimmigrants,theshareof immigrants in managerial/sales jobs. Such immigrants are pivotal to establishing important businessconnections.Theshareofmigrantsinbusinessnetworkoccupationshasalargeandsignificanteffectonexports (but much less on imports), in line with previous studies. Specifically, each business networkimmigrant generates over ten times the value of trade as a non-business network immigrant does.AleksynskaandPeri (2012)show thatbusinessnetworksareespecially trade-enhancing in thecaseoftrade indifferentiatedgoodsandfortradebetweencountrieswithdifferent legalsystems,whileculturalsimilarities(linguistic,colonialorigin)attenuatetheeffectofbusinessnetworksontrade.63

The link between immigration and trade through networks is also affected by the composition of theimmigrantbase,asrecentlyarguedbyEggeretal.(2012).Highlyconcentratedskilledorunskilledmigrantsproducehighertradevolumesthanabalancedcompositionoftheimmigrantbase.Thiscanbeexplainedby the fact that immigrants formstrongernetworkswithin thesameskillgroup thanacrossskillgroups.Theyalsofindevidencethatapolarizationofmigrants(regardlessofwhethertheyareskilledorunskilled)tends to produce more trade in differentiated goods relative to non-differentiated goods. That is, theknowledge-creationeffectofmigrantnetworksisstrongerwhensuchnetworksarepolarized.

Migrantnetworks (inparticular, networksofgraduate students) canalsohaveamore indirecteffectontrade, through the diffusion of similar political ideas. For instance, Spilimbergo (2009) finds a positivecorrelationbetweenpoliticalsystemsinacountryoforiginandinthecountriesinwhichemigrantstudentshavestudied.Sinceformsofgovernmentandtrademaybecorrelated(Yu,2010;seeSectionC.6formoredetails),migrantnetworkscanalsoindirectlyaffecttradethroughtheirimpactonpoliticalsystems.

Untilrecently,evidenceregardingtheroleofthepreferencechannelhasbeenscant.Theearlyliteratureassumedtheimportanceofsuchachannelbecauseofthedifferencebetweentheimmigrantelasticityofimportsand the immigrantelasticityofexports–given that the tradecostchannelaffectsboth importsand exports, while the preference channel only affects the exports. Of late, additional evidence hasemerged.

Bronnenberg et al. (2012) show that US internal migrants tend to consume according to the prevalentchoicesinthestateoforigin.ThesameevidenceisfoundforIndiabyAtkin(2010),whoshowsthatinter-state migrants carry their food tastes with them, consuming food less similar to that consumed in theirhoststateandmoresimilartothatconsumedintheirstateoforigin.Finally,MazzolariandNeumark(2012)show that immigration is associated with increased ethnic diversity of restaurants in California, partlybecauseimmigrantsareconsumerswithpotentiallydifferentdemandcharacteristics,andpartlybecausetheyhaveacomparativeadvantageintheproductionofethnicfoodfromtheircountryoforigin.


131

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

differenttasks,evenlesseducatednativeworkersmaynot see adverse wage consequences from low-skillimmigration.

Immigrationalso impacts innovation inhostcountries.Asnotedabove,theshareofhighlyskilledmigrantsinthe total number of migrants to OECD countries hasincreased dramatically over the last two decades.In the United States between 1995 and 2006,67percentofthenet increasednumberofscientistsand engineers (almost half a million workers) wasforeign-born.65 High-skilled migration can alsocontributetotechnologicalprogressthroughincreasedpatenting, thus helping to develop or to strengthencomparative advantage in technology-intensivesectors.EmpiricalevidencebasedonUSandEUdatasupportsthisidea.66Atthesametime,however,thereis evidence to suggest that immigration appears todisrupttheschoolingofthenativepopulationinsomehostcountries.67

Incountriesoforigin,migrationhas importanteffectsonthe incentivestoaccumulatehumancapital,whichin turnaffectspatternsofcomparativeadvantage.Asdiscussed above, well-educated people in certaindevelopingcountriesareparticularlylikelytoemigrate.This is especially the case in certain middle-incomeeconomieswherepeoplehaveboththeincentivesandthemeanstoemigrate(DocquierandRapoport,2012).Traditionally,thistypeofmigrationhasbeenviewedasdetrimental to the country of origin because of thepositive spill-over effects associated with learning.68However, in certain circumstances it is also possiblethatemigrationresults inanet increase in thesupplyofhumancapital incountriesoforigin,creatinganet“brain gain”. As first explained by Stark and Wang(2002), this is because the prospect of emigratingincreases the returns to schooling, and therefore theincentive to investment in human capital formation.However, if only a fraction of potential migrantsmanage to emigrate, the result is a net increase inhumancapitalinthecountryoforigin.

Beine et al. (2001) show that accumulation ofadditional human capital in the country of origin canmore than compensate for the loss in skill due tomigrantoutflows.69Recently,othershaveargued thatan increase in the possibility of migration might notonlyaffectthelevelbutalsothecompositionofhumancapitalbyencouragingashiftawayfromrent-seekingactivities, which are less conducive to emigration,towards entrepreneurial ones, which are moreconducivetoemigration(Mariani,2007).Themigrationof educated individuals can also imply beneficialtransfer of knowledge, because migrants come backto their home countries to visit, to establish dualresidence, to start businesses and universities, and,sometimes, to stay (return migration). These peoplebring back new ideas and skills, which are crucialingredientstoeconomicgrowth(Freschi,2010;NyarkoandEasterly,2009;The Economist,2011).70

Asarguedabove,migrationcanchangefertilitydecisionsin both source and host countries. Mountford andRapoport (2011) propose a theoretical framework inwhich skilled migration, investment in education andfertilityareanalysedtogether.Inthehostcountry,skilledmigration will have the static effect of reducing theproportionof individualswhochoose tobecomeskilledworkers(becausetheequilibriumwageofskilledworkersdecreases), which will in turn increase the fertility rate.The dynamic effect is the opposite. Intuitively, theproportionofskilledlabourintheeconomywill increaseasaresultofskilledimmigration,whichwill inturnraisethe growth rate and eventually lead to a reduction infertility. If the dynamic effect prevails, the host countrywillaccumulatehumancapitalandhavea lower fertilityrate (and vice versa if the static effect prevails). In thecountry of origin, there is human capital accumulationduetothebraindraineffect(thepossibilityofemigrationincreases the incentive to accumulate human capital,which more than compensates for the loss in humancapital due to emigration). This accumulation of humancapitalleadstoadecreaseinthefertilityrate.71

(ii) Urbanization and trade

Urbanization is one of the most important globaldemographic trends.Asshown inTableC.5, therateofurbanizationincreasedby77percentoverthelastsix decades, rising from 29.6 per cent (0.75 billionpeople) of the global population in 1950 to52.1 per cent (3.6 billion) in 2011. Urbanization isexpected to rise further to 67.1 per cent in 2050.Developedregionsareexpectedtoseetheir levelofurbanizationincreasefrom77.4to86.3percentoverthe same period. In less developed regions, theurbanization rate is projected to increase from46.6 per cent in 2011 to 64.1 per cent in 2050. Inbothgroupsofcountries,urbanareaswillaccountforall expected population growth. Consequently, worldrural population will decline by about 0.3 million by2050(UnitedNations,2012b).

Despitethecommontrendtowardsurbanization,therearestillsignificantdifferencesacrossregions.In2011,Northern America, Latin America and the Caribbean,and Europe had the highest percentage of urbanpopulation(82.2,79.1and72.9percent,respectively).Conversely,AfricaandAsiahadthelowestpercentage(39.6 and 45.0 per cent, respectively). In the comingdecades, urban population growth will be especiallyconcentratedinthesetworegions.AfricaandAsiaareexpected to reachurbanization ratesof57.7percentand 64.4 per cent, respectively, by 2050 (UnitedNations,2012b).

Besidestheshiftinthedistributionofglobalpopulationfrom rural to urban areas, another important trend istheemergenceoflargercities.In2011,themajorityoftheworld’surbanpopulation lived incitieswith fewerthanhalfamillioninhabitants.Inthecomingdecades,however,urbanpopulationwillbemainlyconcentrated


132

incitieswithmorethanhalfamillion inhabitants.Thenumber of mega-cities, defined as cities with morethan 10 million inhabitants, will grow from 23 to 37in the period 2011-25. However, mega-cities will stillaccount fora relatively lowpercentageof theworld’surban population (13.6 per cent in 2025, up from9.9percentin2011).Populationgrowthrateswillvaryconsiderably across mega-cities, with the highestgrowthratesprojectedforLagos inNigeria,DhakainBangladesh and Shenzhen in China. Tokyo, Osaka-Kobe and Moscow will register the lowest growthrates.

Populationgrowthinurbanareascaneitherbeduetonatural increase (birth rates inexcessofdeath rates)or to net internal migration. Studies of 19th-centuryEurope (Williamson, 1988), as well as those on EastAsian countries in recent decades, suggest thaturbanization occurred at the same time asindustrialization and was the result of migration fromrural areas. However, in a number of developingcountries, especially in Sub-Saharan Africa,urbanization rates have increased prior to, orsometimes in the absence of, industrialization.According to Dyson (2011), this can be explained bythe fact that during the demographic transition themaindriverofurbanizationisnotrural-urbanmigrationbutratherthenaturalgrowthofurbancentres.72

Urbanization is among the most striking manifestationsof“lumpiness”–asituationinwhichfactorsofproduction(land, capital, natural resources and various types oflabour)areunequallydistributedwithinacountry(WorldBank, 2009; Puga, 2010).73 In a seminal contribution,CourantandDeardorff(1992)showthat lumpinesscanbe a source of comparative advantage and therefore adeterminant of trade that is distinct from other moretraditionaldeterminantsof trade,suchasdifferences infactorendowmentsandtechnologies.This isbecauseacountry tends to export the good that uses relativelyintensively the factor that is more unevenly distributedacrossitsregions.Consideracountrycomposedoftworegions. Starting from a situation in which factors areevenlydistributedacrossthetworegions,alargeenoughreallocation of one factor – for example, labour –betweenregionswillbringaboutcompletespecialization.

Atthispoint,afurtherreallocationoflabourinthesamedirection can only increase the output of the labour-intensive good in the region producing it, lowering itsautarky relative price. This creates comparativeadvantageinthelabour-intensivegood.74

Various empirical studies have tried to documentwhether lumpiness affects trade patterns. While theearly literature tended to dismiss lumpiness, recentcontributions show that it might be a relevant factor.Most of the studies are indirect tests that try toestablish whether Deardorff’s (1994) “lens condition”isviolated.Thisconditionrequiresfactorendowmentsto vary less across countries than factor inputintensities varyacrossgoods. If the setofpoints (i.e.lens) defined by regional factor abundances passesoutside the set of points defined by goods’ factorintensities, factorpriceequalization is impossibleandlumpiness may affect trade patterns. The lenscondition is found not to be violated for Japan, theUnitedKingdomandIndiabyDebaere(2004)andforOECDcountriesbyDebaereandDemiroglu(2003).

However, more recent work using city-level (asopposed to region-level) data finds that the lensconditionisviolatedinsixEuropeancountries(France,Germany,Italy,theNetherlands,PortugalandSweden),thereby indicating that urban lumpiness might be animportantdeterminantoftradepatterns(Brakmanandvan Marrewijk, 2013).75 Bernard et al. (2010) arguethatfactor lumpinessisalsosignificant inthecaseofMexico. They show that regional concentration ofskilled labour induces skill-abundant regions withinthe country to offer relatively low wages for skilledlabour and thereby specialize in the production ofrelativelyskill-intensivegoods.Asaresult,thecountrybecomes a net importer of labour-intensive products.Inthissense,thecountry’soveralllabourabundanceisunderminedbyregionalheterogeneity.

Urbanization or, more generally, agglomeration canalso influence trade patterns indirectly via its impactonproductivity.76There isampleevidence tosuggestthat workers and firms are more productive in largerand denser cities (Puga, 2010). Estimatedagglomeration gains differ across countries, largely

TableC.5:Urban and rural population, 1950-2050 (billionsandpercent)

1950 1970 2011 2030 2050

Worldpopulation 2.53 3.70 6.97 8.32 9.31

Urban(%) 29.6 36.5 52.1 59.9 67.1

Rural(%) 70.4 63.5 47.9 40.1 32.9

Populationinmoredevelopedregions 0.81 1.01 1.24 1.30 1.31

Urban(%) 54.3 66.3 77.4 81.5 86.3

Rural(%) 45.7 33.7 22.6 18.5 13.7

Populationinlessdevelopedregions 1.72 2.69 5.73 7.03 7.99

Urban(%) 17.4 25.3 46.6 55.8 64.1

Rural(%) 82.6 74.7 53.4 44.2 35.9

Source:UnitedNationsPopulationDivision,World Urbanization Prospects: The 2011 Revision database.


133

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

BoxC.3: Demography and trade: a complex relationship

Therelationbetweendemographyandtradeiscomplicatedbynumerousfactors.First,therecouldbevariablesthataffectbothdemographyand trade.Anexample is thequalityof institutions (asshown inSectionC.6).77Institutions can also have an indirect effect on demography through their impact on economic development(Rodriketal.,2004;BloomandCanning,2004).

Secondly, causality can run in both directions. Galor and Mountford (2006; 2008) claim that trade helpsexplainwhy the timingof thedemographic transitiondifferedbetween technologically advancedand lesstechnologicallyadvancedcountries. In theformer, tradereinforcedspecialization in theproductionofskill-intensive industrial goods, increasing the demand for skilled labour and the incentives to invest in humancapital–which,inturn,reducedfertilityrates.However,inthelattertradeencouragedspecializationintheproductionofunskilled-intensive,non-industrialgoods,raisingthedemandforunskilledlabourandreducingtheincentivesforhumancapitalaccumulation–which,inturn,increasedfertilityrates.

ThecontrastingdemographicexperiencesofBritainand Indiaduring the19th centuryprovidesanecdotalevidence to support this theory (Galor, 2012). During this period, Britain traded manufactured goods forprimaryproductsfromIndia.TheprocessesofindustrializationinBritainledtoasignificantincreaseinthedemandforskilledlabourinthesecondphaseoftheindustrialrevolution,triggeringademographictransitioninthe1870s. Incontrast,thelackofdemandforskilledlabour inIndiadelayedthedemographictransitionuntilthesecondhalfofthe20thcentury.GalorandMountford(2008)providecross-sectionalevidencethattrade (measured as the trade share in GDP in 1985) reduced fertility rates (measured as the averagebetween1985and1990) inOECDcountries,while it increasedfertilityrates innon-OECDcountries(seeFigureC.13).

Moreover,Doetal.(2012)showthatcomparativeadvantagehasanimpactonfertilityrates.Inparticular,countrieswithacomparativeadvantageinfemalelabour-intensivegoodsarecharacterizedbylowerfertilityrates.Thisisbecausefemalewages,andthustheopportunitycostsofchild-rearing,arehigherinthosecountries.78

Causalitymayruninbothdirectionsintherelationshipbetweentradeandmigrationaswell,sinceimmigrantstypicallymovetocountrieswhereformalorinformallinksarealreadyestablishedandwheretradewiththeirhomeland is already present (Briant et al., 2009).79 Using instrumental variable techniques, Briant et al.(2009), Peri and Requena-Silvente (2010) and Bratti et al. (2012) show that immigration leads to trade,althoughtheiranalysesdonotprecludethereversechannelco-existing.80

In thecaseofurbanization, the focushasbeenon theeffectof “lumpiness”– theunequaldistributionoffactors of production within a country – on comparative advantage and trade patterns. A large body ofliterature, however, considers the reverse causal link, investigating the consequences of trade onurbanization.81Amajor researchquestion iswhether tradeopening fostersconcentrationordispersionofeconomicactivitywithinacountry.Intheory,theeffectisambiguousasitdependsontherelativeimportanceofagglomerationanddispersionforces.82Empiricalevidenceshowsthatthedistributionofeconomicactivitypriortotradeopeningcruciallyaffectstheresults.Ingeneral,regionswithbetteraccesstoforeignmarketsbenefit.If,previoustotradeopening,theseregionswerelaggingbehind,thenopeningleadstogeographicalconvergence. If, however, these regionswerealready themostadvanced, then tradeopeningwill result ingeographicaldivergence(Brülhart,2010).

FigureC.13:Effect of trade on fertility rates, by group of countries (thousandsandpercent)

AUS

AUT

BEL

CAN

CHEDEU

DNKESP

FIN

FRA

GBR

GRC

IRL

ISL

ITA

JPN

LUX

NLD

NOR

NZL

PRT

SWE

TUR

USA

-5

0

5

Tota

l fer

tility

rate

19

85

-90

-1 1.5Log trade share in GDP 1985

(a) OECD countries

AGOARG

BDI BENBFA

BGD BGR

BHR

BHS

BLZ

BOL

BRA BRB

BTN

BWA

CAFCHL

CHN

CIVCMR

COG

COL

COM

CPV

CRI

CYP

CZE

DJI

DOMEGYETH FJI

GHA

GIN

GMBGNB

GTM

GUY

HKG

HND

HTI

HUN

IND

ISR

JAM

JOR

KEN

KOR

LAOLBR

LCA

LKALSOMAR

MDGMEXMLI

MLT

MMR

MNG

MOZ

MRT

MUS

MWI

MYS

NAM

NER

NIC

NPL

OMN

PAK

PANPER

PHLPNG

POLPRI

PRY

ROMRUS

RWA

SDN

SEN

SGP

SLB

SLE

SLVSOM

SUR

SWZ

SYR

TCD

TGO

THA

TON

TTOTUNTZA

UGA

URY VCT

VUTWSM

YEM

ZAFZAR

ZMB

ZWE

-4

4

Tota

l fer

tility

rate

19

85

-90

-1.5 1.5Log trade share in GDP 1985

(b) non-OECD countries

Source:GalorandMountford(2008).


134

becauseofcross-countrydifferencesinfactormobility(Au and Henderson, 2006; Combes, 2000), and aregenerally higher for the services sector than formanufacturing. Innovation in knowledge-intensivesectors is especially affected by the geographicalconcentration of economic activity (Audretsch andFeldman, 2004). The implication is that comparativeadvantage in these sectors will also depend onagglomeration.

In summary, recent migration patterns have beencharacterized by significant increases in skilledmigration. This has effects on innovation in the hostcountryandonhumancapitalformationinthecountryoforiginthatcanmakeskilledmigrationbeneficialforboth. Traditional trade models predict that migration(movementoffactors)andtrade(movementofgoods)aresubstitutes.However,withsmallmodificationsthatintroduce, for instance, differences in technologyacross countries, the relationship between trade andmigration becomes complementary. The pro-tradeeffectofmigrantnetworksisagoodexampleofsuchcomplementarity. Finally, internal migration, and inparticularurbanization,canalsohaveeffectsontrade.Recent theories predict that the geographicalconcentrationofafactorofproductionwithinacountrycan give rise to comparative advantage in the goodthatusesitrelativelyintensively.Empiricalevidenceisscant but recent studies suggest that this might bemore than a theoretical possibility. Finally,agglomeration can indirectly affect trade through itsimpactonproductivity.

(d) Conclusions

This section has shown that demographic change isandwillcontinuetobeashapingfactorofinternationaltrade.Ageing,migration,educationalconvergenceandwomen’sgrowingparticipationinthelabourforce–alllinkedtotheunderlyingdemographictransition–helptoshapecountries’comparativeadvantage.Moreover,asthesizeoftheworking-agepopulationincreasesinsome countries and decreases in others – and as aglobal middle class emerges – the size and thecomposition of import demand is also changing, withfurther effects on trade flows. For instance, trade inservices,suchashealthcareandeducation,islikelytoincrease.

The policies that countries adopt to meet thechallengesandopportunitiescreatedbydemographicchange will also have effects on trade patterns.Consider,forexample,thevariouspolicyoptionsfacingEastAsiancountries,suchastheRepublicofKoreaorChina, as they grapple with ageing populations (ILO,2012):developing theappropriateskillspolicies foragreying population; creating the right incentives forincreasinglabourforceparticipationamongwomenaswell as among older workers; accelerating labourproductivity growth in order to counterbalanceprojected low employment and workforce growth

rates; improving the management of labour migrationregimes to help address labour shortages; anddeveloping fiscally sustainable social protectionsystems. Through the various mechanisms discussedin this section, most of these policies are likely toaffect the evolution of comparative advantage andthereforetrade.

Moreover, improving education enrolment rates andthe quality of the educational system will improvecountries’ integration into global supply chains andincrease the sophistication of their exports.Educational policies are particularly important in theAfricancontext,wherethesizeoftheyoungpopulationwillincreasesignificantly.

While it may be relatively straightforward to predictfuture demographic trends, the many theoretical andempirical variables discussed in this section indicatethat it ismoredifficult topredict the tradeeffectsofthese trends. In short, the relationship betweendemographyandtradeiscomplex.BoxC.3concludesthissectionbyofferingsomeinsights intothefactorsbehindthiscomplexity.

2. Investment

The accumulation of physical capital can affect thenature of international trade in a variety of ways.Greaterpublic infrastructure investmentcan facilitatea country’s participation in world markets by, forinstance, reducing trade costs and hence increasingsupply capacity. Such investment in physical capitalcantherefore leadtotheemergenceof“newplayers”in international trade. Investment in roads, ports andother transport infrastructure can also strengthenregional trade, while investment in information andcommunications technology (ICT) infrastructure canenable a larger number of countries to participate inthe ever-expanding international trade in services.Over time,dependingontherateofgrowthofcapitalaccumulation relative to the rate of growth of thelabour force, it is possible for investments ininfrastructure and non-infrastructure physical capital(suchasplant,machineryandequipment) toalter thecomparative advantage of a country already widelyengagedininternationaltrade.

In an economy where factors of production, such ascapital, cannot move across countries, investmentmust be financed by domestic resources. Cross-country resource flows are, however, the currentreality. National Income Accounting shows that acountry that does not generate savings sufficient tofinanceitsowninvestmentmustattractsurplusforeignsavingsintheformofacapital inflow.Suchacountryisanetborrowerfromtheworld.Conversely,acountryinvests abroad when its domestic savings are morethan sufficient to finance domestic investment. Itsendsitssurplussavingsabroadintheformofforeigndirectinvestment(FDI)orinvestmentinforeignstocks,


135

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

bondsorrealestate.Thisstreamofsurplussavingsisreferredtoasacapitaloutflow,makingthecountryanet lender to the rest of the world. Hence, foreigncapitalflowsarethemainsourceoffinancetofill thegap between investment and domestic savings. Thisincludes FDI, portfolio investment and bank lendingfrom abroad. Other external resource flows, such asoverseas development assistance (ODA) andremittancesfrommigrantsalsoplayapart.

Capitalflowsfromabroadcanalsoaffecttradeinwaysother than through their impact on domesticinvestment. FDI, for example, may lead to trade inintermediategoodsbyfacilitatingglobalsupplychains.It may also influence a country’s comparativeadvantage by facilitating the transfer of technology.Portfolio investment and bank lending relationshipsacross countries can strengthen trade flows byreducing information asymmetries between exportersandimporters.Externalresourceflows,moregenerally,may influence a country’s exports by affecting itsexchangerate.

Thissectionfirstillustrateshowinvestmentcanaffectthenatureoftrade,irrespectiveofhowtheinvestmentis financed. It then describes other channels throughwhich different sources of investment finance canaffect trade directly. Finally, it analyses the financingof investment from an empirical standpoint. In doingso, it examines the relationship between domesticresources and domestic investment across countriesandgroupsofcountries. Italsoassessestheorderofmagnitudeanddirectionofexternal resourceflows intheworld.Theaimistoprovideapictureofhow–andwhether – different countries can – or should –enhance their investment rates and use differentinvestment flows to increase their supply capacity,change their comparative advantage and strengthentraderelationships.

(a) Impactofinvestmentonthenatureofinternationaltrade

Sub-sections (i) and (ii) outline two mechanismsthroughwhich investmentaffects thenatureof trade,irrespective of the source of finance used. Whiledomesticresourcesarenaturallyimportant,sotooaresome external finance flows that are likely to have aquantitativelystrongerimpactondomesticinvestmentthanothers.Thisishighlightedlater.Sub-sections(iii),(iv), (v) and (vi) discuss channels through whichdifferent external resource flows can directly affecttrade(i.e.otherthanthroughtheirimpactondomesticinvestment).

(i) Public investment in infrastructure

Itisworthnotingthatcapitalaccumulationintherealmof infrastructurecreation is likely tobeclosely linkedwith public investment, especially in developingeconomies (Jimenez, 1994). Government resources

are therefore crucial to financing this investment. Tothe extent that overseas development assistance,bank lending from abroad and FDI are directedtowards relevant sectors, suchas telecommunication,theymayalsocontributetoinvestmentininfrastructure.It is also important to highlight the fact that publicinvestment in both physical and human capitalinfrastructure is important for the structure of trade.Section C.1 examined how investment in skills andhuman resources can affect trade. This section istherefore limited to a discussion of physical capitalaccumulation.

Investmentinphysicalcapital,suchasroads,portsandICT infrastructure, is likely to reduce trade costs andhence increase countries’ trade participation. In thisway, capital accumulation can enable the emergenceof “new players” in world trade. This is especiallyimportantinthecontextofglobalsupplychains,wherefirmsheadquartered inadvancedeconomiesoffshorecertaintasksinvolvedintheproductionofafinalgoodto developing countries. Given that the decision tooffshore revolves around finding cost-efficientsuppliers of that task worldwide, wage costs are nottheonlyrelevantvariable.Aminimumlevelandqualityof infrastructure, created by investment in physicalcapital,isalsolikelytoplayanimportantrole(BaldwinandLopez-Gonzalez,2012;Kimura,2009;Hewetal.,2009). Production networks, for instance, requirefluidity, low costs and security in the transmission ofinformation.Forthis,ahigh-qualitytelecommunicationssystemisessential(GrossmanandHelpman,2005).

Bettertransportinfrastructurereducestransportcostsandhenceisassociatedwithhighervolumesoftrade.Using data on a cross-section of countries,FigureC.14showsthispositiveassociationinthecaseofchangesinroadnetworkdensityandchangesintheshareoftrade inGDP.Usingmorerigorousstatisticalmethods,NordasandPiermartini(2004)estimatethatdoublingthekilometresofpavedroadsper100squarekilometres increases trade by 13 per cent. Similarly,theyshowthatdoublingthenumberofpavedairportsper squarekilometresof territory inacountrybooststradeby14percent. Investment inbetterqualityandmore reliable ICT infrastructure also leads to areductionintradecostsbyreducingthebarrierswhichinhibit economic exchange over long distances (Finket al., 2005). A more detailed discussion on therelationshipbetweentransportandICTinfrastructure,ontheonehand,andinternationaltradeflows,ontheother,isprovidedinSectionsC.5andC.3,respectively.

The lack of adequate transport infrastructureundoubtedly reduces Africa’s ability to participate inthe world economy. According to Nkuepo (2012), thecontinenthasfewerkilometresofroadnowthanitdidseveral decades ago, with about 70 per cent of theruralpopulation livingmore than twokilometresawayfrom an all-season road. Figure C.15 shows thatbetween1990and2005,India’sroadnetworkalmost


136

FigureC.14:Total road network and trade openness, 1990-2005 (percentagechangeinkilometresandtradetoGDPratio)

Trad

e sh

are

grow

th

Road network growth-100

-50

150

250

200

300

350

-20 100 1200 40 806020

0

50

100

Source:TheWorldBank,World Development Indicators.

Note:Whilemorerecentdataonroadnetworksareavailable,itislimitedtoamuchsmallersub-setofcountries.

FigureC.15:Increases in total road network – top ten countries, 1990-2005 (percentagechangeinkilometres)

0

120

100

80

60

40

20

Indi

a

Kor

ea, R

ep. o

f

Om

an

FYR

Mac

edon

ia

Sau

di A

rabi

a,K

ingd

om o

f

Nig

eria

Nig

er

The

Gam

bia

Pak

ista

n

Bol

ivia

, Plu

rinat

iona

lS

tate

of

Source:TheWorldBank,World Development Indicators.

Note:ValuesforNigeriaandTheGambiahavebeenextrapolatedforoneyearintheperiodunderconsideration.Moregenerally,whilemorerecentdataonroadnetworksareavailable,itislimitedtoamuchsmallersub-setofcountries.Inordertoavoidmakinglarge-scaleextrapolations,wechose2005asthecut-offpoint.

doubled,thelargestpercentageincreaseintheworld.Increases in the road density of certain Africancountries during this 15-year period were alsosignificant. It can be seen in Figure C.15 that thepercentage increase in the road network of Nigeria,NigerandTheGambiawasabout60percentbetween1990and2005.Itislikelythatwithincreasingratesofeconomicgrowth83anda rangeofprospectivepolicyreforms, a larger pool of government resources andmoreefficiency inpublic investmentwillenablemany

more African countries to increase their road densityandhencetheirsupplycapacity.

MostAfricancountriesalsofind ithardtocompete intheworldmarketowing to inadequate, inefficientandvery expensive telecommunication services. This isreflected in Figure C.16, which shows a large gap intelecommunicationinvestmentsbetweenSouthAfricaandthenexttencountriesinthecontinent.Eveninpercapita terms, it shows that along with four islandeconomies, South Africa and other members of theSouthernAfricanCustomsUnion–Botswana,NamibiaandSwaziland–areamongthetencountrieswiththehighest telecommunications investment in Africa.AttractingFDIthroughimprovedregulatoryinstitutionsandpoliciescouldplayanimportantfutureroleinthisregard.Infact,Djiofack-ZebazeandKeck(2009)showthat strong regulatory institutions are a key factoraffecting theperformanceof the telecommunicationssector.

Infrastructureinvestmentisalsolikelytoinfluencetheregionalization of trade in the future. Consider, forinstance, the case of Africa. Limao and Venables(2001)showthat the low levelof tradewithinAfricanin the 1990s is explained to a large extent by theirpoor infrastructure. Even today, the transportation ofgoods by roads within the region is more expensiverelative to other parts of the world. Flying from onecountry to another is expensive, and railwayinfrastructure barely links African countries (Nkuepo,2012). Poor communications infrastructure continuestoberegardedasamajor impediment to tradewithinAfricaaswell(MupelaandSzirmai,2012).

Initiating and encouraging more cooperation ininfrastructure development projects – for example, in


137

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

FigureC.16:Average annual investment in telecommunications in Africa, 1986-2005 (US$millionandpercapitaUS$)

0

1,200

800

1,000

600

400

200

Sou

th A

fric

a

Egy

pt

Mor

occo

Alg

eria

Ken

ya

Nig

eria

Tuni

sia

Côt

e d’

Ivoi

re

Sen

egal

Cam

eroo

n

0

120

80

100

60

40

20

Sey

chel

les

Mau

ritiu

s

Sou

th A

fric

a

Bot

swan

a

Cap

e Ve

rde

Gab

on

Tuni

sia

Nam

ibia

São

Tom

éan

d Pr

ínci

pe

Sw

azila

nd

Source:WorldBank,African Development Indicators.

Note:Angolawasexcludedbecauseofinsufficientdataavailability.WhilemorerecentdataareavailablefromtheInternationalTelecommunicationUnion(ITU),theseincludelessthanone-thirdofthecountriesconsideredaboveeitherbecauseofdataunavailabilityorrelevantharmonizationissues.

telecommunications, transportation,powergenerationand the provision of water – at the regional level willincrease access to these facilities, thereby loweringtransactionscostsandboosting tradeamongAfricancountries in the future (Dupasquier and Osakwe,2006). A future COMESA-SADC-EAC (CommonMarket for Eastern and Southern Africa – SouthernAfrican Development Community – East AfricanCommunity) tripartite preferential trade agreement(PTA) and even a pan-African PTA could thereforeprovideamajorboosttotradewithinAfrica.

Furthermore,investmentinICTinfrastructuremaygivea further impetus to expansion of trade in services.Cross-bordertradeinservices(mode1oftheGeneralAgreementonTradeinServices),for instance, largelydepends on telecommunications as the channel fortransactions. Freund and Weinhold (2004) find thataccess to the internet for trading partners had asignificant impact on US imports of business,professional and technical services. Developingeconomieshithertonot involved inservicestrade inasignificant way can utilize investments in ICTinfrastructure to make initial inroads into thisincreasinglyimportantworldmarket.English-speakingAfricancountries,forexample,couldbecomeoffshorelocations for call-centres and business processoutsourcing. South Africa has already started downthispathdue to thequalityof its telecommunicationsinfrastructure even though high costs remain aproblem.84 Mauritius, another recently successfulcountry, has taken direct regulatory action to ensurethat costs are not a barrier to developing servicesoffshoringbusinesses.85

(ii) Capital accumulation and changing comparative advantage

If a particular sector is more sensitive than others tothequalityof infrastructure,thenpublic infrastructureinvestment can affect a country’s comparativeadvantage.Forexample,YeapleandGolub(2007)find

that the provision of road infrastructure consistentlyappears to be a significant factor in a sector’s totalfactor productivity (TFP) growth and hence in acountry’sproductionspecialization.Theauthorsshowthat road infrastructure appears to be particularlyimportantforproductivitygrowthinthetransportationequipmentsectorandforspecializingintheproductionof textiles and apparel. Good telecommunicationservices may also influence comparative advantageandhencethepatternofinternationalspecialization.

ICT infrastructure is particularly important forinformation-intensive sectors. These are typicallysectorsthatproducegoodswithshortproductcycles,experience rapid fluctuations in consumer tastes,enjoy rapid technology development, and whereinternational vertical fragmentation is common.Consumerelectronics,forexample,ischaracterizedbyall these features. Fashion clothing is an example ofgoods for which tastes change rapidly while theautomotive sector is an example of a sector whereglobal production fragmentation is important (WorldTradeOrganization,2004a).

Investment in non-infrastructure creating physicalcapital,carriedoutlargelybyprivateplayers,canalsoexert an important influence on comparativeadvantage.AccordingtotheHecksher-Ohlinmodeloftrade,countriesshouldproduceandexportgoodsthatuse intensively relatively abundant factors. So for acountry with an abundant supply of unskilled labour,relative to capital, trade based on comparativeadvantage would imply specializing in the productionofunskilledlabour-intensivegoods.

The Rybczynski theorem, however, shows that atconstant relative goods prices, an increase in acountry’s endowment of one factor leads to a morethanproportionalexpansionoftheoutputofthegoodwhich uses that factor intensively and an absolutedeclineoftheoutputoftheothergood.Hence,evenina relatively unskilled labour-intensive economy, an


138

increase in the supply of capital can result in anincrease in the production of the relatively capital-intensive good. Over the medium to long run, theaccumulationofcapitalmaybe largeenough, relativeto thegrowthof the labour force, toalteracountry’scomparativeadvantage,therebymakingcountrieslessspecialized (as alluded to in Section B.2(c)). Thetransformation of Japan from a relatively labour-intensivetoarelativelycapital-intensiveeconomyisacaseinpoint(seeBoxC.4).

Figure C.17 shows that, between 1990 and 2009,severalunskilledlabour-intensiveeconomiessawlargeincreases in their capital-labour ratios. China, VietNamandIndiatopthelistastheircapital-labourratiosincreasedsixfold, fourfoldand threefold, respectively.These and other middle-income countries haverelativelyhighinvestmentrates.Infact,datashowthatunskilled labour-intensive economies, such as China,VietNamandIndia,wereamongthetencountrieswiththe highest average investment rates between 2000and 2010.86 However, many of them also have highpopulation growth rates. Whether these countriestransform themselves into relatively capital-abundanteconomies in the future depends on how the rate ofgrowth of physical capital compares with that of thelabour force. InanemergingeconomysuchasChina,wherepopulationgrowth rateshavesloweddownbutwhere investment in physical capital continuesunabated, this may result in a change in comparativeadvantageinthefuture.

The trade literature suggests that the evolution ofcapital accumulation in an economy, and hencecomparative advantage, is closely linked to itsdomestic savings rates – i.e. a country with a highsavingsrateexportsarelativelycapital-intensivegood(OnikiandUzawa,1965;Stiglitz,1970;GalorandLin,1997; Hu and Shimomura, 2007; Chen et al., 2008).The case of Japan validates this theory. Whiledomestic resources are naturally important fordomesticinvestmentinphysicalcapitalandhenceforcomparativeadvantage,itisworthnotingthatresourceflowsfromabroadcanalsoplayapart(seeBoxC.5foradiscussiononwhichoftheseislikelytohaveastrongeffectondomesticinvestment).

Forinstance,inthecaseofCostaRica,large-scaleFDIbyanumberofmultinationalsestablishedmanufacturingplants in several high-technology electronics sectors,with Intel leading the way in semi-conductor devices(Rodríguez-Clare, 2001). This enabled the country tospecialize in technologically more complex activitiesthan apparel exports. Investment to establish aknowledgecentretodevelopsoftwareandcontributetoIntel’s design processes further strengthened thisprocess of changing comparative advantage. CostaRica’s business-friendly economic and politicalinstitutions, together with its well-educated labourforce,wereinstrumentalinattractingthisFDI(Sanchez-Ancochea,2006).

(iii) Intertwining of trade and FDI

EconomictheoriesofinternationaltradeandFDIhavetended to develop separately. Hence, the traditionaltrademodel,inwhichcomparativeadvantageisbasedon differences in relative factor endowments,87assumes factor immobility among countries. In otherwords, trade and factor mobility are substitutes. Forexample, in lieu of capital from the capital-abundantcountryflowing to thecapital-scarcecountry,capital-intensive goods are exported by the former to thelatter.

However, this hypothesis is somewhat dissociatedfromexistingeconomic reality,which ischaracterizedbyincreasinginternationalfactormobility,mainlyintheform of FDI flows that finance investment (therelationship between trade and the mobility of labouracross countries is discussed in Section C.1).Multinational firms, with their headquarters in onecountry, establish operations under their ownershipand managerial control in another country.88 Giventhattwo-thirdsofworldexportsaregovernedbythesemultinational firms, deciding where to invest issimultaneously deciding from where to trade(UNCTAD,2012).

Totheextentthatlocalproductioninthe“host”countryreplacesexportsfromthe“home”country,FDIandtradecanbesubstitutes.Thisisespeciallytruefor“horizontal”FDI,whichconsistsofinvestmentinproductionfacilitiesabroad to produce the same goods and services asthose produced at home to serve the host countrymarket (Markusen, 1984). Increasingly, however, FDIandtradeareviewedasbeingcomplements(Helpman,1984).ForhorizontalFDI,thismaybebecauseaffiliatesorsubsidiariesareusedas“exportplatforms”–thatis,investment in production capacity results in exportsfrom that country to other third-country markets in itsproximity(Grossmanetal.,2006).

Forexample,evidencesuggeststhathighlevelsofFDIin theautomotive industry contributed significantly to

FigureC.17:Capital-labour ratios, 1990-2009 (percentagechange)

0%

600%

400%

300%

200%

100%

Chi

na

Viet

Nam

Indi

a

Cor

ea,

Rep

. de

Taila

ndia

Chi

le

Sri

Lank

a

Turq

uía

Hon

g K

ong,

Chi

na

Ban

glad

esh

500%

Source:Fouréetal.(2012).

Note:OnlyeconomieswithGDPaboveUS$10billionareconsidered.


139

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

BoxC.4: Investment and changing comparative advantage – the case of Japan

Starting as a labour-abundant country, Japan transformed itself into a leading exporter of capital-intensiveproductsintheperiodaftertheSecondWorldWar.Heller(1976)observesthathighinvestmentrates,causedbysoaringdomesticsavingsandAmericanaidinflows,pushedJapanintoarelativelycapital-abundantpositionbetween 1956 and 1969. Using data on commodity trade statistics, he finds that Japan’s comparativeadvantagehadaccordinglychangedasitsexportswererelativelymorecapitalintensive(seeSectionC.1).Inalaterstudy,BalassaandNoland(1988)findthat theJapanese investment ratecontinuedtobesubstantiallyhigher than thoseofother industrialcountries,suchas theUnitedStates,between1973and1985. In theirexaminationofchangingtradepatterns,theauthorsfindthatrelativetotheperiodbetween1967and1983,Japan’s revealedcomparativeadvantage (RCA) inunskilled labour-intensive industries, suchasapparel andleather,haddiminished.Incontrast,thecountrydevelopedacomparativeadvantageinskilledlabourandhigh-technologyintensiveindustries.SimilarresultsarealsofoundinBalassaandNoland(1989)andLee(1986).

FigureC.18showsthatthechangingshareofmachineryandtransportequipment–regardedasoneofthemost capital-intensive sectors – in Japan’s total exports and the evolving capital-labour ratio in Japanbetween1960and1990arehighlycorrelated.ThisevidencesuggeststhatJapanisagoodexampleofaninvestment-drivenchangeinacountry’srelativefactorendowmentsandcomparativeadvantage.

FigureC.18:Japan’s capital-labour ratio and the share of machinery and transport equipment exports in its total exports, 1960-90

0.2

0.4

0.6

0.8 180

160

140

120

100

80

60

40

0.3

0.5

0.7

20

0

Capital stock per capita Share of machinery and transport equipment exports in total exports

19

60

19

62

Sha

re

Cap

ital s

tock

per

cap

ita

19

64

19

66

19

68

19

70

19

72

19

74

19

76

19

78

19

80

19

82

19

84

19

86

19

88

19

90

Sources:UNComtradeandFouréetal.(2012).

capital goods, design services and research anddevelopment, from the home country. At the sametime, the home country imports varieties of a finalgood from the host country as a result of the supplycapacitycreatedbytheFDI.For instance,ArnoldandJavorcik (2009)find that receivingFDIenhanced theintegration of Indonesian plants into the globaleconomy through increased export intensity andgreater reliance on imports of intermediate inputs.Whatismore,third-countrymarketsmayalsobegintoimportfromthehostcountry.

Consider the electronics industry where FDI inflows,especially from firms in developed and “newlyindustrialized”countries,haveestablishedMalaysiaasaglobalproductionhub.IntermediateinputsareimportedfromthecountryoftheparentfirmintoMalaysia.Atthe

the Czech Republic’s supply capacity and hence itsexportstothird-countrymarketsuntil2008(EconomistIntelligenceUnit,2010).Similarly,Tunea (2006)findsthatNAFTA(NorthAmericanFreeTradeAgreement)-led foreign investment in Mexico’s manufacturingsectorwasdrivenbyitspotentialasanexportplatformfor neighbouring countries. In the absence of FDI,thesemarketsmighthaveremaineduntappedbecauseexporting to them directly from the home countrywouldhaveentailedsignificanttransportcosts.

At the same time, home country operations of theparentfirmcanbelinkedwithhostcountryoperationsvia “vertical”FDI,which involves the fragmentationoftheproductionprocessalongglobalsupplychains. Inthis set-up (see Section B.2(e)), there are increasedexportpossibilitiesfor intermediateproducts,suchas


140

BoxC.5: Contribution of external resource flows to domestic investment

Capital inflows

FDI can affect domestic investment by contributing directly to new plant and equipment (“greenfield”investment) or by acquiring (or merging with) an existing local firm. FDI may also produce investmentspillovers beyond the direct increase in capital stock. For example, it could “crowd-in” the host country’sdomestic investment through linkagesamongfirms–multinationalcorporationsmaypurchasespecializedinputsfromdomesticsuppliers,therebyencouragingnewinvestmentbylocalfirms(Mileva,2008).AccordingtoBorenzsteinetal.(1998),FDIcouldalsospurdomesticinvestmentbyloweringthecostsofadoptingnewtechnologies.ForasampleoftenCIScountriesandAlbania,Mileva(2008)showsthatFDIflowscrowd-indomesticinvestment.Atthesametime,it ispossibleforFDIto“crowd-out”domesticinvestmentbyraisingproductivityandhencewages.

Inflowsofportfolioinvestmentandbanklendingfromabroadcancomplementdomesticsavingsinpromotingdomesticinvestmentbyloweringthecostofcapital(LevineandZervos,1998;Manova,2008a).89Inastudyof11developingcountries,forinstance,Henry(2000)findsthat–onaverage–domesticprivateinvestmentgrowsby22percentagepointsfasterintheperiodafterstockmarketliberalization.Inastudywhichanalysesa larger number of countries, Henry (2003) reaffirms his earlier findings by estimating that the domesticinvestment rate increases by approximately one percentage point every year following capital accountliberalization.

Incontrast,Pal(2006)andMileva(2008)findaweakrelationshipbetweenportfolioinvestmentflowsfromabroadandtherealeconomyinthecaseofIndiaandeconomiesintransition.Thismaybeexplained,inpart,bythefactthatportfolioinvestmentflowsarerelativelymoreshort-terminnature.Moreover,ifforeigncapitalislimitedtostockpurchasesonthesecondarymarket,equityinvestmentincreasesthepriceofthesharesbut not the flow of funds to the companies that wish to increase investment (Kraay and Ventura, 1999).According to Mody and Murshid (2005), multinationals have increasingly focused on acquiring existingassets rather than purchasing newly issued equity. Such capital inflows may still contribute to capitalaccumulationifthenewforeignownersmodernizeorexpandtheiracquisitionsbyinvestinginnewtechnology(Mileva,2008).

Analysingasampleof58developingcountriesbetween1978and1995,BosworthandCollins(1999)showthat while FDI appears to bring about close to a one-for-one increase in domestic investment, there isvirtuallynodiscerniblerelationshipbetweenportfolioinflowsandinvestment,andtheimpactofbanklendingisonlyminor.AccordingtoModyetal.(2003),thismaybeattributabletoaninformationaladvantage(basedon their specialized technical knowledge and market experience), which allows FDI investors to “outbid”other investor-typesfor themostproductiveopportunities. Incountrieswithmissingor inefficientmarkets,foreigninvestorswillprefertooperatedirectlyinsteadofrelyingonlocalfinancialmarkets.

The importance of capital inflows to domestic investment also depends on the subsequent decisions ofdomestic investors. If residual domestic investment opportunities offer low returns, especially since newcapitalinflowscouldindirectlyreducetherisk-freerate,domesticsavingsmayactuallybechannelledoutofthecountry insearchofhigher returnsor lower risk (ModyandMurshid,2005).Suchcapitaloutflowmayactually reduce the resourcesavailable fordomestic investment. It is also likely that countrieswithbetterpoliciesandinstitutions(asdescribedinSectionC.6)arelikelytohavegreatersuccessinabsorbingforeigncapital inflows for domestic investment by creating an environment conducive for the diffusion of newtechnologiesandreducingtheriskofholdingdomesticassets.

Other external resource flows

Intheempiricalliteratureonthesubject,opinionsaredividedconcerningtheeffectofoverseasdevelopmentassistance (ODA) on investment, with results often being a function of the choice of data sample andestimation technique. For instance, while Boone (1996) and Hansen and Tarp (2001) find a statisticallysignificantpositiveimpactofODAoninvestment,DollarandEasterly(1999)andCollierandDollar(2001)donot. It isargued thataidmoneymeant for investment isoftenusedfordisaster relief (DollarandEasterly,1999),financingtaxcuts(Devarajanetal.,1999)orsupportingconsumption(Boone,1994).

Manystudiesfindthatremittancesfrommigrantsarepositivelycorrelatedwithentrepreneurshipandsmallbusinessinvestmentindevelopingeconomies(WoodruffandZenteno,2007;Mesnard,2004).Forinstance,comparing expenditures in Mexican households with and without international migrants, Taylor and


141

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

Mora(2006)findthattheformerspentmoreoninvestmentandlessonconsumptionthanotherhouseholdsatthesameincomelevel.Adams(2005)presentssimilarfindingsforGuatemala.Thereare,however,studieswhichshowthatremittancesmainlycontributetohigherconsumption(BrownandAhlburg,1999).Acentralmethodologicalconcerninthisregardisthatanyobservedrelationshipbetweenremittancesandhouseholdinvestmentmaysimplyreflecttheinfluenceofunobservedthirdfactors.Inarecentstudy,Yang(2008)findsthatexogenousshocks to the incomeofPhilippinemigranthouseholds,manifested inpart viachanges inremittances,havelargeeffectsonrelativelycapital-intensiveentrepreneurialactivity,suchasmanufacturingand transport services.Theauthorargues that remittance receiptsenable investment thatwaspreviouslyinhibitedbycreditconstraints.

same time, the country ranks among of the world’slargestexportersofsemi-conductordevicesandaudio-visual equipment to the FDI-source countries or othermarkets (Malaysian Industrial Development Authority,2006). The same is true for the automobiles industrywhere FDI has resulted in increased exports ofautomobilesfromThailandtodevelopedeconomiesaswellneighboursintheregion(Nagetal.,2007).

Whatifafirmproducesnotonebutseveralfinalgoods?ItresultsinotherchannelsofcomplementaritybetweenFDIandexportsthroughitseffectondemand.First,theestablishmentofaproductiveunitforoneofitsproductsin a foreign market creates a reputation for its brand.This can increase the demand and, consequently, theexportsofotherfinalgoodsforthatmarket(LipseyandWeiss,1984).Secondly,themarketing,distributionanddeliverycapabilitiescreatedbyFDIinoneproductmightenablethehomecountrytoexportallitsfinalproductsthatwouldnotreachcustomersintheforeignmarketinthe absence of FDI (Blonigen, 2001). Thirdly, foreigndemandforafirm’sotherfinalgoodscanbestimulatedthrough the supply of valuable after-sale servicesresulting from FDI, which represents a permanent

commitment to customers in the host country market(HeadandRies,2001).Finally,exports fromthehometo the host country may also increase for the reasonthatFDIstimulatesthehostcountry’spurchasingpowerforimportingbothintermediateandfinalgoodsfromthehomecountry.

Insum,thetheorysuggeststhatgreaterFDIcanleadto more trade. However, can trade also boost FDIflows?Analysissuggeststhatitcan.Exportscanbeasource of information on the host country and henceenhance capital flows (Portes and Rey, 2005). FDImay also follow exports in order to preserve marketsthatwerepreviouslyestablishedbyexports (ObstfeldandTaylor,2004).Tradeassociatedwithcross-bordervertical integration, in particular, may boost FDI as itassuresownershipadvantagesandamarket.

The data show a systematic positive associationbetween trade and FDI, thereby highlighting theircomplementarity (see Figure C.19). Evidence fromparticular sectors and countries reinforces thisfinding. The trade orientation of FDI is well-represented in thedevelopmentexperienceofChina

FigureC.19:World trade and foreign direct investment, 1980-2011 (US$billion)

0

19

80

19

81

19

82

19

83

19

84

19

85

19

86

19

87

19

88

19

89

19

90

19

91

19

92

19

93

19

94

19

95

19

96

19

97

19

98

19

99

20

00

20

01

20

02

20

03

20

04

20

05

20

06

20

07

20

08

20

09

20

10

20

11

5,000

10,000

15,000

20,000

25,000

Wor

ld F

DI i

nflo

ws

Wor

ld e

xpor

ts o

f goo

ds a

nd c

omm

erci

al s

ervi

ces

2,500

1,500

2,000

1,000

500

0

World exports of goods and commercial services World FDI inflows

Sources:WTOSecretariat,UNCTAD.


142

where foreign investment enterprises accounted for58percentoftotalexports in2005(WTO,2010). Itisequallywell-illustratedinothercases.Inthetextilesindustry, for example, FDI from Hong Kong (China)and Chinese Taipei dominates export production inLesotho, Madagascar and Mauritius, while FDI fromtheUnitedStatesdoessointheDominicanRepublic(McNamara, 2008). Furthermore, several empiricalstudies find that more FDI establishing affiliatesabroad is associated with more, rather than less,exports from the parent firm in the home country(Bergsten et al. , 1978; Lipsey and Weiss, 1981;Blomstrom et al. , 1988; Buiges and Jacquemin,1994). Such complementarity has been found to beespeciallytruefor intra-firmexports,highlightingtheimportance of vertical relationships among variousinternationalaffiliates(Pearce,1990).

(iv) FDI, technology diffusion and changing comparative advantage

A country’s position in a global supply chain isgenerally correlated with its comparative advantage.Developing countries complete low value-addedunskilled labour-intensive tasks because they have arelatively abundant supply of unskilled labour. It isadvanced economies where the skill and capital-intensive tasks are completed. In modern economies,however, much comparative advantage is man-made.So is itpossible foracountry thathasacomparativeadvantage inunskilled labour-intensive tasks today tohave a comparative advantage in high-technology-intensivetaskstomorrow?

In Asia, several firms in Japan offshored unskilledlabour-intensivemanufacturingtaskstotheRepublicof Korea, Chinese Taipei, Hong Kong (China) andSingapore, starting in the 1970s (Baldwin, 2012a).Hence, thesecountriesenteredglobalsupplychainsby specializing in component manufacturing andproductassembly.As they industrialized, theybeganto manufacture sophisticated intermediate inputs,which they earlier imported from advancedeconomies.Thesenewlyindustrializedcountriesalsoexpanded into the design and distribution of goodsand hence captured more of the total value added(Wood,2001).

While investment inhighereducation is likely tohaveplayed an important role, the diffusion of technologyand knowledge associated with FDI played a crucialroleinupgrading.Inastudyof105countriesbetween1984 and 2000, for instance, Harding and Javorcik(2012)findapositiverelationshipbetweenFDIandthequality of exports in developing countries. Globalsupply chains have made technology internationallymore mobile by offshoring firm-specific technicalknow-how, especially via investment by multinationalcompanies in the establishment of subsidiariesoverseas.Thishelpedtoenabledevelopingcountries,such as Hong Kong (China), the Republic of Korea,

SingaporeandChineseTaipeitomoveuptheproductladder in terms of capital intensity, technologicalcontent, design and quality. Signs of technologyupgradingandchangingexportorientation, facilitatedbyFDI, arealready visible inChina– it hasbegun toproduce sophisticated intermediate goods andservices that previously would have been imported –andarelikelytoonlygetstrongerinthefuture(Rodrik,2006).

A discussion on the mechanisms through which FDI,both“horizontal”and“vertical”,canleadtotechnologydiffusionisprovidedinSectionC.3.Thefollowingarea few examples. Evidence for direct technologytransfer frommultinationalaffiliates to localsuppliersor technology upgrading due to higher qualityrequirements on intermediate inputs from domesticsuppliers is documented in the case of vertical FDIflows into Lithuania and Indonesia (Javorcik, 2004;BlalockandGertler,2008).

Iacovone et al. (2011) find that following the entry ofWalmex (the Mexican affiliate of Walmart), localMexican retailers started to adopt advancedtechnologies, such as cold chain (a temperature-controlled supply chain), in order to catch up. This isindicative of indirect technology transfer. Knowledgespillovers are also documented in the case of Intel’sFDIinCostaRica.Intelinvestedheavilyinthetrainingof its employees, leading to learning-by-doing andeventhecreationofseveral“spin-off”firms. Intelalsocollaboratedwithpublicuniversitiesinordertoimprovetheircurriculumandteachertrainingintechnicalfields(Rodríguez-Clare,2001).

(v) Information, capital flows from abroad and international trade

Itisarguedthatportfolioinvestmentandbanklendingrelationships between countries can generateinformationthat leadstoanincreaseinbilateraltrade(Lane and Milesi-Ferretti, 2008; Jeanneau and Micu,2002; Portes and Rey, 2005). The relationshipbetween lenders abroadandborrowers at home–orviceversa–can improvetheexchangeof informationbetweenexportersandimporters,therebyencouraginginternational trade. At the same time, existing traderelationshipsmayallowforeigninvestorsandbankstogather information about the destination country andhenceservetoincreaseportfolioinvestmentandbanklendingtothatcountry.Thiscomplementaritybetweenportfolioinvestmentandbanklendingfromabroad,ontheonehand,andtradeflows,ontheother,isdepictedinFiguresC.20andC.21.

Empirical evidence, generated by rigorous statisticalmethods, also supports this complementarity. Usingdata for international portfolio holdings of 67 sourcecountries (including all major international investors)and 200 destination countries, Lane and Milesi-Ferretti (2008) find that bilateral international equity


143

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

positions are strongly correlated with bilateral trade.This evidence is indicative of an information-drivenrelationship between trade and capital flows that isparticularly strong when the collection of informationissimplified.Theauthorsfindthatacommonlanguage,for example, raises equity holdings by 50 per cent.Similarly, Portes and Rey (2005) show that a gravitytypeequationexplains70percentof thevariation inportfolio investment for a sample of developedcountries. They test explicitly for informationasymmetriesusingproxyvariables,suchastelephone

traffic,andshowthatthischannelishighlysignificant.Moreover,theyincludetheseproxiesintradeequationsandshowthattheresultsimprovesignificantly.

Somestudies inthe literatureusemoresophisticatedstatisticaltechniquesinordertoestablishcausalityinthe relationship between trade and capital flowsacrosscountries.For instance,AviatandCoeurdacier(2007)findthata10percentincreaseintradeleadsto a 6 per cent higher level of portfolio investment;causality in the other direction is weaker but still

FigureC.20:World trade and foreign portfolio investment, 2003-10 (US$billion)

Wor

ld e

xpor

ts o

f goo

ds a

nd c

omm

erci

al s

ervi

ces

Wor

ld F

PI i

nflo

ws

0

10,000

25,000

5,000

20,000

15,000

1,400

1,200

1,000

800

600

400

200

0

2003 2008 2009 20102005 2006 20072004

World exports of goods and commercial services World foreign portfolio investment (FPI) inflows

Sources:WorldBankWDI,WTOSecretariat.

FigureC.21:World trade and foreign claims, 1999-2011 (US$billion)

Wor

ld e

xpor

ts o

f goo

ds a

nd c

omm

erci

al s

ervi

ces

Tota

l for

eign

cla

ims

by a

ll re

port

ing

bank

s

0

10,000

25,000

5,000

20,000

15,000

40,000

35,000

30,000

25,000

20,000

15,000

10,000

5,000

0

1999 2004 2005 2006 2007 2008 2009 2010 20112001 2002 20032000

World exports of goods and commercial services Total foreign claims by all reporting banks

Sources:BankofInternationalSettlements,WTOSecretariat.

Note:Foreignclaimsrepresentloansextendedtoresidentsoutsidethecountryinwhichthebanksareheadquartered.


144

significant. Similarly, Jeanneau and Micu (2002) findthat while bilateral trade is significant and highlypositive in explaining bank lending patterns fromadvanced economies (the United States, Japan, theUnitedKingdom,Germany,France, ItalyandSpain)toAsian and Latin American economies (Argentina,Brazil, Chile, Indonesia, the Republic of Korea,Malaysia, Mexico, the Philippines, Thailand andBolivarian Republic of Venezuela), there is alsocausalityintheotherdirection.

(vi) Capital flows, exchange rates and international trade

Capital inflows can lead to an appreciation of theexchange rate in recipient countries, thereby hurtingtheir export competitiveness (Corden and Neary,1982;Agenor,1998;Lartey,2008). Inflowsofforeigncurrencyraisethedemandforbothtradableandnon-tradablegoodsproducedinaneconomy.Inthecontextof a small openeconomy, an increase in thedemandfor tradable goods does not affect their prices sincethese are determined in world markets. At the sametime, the increased demand for non-tradable goodsplacesanupwardpressureontheirpricesandtherebyresults in an appreciation of the real exchange rate.Under a flexible exchange rate mechanism, both thenominal and the real exchange rate appreciate as areaction to the increase in the relative price of non-traded goods. Under a fixed exchange ratearrangement, the expanding money supply increasesdomesticprices,therebyleadingtoarealappreciationof the currency. It is worth noting that in mostcountries, exchange rate appreciation is sporadic,volatile and short-term in nature. Appreciation over alongerperiodoccursonlyinarelativelyfewnumberofcases(SyandTabarraei,2010).

If policy-makers choose to dilute the effect of realexchange rate appreciation by sterilizing incomingresourcesthroughopenmarketoperations,itwillleadtoan increase indomesticdebtalongwithapossibleincrease in the domestic interest rate. This, in turn,may further attract more inflows from abroad andcreate a vicious circle of expected devaluation andcapitalflight,therebyaffectinginvestmentandtradeinthefuture(Calvoetal.,1993).

Several studieshave shown that large capital inflowshave resulted in exchange rate appreciation indeveloping economies (Corden, 1994; Lartey, 2007;Edwards, 1998). For instance, several countries inLatin America and Asia saw their exchange ratesappreciate during the early 1990s when there was asurgeofprivatecapitalinflows(CorboandHernandez,1994). These included Argentina, the Republic ofKorea, Mexico and the Philippines. In a more recentstudy,ADB (2007)finds that realeffectiveexchangerates in the large emerging East Asian economieshave appreciated against the US dollar since 2004,owingtolargerprivatecapitalinflows.

In the context of least-developed countries (LDCs),especially in Africa, several cross-country empiricalstudies find that foreign aid inflows are associatedwithanappreciationoftherealexchangerate(Lartey,2007;Elbadawi,1999).Thisisalsoreflectedincountrystudies on Burkina Faso, Côte d’Ivoire, Senegal andTogo (Adenauer and Vagassky, 1998), Cape Verde(BourdetandFalck,2006),Ghana(Opoku-Afarietal.,2004)andNigeria(Ogun,1998).Thesameholdstruefor several oil-rich countries where exchange rateappreciation has been associated with the influx ofpetro-dollars(The Economist,2007).

Thereis,however,abodyofevidencethatcontradictstheresultsdescribedabove.Forinstance,countriesinLatin America and Asia – Chile, Indonesia andMalaysia–thatreceivedthelargestcapitalinflows(asapercentageofGDP),onaverage,between1989and1992 avoided a significant real exchange rateappreciation (Corbo and Hernandez, 1994). Similarly,empirical evidence shows that foreign aid flows haveoftenbeenassociatedwithexchangeratedepreciation.This includes the findings of Mongardini and Rayner(2009)for36sub-SaharanAfricancountries,IssaandOuattara (2008) for Syria, Li and Rowe (2007) forTanzaniaandSackey(2001)forGhana.

Itisarguedthatcapitalinflowsassociatedwithhigherconsumptionputmorepressureontherelativepriceofdomestic goods than capital inflows associated withhigher investment (Saborowski, 2009). Hence, byensuringthatinflowsaddtotheproductivecapacityofan economy, a well-functioning financial system canattenuatetheupwardpressureontherelativepriceofnon-tradables and therefore on exchange rates. Pro-cyclicalcapitalflowsforinvestmentpurposes,however,canexacerbatemacroeconomicoverheatinganddrivethe real exchange rate to appreciate more. In somedeveloping economies, for instance, pro-cyclicalremittances spent on real estate have resulted inconstruction booms. In light of the above, countrieshaveoftenused restrictivefiscalpolicy to counteracttheexchangerateeffectofcapitalflowsfromabroad(Corbo and Hernandez, 1994). The nature of thecapitalflowmayalsoinfluenceitseffectonexchangerates. For example, the appreciation of the realexchangerateduetoFDIislikelytobelessthanthatdue to more volatile capital flows, such as portfolioinvestment(Lartey,2007).

(b) Financeforinvestment

(i) Domestic resources

Firmslookingtomakeinvestmentsoftendrawontheirretained earnings or other internally generated funds.Any industry with high growth prospects, however, islikely to experience relatively high investment demandcompared with current cash flows and therefore bedependentonexternalfinancing.Thesupplyofloanablefunds comes primarily from household savings


145

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

(seeBoxC.6forabriefaccountofitsdeterminants).Inaddition, central banks can buy securities, oftengovernment bonds, in the open market by paying forthem with money that they create. Given the above, afinancial system that mobilizes and allocates theseresources at low transaction costs to their mostproductiveusesiscrucialforpromotinginvestment(seeBox C.6 for a more detailed discussion). It is worthnoting that public investment may be financed bygovernmentsavings,whicharedefinedastheexcessoftaxreceiptsovertotalexpenditure.

The relationship between domestic savings andinvestment, in quantitative terms, is best captured bythe seminal paper of Feldstein and Horioka (1980).Analysing a sample of 16 OECD countries between1960and1974,theyfindthatthecorrelationbetweenlong-run averages of the saving-output ratio and theinvestment-outputratiowasveryclosetounity.Severalstudiesusing time-seriesdata validate thesefindings(Coakley et al., 1999; Coakley et al., 1996; Mamingi,1997;Miller,1988;Obstfeld,1986;Tesar,1993).Thesame holds true for several cross-country studies(Artis and Bayoumi, 1992; Coakley et al., 1996;Feldstein, 1983; Feldstein and Bacchetta, 1991;Golub, 1990; Obstfeld, 1986, 1995; Penati andDooley,1984;Tesar,1991).

Murphy (1984) finds that the saving–investmentcorrelation was significantly lower (0.59) for the tensmallest countries in his sample than for the sevenlargestcountries(0.98).Similarly,Dooley,FrankelandMathieson (1987) report that the average estimatewas significantly lower in non-OECD economies thanin OECD economies. More recently, empirical studies

whichhaveanalysedvariations,bothacrosscountriesand over time, find that the saving–investmentassociation isclosetounityforOECDeconomiesbutlower for developing countries (Cadoret, 2001;Coakleyetal.,2004;1999;Kim,2001).

FigureC.22also shows that, onaverage, thesavingsratesofmiddle-incomecountrieshaveexceeded thatofhigh-incomecountries for the last twodecades, In2010, middle-income countries had a savings rate of30 per cent, almost double the level of high-incomecountries. In fact, Table C.6 shows that among the15 countries with the highest average savings ratesbetween 2000 and 2010, almost all belong to themiddle-income category. Resource-rich countries inthe Middle East and North Africa – Libya, Qatar, theState of Kuwait and Algeria – occupy the top fourpositions.Goingforward,economicgrowth is likely tocontinuetobehighinthesecountries.Thesameholdstrue for labour-intensive economies in Asia, whereChina,Singapore,MalaysiaandVietNamwere in thetop 15 in the world in the context of savings ratesduringthelastdecade.Withrapidpopulationgrowthinsomeof thesecountries, theactiveworkforce isalsolikely to grow. Hence, high savings rates shouldcontinue to provide sufficient funding sources tosupport thedevelopmentof capitalmarketsandspurinvestment in physical capital. The same cannot besaidforeitherlow-orhigh-incomecountries.

For instance, Table C.6 shows that low-incomecountries, such as Côte d’Ivoire, and advancedeconomies,suchastheUnitedStatesandtheUnitedKingdom,wereamongthe15countrieswiththelowestaverage savings rates between 2000 and 2010.

FigureC.22:Investment and savings rates, 1991-2010 (percentageofGDP)

19

91

19

92

19

93

19

94

19

95

19

96

19

97

19

98

19

99

20

00

20

01

20

02

20

03

20

04

20

05

20

08

20

07

20

06

20

09

20

10

20

11

0

5

10

15

20

25

30

35

% o

f GD

P

High-income investment

High-income savings

Middle-income investment

Middle-income savings Low-income savings

Low-income investment

Source:WorldBank.


146

BoxC.6: Domestic savings and investment

Determinants of household savings

Income is thebasicdeterminantof saving.Thepoorare likely tohave justenough resources tomeet thesocialminimumlevelofconsumption.Incontrast,richerpeoplecanaffordtheluxuryofsavingtoassuretheirfuture consumption. Several empirical studies find that real growth of income, measured by GDP, has apositiveandsignificanteffectonsavings (Fry,1978;1980;Giovannini,1985;1983;Mason,1988;1987).Savingsmayalsodependonfluctuationsinthelevelofincome.

Giventhepredictionsofthe“permanentincome”hypothesis(explainedinSectionC.1)andrecognizingcreditconstraintsfacedbylow-incomehouseholds,rapidbuttransitoryincomegrowthislikelytoraisetheaveragesavings rate if the growth were concentrated in relatively rich households with high saving rates (Collins,1991).Morewealth90wouldtendtoreducesavingoutofcurrentincomebecauseitenhancesanindividual’sabilitytoearnincomeinthefuture(Schmidt-Hebbel,1987;BehrmanandSussangkarn,1989).

Thedemographiccompositionofahouseholdandcountryexertsanimportantinfluenceonsavingsratesaswell.The “life-cyclehypothesis”,described inSectionC.1,predicts thatahigherpercentageofoldpeopleandchildren–thenon-earningsectionofacountry’spopulation–reducesthesavingcapacityofacountry.Severalempiricalstudiesfindthatthedependencyratio,definedasindividualsundertheageof15orover65asashareoftotalpopulation,hadastrongnegativeeffectonsaving(Leff,1969;Mason,1988;1987;Collins,1991;Rossi,1989;WebbandZia,1990).

Wheninterestratesrise,individualsbegintoswitchfromcurrentconsumptiontowardssavingbecausetheformerbecomesrelativelymoreexpensive.Thisisthe“substitutioneffect”.Atthesametime,foranetsaver,anincreaseintheinterestratewouldincreasehisorher(expected)relativeincome, inducinggreatercurrentconsumptionandhencelowersavings.Thisisreferredtoasthe“incomeeffect”.Giventhattheincomeandsubstitutioneffectsof higher interest rates work in opposite directions, the effect of rates of return on savings is ambiguous. Inaddition,interestratescanalsoaffectsavingthroughawealtheffect.Higherrealinterestratesreducethepresent value of future income streams from fixed-interest financial assets. Savings therefore receive aboostevenifthesubstitutionandincomeeffectscanceleachotherout(Schmidt-Hebbeletal.,1992).Muchoftheempiricalliteratureshowsthattherealinterestratehasapositiveeffectonsavingrates(McKinnon,1973;Shaw,1973;Gupta,1987;Balassa,1990).91

Uncertainty about future asset values introduced by inflation could encourage saving for precautionarymotives.Atthesametime,ifincreasesintherateofinflationexceedincreasesinthenominalinterestrate,thiswouldlowertherealrateofreturnandhencediscouragesaving.Theempiricalevidenceisinconclusive(Gupta,1987;Lahiri,1988).

Fiscalpolicychangeswhichraisepublicsavingmayalsoaffectprivatesavingsrates.The“Ricardianequivalence”hypothesis, as reformulated by Barro (1974), states that public debt issues are indistinguishable from taxincreases in thefuture.Thus,achange inpublicsavingshouldbeoffsetbyanequalandoppositechange inprivatesaving.Thehypothesishasbeenwidelyrejectedinempiricalstudies,withthepervasivenessofborrowingconstraintscitedasthemainreasonforhouseholdsnotevenlyspreadingtheirconsumption-savingsbehaviourovertheirlifetime(HaqueandMontiel,1989;Rossi,1988;Schmidt-HebbelandCorbo,1991).

Cultural attributes may also have a significant impact on the level of savings. Using cross-country data,ShohamandMalul(2012)findthatasthelevelofuncertaintyavoidanceandcollectivismincreases,thelevelofnationalsavingsalsoincreases.

From savings to investment

The banking sector is the principal savings-investment conduit in most financial markets and therefore iscentral to themobilizationofdomestic resources for development.Unfortunately, it hasoftennot cateredwelltotheinvestmentneedsofsmallandmedium-sizedenterprises(SMEs)andthoseintheinformalsector,especially in developing countries (Zeldes, 1989). For instance, the top five banks serving SMEs in non-OECDcountriesreachonly20percentofformalmicroenterprisesandSMEs.InSub-SaharanAfrica,thisnumberisevenlower,at5percent(Dalberg,2011).Publicsectorbanks,thepostalsystemandmicrofinanceschemeshaveplayedaroleinmobilizingresourcesforgroupswholackcollateral.

Domesticsavingsmayalsospurinvestmentbyfirmsthroughholdingsinstocks,bondsandrelatedfinancialinstruments. Inmostdevelopingcountries,owingtoaweak legal frameworkand lowparticipationratesofinstitutional investors, suchasmutual funds, pension fundsor insuranceschemes, thesemarketsare still


147

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

TableC.6:Average annual savings rate, 2000-2010 (percentageofGDP)

Top 15 Bottom 15

Libya 59.81 Serbia 10.15

Qatar 55.81 Iceland 10.38

Kuwait,theStateof 48.36 Côted'Ivoire 11.64

Algeria 47.88 ElSalvador 12.07

China 46.90 Cyprus 12.12

Singapore 42.27 LebaneseRepublic 12.46

Iran 40.34 Greece 12.87

SaudiArabia,Kingdomof 36.92 BosniaandHerzegovina 13.05

Malaysia 35.55 Portugal 13.88

Azerbaijan 35.51 Guatemala 14.29

Norway 35.32 UnitedStates 14.61

TrinidadandTobago 34.27 Cameroon 14.67

Venezuela,BolivarianRepublicof 33.92 UnitedKingdom 14.72

Oman 32.93 DominicanRepublic 14.89

VietNam 32.70 Lithuania 15.15

Source:InternationalMonetaryFund,World Economic Outlook Database ,October2012.

Note:CountrieswithanaverageGDP,between2000and2010,belowcurrentUS$10billionwereexcluded.

Without sufficiently broad-based economic growth, agrowingmiddleclass thatcanpropelsavings rates inlow-incomecountriesisunlikelytoemergeinthenearfuture. In some advanced economies, such as theUnitedStates,lowinterestrates,prospectsofinflation,stagnant incomes owing to the crisis and culturalfactorsare likely tohinderan increase in their futuresavingsrates.

(ii) External resource flows

Overseas development assistance and migrant remittances

FigureC.22showedthatthegapbetweentherateofdomestic savings and domestic investment for low-income countries has been consistently high in therecentpast,wideningconsiderablybetween2002and2010. In 2010, the savings rate in low-incomecountries, on average, was about one-third theinvestment rate. Figure C.23 shows that overseasdevelopmentassistance(ODA)islikelytohaveplayedapartinfinancingthissavings-investmentgapinlow-incomecountries.

Given the limits to the growth of ODA in the future,owingtotherecessionarysituationinseveraladvancedeconomies, the future importance of other resource

FigureC.23:Overseas development assistance (ODA) and investment in low-income countries, 1990-2009 (US$billion)

O

O 10 20 30 40 50

20

40

60

80

100

120

Inve

stm

ent

Aid inflow

Source:WorldBank.

relativelyunderdeveloped.Economiesintransitionareacaseinpoint(Mileva,2008).Withrecentderegulationand liberalization measures, however, stock and bond markets are becoming an increasingly importantmeansofmobilizing funds inseveralemergingeconomies, including forhighgrowthpotentialSMEs (BIS,2012;Dalberg,2011).

Itisworthnotingthattotheextentthatpurchasesofstockstakeplaceonthesecondarymarketanddonotconstitutethepurchaseofnewlyissuedequity,increasedstockholdingsareunlikelytoincreasetheflowofcapitaltofirmsthatwishtoincreaseinvestment(KraayandVentura,1999).Duringtherecentfinancialcrisis,themostseriouslyaffectedfirmswerethoselistedonstockmarketswithsmallcapitalization–whichsufferfroma lackof investor interest–andSMEs–which suffer from the reluctanceofbanks toapprovenewloans or to roll over existing credit lines (OECD, 2012c; Dalberg, 2011). Section D.3 shows that this alsoholdstrueforthecaseoftradefinance.


148

flows from abroad in raising investment rates in low-income countries cannot be under-estimated. This isparticularly significant because data reveal that low-incomecountries, suchasMyanmarandKenya,wereamong the ten countries with the lowest averageinvestment rates between 2000 and 2010. Certainmiddle-income developing economies, such as Côted’Ivoire, Angola, Cameroon, the Plurinational State ofBoliviaandYemen,werealsoincludedinthisgroup.92Thissuggeststhatprivateexternalresourceflowsarelikely to be important for enhancing physical capitalaccumulationinmiddle-incomecountriesaswell.

Officially recorded migrant remittances to developingcountries, estimated at US$ 406 billion in 2012, arenowmorethanthreetimesthesizeofODA.Comparedwith other private capital flows, remittances haveshowed remarkable resilience during the recentfinancialcrisis(WorldBank,2012a).

In2012,largeemergingeconomies,suchasIndia,China,thePhilippines,Mexico,EgyptandVietNam,wereamongthetoptenrecipientsofmigrantremittancesintheworld(World Bank, 2012a). Figure C.24 shows that as apercentage of GDP, however, low-income countries,includingTajikistan,Haiti,theKyrgyzRepublicandNepal,wereamongthetoptenrecipientsofmigrantremittancesover the last decade. No Sub-Saharan African countryappearsinthislist.Thismayberelatedtothehighcostofsendingremittances.Forexample,accordingtoRathaetal. (2008), the average cost of sending US$ 200fromLondontoLagos,Nigeria, inmid-2006wasabout14percentoftheamount.Theirestimatessuggestthathalving remittance costs from 14 to 7 per cent for theLondon-Lagos corridor would increase remittances by11percent.

Remittance costs could be reduced by loweringremittance fees and by improving access to banking

forremittancesendersandrecipients.Thisisrelevantbecauseforecastsshowthatthegrowthofremittancesisexpectedtobestrongerinthenearfuture,especiallyin regions that rely on remittances from the UnitedStates, the Russian Federation and the Middle East(World Bank, 2012a). Importantly, however, whilemigrantremittancescanenableinvestmentinphysicalequipmenttoinitiateasmallhouseholdbusiness,theyareunlikelytobeabletosustaincapitalinvestmentbylargercompanies.

Portfolio investment and bank lending from abroad

Whenaneconomyliberalizesitscapitalaccount,itwillseeanincreaseinportfolioinvestmentflowsandbanklending if the marginal returns to capital are high inrelationtotherestoftheworld.Ingeneral,thiswouldmean that capital moves from capital-abundantcountries with low rates of return to capital-scarcecountrieswithhighratesofreturn.Thelargeinflowofprivate capital into emerging economies, starting inthe 1990s, can partly be explained by this rate ofreturndifferential.At thesame time, risk (bothactualand perceived) could narrow this differential ineffective terms. Hence, the increased inflow ofportfolioinvestmentandcommercialbanklendingmayalsobeattributabletofactorsthatreducedrisk(Modyand Murshid, 2005) – policy reforms, regulatorychangesandmorestablemacroeconomicpolicies.

Figure C.25 shows that the top ten recipients ofportfolio investment inflows among developingeconomiesduringthelastdecadewerealmostentirelyinAsiaorLatinAmerica.FigureC.26shows that thesame holds true for bank lending from abroad. Thecontinuedimportanceoftheseinvestmentflowsinthefuturewillundoubtedlybe influencedbythehealthofglobal financial markets as well as the ability ofrecipient countries to strengthen independent

FigureC.24:Top ten recipients of remittances from migrants as a share of GDP, 2000-10 (percentage)

FigureC.25:Average annual foreign portfolio liabilities of developing countries – top ten, 2001-10 (US$billion)

0

60

70

50

40

30

20

10

Tajik

ista

n

Tong

a

Hai

ti

Kyr

gyz

Rep

ublic

Mol

dova

,R

ep. o

f

Guy

ana

Jord

an

Alb

ania

Bos

nia

and

Her

zego

vina

Nep

al

0

300

250

200

150

100

50

Hon

g K

ong,

Chi

na

Mex

ico

Bra

zil

Kor

ea,

Rep

. of

Sin

gapo

re

Chi

na

Indi

a

Sou

thA

fric

a

Chi

nese

Taip

ei

Rus

sian

Fede

ratio

n

Source:WorldBank,World Development Indicators.

Source:InternationalMonetaryFund,Coordinated Portfolio Investment Survey.

Note:EconomieswithanaverageGDP,between2000and2010,belowcurrentUS$10billionwereexcluded.


149

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

regulation, improve transparency and conform torelevant international accounting and auditing rules.For developing countries in Africa, for example,the establishment of a strong legal framework andgreater reliance on market-based credit assessmentmethodologies would be necessary first steps tocreatecapitalmarketsthatcanattractforeignportfolioinvestmentandbanklendingfromabroad.

Foreign direct investment (FDI)

Traditionally,FDIconsistedof intra-industry investmentflows between similar developed countries (Forte,2004). The latest World Investment Prospects Surveysuggests that theEuropeanUnionandNorthAmerica

will remain among the most important regions for FDIby multinational companies in the medium run(UNCTAD, 2012). During the 1990s, FDI directed atdeveloping countries began to grow substantially (seeFigure C.27). This largely represented investments byadvanced economies’ firms in developing countries,wherebytheformeroffshoredunskilledlabour-intensivepartsoftheproductionprocesstothelatterinordertotake advantage of lower costs (Helpman, 1984).93 Inaddition tosuch “vertical”FDI, advancedcountryfirmsalso viewed developing countries with large marketsand significant barriers to trade as appropriatedestinationsfor“horizontalFDI”(Dunning,1980).

Much like foreign portfolio flows and bank lendingfrom abroad, FDI flows into developing economieswere largely confined to Asia and Latin America.TableC.7showsthatwiththeexceptionofTurkey,thetop 15 developing country recipients of FDI inflowsduring the last two decades were in these twocontinents. East Asia did particularly well, with asmanyas six countries in the top15andChinaat thetopofthetable.Thismaybeexplained,inpart,bytheavailability of adequate supporting infrastructure andthe quality of institutions because they reducetransactioncosts(seeSectionC.6).

Countries in South-East Asia, for example, haveconcentrated their public resources on thedevelopment of infrastructure, including roads, ports,electricityandtelecommunicationservices(AndoandKimura, 2005). The World Investment ProspectsSurveyoutlinesthecontinuedimportanceofAsiaandLatin America, as respondents listed China, India,IndonesiaandBrazilasfourofthetopfivemostlikelydestinations for their FDI in the medium term(UNCTAD,2012).

Even inAfrica,wherepublic infrastructure investmentis relatively inefficient, improving infrastructure has a

FigureC.26:Top ten recipients of bank lending from abroad amongst developing countries, 2001-10 (US$billion)

0

350

400

300

250

200

150

100

50

Hon

g K

ong,

Chi

na

Mex

ico

Bra

zil

Kor

ea,

Rep

. of

Sin

gapo

re

Chi

na

Indi

a

Turk

ey

Chi

nese

Taip

ei

Rus

sian

Fede

ratio

n

Source:BankforInternationalSettlements.

Note:EconomieswithanaverageGDP,between2000and2010,belowcurrentUS$10billionwereexcluded.

FigureC.27:Inflows of foreign direct investment, 1980-2010 (US$billion)

19

80

19

82

19

84

19

86

19

88

19

90

19

92

19

94

19

96

19

98

20

00

20

02

20

04

20

06

20

08

20

10

Developing economies Developed economies

0

200

400

600

800

1,000

1,200

1,400

Source:UNCTAD.


150

positive impact on FDI inflows (Asiedu, 2002;Morrisset,2000).Itisalsoarguedthathighdomesticprivateinvestmentisasignalforhighreturnstocapital,which attracts foreign investment. For instance,analysing a sample of 38 Sub-Saharan Africancountries between 1970 and 2005, Ndikumana andVerick (2008) find that domestic private investmenthas a strong positive impact on FDI inflows. ThissuggeststhateffortstoimproveincentivesforprivateinvestmentthroughimprovingthequalityofinstitutionswillresultinforeigninvestorsviewingAfricancountriesmore favourably in the future. In fact, the WorldInvestmentProspectsSurveyshowsthatFDIinAfricaisexpected topickupover themedium runowing tostronger economic growth, on-going policy reformsandhighcommodityprices(UNCTAD,2012).

In recent times, high savings rates, increased capitalintensity and technological progress have resulted incertain developing economies becoming sources ofFDI as well. Figure C.28 shows a steady increase inFDI outflows from developing economies between2003and2010.ThebulkofthisFDIrepresentsflowsfrom emerging economies to low-income countries,contributingtoincreasedinvestmentratesinthelatter(WorldBank,2011a).

TableC.7showsthatthetopfivesourcesofFDIamongdevelopingeconomiesoverthelasttwodecadesareinEastAsia(withHongKong,ChinaandChinaoccupyingthetoptwopositions).OtherimportantsourceregionsincludeIndiaaswellascountriesinLatinAmericaandthe Middle East. Furthermore, much of the FDIbetweendevelopingcountries is intra-regional (WorldBank, 2011a). Inter-regional FDI among developingeconomies goes primarily from Asia to Africa. ChinaandMalaysiaareamongthetoptensourcesofFDIinAfrica(UNCTAD,2006).

TheWorldInvestmentProspectsSurveyreportsthat,inmarked contrast to developed countries, nearly one-quarterofrespondentsindevelopingeconomiesforesawa decline in their FDI budgets in 2013 and 2014(UNCTAD,2012).Thismaybeexplainedbythefactthatmultinational companies from developing economiescontinuedtoinvestatnearrecordlevelsduringthecrisisandmay focuson rationalizing their investments in themedium term. In the long run, however, high expectedgrowthinemergingeconomies,afamiliaritywithsimilarpolicy environments and the overall strengthening oftrade links between developing economies is likely toenhanceFDIbetweenthesecountries.

(c) Conclusions

Investmentininfrastructurecanleadtotheemergenceof“newplayers” inworldtradeinthefuture.Thismaybe particularly important for low-income countries inAfrica, hitherto less involved in global productionnetworks. It may also change the nature of trade inother ways. Better transport infrastructure acrossneighbouring countries, such as road connectivity,could strengthen regional trade in Africa. Moreextensive ICT infrastructure could further expandservices trade and alter the pattern of internationalspecialization.English-speakingAfricancountries, forexample, could mark a presence in the area ofbusinessprocessoutsourcing.

Governments in thesecountriesmust therefore focuson scaling up and improving the quality of publicinfrastructure. This may involve enhancing domesticsavingsrates.Theimplementationofgrowth-promotingstrategies that raise household incomes would becentral topromotingsavings.Altering taxsystemsandmacroeconomicpoliciesmayalsoplayapart.Ensuringthat savings are translated into investment throughimprovingtheefficiencyofcapitalmarketsislikelytobe

TableC.7:Average annual FDI flows of the top 15 developing countries, 1990-2011 (US$million)

Inward FDI Outward FDI

China 55,253 HongKong,China 33,146

HongKong,China 28,758 China 15,473

Brazil 20,635 Singapore 10,435

Singapore 19,113 Korea,Republicof 7,423

Mexico 16,378 ChineseTaipei 5,899

India 10,370 India 4,922

SaudiArabia,Kingdomof 7,872 Malaysia 4,291

Chile 6,537 Brazil 3,660

Argentina 6,089 Mexico 3,121

Turkey 5,578 Chile 2,986

Thailand 5,286 UnitedArabEmirates 2,621

Malaysia 5,055 Kuwait,theStateof 2,135

Korea,Republicof 4,463 Thailand 1,551

Colombia 4,262 Colombia 1,446

UnitedArabEmirates 3,843 Panama 1,392

Source:UNCTAD.

Note:EconomieswithanaverageGDP,between2000-2010,belowcurrentUS$10billionwereexcluded.Indonesiawasexcludedbecauseoflargegapsinthedata.


151

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

equally important. Governments could utilize overseasdevelopment assistance, FDI and bank lending fromabroad to increase infrastructure investment as well.The WTO’s Aid for Trade initiative for developingcountries is also important in this regard as it canincrease a country’s supply capacity and hence itsparticipationintheworldmarket(seeSectionE).

Greater public and private investment in physicalcapital, financedbydomestic savingsorcapitalflowsfrom abroad, may also influence the comparativeadvantageofcountries.Thereisapossibilityforsomeunskilled labour-intensive economies, such as Chile,China and Turkey, to become capital-intensiveeconomiesinthemediumtolongrun.Savingsratesinmanyoftheseeconomiesarealreadyhigh.

Therefore, in order tomoveup theproduct ladder (intermsofcapitalandtechnologyintensity),governmentsmust concentrate on creating adequate investmentopportunities for both domestic and foreign capital.This lack of opportunity is perhaps reflected in theincreasing outflow of FDI from certain developingeconomies,suchasChina,Singapore,theRepublicofKorea and India, to other developing and evendeveloped economies. Of course, it is important tohighlight the fact that outward FDI from developingeconomies is associated with the emergence ofdeveloping country-based multinational companieswhich, by enhancing capital and technology intensity,canitselfinfluencecomparativeadvantage.

It ishard topredicthowcapitalflowsacrosscountries(andthereforetheircontributiontocapitalaccumulation)will evolve in the future. Existing forecasts from the

World Investment Prospects Survey, for example,suggest that FDI flows are expected to increase at amoderate but steady pace over the medium term(UNCTAD,2012).Thisbaselinescenario,however,doesnotconsiderthepossibilityofnegativemacroeconomicshocks.

It is possible that the fragility of the world economy,the volatility of the business environment anduncertainties related to the sovereign debt crisis willnegatively impact FDI flows in the medium term.Nevertheless, developing and strengtheninginstitutions, such as a sound legal framework, wouldundoubtedly be important to attracting FDI.Preferential trade agreements with provisions for“deep integration” can play an important role in thisregard. Establishing capital markets with sufficientdepthisalsocrucialtoattractingadditionalsourcesofinvestment finance, such as foreign portfolioinvestment and bank lending from abroad. This holdstrueforbothlow-andmiddle-incomecountries.

Reforms in the banking sector need to encouragefinancial institutions to move towards sound creditassessment methodologies. At the same time, stockand bond markets can play a larger role in domesticresourcemobilization.Thiswouldrequireastronglegalframework, transparency requirements, financialaccountingandauditingrulesofinternationalstandard.The enforcement capabilities of independentregulation are also likely to play a part in reducingsystemicriskandprotectinginvestors’interests.

It is worth noting that external resource flows fromabroad influence thenatureof tradenotonly through

FigureC.28:Outflows of foreign direct investment, 1980-2010 (US$billion)

19

80

19

82

19

84

19

86

19

88

19

90

19

92

19

94

19

96

19

98

20

00

20

02

20

04

20

06

20

08

20

10

Developing economies Developed economies

0

200

400

600

800

1,000

1,200

1,400

1,600

1,800

2,000

Source:UNCTAD.


152

their impactondomestic investmentbutalsodirectly.For instance, the trade literature suggests thatportfolio investment and bank lending relationshipsacrosscountriescanincreasetradeflowsbyreducinginformation asymmetries between exporters andimporters. Similarly, FDI flows complement trade byfacilitating global supply chains – increasing exportsof intermediateproductsandservices from thehomecountry and those of the final good from the hostcountry. Moreover, exports from the host country tothirdcountrymarketsmayincrease.FDIflowscanalsoaffect the comparative advantage of developingeconomies by facilitating the transfer of technologyacross countries. China is an example of suchtechnologyupgrading.

In fact, to the extent that investment and trade arecomplementary,aninternationalsystemofinvestmentrules can increase the flow of foreign investment bypromoting predictability and security of access forforeign investors (see Section E). So can bilateral orregional agreements, which are being increasinglyused to govern international investment. Theseagreements, however, run the risk of affecting the“level playing field”94 in the future by creatingregulatorydivergence.Asetofmultilateralinvestmentrules could ensure a more efficient internationalallocation of resources (with investment not divertedbecause of preferential treatment) across borders,which in turn should help trade. It could also bringgreater parity between big and small countries,reducing the power imbalance which may arise if alarge country negotiates with a small country on abilateralinvestmentagreement.

3. Technology

Technological differences between countries are animportant determinant of income levels and trade.Empiricalresearchhasshownthattheaccumulationofphysical and human capital can only partially explaindifferentincomelevelsacrosscountries(EasterlyandLevine, 2001; Prescott, 1998) and different tradepatterns. The residual is commonly attributed totechnologicaldifferencesbetweencountries,wherebytechnologyisdefinedastheinformationorknowledgerequiredforproduction.

Technologicalprogressisundoubtedlythemajorfactorexplaining the fast growth in income in the 19th and20th centuries. Electrification, the telephone, theinternal combustion engine and other breakthroughshave dramatically changed the way the world works(seeSectionB.1).Likewise,technologicalprogresswillbe amajor factor in explaining the futurepatternsoftrade and growth. Simulations about the future ofglobaltradediscussedinSectionB.3highlightthattheassumptions about the future path of technologicalprogressplaybyfarthelargestroleinaffectingoveralloutcomes.

Acountry’s technological level isdeterminednotonlyby domestic innovation but also by the diffusion oftechnology fromabroad.Typically,while the former isparticularly important for high-income countries, thelattermostlyaffectstechnologicalprogressinmiddle-and low-income countries. This section first looks atpatternsofinnovationandtechnologytransfer.Thenitdiscusses how technological changes affect trade.Thirdly, it looks at the determinants of technologicalprogress. Finally, it explores what these trends implyforthefutureoftradeandtradepolicy.

(a) Technologypatterns

Section B found that there are important emergingplayersininternationalmarkets,thattradeisbecomingmoreregionalizedandthatit ishighlyconcentratedinfew global companies. This section explores whetherinnovation and technology transfers can help toexplain these patterns. In particular, it exploreswhether there isevidenceofemergingnewcountries(that have significantly accelerated their capacity toinnovate or absorb existing technologies) and newsectors(wheretechnologicalknowledgehasincreasedfaster), whether innovation and technology transfersare more localized, regionalized or globalized than inthepast andwhether there is a relationshipbetweenthesetrendsandoffshoring.Finally,italsolooksattheroleoflargemultinationalcompaniesversussmallandmedium-sized enterprises in driving technologicalprogress.

(i) Measuring technological progress

Measuring technological progress is a difficult andimperfect field of study. Widely used measures oftechnologicalprogressincludetotalfactorproductivity,research and development (R&D) expenditure andpatent applications (Keller, 2010). However, eachmeasure captures a different and incomplete pictureoftechnologicalprogress.

Totalfactorproductivity(TFP)measuresaneconomy’sefficiency in transforming inputs into outputs.Empirically, total factor productivity is defined as theoutputperunitofcombinedinputs(usually,aweightedsum of capital and labour) and is calculated as adifference between a country’s GDP and thecontributionofcapitalandlabour.Theresidualoutputthat is not explained by capital and labour inputs isconsidered “technology”. This approach suffers fromimportantlimitationsduetobothalackofdataanditspoor quality. For example, estimations of TFP mayattribute to technology what should be explained bylabour or capital (physical and human capital), werethedataofbetterquality.

R&Dexpendituremeasurestheinputintotechnologicalinnovationactivity.Adrawbackofthisapproachisthatnot all research investments generate innovations;95

and even when they do, the rate of return can vary


153

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

significantly depending on the specific investor andthe way investments were made. For example, thereturn to publicly funded R&D is typically lower thanthe return to privately funded R&D (Keller, 2010).Moreover, since data are typically collected on ageographicalbasis,itoftenfailstodistinguishbetweenR&D investment in domestic firms and in foreign-ownedaffiliates.

Measuring patent applications addresses several ofthe limitations of other approaches. Unlikecomparisons of R&D expenditure, patent applicationdata captures the outputs of the innovative process(the invention) rather than its inputs (the research).96

Thisapproachalsodistinguishesbetweeninnovationsgeneratedbyresidentsandnon-residents.97However,asimplecountofpatentsmaybeamisleadingindicatorof country’s level of technology for several reasons.First, there is not necessarily a direct correlationbetweeninventionsandinnovations.Notallinnovationsarepatented.Forexample,patentstendnottocaptureinnovationsinservicesororganizationalmethods.Norare all inventions patented in the country where theyweregenerated.

An invention produced in a developing country, forexample,mightnotbepatentedthereifitislikelythatthe technology will be manufactured or producedelsewhere.Patentsalso tend tohavewidelydifferentscientificandcommercialvalues–typicallyarelativelysmallnumberofpatentsaccountsforalargeshareofthevalueofthepatentstock98–meaningthatthereisatenuouslinkbetweenacountry’snumberofpatentsanditstechnologicaloutput.

Regardingtechnologytransfers,therearetwoaspectsthat can be measured: the purchase of technology(seeBoxC.7)99and technologyspillovers. Included inmeasures of technology purchases are royaltypayments, R&D services trade, trade in technology-intensive goods, the share of foreign-ownedemployment in total employment, and foreign directinvestment.Ineachcase,acertaintechnologyismadeavailable to the importing country in exchange forpayment– i.e.a licensingfee,awage,or thepriceofthe good. This measurement implicitly assumes thatthe technology embodied in these imports is notpermanently available to domestic producers. If theimportofthatgoodorservicestopsforanyreason,orthe licence expires, the productivity gains are alsoassumed to disappear, as the importing country isunable to produce the knowledge embodied in thegood,serviceorlicenceonitsown.

Alimitationofmeasuringtechnologytransferintermsofthemonetaryvalueofthemarkettransactionisthatitdoesnotaccountfortechnologyspillovers–i.e.thetechnology that is absorbedby the importing countrywithoutpayment.Thismayhappenthroughavarietyofchannels, including worker training, interaction withsuppliers or reverse engineering. When an importing

country acquires technology this way, the knowledgeisretainedeveniftheactofimportingisinterruptedorstopped.

Spillovers are difficult to distinguish empirically fromknowledge flows, although there are at least twoconceptualdifferences.First,knowledgeflows,unlikespillovers,donotnecessarilyinvolveexternalities;andsecondly, they are consistent with a two-wayinteraction between actors rather than involving theone-way transfer of technology from one actor toanother. One common way to estimate technologyspillovers is to study the impact of foreign R&D onproductivityor, alternatively, the impactof technologydeveloped abroad on the rate of innovation of thehome country. Evidence on technology spillover isdiscussedlaterinthesection.

(ii) The geography of technological progress

New leaders

Inrecentyears,therehavebeenimportantchangesinthe geography of innovation. Although thetechnological gap between high- and low-incomecountries persists, R&D investments have becomemore globalized (Fu and Soete, 2010; Lundvall et al.,2009).

FigureC.29showsthedistributionofbusinessR&Dinasampleof37countriesfor1999and2010.100Itcanbeseenthatoverthesampleperiod,R&Dexpenditureshave become less concentrated. For example, whilecountriesrepresenting20percentoftotalpopulationaccounted forabout70percentofR&Dexpenditurein 1999, these countries accounted for only about40percentofR&Din2010.

Most importantly, certain countries that traditionallyhave served simply as production platforms fordevelopedcountries increasinglybasetheireconomicgrowth on their own capacity to innovate andcontribute to the technology pool (Mahmood andSingh, 2003). For example, Table C.8 provides thetotalnumberofpatentapplicationsbycountryoforiginoftheapplicant.ItshowsthatthecontributionofChinaand other Asian countries, such as Singapore, Indiaand the Republic of Korea, to the “pool” oftechnological innovation has significantly increasedfrom1985to2010.101

One possible explanation for the growing importanceoftheseAsiancountriesininnovationistherelocationof significant manufacturing capacity to them(including the development of new and existingdomestic industriesaswell as the locationof foreignsubsidiaries). As Pisano and Shih (2012) point out,producersbenefitfromtheinteractionwithinnovatorsandviceversa.ThetransferfromR&Dintoproductioncan be complex and require significant coordinationbetween those who design a good and those who


154

BoxC.7: Limits of traditional measures of market- and intellectual-property-based technology transfer: a statistical perspective

Changingeconomicenvironmentsandbusinesspracticesrequirestatisticalframeworkstoadapt.Methodologicalresearch has helped clarify a number of conceptual issues which were left untackled in previous statisticalframeworks. Consequently, the 1993 System of National Accounts and the fifth edition of the Balance ofPayments Manual (BPM5) were both revised in 2008 to reflect better the economy and trading structure.Subsequently, the Manual on Statistics of International Trade in Services was also revised to guaranteeconsistencywiththemainframeworksandtoaddconceptualclarificationonaspectsthatwerenotfullyelaboratedinthepreviousversion,suchasthemeasurementofinternationalsupplyofservicesbymodeofsupply.

Thenewguidelines–thesixtheditionoftheBalanceofPaymentsManual(BPM6)andthe2010ManualonStatisticsofInternationalTradeinServices(MSITS2010)–alsoprovideclearerconceptualguidanceastohowtoclassifyandmeasuretransactionsrelatedto intellectualproperty,andinparticularthoserelatedtotechnologytransfer.Thecategory“royaltiesandlicencefees”hasbeenreplacedby“chargesfortheuseofintellectual property not included elsewhere” and the item “research & development services” has beenbrokendownbyadditionalcategoriestoallowforaclearerconceptualmeasurement.

Transactions relating to the right to use the results of research and development102 are covered undercharges for the use of intellectual property not included elsewhere. Transactions related to research anddevelopment services as well as the outright sales of property rights arising from research (e.g. patents,industrialprocessesanddesigns,copyrightsarisingfromresearchanddevelopment)arecoveredundertheresearch and development services item. MSITS 2010 proposes a breakdown of this item into “workundertakenonasystematicbasistoincreasethestockofknowledge”(reflectingthecoverageofresearchanddevelopmentwithinthenationalaccounts)and“other”.Theformerisfurtherbrokendowninto“provisionofcustomizedandnon-customizedresearchanddevelopmentservices”and“saleofproprietaryrightsarisingfromresearchanddevelopment”whichisitselfbrokendowninto“patents,copyrightsarisingfromresearchanddevelopment,industrialprocessesanddesigns”and“other”.

Tocollecttherespectiveinformationrequiresdraftingofappropriateguidancefordatacollectionsystems,suchas the international transaction reporting system or general trade in service surveys. For example, whenconsidering multinational enterprises, many of the technology transfer transactions take place within thisparticular group of firms and consequently the valuation of trade (i.e. transactions) may be distorted as thepricingusedmaysignificantlybe influencedby taxpolicies in the locationswhere thesemultinationalshaveestablishedaffiliatesandthereforemaysignificantlydifferfromtheactual“real”-marketvalueoftransactions.Economicownershipofintellectualpropertyassetsmaybeanadditionalbarriertotheappropriatemeasurementoftransactions.Indeed,multinationalsmaychoosetoregistertheirpatentorindustrialprocessinonecountryrather than another based on “tax evasion” strategies. Consequently, the country of registration is notnecessarilythesameastheoneoftheeconomicowneroftheintellectualproperty–thesameinventionmaybe patented in multiple countries. For example, statistics from the World Intellectual Property Organization(WIPO)reportthatabout40-50percentofallpatentsareso-calledsecondaryfilings.

Inothercases,firmsmaynotbeaffiliatedbutacliententerprisemayoutsourcecompletelytheproductionofaproduct(i.e.virtualmanufacturing),providingalltheknowledgetothemanufacturerfortheproductionofthese goods. Again, it is unclear how the relevant transactions, and in particular the ones pertaining toknowledgetransfer,shouldbeaccountedforornot,asthismaydiffersignificantlyaccordingtothedifferenttypes of arrangements which are adopted. In other words, although international guidelines clarify theconceptual classification of transactions, they fail to provide clear recommendations as to how to clearlycompiletherespectivestatistics.ThecompilationguidancethatiscurrentlybeingdraftedbytheUNexpertgrouponcompilationoftradeinservicesstatistics103should,however,helpclarifythesituation.Inaddition,aTaskForceonGlobalProductionhasbeenestablishedbytheConferenceofEuropeanStatisticianstosetupclearerguidelinesinrelationtoglobalproductionarrangementsfromtheperspectiveofnationalaccountsaswellasfromtheperspectiveofthetradeinservicesandbalanceofpaymentsstatistics.

Followingtheestablishmentofthesemoredetailedguidelines,itisexpectedthatsomespeciallytargetedsurveysshould help improve the situation, in particular when it comes to the more detailed information sought.Nevertheless, many countries/compilers will probably not be in a position to collect accurately such detailedinformation,oftenforbudgetaryreasons.Asolutioncouldbetocomplementtheirmoregeneraldatacollectionsystemswiththeinformationcollectedanddisseminatedbythosecountrieswhichwillengageinamoredetailedand sophisticated data collection system (probably for countries which have a particular interest in thisinformationbecauseofsignificant researchanddevelopmentactivities).However, thiscanonly function ifthere isefficient cooperationbetweencompilersofdifferent countries. In addition, itwill benecessary tohavedetailedbilateralinformationpublishedbycountriesthatwillbeengagingindetaileddatacollection.


155

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

manufacture it. Similarly, designing a product may bedifficult if the designer does not understand howproduction works. Thus, as manufacturing shifts toAsia,itislikelythatsotoowillknow-how,researchandeventuallyinnovation.

However,thegrowingimportanceofAsiancountriesininnovation is not driven by multinational firms alone.Forexample,thegreatmajorityofpatentsandthebulkof R&D activity in China are generated by Chineseentities. R&D conducted by foreign subsidiaries stillrepresents a relatively small share.104 As shown inFigure C.30, the number of patent applications byresidents105 in Asia has increased significantly since1995, as have applications in the OECD from non-residents.

Internationalization of technological progress

As discussed above, technological progress isdeterminednotonlybydomesticinnovationbutalsoby

international technology spillovers. In developingcountries,wheredomesticinnovationis low,spilloversacquire relatively greater importance. Understandingtheirgeographicalextent– i.e.whether thespilloversarelocalizedorglobal–iscrucialtodeterminingtheirnature and impact. Indeed, the prevalence ofinternational technology spillovers is a majordeterminant of the world’s income distribution. Whileglobal technological spillovers promote incomeconvergenceworldwide,localspilloversdonot.

In general, the empirical evidence supports the viewthat spillovers tend to be local – i.e. stronger withinthan across countries. Using patent citations as ameasureoftechnologicalspillovers,Jaffeetal.(1993)find that US patents are more often cited in otherUSpatentsthaninotherforeignpatents(Branstetter,2001; Eaton and Kortum, 1999). Looking at a widerset of countries, Keller (2002) also finds thatinternational technological spillovers are conditionalon geographical distance. Measuring the impact of

TableC.8:Patent applications by country of origin, 1985-2010 (top30countries)

Number of patent applications Global ranking

US office only US office only

Origin 2010 2010 1985 2010 2010 1985

Japan 468,320 84,017 274,404 1 2 1

UnitedStates 432,911 241,977 64,308 2 1 3

China 308,318 8,162 4,066 3 9 10

Korea,Republicof 178,644 26,040 2,703 4 4 15

Germany 173,532 27,702 32,574 5 3 4

France 65,623 10,357 12,240 6 8 6

UnitedKingdom 50,865 11,038 19,846 7 7 5

Switzerland 39,393 4,017 3,344 8 13 13

Netherlands 33,388 4,463 1,994 9 11 20

RussianFederation 32,835 606 3 10 26 71

Italy 27,910 4,156 2,137 11 12 18

Canada 24,209 11,685 2,110 12 6 19

Sweden 22,443 3,840 3,871 13 14 12

India 14,862 3,789 982 14 15 27

Finland 13,046 2,772 1,732 15 17 23

Belgium 11,804 2,084 807 16 18 30

Australia 11,556 3,739 21 17 16 54

Denmark 11,233 1,773 872 18 19 29

Austria 11,062 1,661 2,282 19 20 16

Israel 10,928 5,149 800 20 10 31

Spain 10,733 1,422 2,163 21 22 17

DemocraticPeople'sRepublicofKorea

8,055 - - 22 - -

Norway 5,595 936 928 23 24 28

Singapore 4,229 1,540 4 24 21 69

Brazil 4,212 568 1,954 25 27 22

Turkey 4,211 150 132 26 36 38

Ireland 4,102 796 730 27 25 32

Poland 4,061 185 5,124 28 35 8

NewZealand 3,223 541 1,010 29 28 26

Ukraine 3,038 64 - 30 48 -

Source:Authors’calculationbasedondatafromWIPOIPStatistics,athttp://ipstatsdb.wipo.org/ipstatv2/ipstats/patentsSearch,accessedinMarch2013.

US office only


156

FigureC.29:R&D distribution, 1990-2010 (cumulativeshares)

Per

cent

age

of to

tal B

ER

D

Percentage of total population (37 countries)

Business R&D Expenditure, 1999Business R&D Expenditure, 2010

50

60

70

80

90

100

40

30

20

10

0

0 10 20 30 40 50 60 70 80 90 100

Source:Authors’computationsbasedonR&DdatafromOECDScience,TechnologyandR&DStatistics,athttp://www.oecd-ilibrary.org/science-and-technology/data/oecd-science-technology-and-r-d-statistics_strd-data-en,andpopulationdatafromWorldDevelopmentIndicators(WDI),athttp://databank.worldbank.org/

FigureC.30:Patent applications from residents and non-residents, 1995-2010

0

20

10

19

95

19

96

19

97

19

98

19

99

20

00

20

01

20

02

20

03

20

04

20

05

20

06

20

07

20

08

20

09

100,000

200,000

300,000

400,000

500,000

600,000

700,000

800,000

900,000

OECD R OECD NR Asia RAsia NR LAC NRLAC R

Source:Authors’computationsbasedondatafromWIPOIPStatisticsDatabase,athttp://ipstatsdb.wipo.org/ipstatv2/ipstats/patentsSearch.

Note:“R”indicatesresidentsand“NR”non-residents.ItalyisexcludedfromtheOECDgroupduetothelimiteddataavailability.TheregionsofAsiaisrepresentedbytoppatentapplicants.TheregionAsiaincludesChina,JapanandRepublicofKorea;LatinAmericaandtheCaribbean(LAC)includesBrazil,ChileandMexico.

R&D expenditure in the five OECD countries onindustry-level productivity of another nine OECDcountries, he finds that the impact decreases withdistance. The degree of localization, however, hasdecreased over time. For the period 1973-83, Keller(2002) estimates that at a distance of 2,000kilometres between the senders and the receivers oftechnologicalknowledgeonly5percentwasactuallyabsorbed. However, for the period 1986-95, he findsthispercentageincreasedto50percent.106

Onepossibleexplanationforthewideninggeographicalradiusoftechnologyspilloversistheinternationalizationoftheinnovationprocess,includingthegrowingmobilityof experts and expertise, the increasing number ofinternational co-authorships and the rising share ofpatents that list inventors frommore thanonecountry(WIPO,2011).AsshowninFigureC.31,oneofthemostinteresting recent developments is the increasedincidence of co-authorship between developed anddevelopingcountryscientistsandresearchers.

AsecondpossiblefactordrivingthegrowingradiusofR&D spillovers is the increased importance ofproduction networks. The international fragmentationof the production process increases cross-borderinteractions, which in turn increases technologicalspillovers.FollowingtheapproachsuggestedbyKeller

(2002)andBottazziandPeri (2003), this reportalsocalculates how R&D spillovers decline with distancebut itdistinguishesbetweencountries thatarehighlyintegratedandthosethatarenot.107

AsshowninFigureC.32,R&Dspilloversfromverticallyintegrated countries remain more significant overlonger distances than R&D spillovers from countriesthat are on average less vertically integrated.Specifically, a 10 per cent increase in foreign R&Dspending incountries thatare locatedwithin300kmtranslates on average into 0.04 per cent increase inpatenting in the home country. However, the result ishigherforverticallyintegratedcountry-pairs,forwhicha 0.08 per cent increase in patenting at home isestimated if the foreign country is highly verticallyintegrated with the home countries (details of themethodologyusedareprovidedinBoxC.8).

Although production networks may have helped towiden the radius of technology spillovers, thesenetworks tend tobe regional rather thanglobal– i.e.they tend to increase trade and investment flowsbetween closer international locations than betweenlocations farther apart. It follows that technologicaldiffusion may also have become more regionalizedrather than more globalized – an observation that issupported by the results of this report. As shown in


157

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

FigureC.31:International co-authorship of science and engineering articles, 1995-2010 (inthousands)

1995 2010

Developing-Developing

1.511.3

17.5

70.3

6.6

22.5

Developed-Developing

Developed-Developed

Source:Authors’computations,basedondatafromAppendixTable5-41ofScienceandEngineeringIndicator2012,NationalScienceFoundation(NSF).Retrievedat:http://www.nsf.gov/statistics/seind12/append/c5/at05-41.xls

Figure C.33, technology spillovers are much higheramongcountrieswithinaregionthanoutsideit.

Observed patterns of trade in high-technologyproducts also support the idea that technologyspillovers may have regionalized. Figure C.34 showsthe percentage of trade in high-technology productswithinaregionversusbetweenregions.Interpretinganincrease in trade inhigh-techproductsasameasureof stronger technological spillovers, the increasingshare of trade in high-technology goods within aregion seems to point to an intensification oftechnologydiffusionattheregionallevel.

One implication of the regionalization of technologyspillovers is the possible emergence of “convergence

clubs”, that is groups of countries that becomeincreasinglysimilarintermsoftechnologylevels,trademore among themselves, share similar economicinterests and possibly engage in building strongerregionalinstitutions.

(iii) The changing nature of technological progress

Sectoral distribution

R&Dspendingishighlyconcentrated.Nearly90percentof R&D investment takes place in the manufacturingsector, and within this sector over 90 per cent ofinvestments occur in just four industries: chemicalproducts, electrical and non-electrical machinery(covering information communication technology– ICT)and transportation equipment (see Table C.9 andAppendixTableC.1fordetailsoncomposition).

AlthoughmostR&Dtakesplace inthemanufacturingsector, R&D in the services sector has experiencedthe fastest growth since the early 1990s. Table C.9shows that R&D expenditure in services increasedfrom6.7percentoftotalbusinessR&Dbetween1990and95tonearly17percentbetween2005and2010.Within services, business services saw the biggestincrease in R&D expenditure over the period (seeAppendix Table C.2). In general, knowledge-intensivebusinessservices(KIBS)areemergingaskeydriversof knowledge accumulation and may in the long runreplace manufacturing as the engine of globalinnovation. Eurostat’s 2008 Community InnovationSurvey108showsthattheproportionofinnovativefirmsin some categories of KIBS, including 59 per cent

FigureC.32:R&D spillovers by distance and degree of vertical integration

Vertically integrated countries All countries

0 km

0.02

0.04

0.06

0.08

0.1

0.12

0

20

0

60

0

1,0

00

1,4

00

1,8

00

2,2

00

2,6

00

3,0

00

3,2

00

3,4

00

3,6

00

3,8

00

4,0

00

4,2

00

4,4

00

40

0

80

0

% c

hang

e in

num

ber o

f pat

ents

du

e to

10

% c

hang

e in

fore

ign

R&

D s

pend

ing

1,2

00

1,6

00

2,0

00

2,4

00

2,8

00

Source:Authors’calculationsbasedonPiermartiniandRubinova(2013).SeeBoxC.8fordetails.

Note:“Verticallyintegratedcountries”aredefinedasthosecountry-pairswithashareoftradeinintermediatesabovethemedian.


158

BoxC.8: Production network and the geography of technological spillovers: methodology

The estimates for international R&D spillovers in Figures C.31 and C.32 were obtained through aneconometricmodel (PiermartiniandRubinova,2013).Following theeconomic literature (Keller,2002;andBottazzi andPeri,2003), acountry’spatentsapplicationswere related to itsR&Dspendingand theR&Dundertakeninforeigncountries.Intuitively,ifthelevelofforeignR&Dmattersfordomesticinnovation,someofthetechnologycreatedabroadmustcrossinternationalborders.

Inparticular,usingapanelof41countriesovertheperiod1996-2007,thefollowingequationwasemployed:

ln(Patents)it=α+β∙ln(R&D)it+γ∙PoolR&D it+Xit ∙δ+t+e it

where Patents indicates the number of patent applications of country i at time t, R&D denotes domesticbusinessexpenditureinR&DandPoolR&D isthepoolofR&Davailabletothehomecountryandgeneratedabroad.All variablesare in logarithms (ln). Inparticular, the variablePoolR&D is calculatedasaweightedaverageofallforeigncountries’R&Dexpenditure,wheretheweightsarethedistancesbetweenthehomecountryandeachforeigncountry.Theformulausedtoconstructthisvariableis:PoolR&D it=∑j≠i ln(R&D)jt∙exp(-distance ij)

Asetofcontrolvariables(denotedbyXintheequationabove)wasalsointroduced.Theseincludepopulationand real GDP per capita to control for a country’s market size, the share of tertiary graduates in totalpopulation tocapture thecountry’scapacity togenerate innovation,andthe levelofpatentprotectionandtheoriginofthelegalsystemtocontrolforqualityofinstitutions–animportantdeterminantoftheincentivestoinnovate.

Thecoefficientofinterestisγ.Thisindicatesthepercentagechangeindomesticpatentingactivityduetoa1 per cent increase in the pool of foreign R&D. A positive value of this coefficient suggests internationaltechnologicalspillovers.

To test how production networks affect international spillovers, the variable PoolR&D was split intotwocomponents.OneisthepoolofR&Dfromcountriesthatarehighlyverticallyintegratedwiththehomecountry.TheotheristhepoolofR&Dfromtherestoftheworld.Theverticalconnectionwasdefinedonthebasisoftheshareofmachinerypartsandcomponentscountry i importsfromcountry j relativetothetotalmachinery imports of country i. Thus, country j is identified as a country highly vertically integrated withcountryiifitsexportsofintermediatestocountry i areabovethemedian.

When the R&D spending pool was split, it was found that R&D spending only from countries that areimportantinputsuppliershasapositiveandsignificanteffectonthehomecountry’spatentingactivity.

in information and communication and 52 per centin finance and insurance activities, exceeded the51 per cent share in the manufacturing sector(Meliciani,2013).

Data on patent applications highlight the significantcontribution of ICT-related technologies to innovationoverthepastthreedecades.TableC.10showsthetentechnologyfieldsthatexperiencedthefastestgrowthin termsofpatentapplicationsover theperiod1980-2010.Amongthesetoptentechnologyfields,fivearerelated to ICT development – namely, IT methods formanagement, digital communication, computertechnology,semiconductorsandtelecommunications.

The importance of the ICT sector in innovation overthe past decades has led many to identify the ICTrevolution as the third period of industrial innovation.This revolution began in 1960 and followed twopreviouswavesofinnovations.Thefirst,between1750and1830,createdsteamengines,cottonspinningandrailroads. The second, between 1870 and 1900,

produced electricity, the internal combustion engineandrunningwaterwithindoorplumbing.Jorgensonetal. (2005) extensively study the contribution of IT toproductivityandgrowth.Theyestimatethatasagroup,IT-producingindustriescontributedmoretothegrowthof total factor productivity between 1977 and 2000thanallotherindustriescombined.

However, other economists have questioned whetherICT innovations have had as profound an impact oneconomicgrowthasprevioustechnologicaladvances,such as steam power or electrification. In a recentpaper, Gordon (2012) argues that the ICT revolutionhas not fundamentally changed living standards andthat its economic impact is already diminishing. Insupport of his argument, he notes the slowdown inUSproductivitygrowthsincethe1970s.

Others highlight other explanations for the USproductivity slowdown and are more optimistic aboutthepotentialgrowthimpactoftheICTrevolution.First,energy price shocks in the 1970s and 2000s may


159

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

FigureC.33:Technology spillovers within a region versus spillovers into other regions

0

0.2

0.1

EU27 CIS North America South America West Asia East Asia

0.3

0.4

0.5

% c

hang

e in

num

ber o

f pat

ents

due

to 1

0% c

hang

ein

fore

ign

R&

D s

pend

ing

at 3

00 k

m

0.6

0.7

0.8

0.9

Intra-region spillovers Extra-region outward spillovers Extra-region inward spillovers

Source:Authors’calculationsbasedonPiermartiniandRubinova,2013.SeeBoxC.8fordetails.

FigureC.34:Exports of high-tech products within a region versus between regions, 1998 and 2011 (US$billionandpercentage)

1998 2011

North America

150 155

28%

72%

34%

66%

1998 2011

Europe

221

564

62%

38%

58%

42%

74%

Extra-region Intra-region

1998 2011

South and Central Americaand the Caribbean

4.3

16.3

33%67%

32%

68% 1998 2011

Africa

0.9

2.2

70% 35%65%

1998 2011

Middle East

3.3

6.9

99.7% 97%

1998 2011

Commonwealth ofIndependent States

2.6

5.4

81% 27%

73%

1998 2011

Asia

140

540

46%

54%

59%

41%

Source:Authors’computation,basedondatafromWorldIntegratedTradeSolution(WITS),WorldBank,athttp://wits.worldbank.org/wits/.Thedefinitionofhigh-techproductsfollowsOECDWorkingPartyonInternationalTradeinGoodsandTradeinServiceStatistics(2008).

Note:BoundariesdonotimplyanyjudgementonthepartoftheWTOastothelegalstatusofanyfrontierorterritory.

betterexplaintheproductivityslowdown,whichstartedin the 1970s and gradually spread to the widereconomy via the most energy-intensive sectors.Secondly, the information technology revolution maystill be in its early phase, with its major economicimpactsyettobefelt.Thepast twomajor technologywaves, in the early 19th century and in the early20th century, required almost a century before theirimpactfullydiffusedthroughouttheeconomy.

Moreover, the influence of technology and innovationon the economy may be cumulative. While doublingtechnologicalcapacitymaynotmattermuchwhentheinitial level is low, it can have huge effects when thelevel rises.109 The exponential growth of the internetoverthepasttwodecades,asthesynergiesbetweenexistingcommunicationsandinformationtechnologiesarerecognizedandexploited,illustratesthiscumulativeeffect.


160

TableC.9:Business R&D by sector, 1990-2010Sector 1990-95 average 2005-10 average

Value Share of total BERD (%) Value Share of total BERD (%)

Agriculture,huntingandforestry 578.5 0.4 606.1 0.2

Manufacturing 126,442.8 88.9 200,273.1 81.0

Services 9,470.8 6.7 41,703.0 16.9

Source:Authors’computations,basedondatafromOECDScience,TechnologyandR&DStatistics,athttp://www.oecd-ilibrary.org/science-and-technology/data/oecd-science-technology-and-r-d-statistics_strd-data-en

Note:Totalover24countries;valuesinUS$million,PPPadjusted,2005constantprices,shareinpercentage.Forthepurposeofconsistencyandcomparability,theaggregationofbusinessR&Dbysectorsisdoneusingonlycountrieswithdataforallthreesectors,bothintheperiodof1990-95and theperiodof2005-10.Asa result,24countriesare in the sample– i.e.Austria,Canada,CzechRepublic,Denmark,Finland,Germany, Greece, Hungary, Iceland, Ireland, Japan, the Republic of Korea, Mexico, Netherlands, New Zealand, Norway, Portugal, Romania,Singapore,SlovakRepublic,Slovenia,Spain,SwedenandTurkey.

TableC.10:Patent publication by technology field (ten fastest growing), 1980-2010 (percentage)

Technology fieldAverage

growth rate 1980-2010

Share of world total patent publication

in 2010

Micro-structuralandnano-technology

98 0.17

ITmethodsformanagement

58 1.31

Digitalcommunication 39 4.27

Computertechnology 26 7.37

Biotechnology 24 2.28

Semiconductors 22 4.35

Medicaltechnology 20 4.41

Telecommunications 18 3.20

Analysisofbiologicalmaterials

17 0.67

Audiovisualtechnology 16 4.57

Source:Authors’computations,basedondatafromWIPOIPStatisticsDatabase,athttp://ipstatsdb.wipo.org/ipstats/patentsSearch

The role of multinationals versus SMEs

Most R&D spending is conducted by firms based inOECD countries; multinational firms in particular aremajor drivers of R&D spending.110 Available data for1999showthatintheUnitedStates83percentofallmanufacturing R&D was conducted by parentcompaniesofUSmultinationals(NSF,2005).

Sofar,smallandmedium-sizedenterprises(SMEs)haveunderperformedrelativetolargerfirms,bothintermsofR&D spending and innovation. In a recent report onSMEsandinnovation,theOECD(2010a)observesthat“SMEsinnovatelessthanlargefirmsacrossarangeofcategories including product innovation, processinnovation,non-technologicalinnovation,new-to-marketproduct innovation and collaboration in innovationactivities”. This observed gap still persists even afteradjustingforfirmsize–i.e.SMEshavelowerinnovationratesperemployeethanlargerfirms(Audretsch,1995).However, thisstatisticalgap tends toobscure the factthatthereissubstantialinteractionbetweenlargefirmsand SMEs in innovation. SMEs that have producedbreakthrough innovations are often acquired by largefirms which then build upon and commercialize theinitialinnovation.

There are reasons to expect that SMEs will becomeincreasingly important in the global landscape ofinnovation. Recent developments in productiontechnologies and consumer tastes suggest thateconomiesofscaleandscopeinR&Dandproduction– the competitive edge of larger firms – will becomeless significant and advantageous in the future. TheOECD (2010a) points to two particular trends whichmayreducetheimportanceofeconomiesofscaleandscope, and potentially empower SMEs. First, someinnovations,suchas3Dprinting,willmake itpossibleforSMEsacrossnumerousindustriestoproduceonasmall scale as efficiently as large-scale production.Secondly, as global consumers’ incomes rise, theirdesireforvarietyincreasesaswell.ThisincreasesthescopeforSMEstofillnichemarkets.

Both trends mean that the multinationals’ currentadvantage in producing standardized products on alargescaleatalowcostmaydiminishinthefuture.Asaconsequence,onemayexpect thatsmall innovatingfirms will be more likely to commercialize their owninnovationsandtoinvestmoreinadditionalinnovations.

(b) Technologyandtrade:Atwo-wayrelationship

Traditionaleconomictheoryviewedacountry’sleveloftechnology as an exogenous explanatory variable oftrade – that is to say, technology is taken as a givenfactor shaping other economic variables, includingexports and imports. However, in the real world,technological change is not drawn randomly from aglobalpoolof innovationbut rather is theoutcomeofeconomic forces. When firms decide how much toinvest in R&D, they consider the expected economicreturns from innovation. The greater the expectedrewards for a dollar spent in R&D, the greater theirincentivetoinvestininnovation.

Severalfactorsaffectfirms’incentivestoinnovate,oneof which is trade. Thus, to understand howtechnological progress will affect future patterns oftrade, it is also important to understand how tradeitselfaffectstechnologicalprogress.

Thissectionfirst looksathowtechnologicalprogressaffects trade, then it discusses how trade and otherfactorsshapetechnologicalprogress.


161

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

(i) How does technology affect trade?

Shaping comparative advantage

Economic theory views technology as a factordetermining the patterns of trade. According totraditional theory, tradeoccursbecausecountriesaredifferentandoneofthesedifferencesistechnology.Inshaping comparative advantage, technologicaldifferences between countries help to shape thepatterns of trade. In the simplest Ricardian model, acountry exports the good which it is relatively moreefficientatproducingthanitstradingpartner–thatis,thegoodwiththelowestopportunitycost.

Until recently, trade theory ignored differences acrossfirms, and trade models assumed that all firms in acountry shared the same technology. However, thesetraditionalmodelsfailedtoexplainevidencethatnotallfirmsexport,and thatexportingfirms tend tobe largerand more productive than non-exporting firms. In thenewtrademodels,firm-specifictechnologicalknowledgeisseenasakeydeterminantofwhetherafirmexportsorjustservesthedomesticmarket(Melitz,2003).

A firm’s relative productivity also helps to explainwhetheritwillexportitsproductsorsellthemthroughaforeignsubsidiary–i.e.throughsocalled“horizontal”FDI (Helpman et al., 2004). The assumption is thatexporting involves lower fixed costs than FDI, whileFDI requires lower variable costs than exporting.Becauseofexistingfixedcostsofexporting,only themost productive firms will export, and among theseonlythemostproductivewillengageinFDI.

Inaworldwherefirmsproducefinalgoodsbyassemblinga range of intermediate goods, technology is also animportantdeterminantofwhetheracertaininputortaskis produced domestically and exported or whether it isoffshored.Ingeneral,trademodelsofverticallyintegratedfirms assume that technology can be transferred fromtheparent to theaffiliatecompany(this includesrecentmodelsof trade in tasks,which thenoccurs in linewithcomparativeadvantageinfactorsofproduction).

However,whentechnologytransfer iscostly foragivenmarket, technologicallycomplex inputswillbeproducedat home and exported, and only the more standardizedinputs will be produced abroad. This is because morecomplex inputsmay involvehighercostsof transferringthe technological information needed for offshoreproduction. Moreover, if the technology involved in theproductionofintermediategoodscanonlybetransferredthroughface-to-facecommunication,inputsimportedbyanaffiliate from itsparentcompanywillbe increasinglytechnologically complex as the distance between theparent company and the affiliate increases. In fact, USexports show a positive relationship between thecomplexity of exports (measured as the average R&Dintensityofexports)andthegeographicaldistancetothedestinationmarkets(KellerandYeaple,2009;2012).

The traditional Ricardian models of trade, as well asheterogeneous firms’ models, do not account fortechnological spillovers. Imports embody foreigntechnology but they do not change the importer’stechnological know-how. Similarly, in the traditionalmodel of vertically integrated firms, there are notechnological spillovers from the affiliate company tothedomesticfirms.However,evidenceclearlysupportstheviewthatknowledgespilloversexist.

What does this imply for trade patterns? If countries’accesstotechnologywereidentical–i.e.iftechnologicaldiffusionwereperfectandglobal– tradewouldoccuronly on the basis of relative factor abundance ratherthan technological differences (Heckscher-Ohlintheory). However, clearly, technological diffusion isneither perfect nor global. Thus, understanding thegeographical extent of technology transfers and theirimpact is essential to understanding which factorsshapetrade–relativefactorabundanceortechnologicaldifferences.

The concepts associated with the new economicgeography can provide important additional insightsinto the way technology diffusion has an impact onproductionand tradepatterns (Krugman,1991;Headand Mayer, 2004; Krugman, 1998). Since technologyspillovers are greater among firms located in closeproximitytooneanother–helpingtodrivedowntheirproductioncostsandmake themmorecompetitive ininternational markets – these spillovers indirectlycreate agglomeration forces that shape trade. Tobenefitfromtechnologyspillovers, industrieswill tendto be concentrated in certain places, especially in acountry with a large domestic market for the goodbeing produced. Locating in a large market will alsobenefitfirmsbyreducingtransportandtradecosts. Itfollowsthat,underthesecircumstances,acountrywillexport the product for which it has a home marketadvantage–thatistosay,theproductforwhichithasthelargestdomesticdemand.111

Reducing trade costs

Trade costs are generally estimated to be a moresignificant obstacle to trade than policy barriers. In2004,forexample,aggregateexpenditureonshippingonlywas three timeshigher than theaggregate tariffduty paid (Anderson and Van Wincoop, 2004). Thus,any change in trade costs is likely to affect tradesignificantly.

Technological innovation has had a major impact ontrade costs. The introduction of containerization andjet engines has significantly reduced sea and airtransport costs. More recently, the use of radiofrequencies, identification tags and the internet hasallowed firms to keep track of where a product is atany time. This has significantly improved logisticsservicesandmadepossiblethedevelopmentofamoreefficient multi-modal transport system. As will be


162

discussed in Section C.4, the reduction of transportcostshasasignificantimpactonboththevolumeandcompositionoftrade.

Technological advances have also significantlyreduced communication costs. Exporters needinformation on profitable trading opportunities.Importers need information on suppliers ofintermediategoods,onproductspecifications,andonscheduling production processes. The telephone isstillaprimarymeansofcommunicationbuttheinternetis becoming an increasingly important, versatile andlow-cost communications tool. Mobile phones arebecoming increasingly important to commerce,especially in developing countries, because theyrequirelessinfrastructureandareuntiedtolocation.

Loweringthecostofcommunicationsaffectstrade inseveral ways. First, it can lower the variable costs oftrade and thus help to increase trade volumes – inmuch the same way that lowering tariffs increasestradevolumes.Secondly,itcanlowerthefixedcostsoftrade by improving exporters’ or importers’ access toinformation – everything from market intelligence topotential trade partners. As discussed above, whenfixed entry costs are high, only the most productivefirms can export, so a reduction in communicationcosts can be expected to encourage smaller, lessproductivefirmstoenterinternationalmarkets.

ExaminingeBaytransactions,arecentstudybyLendleet al. (2012) shows that while most “offline” sellersexportonlyoneproducttoonemarket,mostsellersoneBay export to more than five markets and in morethan five product categories, suggesting that theinternet has significantly lowered cross-border tradecosts for small business, especially the cost ofmatching buyers and sellers. Thirdly, lowercommunication costs can affect the composition oftrade. Because some tradable sectors are moreinformation-sensitivethanothers–suchasgoodswithshort product cycles (e.g. consumer electronics) orones that feed into complex production chains (e.g.automotive parts) – falling communications costs willdisproportionately benefit them. Fink et al. (2003)showthattheimpactofareductionincommunicationcosts is as much as one-third higher for trade indifferentiated goods (e.g. technologically advancedmanufacturing goods) than for trade in homogenousproducts (e.g. agriculture or standardizedmanufacturinggoods).

ICT and trade

The development and diffusion of ICT has had aparticularly powerful impact on trade – including thegrowingimportanceofintermediategoodsintrade,ofservices trade, of e-commerce and of developingcountries. ICT has been an essential prerequisite tothe rapid growth of global supply chains by makingproduction coordination across borders easier.

Production chains require deep and continuouscoordination between headquarters and affiliateactivities. Sharing information between terminaloperators, shippers and customs brokers and a widerange of other actors is essential for the efficientmanagement of production networks, where just-in-time delivery is required. One by-product of the ICTrevolutionisthatworldtradeinpartsandcomponentshas increased much faster than total merchandisetradesincetheearly1990s.112

ICT developments have also underpinned the growthof services trade, including the offshoring of serviceactivities, such as data processing, research anddevelopment and business processes to lower-costlocationsaroundtheworld.Serviceswhichwerenon–tradable in the past – or tradable only at very highcosts – have become highly tradable today. This isparticularly true of knowledge-intensive businessservices (KIBS): legal services, accounting, taxconsultancy, market research, auditing, managementconsultancy, architectural, engineering and technicalconsultancy, technical testing and analyses,advertisingandotherbusinessactivities.

KIBS’sshareofworldtradegrewatanannualrateof8 per cent between 1990 and 2000, and at 10 percent between 2000 and 2010 (National ScienceBoard,2012),innosmallpartbecauseoftheimpactofICT developments (see Section B). ParticularlysignificantisthegrowthofKIBSinemergingmarkets.Since 1990, China, India, Indonesia and the RussianFederationhaveexperiencedparticularly highgrowthin termsof theirshareofglobalvalue-added inKIBS.Specifically, China reached 5.5 per cent of globalvalue-added inKIBS in2010,up from1.6percent in1990;Indiaaccountedfor2percentin2010,upfrom0.8percentin1990(Meliciani,2013).

ICTdevelopmentshavealsochangedthenatureoftheproductsthataretraded–fromtradeinphysicalgoodstotradeindigitalgoods,fromtradein“atoms”totradein“bits”.Musicandfilmmarkets,forexample,arebeingcompletely transformed by e-commerce anddownloading, making trade in physical CDs or DVDsincreasinglyobsolete.Blinder(2006)suggeststhatasthe distinction between tradable and non-tradablegoods and services becomes increasingly blurred, sotoo will trade theory predictions based on thetraditional factor endowment of skilled and unskilledlabour. In particular, he argues that as an economybecomesmoreservice-oriented,thenewtradetheoryshould focus on personal versus impersonal servicesas a source of comparative advantage, as the lattercanbeeasilyoffshoredwhiletheformercannot.

Given current trends, it is also likely that ICTinfrastructure will become an increasingly importantfactorshapingtradeflowsinthefuture.113Forexample,developing countries’ potential to “leap-frog” to thenext levelof ICT infrastructure–asmanyarealready


163

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

doing in the case of mobile phone technology – maybe a source of competitive advantage vis-a-visdevelopedcountrieswhichareburdenedwiththesunkcosts of traditional communication infrastructures.Although the so-called “digital divide” between highandlow-incomecountriesisstill large,thereareclearsigns that it is narrowing (see Figure C.35 andAppendix Table C.3). Over the last two decades, thegrowthinfixedlineandmobileconnectionsaswellasin the number of internet hosts has been faster indeveloping than in developed countries. One reasonforthisisthatwhilefixed-linecommunicationsrequiresa substantial investment in infrastructure, the initialinvestmentinmobilenetworksisrelativelymodest.

Other indirect channels

Technological changes also affect trade indirectlythroughtheirimpactonotherfactorsshapingtrade.

The ICT revolution provides a clear example of themany dimensions in which technology’s impact ontrade can be analysed. The use of the internet forbanking, for buying and selling goods, for organizingtravelandaccommodationarea fewexamplesof themany ways in which ICT developments increaseinternational competition, reduce trade costs andcreate new markets. However, the effects of the ICTrevolution go well beyond their direct impact onservicestradeandproductmarketcompetition.

First, ICT has significantly changed the way that thelabour market operates. The internet reduces searchcosts for a new job and vastly expands the

geographicalscope,creatingabettermatchbetweenemployersandemployees,andreducing thefrictionalrateofunemployment.Italsoallowsindividualstoworkor conduct business outside the office or companypremises.Newandmoreflexibleformsofemploymentare opening up, improving employment prospects,especially fordisadvantagedpeople in the jobmarket(e.g. older workers and women).114 These and otherimpactsoflaboursupplyontradeareanalysedinmoredetailinSectionC.1.

Secondly, ICT has an impact on human capitalaccumulation.Forexample,onlineuniversitiesofferanexpandedarrayofcourseoptions–fromprofessionalcourses to post-educational programmes – andincreaseaccesstospecializedknowledgefromremotelocations. The effects of human capital accumulationontradeareanalysedinSectionC.1.

In general, technology changes and technologytransfer have a strong impact on income distributionand inequality. The effects of income inequality ontradeareanalysedinSectionD.1.

Insum,thediffusionofICTworldwidemaybeexpectedtoyieldsignificantchanges in international trade.Notonlywillthetrendtowardslowercommunicationcostsandincreasedtradevolumesbelikelytocontinue,butchanges in the patterns and nature of trade are alsoinevitable. First, the importance of e-commerce overother forms of trade will continue to increase.Secondly, the role of SMEs in exporting can beexpected to assume greater importance. Thirdly, astheICTnetworkexpands,newplayerswillbelikelytoemerge in information-intensive sectors, such asconsumer electronics and automotives, and therelative importance of factors of comparativeadvantagewillchange.

(ii) How does trade affect technological progress?

To understand how technological progress will affectfuture patterns of trade, it is also important tounderstand the factors shaping technologicalprogress. One of these factors is trade itself. Tradeaffectstechnologicalprogressintwoways:throughitseffect on the incentive to innovate and throughtechnologytransfers.

Trade and innovation

Trade affects firms’ incentive to innovate through itseffect on the size of the market in which a firmoperatesaswellasthrough itseffectoncompetition,technology transfersand institutions.FirmsspendonR&Dtoincreaseprofitsortokeepupwithcompetitionfrom other innovating firms. All else being equal, thelargerthemarket,thelargerthefirm’sexpectedprofitsfrominnovation.Byincreasingthesizeofthemarketinwhich a firm operates, trade provides firms with the

FigureC.35:Annual growth in telecommunications infrastructure by income group, 1995-2011

-5%

5%

15%

25%

35%

45%

55%

65%

75%

Low-income countries

Fixed-lines Mobile cellularsubscription

Internet users

Middle-income countries

High-income countries

Source:Authors’computations,basedondatafromWorldBank,World Development Indicators.


164

opportunity for greater profits, thus increasing theirincentivetoinvestinR&Dandthereforetheprobabilityofinnovation–i.e.therearepositivescaleeffects.

Trade also increases competition. The effects ofcompetitiononinnovation,however,arelessclearcut.On the one hand, by reducing the monopoly rents115

associatedwithinnovation,competitionisexpectedtoreduceincentivestoinnovate(Schumpeter,1942).Onthe other hand, more rigorous competition may givefirms a greater incentive to innovate because if acompetitor innovatesfirst, rivalfirmsare likely to losemarket share, experience losses and possibly beforcedtoexitthemarket.

Tradecanalsoaffectinnovationincentivesthroughitseffectson technological spillovers.Again, theeffectsof technological spillovers on the innovation rate arenot one-way. While imitation may foster R&Dinvestmentsinanefforttoriseabovethecompetition(Helpman, 1993), the reduced returns to innovationmayreducefirms’ incentivetoengage inR&Dactivity–i.e.thereareambiguouseffectsofimitation.

Finally,tradeshapestheinstitutionalframework,whichinturnshapestheeconomicincentivesoffirms.Aswillbe discussed in Section C.6, there is a positivecorrelation between trade and quality of institutions,and countries with better institutions tend to investmore in education and infrastructure. These linkagesgenerateapositiverelationshipbetweentradeandthereturnstoinnovation,thusfosteringfirms’incentivetoinvestinR&D.

The economic literature on the empirical relationshipbetweentradeandtechnologicalprogressisextensive,andincludesbothcountry-levelandfirm-levelstudies.Ingeneral, empiricalevidencebasedoncountry-leveldata shows a correlation between trade andinnovation.116However,onegeneralcriticismof thesestudiesisthattheydonotmanagetodistinguishfullybetweencauseandeffect.Thisisbecauseitisdifficultto disentangle trade policy changes from otherdomesticpolicy changesundertakenbygovernmentsthat simultaneously affect growth (Rodriguez andRodrik,2001).

More recent studies based on firm-level data alsosupport theviewthat trade increasesthe incentivetoinnovate. Focusing on trade opening betweenArgentinaandBrazilbetween1992and1996,Bustos(2011)findsthatArgentinianfirms insectorswiththelargest market access gains were more likely toincrease technology spending thanfirmsoperating insectorswheretradeopeningwaslessambitious.117

Trade and technology transfers

Totheextentthattechnicalknowledgeisembodiedina product, it also travels with the product. In otherwords, imports of technologically advanced goods

provide firms with access to the technologiesembodied in the imported good. Such imports canincrease productivity both by using the good inproduction processes and by providing opportunitiesfor “reverse engineering” – learning about how animported product is produced and imitating it. To theextentthattheexpenseofreverseengineeringislessthan the expense of developing the technologyindependently, the importing country derives a gainfromimporting–orfromtechnologicalspillover.

In addition, international trade provides a channel ofcommunicationthatencouragescross-borderlearningofproductionmethods,productiondesignandmarketconditions.Throughexporting,firmsalso interactwithforeigncustomers.Thesecustomersmayalsodemandhigher quality standards than domestic customerswhileatthesametimeprovidinginformationonhowtomeetthosehigherstandards.Thus,exportingbecomesa channel for technology transmission for “learning-by-exporting”.

Several empirical studies confirm that imports areanimportantchanneloftechnologydiffusion.Inparticular,the extent of technological spillovers appears to belinked to the composition of imports. Technologytransfer is higher when imports come from industrialcountries and are presumably characterized by ahigher embodied technological content than importsfrom developing countries (Coe and Helpman, 1995;Coe and Hoffmaister, 1999; Keller, 2000).Furthermore, technology spillovers are stronger forimports of capital goods, machinery and ICT goods(Coe et al., 1997; Gera et al., 1999; Xu and Wang,1999; Acharya and Keller, 2009; Van Meijl and vanTongeren,1998).AstudybyAmitiandKonings(2007)alsoshows that technology transfersarestronger forimportsofinputsthanfinalproducts.

Empirical evidence supporting the existence of alearning-by-exporting effect is less extensive. This ispartly because studies attempting to assess thisimpact face two methodological hurdles: first, how tocontrol forself-selectionof themostproductivefirmsinto export markets. Does exporting – and learning-by-exporting–makefirmsmoreproductiveordoonlythemostproductivefirmsexport?Theseconddilemmais how to distinguish between productivity gains thatare the resultof learning-by-exportingandgains thatare the result of high profits from exporting and ofgreater incentives to invest in technology (i.e. scaleeffects).Despitetheextensiveevidencethatexportingfirms tend to be more productive than firms onlyserving the domestic market (Bernard and Jensen,1999), this might simply reflect the fact that onlythemostproductivefirms(self-selection)exportinthefirstplace.

However, some evidence based on micro-level datasupports the learning-by-exporting hypothesis. Forexample, using firm-level data for Slovenian


165

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

manufacturing, De Loecker (2007) finds that theproductivity of exporting firms increases once theystartexportingandthattheproductivitygapbetweenexporters and their domestic counterparts increasesover time.118However,whilefirm-levelevidence takesintoaccountself-selection, itstilldoesnotdistinguishbetweenwhetherproductivitygainsarisemainly fromtechnology transfers or from higher incentives toinnovate.

(iii) What other factors affect technological progress?

One determinant of technological progress is thestrengthofintellectualproperty(IP)rights.Theoreticalarguments on the relationship between IP protectionandtechnologicalprogressaremixed.119Advocatesofstronger IP protection claim it will lead to moreinnovationbyincreasingfirms’rewardsforundertakingresearch.Moreover,evenifmuchoftheresearchtakesplace in advanced economies, stronger IP protectionwillfacilitatetechnologytransferbyencouragingmoreFDI, especially from high-technology firms.120 OthersarguethatstrongIPprotectionwillreducetechnologytransfers and may even reduce the incentive toinnovatebyentrenchingmonopoliesandbydiminishingthecompetitive-threatincentivetoinnovate.

Theempiricalevidenceisequallymixed.Forexample,Coeetal. (2009)findthatstrongpatentprotection isassociated with higher levels of total factorproductivity, higher returns to domestic R&D, andlarger international R&D spillovers. Using data onUS multinationals, Branstetter, Fisman and Foley(2006)alsosupporttheviewthatthereisanincreasein technology transfers to countries that strengthentheir IP regime. However, several studies, such asBessen and Maskin (2000), Lerner (2002a, 2002b),Sakakibara and Branstetter (2001) and Scherer andWeisburst (1995), suggest that there is a negativecorrelation between strengthening IP protection, ontheonehand,andincreasinginnovationortechnologydiffusionontheother.

Other importantdeterminantsof technology transfersareFDIflows,movementofpeople121anddirecttradein knowledge through technology purchases orlicensing. Any policy that affects these channels hasan impact on technology transfer. For example,Hovhannisyan and Keller (2012) show that businesstravel plays an important role in diffusing innovationand suggest that lifting limits on the cross-bordermovement people as well as liberalizing internationalpassenger air travel could have additional benefits intermsofincreasinginnovation.

A large body of literature analyses the potentialspillover effects of FDI. FDI can increase technologytransferbyencouraginginteractionbetweendomesticand foreign firms. One channel is vertical FDIspillovers. Linkages between upstream and

downstream producers can encourage the directtransfer of technology from the multinational to thelocal buyer. Higher-quality requirements onintermediate inputs from suppliers can also result intechnology transfers. Another channel is horizontalFDIspillover.Geographicalproximity tomultinationalscan reduce the costs of learning or adopting a newbusiness technology within the same industry. Whileolder empirical studies suggest that technologyspillovers were associated with more vertical ratherthanhorizontalFDI,more recentempiricalworkfindssignificant technology spillovers from horizontal FDItoo(KellerandYeaple,2009).122

Itisimportanttonotethattheinternationaldiffusionoftechnologyisnotautomatic.Technologytransferisnotjust a question of “supply” but of “demand” – and, inparticular,ofafirm’sorcountry’s“absorptive”capacity.Forexample,inorderfortechnologytobetransferredthrough the use of specialized and advancedmachineries invented abroad, it is necessary forworkerstohavetheskillsneededtousethemachineryand organize the production process. Mayer (2001)shows that it is the combination of the know-how oftheworkforceandtheimportsofmachinerywhichhasa positive effect on economic growth. Even reverseengineering requires skills. The capacity to absorbinternationalR&Dspilloversdiffersacrosscountries.Arecent study by Coe et al. (2009) suggests thattechnological spillovers increase with the ease ofdoing business in a country and the quality of itstertiaryeducationalsystem.

(c) Whatdothesetrendsmeanforthefutureoftrade?

Severaltrendsarediscerniblefromthisanalysisoftheglobalpatternsof innovationand technology transfer.One is the emergence of new players among thecountries that are driving technological progress.Some countries have significantly accelerated theirinnovative ability and capacity to absorb existingtechnologies.AmongtheseareChina,theRepublicofKorea and Singapore. However, there are alsocountries, especially in Africa, that continue to lagbehind. The low quality of education and theinstitutional framework in these countries is primarilyresponsiblefortheirlowabsorptivecapacity.

A second trend is the regionalization of technologytransfers. By reducing coordination costs, the ICTrevolution has fostered the development of supplychains. Supply chains embody several relateddimensions of international economic relationships –investment,competitionandmovementofpeople–allof which intensify technology transfers. However,supplychainsdonotincreasetheflowoftechnologicalknowledgeatthegloballevel.Theyincreaseitamongcountries with regional networks, thus encouragingtheformationofregional“convergenceclubs”.


166

Thirdly,ICTdevelopmentshavesignificantlyincreasedthe share of services in world trade. In particular,knowledge-intensive business services (KIBS) areemerging as key drivers of knowledge accumulation.These trends – together with reduced productivitygrowth in manufacturing – may point to a potentialshift frommanufacturingtoservicesastheengineofglobalinnovation.

Finally,SMEsappeartobenefitfromimprovedaccessto the international market. By dramatically reducinginformation, transaction and searching and matchingcosts,theICTrevolutionhassignificantlyreducedthefixed costs of entering markets, thus increasingopportunitiesforSMEs’participation.

What should we expect from these trends?

The emergence of new global players, together withtechnological convergence at the regional level, islikely to lead to the emergence of economic actorsthat no longer see countries as the unit of referencefor international relationships. This could haveimportantconsequencesintermsofhownegotiationsareconductedatthemultilaterallevel.

Secondly, technologicaladvancementshavebeenkeytothedevelopmentofsupplychains.Supplychains,inturn, have encouraged technology transfer andconvergence across countries. If the process ofproduction fragmentation continues or intensifies,governments will be pressured to adopt policies thatfacilitate domestic industries’ integration intoproduction chains. These can take the form of R&Dsubsidies, infrastructure investments and stronger IPprotectiontoencourageFDIinflows.

Furthermore, the globalization of R&D, thefragmentation of production processes and thediffusionofdigitaltechnologiesarecreatingamismatchbetween the geographical scope of economic agentsand the regulatory regime under which they operate.For example, while the internet allows consumers toshop globally, IP protection and competition laws areadministeredandenforcednationally.

Toallow for the full potential ofe-commerceand theglobalization of production to materialize, IP andcompetition regimes will need to adapt. Pressures toextend rules beyond national borders are alreadymanifested in themultiplicationof “deep”preferentialtrade agreements that include IP and competitionpolicy provisions. More generally, the link betweentrade and technological progress points to the needfor freer flow of goods, services and ideas at themultilateral level. If technology spillovers result fromtrade,forexample,coordinatedactiontoreducetradeobstacleswouldincreaseeconomicwell-being.ThisisfurtherdiscussedinSectionE.

Thirdly,whiletheanalysisoftradepatternsinSectionBreveals a relocation of labour-intensive activities to

developing countries and the emergence of a smallnumber of firms as global trade players, recentinnovations,suchas3Dprintingandrobotics,arelikelytochallengethisstatusquo.3Dprintingisaprocessofmaking a three-dimensional solid object from a digitalmodel by adding material layer by layer. With only rawmaterials and encrypted data streams required formanufacturing, and as production becomes moreindividualized, access to these new technologies maymakeitfareasierforSMEstoenterexportmarkets.Inaddition,byreducingtheimportanceoflabourcostsforcomparative advantage, robotics and 3D printing mayalso shift manufacturing, together with whole supplychains,backtodevelopedcountries.

As of 2012, 3D printing technology is used forprototyping and manufacturing in sectors such asconstruction,aerospace,jewelleryandhealthcare.But,it is foreseeable thatas theprintingspeed increases,itsusemayspreadtohouseholds.Ifthishappens,onemayevenpredictareductionofglobaltradeincertaintypes of goods, if end users can easily manufacturethem.

However, traditional production methods (sometimereferred toassubtractiveprocesses)and3Dprintingare likely to complement each other rather thancompete.3Dprintingmayproveadvantageousfortheproduction of components characterized by internalvoids,suchastubes.Butforproductionprocessesthatstartwithasolidmassfromwhichmaterialisremovedin order to obtain the desired design, traditionalmanufacturing may continue to prevail. The effect of3D printing is therefore likely to vary significantlyacrosssectors.

4. Energyandothernaturalresources

Like labourandcapital, natural resourcesare factorsof production that serve as inputs in goods andservices production. While there is a broad range ofnatural resources that could be discussed, the focusherewillbeonenergyand,toalimitedextent,onlandand water, which are the natural resources typicallyincluded in aggregate production functions (fordiscussion of trade and a wider variety of naturalresources, see the 2010 World Trade Report (WTO,2010)andRutaandVenables(2012).

Thesectioncoversfourthemes–unevengeographicaldistribution, volatility of prices, exhaustibility andinnovation,andnegativeenvironmentalexternalities–that correspond to fundamental characteristics ofnaturalresourcesandwhichcanaffectbothproductionandthepatternoftrade.

Part(a)discussestheunevengeographicaldistributionof natural resources, which affects countries’comparative advantage and hence the pattern of


167

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

international trade.Differences in factorendowmentsconfermarketpoweron resource-abundantcountriesand have geopolitical implications. Part (b) describeshow increases in natural resource prices can havemajorcontractionaryeffectsoneconomies,whichcanin turn dampen international trade. Natural resourcepricesalso tend tobe volatile.Thishasan impactontradebyincreasingtheuncertaintyfacedbyimportersandexporters.

Part (c) takes up the issue that natural resources arepotentially exhaustible, which can act as a brake tofuture economic expansion. This also implies thatcomparative advantage conferred by nature can bedissipated. It also discusses the role of innovation inincreasing efficiency in the use of natural resources,discovering new supplies and developing alternatives.Thismeanshumaninnovationcanoffsetlimitednaturalresources.Part(d)examineshownaturalresourceusecanbesubjecttoenvironmentalpressuresandtheroleof public policy in that context. Part (e) presentspossible scenarios in the future evolution of naturalresourcesupplyandcostsandinternationaltrade.Part(f)offerssomeconcludingobservations.

(a) Unevengeographicaldistributionofnaturalresources

This section presents evidence of the unevengeographical distribution of natural resources in thecase of energy, water and land and discusses theimplications for the pattern of trade. It then describeshow concentration in resource endowments confersmarketpoweronsomesupplyingcountriesandhowthis

couldbeexploitedthroughtheuseofexportrestrictions.Finally,therepercussionsforgeopoliticsareconsideredasresource-abundantcountriesmakeuseofmonopolypower to pursue their international interests andresource-scarce countries prioritize the pursuit ofresourcesecurityintheirinternationalrelations.

(i) Resource abundance and trade patterns

Oil, coal and natural gas were the sources of almost90percentofworldwideenergyuse in2011,ascanbe seen in Figure C.36. Of these, oil is the mostimportant,accountingforathirdoftotalenergyusein2011. However, this is down from its peak of 48 percent in themid-1970s(aroundthetimeof thefirstoilcrisis). Coal had about as large a share as oil in themid-1960s but then underwent a long decline. Thiswas reversed at the turn of the millennium, with thesurgeincoalconsumptionbyChinaandIndia.Naturalgas has risen in importance, with its share climbingfrom16percentin1965to24percentin2011.Thisincrease is likely to continue because of newdiscoveriesandextractionmethods inNorthAmerica(seethediscussiononshaleoilbelow).

The contribution of nuclear energy, hydroelectricityandotherrenewablesourcesissmallbuttheshareofrenewableshaspickedupinthelastdecade,driveninpartbyhigherenergyprices(seethediscussionbelowonchangesinenergyprices).

The standard Heckscher-Ohlin theory predicts thatcountries which are relatively abundant in a factor ofproductionwillexport thecommoditywhichuses that

FigureC.36:World energy consumption by type, 1965-2011 (percentage)

0

1965 1970 1975 1980 1985 1990 1995 2000 2005 2010

5

10

15

20

25

30

35

40

45

50

Oil Coal Natural gas Hydroelectricity RenewablesNuclear energy

Source:BP,Statistical Review of World Energy,June2012.

Note:Oilconsumptionismeasuredinmilliontonnes;otherfuelsinmilliontonnesofoilequivalent.Renewablesarebasedongrossgenerationfromrenewablesourcesincludingwind,geothermal,solar,biomassandwaste.Renewablesconsumptionisconvertedintomilliontonnesofoilequivalentonthebasisofthermalequivalenceassuming38percentconversionefficiencyinamodernthermalpowerstation.


168

factorintensively.Acontemporaryvariantofthisstoryargues that a country will capture larger shares ofworldproductionand trade in commodities thatmoreintensively use its abundant factor (Romalis, 2004).The factor-proportion explanation has traditionallyassumed that factors of production are non-exhaustible (such as Ricardo’s “indestructible powersofthesoil”).KempandVanLong(1984)showthatthepredictionof theHeckscher-Ohlin theoryalsoappliestosituationswhenallof thefactorsofproductionareexhaustible as well as when exhaustible factors arecombinedwithnon-exhaustiblefactors.

The theory is about relative rather than absolute factorabundanceandlinksthattoexportsofproductsthatareintensive in those factors rather than to exports of theresource itself. Notwithstanding these caveats, TablesC.11 to C.13 corroborate the relationship betweencountries’ endowments of natural resources and theirexportperformance.ThecountrieslistedinTableC.11–themostprominentbeingtheKingdomofSaudiArabia,CanadaandIran–have95percentoftheworld’sprovedreservesofcrudeoiland86percentoftotaloilexportsin2010.ThosecountriesinTableC.12–withtheRussianFederation,IranandQatarhavingthelargestreserves–accountfor91percentofprovedreservesofnaturalgasand77percentofallnaturalgasexportsin2010.Finally,the countries shown in Table C.13 – with the UnitedStates,theRussianFederationandChinabeingthetopthree – have 96 per cent of total recoverable coal and93percentoftotalexportsofcoalin2010.

Water and land

The availability of land suitable for agriculturalproduction, especially arable land, determines thepatternsofagriculturalproductionandthedependencyof countries on imports of crops. Figure C.37 showscountrieslistedintermsoftheshareoflandareathatisarable.Mostoftheworld’sarablelandissituatedinSouthern and Eastern Asia, North America and Sub-Saharan Africa. The share of total land area that isarablevariesconsiderablybyregion,from23percentin Western and Central Europe to only 4 per cent inNorthAfrica.

There is a positive correlation between the size of acountry’s per capita arable land endowment and thevalue of its agricultural exports. This is depicted inFigureC.38,whichusesdatafor195countriesin2008.

Freshwaterisarenewablebutscarceresourcewithanuneven geographical distribution. Figure C.39 showsthis geographical pattern using renewable waterresourcespercapitaasameasureofavailability.Themostwater-abundant regionsareSub-SaharanAfricaand South America. The regions of North Africa,CentralAsiaandtheMiddleEastareattheotherendofthespectrum,withseverelylimitedwaterresources.

FigureC.40showshowwaterresourceavailabilityhaschanged over time. Reflecting their more rapidpopulation growth, water resources per capita in

FigureC.37:Arable land as a percentage of total land area, 2011

<2 2-5 5-10 10-15 15-20 20-30 30-40 40-50 >50

Source:FAO,Aquastat,http://www.fao.org/nr/water/aquastat/data/query/index.html?lang=en

Note:Arablelandisthelandundertemporarycrops,meadows,gardensandfallow.Itdoesnotincludeareasunderpermanentcropssuchascoffeeorcocoa.Surfacesinwhite:Dataunavailableatthetimeofwriting.ColoursandboundariesdonotimplyanyjudgementonthepartoftheWTOastothelegalstatusofanyfrontierorterritory.


169

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

TableC.11:Countries with the largest proved reserves of crude oil, 2008

Country

Proved reserves

(billions of barrels)

Share of world exports

of oil (2010)

SaudiArabia,Kingdomof 267 16.0%

Canada 178 3.4%

Iran 136 5.6%

Iraq 115 4.5%

Kuwait,theStateof 104 3.3%

Venezuela,BolivarianRepublicof

99 3.8%

UnitedArabEmirates 98 5.0%

RussianFederation 60 11.4%

Libya 44 3.2%

Nigeria 36 5.5%

Kazakhstan 30 3.3%

UnitedStates 21 0.1%

China 16 0.1%

Qatar 15 2.6%

Brazil 13 1.4%

Algeria 12 2.6%

Mexico 11 3.4%

Angola 9 4.5%

Azerbaijan 7 2.1%

Norway 7 3.7%

Shareofworldtotal 95.2% 85.6%

Source:USInformationEnergyAdministration.

Note:Amountofrecoverablecoalisbasedon2008data.

TableC.13:Countries with the largest total reserves of recoverable coal, 2008

CountryRecoverable coal (million short tons)

Share of world exports of coal (2010)

UnitedStates 260,551 6.9%

RussianFederation 173,074 10.1%

China 126,215 1.9%

Australia 84,217 27.1%

India 66,800 0.2%

Germany 44,863 0%

Ukraine 37,339 0.6%

Kazakhstan 37,038 3.0%

SouthAfrica 33,241 6.3%

Serbia 15,179 0%

Colombia 7,436 6.3%

Canada 7,255 3.0%

Poland 6,293 1.5%

Indonesia 6,095 26.1%



Note:Amountofrecoverablecoalisbasedon2008data.

TableC.12:Countries with the largest proved reserves of natural gas, 2009

Country

Proved reserves

(trillions of cubic feet)

Share of world exports of natural gas

(2010)

RussianFederation 1,680 22.3%

Iran 992 0.9%

Qatar 892 11.5%

UnitedStates 273 4.3%

SaudiArabia,Kingdomof 258 0%

UnitedArabEmirates 214 0.5%

Nigeria 184 2.6%

Venezuela,BolivarianRepublicof

171 0%

Algeria 159 5.3%

Iraq 112 0%

Indonesia 106 3.9%

Turkmenistan 94 0%

Kazakhstan 85 1.0%

Malaysia 83 3.3%

Norway 82 9.8%

China 80 0.3%

Uzbekistan 65 1.2%

Kuwait,theStateof 63 0%

Egypt 59 1.1%

Canada 58 8.9%



Note:Amountofprovedreservesbasedon2009data.

Sub-SaharanAfricahavedeclinedatthehighestrate,followedbySouthernandEasternAsia.

Agriculture accounts for 69 per cent of globalfreshwater withdrawals and 90 per cent of itsconsumptiveuse,i.e.waterlostduetoevaporationandtranspiration(FAO,2012).Thus,onemightreasonablyassume that the geographical distribution of waterobservedinthepreviousfigureswillbereflectedinthepatternofagriculturaltrade.

However, water endowments do not seem to have astrong influence on agricultural trade patterns.Hoekstra (2010) attributes this to heavy governmentinterventioninagriculturethrough,amongotherthings,subsidies, tariffs and sanitary and phytosanitarymeasuresaswellasindomesticwatermarketswheretheresource isseverelyunder-priced.All thesepolicydistortionsworktoblunttheeffectofendowmentsonagriculturaltrade.Hesuggeststhatwaterendowmentsaffect trade patterns only in cases where there isabsolute water shortage, which forces water-scarcecountries to import water-intensive products becausetheysimplycannotbeproduceddomestically.

However, recent work by Blackhurst et al. (2010) andDebaere(2012)suggestthatmanufacturingsurpassesagriculture in total water use if one accounts for thewater-intensivenessofthepowerusedinmanufacturing.

Debaere (2012) finds that countries that are relativelywater abundant tend to export more water-intensiveproducts.Hisresultssupportthehypothesisthatwaterisasourceofcomparativeadvantage.However,healsofinds that water contributes significantly less to the


170

pattern of exports than the traditional productionfactors,suchaslabourandphysicalcapital.

(ii) Market power and geopolitics

The uneven geographical distribution of naturalresources means that some resource-abundantcountries will have market power in trade. This maycreate a temptation to exploit that market powerthrough the use of export restrictions. By reducingsupplyofthenaturalresourceininternationalmarkets,theworldpriceoftheresourcerises,creatingaterms-of-tradegainfortheexportingcountryandaterms-of-tradelossfortheimportingcountries.

While the temptation to exploit market power couldapply to other sectors as well, there is evidence thatexporttaxesandotherrestrictionsaremorefrequentlyappliedtonaturalresourcesthantoothermerchandisegoods.While just5percentofworld trade iscoveredby export taxes, the share more than doubles to11percentfornaturalresourceproducts(WTO,2010).OfallexportrestrictionsnotifiedtotheWTO,morethana third–some2,577outofa totalof7,328notified–havebeenappliedtonaturalresourceproducts.

Theunevendistributionofnatural resourcescanalsohavegeopolitical impacts– i.e.,monopolypowerovernatural resource supplies in some countries and

FigureC.38:Agricultural exports and endowment of arable land per capita, 2008

FigureC.39:Renewable water resources per capita by region, 2011 (1,000m3/inhabitant/year)

0

2

-10 -8 -6 -4 -2 0 2

4

6

10

14

20

18

Log

of a

gric

ultu

ral e

xpor

ts

Log of arable land per capita

8

12

16

<400 400-1,000 1,000-3,000 3,000-6,000 6,000-12,000 12,000-30,00030,000-60,000 60,000-100,000 >100,000

Source:World Development IndicatorsandUNComtrade.


Note:Surfacesinwhite:dataunavailableatthetimeofwriting.ColoursandboundariesdonotimplyanyjudgementonthepartoftheWTOastothelegalstatusofanyfrontierorterritory.


171

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

scarcity of these resources in other countries canaffecttheirpolitical,military,anddiplomaticbehaviour.

Countrieswithabundantsuppliescanusecontroloverthese resources to support their international goalsand causes. In the 1973 Arab-Israeli war, MiddleEasternmembersoftheOrganizationofthePetroleumExportingCountries(OPEC) launchedanoilembargoagainst Western countries supporting Israel. Theenergyinfrastructureofmajorsupplyingcountriescanalso become so crucial to the global economy thatthey become targets. There have been persistentattacks on energy infrastructure by insurgent groupsinAlgeria,Colombia,theNigerDeltaandIraq(Lacherand Kumetat, 2011). In early 2013, a massive naturalgas facility in Algeria became the target of a violenttakeoverbyterroristforces.

Even if energy supply is not threatened, geopoliticaltensionsbetweencountriescanpromptsometoincuradditionalcosts.Forexample, inorder tonotbecomeoverlydependentonnaturalresourcetransitcountries,someofwhichwereoncepartoftheSovietUnion,theRussianFederationconstructednewoutletsfor itsoiltoEuropethroughtheBalticPipelineSystem(Laurila,2002). It is also started building a major new gaspipeline under the Black Sea to transport gas toSouthernEurope.123

Countries threatened by scarcity make securingaccess to natural resource supply a priority of theirinternational relations. China’s state oil companieshaveseveraloilsupplycontractswithforeignfirmsandcountries. The major Chinese oil companies haveacquired a variety of holdings in Angola, Azerbaijan,

Canada, Chad, Indonesia, Iraq, Iran, Kazakhstan,Myanmar (Burma), Nigeria, Peru, the RussianFederation, the Kingdom of Saudi Arabia, Sudan,Turkmenistan,UzbekistanandtheBolivarianRepublicofVenezuela(Hayward,2009;U.S.GAO,2013).

Foreign investment in farm land has increasedsignificantlyover thepast fewyears.Theseflowsareglobal in scope, involving 62 countries, where suchacquisitions have occurred, and 41 countries whoseenterpriseshavemadeforeignlandinvestments(Rullietal.,2013).Althoughexactfiguresarehardtoobtain,the latest estimates indicate that these farm dealsrange between 47 million (Rulli et al., 2013) and56millionhectares(Deiningeretal.,2011).

Table C.14 lists the top ten investors or acquirers offoreign farm land as well as the top ten destinationcountries for these investments. Although countrieswherearablelandandwaterareparticularlyscarce(e.g.countries in the Middle East and countries with agrowing demand for food, energy and raw materials,such as China and India) are active players, the topinvestorsarecompaniesfromtheUnitedKingdomandtheUnitedStates.ThedestinationsoftheseinvestmentsarecountriesinAfrica,South-EastAsia,SouthAmericaaswellas theRussianFederationandUkraine.Theseinvestments frequently take the form of long-termleases, outright purchases or contracts, with theacquiredlandbeingdevotedtoraisingcropsforfoodorbiofuel(vonBraunandMeinzen-Dick,2009).

Totheextentthatforeigninvestorsareabletoincreaseagricultural productivity in land and water-abundantcountries, there are economic benefits from such

FigureC.40:Renewable water resources per capita by region, 1967-2011 (1,000m3/inhabitant/year)

0

1,000

2,000

500

1,500

Ren

ewab

le w

ater

reso

urce

s by

regi

on

0

1,000

8,000

6,000

Tota

l ren

ewab

le w

ater

reso

urce

s

5,000

4,000

3,000

2,000

7,000

Northern Africa Sub-Saharan Africa Northern America Central America and the CaribbeanSouthern America Southern and Eastern Asia Central Asia Middle EastWestern and Central Europe Australia and New Zealand Total

1967 1972 1977 1982 1987 1992 1997 2002 2007 2011


Note:Renewablewaterresourcesincluderenewablesurfaceandgroundwater,internalandexternal.


172

TableC.14:Top ten destinations and countries of origin for foreign investment in land

Country of land acquisitionArea acquired

(millions of hectares)

Country acquiring foreign landSize of acquisition

(millions of hectares)

DemocraticRep.oftheCongo 8.1 UnitedKingdom 4.4

Indonesia 7.1 UnitedStates 3.7

Philippines 5.2 China 3.4

Sudan 4.7 UnitedArabEmirates 2.7

Australia 4.6 Israel 2.0

RussianFederation 2.8 Egypt 1.4

Brazil 2.3 Korea,Rep.of 1.3

Tanzania 2.0 India 1.2

Mozambique 1.5 SouthAfrica 1.1

Ukraine 1.2 Malaysia 1.0

Source:Rullietal.,2013.

investments.Amajorconcern,however,isthatpropertyrights are often weakly enforced in the countrieswhere such acquisitions have been made, raising thepossibilitythatthelocalownersmayhavebeenunfairlyor illegally displaced (Deininger et al., 2011). Clearly,the often negative attention these activities haveattracted underscore how increased competition fornatural resource supplies can raise internationaltensions,especially if thenatural resource isseenasvitalforfoodornationalsecuritybyotherstates.

In conclusion, differences in natural resourceendowmentsappeartoexplaintradepatternsrelativelywell. In addition, more concentrated control overnaturalresourcesconfersmarketpower,whichcanbeenhancedthroughtheuseofrestrictivetradepolicies.Concentration may also enable resource-abundantcountriestouseittopursuenon-economicobjectives.Countries faced with acute resource scarcity may inturn pursue natural resource security at the expenseof international relations. To the extent that thesegeopolitical factorscreateorexacerbate internationaltension, they can increase the price of naturalresources beyond what would have been created bymonopolypowerandalsoincreasepricevolatility.Bothof these can have harmful effects on the globaleconomyandtrade(seethediscussionbelow).

(b) Changesinnaturalresourcepricesandvolatility

Aswasnotedearlier,naturalresourcepricestendtobevolatile. The following section focuses on energy –rather than landorwater–pricechangesandvolatilitybecause of data availability. The world economy ischaracterized by a group of net oil-importing industrialeconomieswhichabsorbalargeshareofglobaloutput,ononeside,andagroupofnetoil-exportingcountries,ontheother.Changesinthepriceofoilrepresentlargeterms-of-trade shocks that adversely affect industrialeconomies while benefiting oil exporters (Backus andCrucini, 2000). Because of the size of industrialcountriesandtheimportanceofoil,theseterms-of-tradechangesreverberatethroughouttheglobaleconomy.

Energyisamajorfactorofproductionanditisdifficulttosubstitutecapitaloranotherfactorofproductionforoilintheshortrun,soanincreaseintheoilpricewillreduceproduction of net energy-importing countries and slowtheirrateofeconomicgrowth(Hamilton,2009).This,inturn,leadstoalowerrateofimportexpansion.Ofcourse,higher oil prices should expand output and increaseGDP growth of net energy exporters (Korhonen andLedyaeva,2010).However,fortheglobaleconomyasawhole,theevidencesuggeststhatthenegativeeffectonoutputandtradetendstodominate.

Secondly,changesinthecostofenergycanalterthecommodity composition of a country’s export andimportsdependingontheirenergyintensity(Sadorsky,2012). Although there is no standard list of energy-intensive products or industries, information from theenergybalancesheetoftheEuropeanUnion(Eurostat,2011)pointstothesebeingaluminium, ironandsteel,chemicals, glass, pottery and building material (e.g.cement),andpulpandpaper.Allthingsbeingequal,anincreaseinenergypriceswillraisethepricesoftheseenergy-intensiveproducts. Itwillconsequentlyreducedemand for themanddecrease theirshare (in realorvolumeterms)ininternationaltrade.Theextentofthiseffectwilldependon,amongotherfactors,theabilityofproducers tosubstituteother factorsofproductionfor energy and the elasticity of demand for theseproducts – the responsiveness of buyers to higherprices. The more inelastic the demand, the less theimpactofhigherenergyprices.

Another salient feature of global energy markets ispricevolatility.FigureC.41shows theweeklynominalcrudeoilspotprice(i.e.thepriceofcrudeoiltradedona “spot” market and available for almost immediatedelivery) between 1987 and 2012 and the square oftheweeklyreturnofoilpricesthatisusedasaroughmeasure of volatility.124 Based on this, price volatilitytendstoclusteratspecificpointsintime.Someofthelargerisesorfallsinthespotpriceofcrudeoilcanbelinked to specific instancesof economic andpoliticalcrises, which have the potential to disrupt globalenergysupplyordemandsignificantly.


173

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

The highest peak in short-term volatility occurred in1990-91and is linked to Iraq’s invasionofKuwaitandthefirstGulfWar.OtherperiodswithlargefluctuationsinoilpricesweretheterroristattackontheWorldTradeCentre in the United States in 2001 and the Iraq warthat began in March 2003. Both the commodity pricespike of 2007-08 and the financial crisis in 2008-09,whichledtothebiggestdropinoilpricesthatthespotmarkethaseverexperienced,arealsoevident.

The underlying reasons for volatility are complex,involving demand and supply factors and shocks toboth. Evidence from Hamilton (2009) and Smith(2009) point to the role of low demand and supplyelasticities,particularlyintheshortrun.Alonger-termexplanation has been provided by Dvir and Rogoff(2009), who contend that volatility spikes wheneverperiods of rapid industrialization have coincided withuncertainty regarding access to energy supply. Theypoint to 1861-78 and 1973-2009 to support theirargument.Thesewereperiodsofrapidindustrialization–bytheUnitedStatesinthefirstperiodandEastAsiain the latter period – as well as periods of supplyuncertainty due to the monopoly of railroads ontransportation in the United States and to OPEC’sabilitytorestrictaccesstosupplyinthelatterperiod.

Manypopularaccountsoftheincreaseinoilpricesinthe lastdecadeattribute it to thegrowingappetiteofemerging China and India for energy to power theirrapid development. Beyond these explanations, anumber of authors have argued that speculation hasplayedaroleintherecentincreasesincommodityandnatural resource prices (Masters, 2008; Caballero etal., 2008; Robles et al., 2009). This is discussed insomedetailinthe2010World Trade Reportonnaturalresources(WTO,2010).

Volatilityofoilpricescanreducetradeflowsbecause itincreases the risks faced by importers (Chen and Hsu,2012). In oil-importing countries, fluctuations in currentprices create uncertainty about the future trend for oilprices, leading households to postpone purchases ofconsumer durables and firms to postpone investmentdecisions (Elder and Serletis, 2010; Henriques andSadorsky, 2011). This reduction in spending by bothhouseholds and firms reduces aggregate demand andhencetotalimportsaswell.TheempiricalstudybyChenand Hsu indicates that total exports by oil-importingcountriesalsofallasaresultofoilpricevolatility.

(c) Exhaustibilityandtheroleofinnovation

FollowingSweeney(1993),exhaustiblenaturalresourcescanbedefinedas thosewhoseadjustmentspeed–orrenewability– issoslow that theycanmeaningfullybeconceived of as being made available once and onlyoncebynature.125Oilornaturalgasdepositsaretypicalexamplesofexhaustiblenaturalresources.

The exhaustibility of some natural resources hasfrequentlycausedalarm.In1972,theClubofRome–aglobalthinktank–famouslyclaimedthatpressuresfromeconomicactivitiesandpopulationgrowthwouldleadtothecollapseoftheeconomyandtheenvironmentgiventhefinitesuppliesofnatural resources(Meadowsetal.,1972).Othershaveproposed“peak”theories,wheretheextractionofexhaustibleresourcesispredictedtofollowa bell-shaped curve, initially increasing exponentially,reaching a peak, and then declining exponentially untiltheresourcesaretotallyexhausted(Hubbert,1956).

Itcanbearguedthattheideathattherateofextractionof an exhaustible resource eventually reaches a

FigureC.41:Weekly nominal oil spot prices and squared returns (Brent, Europe), 1987-2012 (US$/barrelandweeklyreturn2)

1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0

20

40

60

80

Nom

. wee

kly

Bre

nt

Wee

kly

retu

rn2100

120

140

160

Return2 Nom. weekly Brent

Source:USEnergyInformationAdministration(EIA),owncalculations.

Note:Brentcrudeoil(inUS$/barrel)istradedattheIntercontinentalExchange(ICE)FuturesinLondon.WeeklypricesasofclosingtimeonMondays.


174

maximum, after which it declines is basicallytautological (Hamilton, 2012). What makes the peaktheorysoundapocalypticistheimpliedpredictionthatthe peak is either behind us or about to come. Bycomparison, economists tend to be more sanguineabout the exhaustibility of natural resources andconcerned with other questions. How will markets,whether competitive or not, determine the rate ofextraction of an exhaustible resource (Hotelling,1931)? What is the optimal way of taking inter-generationalequityintoaccount,i.e.howmuchshouldcurrentgenerationsconsumeandhowmuchshouldbeleft behind to future generations (Solow, 1974;Hartwick,1977;Chichilnisky,1996)?Thereareseveralreasonsforthis.

First, the total amount of the exhaustible resource isnot known for certain, so given sufficient economicincentives, reserves can be maintained or increasedthroughtheexploitationofdepositsinitiallyconsideredas not economically accessible (Pindyck, 1978).Secondly, history has shown that technologicalinnovationoffersapotentresponsetotheproblemofexhaustibility. Innovations can increase efficiency intheuseofanexhaustibleresourcesothattheamountrequired to produce a unit of output is reduced overtime. New methods of exploration can increase thelikelihoodofmakingnewgeologicaldiscoveries(ArrowandChang,1982).Innovationcanlowertheextractioncosts of the resource (Hamilton, 2012). Finally,technology may advance enough so that it becomespossible for reproducible or renewable resources totake theplaceof theexhaustible resource (DasguptaandHeal,1974).Ultimately,itisanopenquestionastohow long innovation will allow us to keep one stepaheadofnaturalresourceexhaustion.

Usingenergyasanexample,FigureC.42showslong-run trends in energy intensity, which measures howmanyunitsofenergyareneededtoproduceaunitofGDP.Thelowertheindicator,themoreenergyefficientan economy is. It can be seen that global energyintensityhasbeendecreasingeveryyearsince1970.Thisistrueforlargeadvancedeconomies,suchastheUnited States, but interestingly, even emergingcountries, such as China and India, have exhibitedfalling energy intensities in the last ten to 20 years,which were assumed to be periods of extensiveenergyuse.

Geller et al. (2006) estimate that without energyefficiency improvements, the OECD countries wouldhaveusedapproximately49percentmoreenergythanwasactuallyconsumedasof1998.Theyattributethisincreased efficiency to, among other measures, thedevelopmentandcommercializationofanumberofnewenergy efficiency technologies (e.g. energy-efficientbuilding technologies, appliances, electronic lightingballasts,etc.).Technological improvementsalsoplayedan important role in reducing China’s energy intensity(He and Zhang, 2006). Kiang et al. (2011) estimates

that technology improvements accounted for 40 to60percentofChina’senergysavings.

TheriseofshalegasintheUnitedStatesisagoodcasestudy of how technological change can dramaticallyaugment thesupplyofanexhaustiblenatural resource.Shale gas refers to natural gas that is trapped withinfine-grainedsedimentaryrocks.Acombinationofotherinnovationswasneededbeforethesedepositscouldbecommercially exploited. The extraction technology –hydraulic fracturing (“fracking”) – requires pumpingwater,chemicalsandsandundergroundtoopencracksintherockandallownaturalgastobereleasedfromtheshale.126However,itcouldonlybeusedproductivelyandpredictablyonceknow-howtomapshaleexpansesandto drill horizontally in rock formations was developed(Trembathet al., 2012).Asa result of theseadvances,shale gas production in the United States has grownnearlytwenty-twofoldsincethe1990s.

A more dramatic illustration of how technologicalchangecoulddelayoroffsetexhaustibility isshown inFiguresC.43andC.44.FigureC.43showsthestockofproven oil reserves and the ratio of these reserves toworld oil consumption over the last three decades.Duringthatperiod,thestockofprovenreservesrosebymore than 140 per cent while the ratio of reserves toglobal consumption actually rose from 11 to 19.A similar picture canbedrawn for the caseof naturalgas, which is shown in Figure C.44. Proven reservesrose by about 160 per cent in the last three decades

FigureC.42:Energy intensity, 1970-2011

19

70

19

75

19

80

19

85

19

90

19

95

20

00

20

05

20

10

0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

US China India World

Source:BP,Statistical Review of World Energy,June2012;WorldBank’sWorld Development Indicators(WDI);owncalculations.

Note:Theunitofenergyintensityisdefinedaskilogrammesofoilequivalentperconstant2011US$.CurrentGDPinUS$hasbeendeflatedto2011US$levelsbyusingtheworldGDPdeflatorprovidedbytheWDI.


175

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

whilethereservetoconsumptionratiocontinuedtorise.Thepatternof risingglobal reservescanbeshown tohold for a wider range of exhaustible resources –bauxite,copper,ironandzinc(Lomborg,2012).

Risingenergyandnaturalresourcespricesasaresultof scarcitywill create incentives for firms to invest ininnovation.However,thelevelofR&Dinvestmentmaystill be lower than the social optimum because thepotential payoffs will not be realized for decades,which is beyond the planning horizons of most firms(Sathaye et al., 2007). This market failure – thedivergencebetweenprivateandsocialbenefits–canwarranttheuseofR&Dsubsidiesintheenergysectortoincreaseinnovationandfindtechnologicalsolutionstotheexhaustibilityofnaturalresources.

Geller et al. (2006) attribute a prominent role togovernment-fundedR&Dinthelong-termimprovementin energy efficiency in OECD countries. Similarly,Trembathetal.(2012)havepointedtothecrucialroleplayed by the US government in the successfuldevelopmentofshalegas. Innovationandprogress inthedevelopmentofhydraulicfracturingandotherkeygas recovery technologies came about from public-privateresearchandcommercializationefforts.Atthesametime,subsidiesgivegovernmentsgreaterleewayto pursue a kind of industrial policy, where the newobjects of largesse are future “winners”, such asbiofuels,solarandwind.127Thisgivesrisetotheriskofindustriesbeingpromotednotforpublicpolicyreasonsbutinordertobenefitdomesticproducergroups.

Whatarethetradeimplicationsoftheexhaustibilityofnatural resources? First, it means that a country

favourablyendowedwitha largestockofexhaustiblenaturalresourcescouldseeitscomparativeadvantageerode over time. While empirical analysis of thedynamicsoftradespecializationwithrespecttonaturalresources has received little attention,128 there areseveral studies that appear to demonstrate shifts incomparative advantage in natural resource-exportingcountries.

Leamer (1984) shows that between 1958 and 1975,countries such as Australia, the Dominican Republic,Honduras, Paraguay and the Philippines experienceda significant increase in mineral extraction while thelevel of mineral extraction of Cyprus, Ghana andYugoslaviadecreasedsignificantly.Davis (1995)findsthat between 1973 and 1991, Botswana, theDemocratic Republic of the Congo, Angola, Guinea,Niger, Papua New Guinea, the Syrian Arab Republic,South Africa, Cameroon, Togo and Ecuador gainedcomparative advantage in mineral extraction whileTunisiadiversifiedawayfrommineralactivities.

ArecentpaperbyAlvarezandFuentes(2012),usingalarge sample of countries between 1962 and 2000,finds that comparative advantage in raw materialstends to be less persistent than in manufacturedgoods. It is, however, unclear to what extent thesechanges are the result of the exhaustion in naturalresource endowments or the result of other factors,includingchangesinpolicy(Davis,2010).Furthermore,as this discussion of exhaustibility has highlighted,technological change is a potent force that canbeharnessedbynatural resource-abundantcountriesthat wish to maintain comparative advantage inthatsector.

FigureC.43:Proven oil reserves and reserves-to-consumption ratio, 1980-2011

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010

0

200

400

600

800

1,000

1,200

1,400

1,600

1,800

0

2

4

6

8

10

12

14

16

18

20

Proven oil reserves Oil reserves-to-consumption ratio

Oil

rese

rves

-to-

cons

umpt

ion

Bill

ions

of b

arre

ls

Source:BP(2012).


176

FigureC.44:Proven natural gas reserves and reserves-to-consumption ratio, 1980-2011

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010

0

50

100

150

200

250

0

10

20

30

40

50

60

70

80

Proven natural gas reserves Natural gas reserves-to-consumption ratio

Trill

ions

of c

ubic

met

res

Res

erve

s-to

-con

sum

ptio

n ra

tio

Source:BP(2012).

(d) Environmentalcosts

Sometimes, the process of extracting naturalresourcesortheirconsumptioncanhaveenvironmentalcosts. For instance, the current technology forextractingshalegas–hydraulicfracturing(fracking)–creates a number of environmental risks. Hydraulicfracturing fluid could leak into and contaminategroundwater.Methanecouldbeaccidentally releasedwhile exploiting shale gas reserves. Fracking itselfcouldcausesmallearthquakes.

At present, the most serious example of negativeenvironmental impacts associated with naturalresource use is the burning of fossil fuels. Thisproduces carbon emissions which accumulate in theatmosphereandcanremainthereforcenturies.Thesecarbonemissionsarethemainreasonfortheobserved(and projected) increase in average globaltemperatures (Intergovernmental Panel on ClimateChange, 2007). Climate change has adverseconsequences at a physical level (rising sea levels,changes in ice cover and frequency of extremeweather events) and biological level (agriculture,forestry and human health). It is believed that thedoubling of greenhouse gas emissions in theatmosphere (to450partspermillion) relative topre-industrialtimeswillincreasetheserisksdramatically.

As a result, many countries have taken steps,sometimes unilaterally and sometimes with others, tomitigate the adverse consequences of using fossilfuels. They include taxes on fuels, emission tradingschemes covering sectors that are consideredemission-intensive, increasing energy efficiency,spurringeffortstofindalternativestofossilfuels,etc.

The joint report of the WTO and the United NationsEnvironmentProgramme(UNEP)ontradeandclimatechange (WTO and UNEP, 2009) contains acomprehensive account of various national andinternationalinitiatives.

SectionD.2ofthisreportwilldiscussthetradeeffectsof environmental policies, including those arising fromclimate change mitigation efforts. However, it isimportant tomention twopoints.First, climatechangepolicy will be crucial to the evolution of energy pricesandtothefuturemixofenergysources.Secondly,theremay be continued differences in the stringency ofclimatechangepoliciesadoptedbygovernments, thuspossibly creating cost differences between countriesespeciallyinenergy-intensivesectors.Thesepointsaretakenupbelowinthediscussionoffuturescenarios.

(e) Futureofnaturalresourcesandtrade

The focus of the following section is on water andenergysincemuchmoreworkhasbeendoneonthesenatural resources thanon land.Basedonprojectionsmade by the OECD (2012c), the International EnergyAgency (2012), the US Energy InformationAdministration (2012)and leadingenergycompanies,such as BP (2012b), the potential implications offuture supply and demand developments of naturalresourcesforinternationaltradeaswellastradepolicyareconsidered.

Water and international trade

The OECD (2012c) projects that global demand forwater will rise by 55 per cent between 2000 and2050. This growth in demand will come mainly from


177

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

manufacturing, electricity and domestic use.Increasingly, the future will see agriculture and theenvironment competing for water with cities, energysuppliersandindustries(seeFigureC.45).Inthefaceofthesecompetingdemands,therewillbelittlescopefor increasing water for irrigation, and thus, also forfoodandagriculture.

This pressure on water resources has two possibleimplicationsforinternationaltrade.Thefirstimplicationis with respect to the pattern of agricultural tradeamong countries. The OECD’s environmental outlookprojects that,by the turnof thiscentury, therewillbesevere water shortages for the entire populations ofSouth Asia and the Middle East and large shares ofChina’sandNorthAfrica’spopulation.

As noted before in the discussion on the unevengeographical distribution of natural resources, underconditions of severe water shortage, water-scarcecountries will be forced to import water-intensiveproducts. This suggests that food and agriculturalproducts will become a larger share of the futureimportsof thecountries inwater-scarce regions.Thesecondimplicationpertainstotheproductcompositionof international trade, and in particular, to thepossibility that the long-term decline in the share offood and agricultural products in international trade,whichwasdiscussed inSectionB,mightbearrestedorevenreversed.

Energy and international trade

Given the rising trajectory of world GDP andpopulation, energy needs are projected to rise bynearly a third by the year 2035 (International EnergyAgency, 2012b). Most of this growth will come fromincreased energy demand by emerging economies,whereastherewillbenosignificantchangesinenergyconsumption by developed nations (BP, 2012b;InternationalEnergyAgency,2012).

Theenergymix isexpectedtochange,withthesharesofcoalandoildecliningwhilethesharesofnaturalgasandrenewablesourcesareexpectedtorise.Inparticular,USnaturalgasproductionisexpectedtoincreasefrom21.6trillioncubicfeetin2010to27.9trillioncubicfeetin 2035 (US Energy Information Administration, 2012).Almost all of this increase will be due to shale gasproduction,whichwill grow from5 trillion cubic feet in2010to13.6trillioncubicfeetin2035.Fossilfuelswillcontinuetomeetthebulkoftheworld’senergyneedsinthefuture,makingup75percentoftheworld’ssourceofenergy(seeFigureC.46).Naturalgaswillcontributethe most to the estimated growth in energy demand.While the share of renewables in total energyconsumptionwillriseto15percentby2035,itwillnotbeabletosatisfygrowingenergydemandonitsown.

An important concern for international trade is thefuture evolution of energy prices. The International

FigureC.45:Global water demand: baseline scenario, 2000 and 2050 (cubickilometre)

0

1,000

2,000

3,000

4,000

5,000

6,000

km3

irrigation domestic livestock manufacturing electricity

2000 2050

OECD

2000 2050

BRICS

2000 2050

Rest of World

2000 2050

World

Source:OECD(2012b).

Note:Thisgraphonlymeasuresbluewaterdemandanddoesnotincluderain-fedagriculture.(BRICS:Brazil,theRussianFederation,India,China,SouthAfrica.)


178

Energy Agency’s latest outlook considers threescenarios in detail. These are differentiated by thekinds of policies it is assumed countries will adopt,either unilaterally or as part of internationalagreements.Thepoliciesofconcernarethoserelatedto renewable energy, energy efficiency, fossil fuelsubsidiesandmitigationofclimatechange.

The “New Policies” scenario, which is the baselineprojection in the International Energy Agency (IEA)report,assumesthatpoliciesthatareinplacenowwillcontinue in the future and, more importantly, thatannounced (but not yet implemented) governmentalpolicy actions will be realized in the near future. The“CurrentPolicies” scenarioassumes thatonly currentpolicies, and not announced policies, will be in force(“business as usual”). The third scenario is the so-called “450 Scenario”,129 in which it is assumed thatnew national and supranational policy actions will beadopted to limit the global average temperatureincrease to 2°C. Here, it is assumed that all OECDcountries will eventually apply carbon taxes onCO2emissions.

FigureC.47showstheresultingprojectionsoftherealpriceofcrudeoil imports,whichservesasaproxy forinternational prices of petroleum. These have beennormalized to 100 in the base year 2011. The futuretrendforenergypricesishighest(risingby35percent)under the business as usual assumption. The “NewPolicies”scenariowouldseeamoremodestincreaseinenergyprices(theywouldriseby16percent)by2035.This is because implementation of climate-friendlypolicies in the New Policies scenario means energydemand is lower than in theCurrentPoliciesscenario.Therefore, there is less need to exploit very costlyreservesandthuspricesarelower.Finally,theforecastchange in energy prices for the 450 Scenario isnegative. The assumption is that stronger abatement

policies (relative to thefirst twoscenarios) succeed insubstantially limiting energy demand so that energyprices actually fall (by 8 per cent) below their levelin2011.

Intermsofthelikelyeffectoninternationaltrade,therapid development of shale gas in the United Stateswill createa “sea-change” inglobal energyflowsandthe pattern of international trade in oil (InternationalEnergy Agency, 2012). The United States will re-emerge as a major producer and exporter of energyrather than just consumer and importer of energy.Itwillbecomeanetexporterofnaturalgasby2020.Asaresult,NorthAmericawillbecomeself-sufficientinenergyandanetoilexporterby2035.

Anothercountrythatwillhavealargeimpactonenergymarkets is Iraq,with the IEAprojecting that itwill bethe largest source of global oil export growth up to2035. This will represent a dramatically successfulrehabilitation of its energy sector driven by thecountry’s ample reserves, low extraction costs andinvestor-friendlypolicies.Boththesechangeswillthusrequire Middle East oil to find an alternative to theNorth America market, with the most likely scenariobeing that itwillbe redirected toconsumers inAsianmarkets.

The higher energy prices predicted in the future maylead to shifts in the composition of trade as well.Assumingthere isonly limitedscopeforsubstitution inproduction towards other factors, such as capital or

FigureC.46:Energy mix, 2010 and 2035

FigureC.47:Projections for energy prices up to 2035 (percentage)

0%

35%

30%

25%

20%

15%

10%

5%

2010 2035

Oil

Coa

l

Nat

ural

gas

Nuc

lear

Hyd

ro

Ren

ewab

les

80

90

100

110

120

130

140

150

20

11

20

14

20

17

20

20

20

23

20

26

20

29

20

32

20

35

New policies Current policies 450 Scenario

Source:InternationalEnergyAgency(2012).

Note:Theenergymixprojectedfor2035correspondstothe“NewPolicies”scenariointheIEAReport.

Source:InternationalEnergyAgency(2012).

Note:Realimportpricesforcrudeoilareexpressedin2011US$andhavebeennormalizedto100fortheyear2011.“NewPolicies”scenarioassumesthatpoliciesthatareinplacenowwillcontinueinthefutureandthatannounced(butnotyetimplemented)governmentalpoliciesactionswillberealizedinthenearfuture.The“CurrentPolicies”scenarioassumesthatonlycurrentpolicieswillbeinforce.The“450Scenario”assumesthatnewnationalandsupranationalpolicyactionswillbeadoptedtolimittheglobalaveragetemperatureincreaseto2°C.


179

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

labour, energy-intensive industries will be penalizedmore than other sectors by rising energy costs.Furthermore, it is likely that countries will differ in thestringency of their climate change mitigation policies.This means that countries with tougher environmentalpoliciesmightseeadeteriorationintheircompetitivenessin energy-intensive sectors relative to countries withmuch weaker regulations. Finally, the projected rise inenergy prices also has an important bearing on fuelcosts and therefore on transportation costs. Theseimpactsandwhat they imply for international tradeareconsideredmorefullyinSectionC.5.

Beyond these impacts, a number of policy issuesdiscussedabovearelikelytocontinuetoberelevantoreven grow in importance in the future. They includethe use of export restrictions by resource-abundantcountries to enhance their market power ininternational trade, the use of subsidies to provideincentives inthesearchforalternativestofossil fuelsand their possible misuse for industrial policy,agricultural protection and the pricing of naturalresources such as water, and varying adoption ofclimatechangemitigationmeasures.

(f) Conclusions

One of the patterns identified in Section B.2(c) wasthe highly concentrated exports of natural resource-abundant developing countries. An important lessonthat these countries can draw is that comparativeadvantage built on exhaustible resources can befragile.For thosecountriesandtheworldasawhole,investmentsinR&Darecrucialiftheseadvantagesaretobemaintainedovertime.

In the case of energy trade, a major shake-up in thenext two decades is likely, with the re-emergence oftheUnitedStatesand,toalesserextent,Iraqinglobalenergy production and trade. Middle East oil exportswill shift decisively toAsia.Higher populationgrowthand a much larger global economy will push energypricesupasdemandincreases,possiblyreducingtheshare of energy-intensive products in world trade.Severe water shortages in South Asia, the MiddleEast,NorthAfricaandChinaarelikelytoleadtorisingfood and agricultural imports in those water-scarceregions.Thiswillprobably result inacontinued focuson a number of trade policy issues in the naturalresources sector, with the most prominent beingexport restrictions and subsidies. Policies in otherareas,suchastrade-distortingmeasuresinagricultureand varying application of climate change mitigationmeasures,arelikelytoplayimportantrolesaswell.

5. Transportationcosts

The cost of transporting goods from producers tousers affects the volume, direction and pattern oftrade. It determines where the line between tradableand non-tradable goods is drawn and shapes which

firms are able to participate in trade and how theyorganize their production internationally. The cost oftransportationisinturninfluencedbyawiderangeoffundamental determinants. These include thegeographical features of countries, the quantity andquality of the physical infrastructure that supporttransportationservices,theproceduresandformalitiesused to control the movement of goods from onecountry to another, the extent of competition in thetransportation sector, the pace of technologicalinnovation in the sector and the cost of fuel (Beharand Venables, 2010). The characteristics of theproducts being shipped also affect transportationcosts.

Part (a) of this section discusses how transportationcostscanaffectinternationaltrade.Part(b)examinesin detail each of the major determinants oftransportation cost and their importance. Part (c)concludes by presenting possible scenarios in theevolutionoftransportationcosts.

(a) Howtransportationcostsaffecttrade

After decades of significant tariff cuts around theglobe, which can partly be attributed to successfulnegotiationswithin theGeneralAgreementonTariffsandTrade(GATT)andtheWTO,averagetariffbarriersarenow lower thanmanycomponentsof tradecosts,includingtransportationcosts.ThisisdocumentedinacomprehensivesurveyoftradecostsbyAndersonandvanWincoop(2004)(seeSectionB).

Taking the United States as an example, Table C.15compares its ad valorem transportation costs with itsaverage ad valorem tariff rates, weighted by importvalues.ThefiguresfortheUnitedStatesareinlinewiththeconclusionsdrawnbyAndersonand vanWincoop;USad valoremtariffratesinmostcasesarelowerthanad valorem transport costs. The measure of transportcosts used in these calculations only includes theinternational part of transportation. If inlandtransportation isalso included, the totalcosts involvedwill be even higher. The reversal in importance oftransportationcostsandtariff rateshighlights thewayinwhich transportationcosts is similar toprotectionistpolicymeasuresinthattheyleadtoan“anti-tradebias”– a greater incentive to produce for, and rely on, thedomesticratherthantheworldmarket.130

Following Samuelson (1954), most trade models thatinclude transportation costs assume they areproportional to the price of the traded good(transportation costs are the “iceberg costs”). As aresult, transportation costs drive a wedge betweenoriginanddestinationpricesbut theydonotproducechangesintherelativepricesofgoods.Consequently,highertransportationcostsreducethevolumeoftradebut do not necessarily change the composition oftrade. However, if all or a significant part oftransportationcostsisadditive–i.e.chargedonaper


180

unitbasis rather thanpurelyproportional to thepriceofthetradedgood–thentheconclusionthatrelativepricesareleftunchangednolongerholds.Inparticular,transportation costs can be expected to havepronouncedeffectsontherelativepricesofbothhigh-qualityandlow-qualityproductsaswellasgoodswithdifferentweight-to-valueratios.

Since a higher-quality good will typically sell for ahigher price than the low-quality good, fixedtransportation costs per shipment will make up abigger share of the price of the low-quality good. Anincreaseintransportationcostswillconsequentlyraisethepriceofthelow-qualitygoodproportionatelymorethanthatofthehigh-qualitygood.Thiswillencourageconsumers in export markets to switch towards thehigh-quality good, thereby increasing its share ininternational trade.131 A greater share of the low-quality good will be left in the home market (seeBox C.9). Conversely, a reduction in transportationcosts will lead to an increase in the share of low-qualityproductsininternationaltrade.Thegreaterthedisparity in prices between high-quality and low-quality goods, the bigger will be the impact oftransportationcostsonthepatternoftrade.

Hummels and Skiba (2004) test whether data ontransportation costs are more consistent with theadditive rather than the iceberg story and whethertransportation costs alter relative prices of high andlow-qualityproducts.Theirstudyisbasedonimports,atthe six-digit level of the Harmonized System, of anumberofLatinAmericancountries–Argentina,Brazil,Chile,ParaguayandUruguay–and theUnitedStatesfrom all countries. The study finds that transportationcosts are not proportional to price (not of the icebergform)andclosertobeingadditive132andthattheshareof high-quality goods relative to low-quality goodsincreasedwhenperunitfreightratesrose.

Beyond quality differences, another characteristic oftradedproductswhichturnsouttobeimportantisthe

value-to-weight ratio (Hummels, 2007). Box C.10discusses some estimates of the value-to-weightratiosofEUandUSimportsdisaggregatedbymodeoftransportation. All things being equal, transportationcostswillhaveasmallerimpactonthelandedpriceofthegoodwithahighvalue-to-weightratio.Toseethis,comparetheeffectofshippingametrictonofironoreworth US$ 120 to a metric ton of gold bullion worthUS$54.7million.133Sincetheyhavethesameweight,shipping costs will be very similar; only the higherinsurance costs will probably be different for theseshipments.However,giventhevastlydifferentvalueofa metric ton of these products in ad valorem terms,transportation costs will have a bigger impact onthe delivered price of iron ore compared with thedelivered price of gold. Given these relative priceeffects, higher transportation costs will tend toincrease the share of goods with higher value-to-weightratioininternationaltrade.

A recent paper by McGowan and Milner (2011)provides some corroborating empirical evidence ofhowthecompositionoftradeisaffectedbyincreasedtrade costs (of which transport cost is an importantelement). They focus on “trade cost intensiveindustries” which produce goods that have a largeshare of imported intermediates. These industriesinclude:coke,petrolandnuclearfuel;pulp,paperandpaper products; and electrical machinery. Using asample of 37 industrialized and transition countries,theyfind that industries located incountrieswith lowtradecostscapturesignificantlyhighersharesofworldexports,with thiseffectbeingstronger in tradecost-intensiveindustries.

Another channel through which changes intransportationcostscanaffectthepatternoftradeisthroughitsimpactonthe“extensivemarginoftrade”–the increase in the number of products a countrytrades. Not all products that a country produces areexported. However, by reducing the wedge betweenprices at the origin and destination, declining

TableC.15:Ad valorem transport costs for US imports, 1996 and 2011All modes of

transportSeaborne Airborne Other modes Tariffs

1996

Totalmerchandise 3.35 4.55 2.90 1.84 2.49

Agriculturalproducts 6.93 8.32 20.92 3.87 2.94

Fuelsandminingproducts 5.40 6.51 0.94 3.27 0.47

Manufacturingproducts 2.84 3.73 2.82 1.45 2.76

2011

Totalmerchandise 2.63 3.48 2.34 1.11 1.38



Manufacturingproducts 2.75 3.96 2.39 0.96 1.59

Source:USCensusBureau’sUSImports of Merchandise,owncalculations.

Note:Average forallmodesandeveryotheraggregation isweightedby imports (dataoriginally inHS10-digitdisaggregation).Theaveragetariffrateisconstructedbyweightingindividualtarifflines(aggregatedbyTRAINSattheHS6-digitlevel)withrespectiveimportvalues.


181

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

detransportationcostscan increase the rangeofgoods

available for international commerce, making goodsthatarecurrentlynon-tradedtradable.

Moreiraetal. (2008)provideestimatesof thepotentialdiversification arising from reductions in transportationcostsfornineLatinAmericancountries–Argentina,thePlurinational State of Bolivia, Brazil, Chile, Colombia,Ecuador, Paraguay, Peru and Uruguay. They note thatthe degree of diversification of these countries’ exportbundles is smaller than would be predicted from theirsize and below or just about average in terms of theirlevelsofdevelopment.Theauthorsmeasurethedegreeofproductdiversificationbythenumberoftarifflinesatthesix-digitHS level thatshowpositive tradeflows foreachpairofcountries.Theyestimatethata10percentdecline inaveragetransportcostswouldbeassociatedwith an expansion of more than 10 per cent in thenumber of products exported and with a 9 per centincreaseinthenumberofproductsimported.Obviously,therearegoingtobedifferencesacrossthesecountriesin the extent to which falling transportation costsgeneratechangesintheextensivemarginoftrade.Theyestimate that larger economies such as Argentina andBrazilwould increasethenumberofproductsexportedtoothercountriesintheregionbybetween210and253itemswhereassmallereconomiessuchasColombiaandPeruwouldseeanincreaseofabout50items.

Beyondsimplymovinggoodsfromorigintodestination,transportationserviceshavea temporaldimensionaswell – i.e. the time it takes to deliver a good to itsdestination. Figure C.48 shows that the time needed

to export varies considerably by country and level ofdevelopment. The time needed to export is muchshorter in Europe, North America and Australiacompared with most African and landlocked CentralAsian countries. For the former group, it takes lessthan12daysonaveragetomakeacontainerreadytoleave the country by ship including inlandtransportation, customs clearance and loading. FormostCentralAsianandAfricancountries, thetimetoexportsuchacontainerislongerthan25days.

Thereareseveralwaystothinkaboutthecostoftimeorofdelaysinthecontextoftrade.First,onecanthinkofthecost intermsoftheworkingcapitalthat istiedupwhileshipmentswaitintheholdsofships.Withthisperspective,thecostoftimeisjusttheinterestcostoftheseshipments.Asecondwaytothinkaboutthecostof time is as the rate of depreciation or technicalobsolescence of the tradable good, which could bequite significant for fresh produce, fashion itemssubject to fads or consumer electronics (e.g. smartphones) where innovation is extremely rapid. A thirdandqualitativelydifferentway to thinkabout thecostof time is in terms of uncertainty (Harrigan andVenables,2006).

Thereareatleasttwosourcesofuncertainty.Thefirstarisesfromthewaythatmuchofglobalproduction isorganized. The rise of global supply chains (seeSection B.2(e)), just-in-time inventory managementand lean retailing is making a broader range ofproductsmoretime-sensitive.Forglobalsupplychainsthat depend on manufacturing final products from a

BoxC.9: The mysterious case of the missing delicious red apples

Before itbecameassociatedwithcorporatebehemoths likeAmazon,Boeing,MicrosoftandStarbucks,aswell as the cultural phenomenon that was grunge music, the US state of Washington was famous for itsapples.Tosomeiratestateresidentsthough,itappearedthatonlythesmallandold-lookingonesremainedinthestate,whilealltheredanddeliciousappleswerebeingshippedoutofstate.TothestateresidentswhowrotetotheirlocalnewspapertheSeattle Timesexpressingtheirdisappointment,itwasamysterythathadnoobviousexplanation.

However, the answer to this mystery had long been part of the lore in the economics department at theUniversityofWashingtonandwasevenpartofclassroomdiscussionsandexams.Theanswertothemysteryrelied on the fact that a per unit transportation charge applicable to both high-quality and low-qualityproductslowerstherelativepriceofthehigh-qualityproductatthepointofdestination.Thisleadsconsumersatthedestinationtopurchaseagreaterproportionofthehigh-qualityproductthanconsumersintheplaceoforigin.TheexplanationprovidedbytheeconomistsoftheUniversityofWashingtontothereadersofthe28October1975editionoftheSeattle Timesisreproducedbelow:

“Suppose, forexample,agoodapplecosts10centsandapoorapple5cents locally.Then,sincethedecisiontoeatonegoodapplecoststhesameaseatingtwopoorapples,wecansay that a good apple in essence costs two poor apples. Two good apples cost four poorapples.Supposenowthatitcosts5centsperapple(anyapple)toshipapplesEast.Then,intheEast,goodappleswillcost15centseachandpoorones10centseach.Butnoweatingtwo good apples will cost three, not four poor apples. Though both prices are higher, goodapples have become relatively cheaper, and a higher percentage of good apples will beconsumedintheEastthanhere.Itisnoconspiracy,justthelawofdemand.”

Source:BorcherdingandSilberberg(1978).


182

BoxC.10: Value-to-weight ratios of EU and US imports

The value-to-weight ratio of traded goods has been increasing for all modes of transportation but moststrongly for air transportation during the late 20th century (Hummels, 2007). Table C.16 illustrates thisrelationship forEUandUS imports in the last ten years,with the increasestronger for sea, rail and roadtransportation. This might be related to the surge in jet fuel costs in particular, which shifted parts ofinternationaltradebacktothesemodesoftransportation.

However, thehugedifferences inthevalue-to-weightratiobetweenair transportationandothermodesaswellasbetweendifferentproductgroupsseem tobesimilar forboth theEuropeanUnionand theUnitedStates.Onaverage,goodsthataremovedbyplanesinsteadofvesselsareabout100timesmorevaluableintermsofthisratio.Mostofthesedifferencescanbeattributedtothetradeinmanufacturedgoods,whichisresponsibleforamajorpartofworldtrade.

TableC.16:Value-to-weight ratios for EU and US imports, 2001 and 2011

Sea Air Rail RoadOther modes of transport

EU 2001

Totalmerchandise 364 80,323 164 2,676 448

Agriculturalproducts 486 4,828 111 627 629

Fuelsandminingproducts 140 18,759 87 505 166

Manufacturingproducts 2,042 107,911 498 5,198 4,645

EU 2011

Totalmerchandise 1,080 123,546 398 5,184 952

Agriculturalproducts 1,142 8,140 267 1,349 1,482

Fuelsandminingproducts 600 53,606 248 1,776 677

Manufacturingproducts 3,935 146,445 1,210 9,100 7,178

US 2001

Totalmerchandise 612 85,377 - - -

Agriculturalproducts 981 5,159 - - -

Fuelsandminingproducts 155 281,670 - - -

Manufacturingproducts 2,561 96,087 - - -

US 2011

Totalmerchandise 1,497 133,167 - - -

Agriculturalproducts 1,969 7,804 - - -

Fuelsandminingproducts 704 293,260 - - -

Manufacturingproducts 4,495 140,344 - - -

Source:Global Trade Atlas,maintainedbyGTIS(GlobalTradeInformationServices),owncalculations.

Note:Value-to-weightratiosareshowninUS$permetricton.Averagesforaggregationsareweightedbyimports(dataoriginallyinHS6-digitandHS10-digitdisaggregation,respectively).Onlyexternalimports(fromoutsidetheEU)areused.TheUSonlyreportsconsistentweightdataforitsmaritimeandairborneimports,othermodescannotbecomputed.

largearray of parts and components, unsynchronizeddeliveriescandisrupttheentireproductionprocess.134Uncertainty about exact delivery times can reducetradeascompaniesmightsourcemoreoftheir inputslocallytoreducetheriskofproductioninterruption.

Asecondsourceofuncertaintyarisesfromvolatilityinproductdemand(HummelsandSchaur,2010).Ifafirmfails to correctly guess the tastes of foreignconsumers,itwillbesaddledwithproductsthatnoonewants.Ifthefirmdecidestobecautiousbyproducingonly a limited amount of a given design, it will fail totakefulladvantageofthemarketopportunityevenifitguessedrightaboutforeignconsumertastes.Thefirmcan avoid this dilemma if it can move its productsrapidlytointernationalmarkets,allowingittotimeandadjustitsproductiontomatchforeigntastes.

Empirical work that attempts to measure the cost oftime delays approaches the issue in several ways.Some studies estimate the cost in terms of thereduction in tradevolume.Djankovetal. (2010)showthatanadditionalday in theaveragetimetoexport–meaning the time a shipment requires to move fromthe company grounds to being actually exported –reducesexportsbymorethan1percent.Otherstrytomeasure the percentage increase in the price of thetradablegoodatthepointofdestination.

Hummels and Schaur (2010) estimate that each dayspentintransitisequivalenttocharginganad valoremtariff rateof0.6percentto2.3percent.Tradeflowsthatconsistofpartsandcomponentswerefoundtobe60percentmoretimesensitiveandhencemorelikelyto be transported by airplanes. Air transportation is


183

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

obviously faster but also more costly than seatransportation. Other studies consider the likelihoodthatcountriesmaynotbeabletoenterspecificexportmarketsorparticipate inglobalsupplychains if thereare lengthy delays in trade shipments (Nordas et al.,2006;HummelsandSchaur,2012).TheHummelsandSchaurstudyestimatesthatadelayofthreedayscanreduce the probability to export by 13 per cent (seeSectionB.2(a)).

All inall, theseestimatespaintasimilarpicture–thecost of time delays in international trade is high. Theestimatesinthesestudiessuggestthatadelayofoneweek inshipmentscan reduce the volumeofexportsbyasmuchas7percentorraisethedeliveredpriceofgoodsby16percent.Forexceptionallytime-sensitivegoods,suchaspartsandcomponents,thevolumecanbereducedbyasmuchas26percent.

(b) Determinantsoftransportationcosts

What factors are likely to influence transportationcosts? The possible determinants include productcharacteristics, geography, infrastructure, marketcompetition, technological change, trade facilitationandfuelcosts.

(i) Product characteristics

As discussed above, ad valorem transportation costsdifferdependingonthecharacteristicsoftheproductbeing shipped. Two features particularly relevant inthisregardarethequalityoftheproductanditsvalue-to-weight ratio. All things being equal, ad valoremtransportation costs will be lower for high-quality

goods and for goods which have a higher value-to-weightratio.

(ii) Geography: landlocked countries and distance to markets

The geographical characteristics of countries canhave a significant bearing on transportation costsand hence on countries’ ability to participate ininternational trade. One of the most salient of thesegeographicalfeaturesisaccesstoanoceanorocean-accessiblesea.

There are more than 40 landlocked countries in theworld. Of these, 31 are developing countries, with16 of them being least developed.135 An importantreasonwhybeing landlocked isdisadvantageous toacountry’stradeisthatthecountrybecomesdependenton the transitstates (Arvisetal.,2007)and thus thelocation, size and quality of the transportationinfrastructure to support trade are not fully under itscontrol.Neitherarethepoliciesorregulationsthatwillapplytothetransportationandlogisticssectors.Thesehave to be negotiated with the transit states and theoutcome is not necessarily what the landlockedcountrywouldhavechosen.Also,thetransitcountriesmayhavepoliticalandeconomic incentivesto imposecostsonthelandlockedcountries(Gallupetal.,1999).

Usingthedifferencebetweenc.i.f(cost,insuranceandfreight) and f.o.b (free on board)136 values as ameasure of transportation costs, Radelet and Sachs(1998)findthatlandlockedcountriesface63percenthighercosts.Moreiraetal. (2008)showthatthecosttoimportgoodsintoParaguay,alandlockedcountry,is

FigureC.48:Time needed to export goods (indays)

Less than 12 12 to 19 19 to 25 Greater than 25

Source:WorldBank,Doing Business – Trading Across Borders(2013).

Note:Countriesaregroupedaccordingtothetimetoexport(indays)astandardizedcontainer,meaninghowlongagoodonaveragetakesfromtheproductionsitetobeingreadytoleavetheexportingcountrybyship.Surfacesinwhite:Dataunavailableatthetimeofwriting.ColoursandboundariesdonotimplyanyjudgementonthepartoftheWTOastothelegalstatusofanyfrontierorterritory.


184

about twice as high as the average for other LatinAmericancountriesthathaveaccesstotheAtlanticorPacificOcean.

Using a different measure of transportation costs –shippingrates–LimaoandVenables(2001)estimatethat being landlocked increases transportation costsby 55 per cent, which is similar in magnitude to theestimate found by Radelet and Sachs. As aconsequence of this, they estimate that beinglandlockedreducestradevolumebyabout40percenton average. At the same time, recent research(Borchert et al., 2012) cautions against imputing allthese estimated negative effects to geography asmany landlocked countries also restrict trade inservice sectors – e.g. telecommunications and airtransportation – that connect them with the rest oftheworld.

Another important geographical feature that affectstransportation costs is a country’s distance to othermarkets and to transportation routes.137 Hummels(2007)estimatesthata10percentriseinthedistancebetween the exporting country and the destinationport within the United States increases thecorrespondingtransportationcostsby2.7percentforair and 1.5 per cent for sea shipments. Most otherstudies do not directly try to estimate the effect ofdistanceonshippingorairlinetransportationcharges;instead, the effect of distance is measured by howmuchitreducestradevolumes.Thesestudiesshowahigh and persistent negative impact, suggesting thatclaimsabout thedeclining impactofdistancemaybepremature.

DisdierandHead(2008)reviewmorethan1,400gravitymodelestimatestosystematicallyanalysetheeffectofdistanceontrade.138Theobjectivesofthisanalysisareto determine the central tendency of the resultsindicatedaboveaswellastoidentifysourcesofvariationin the results. On the first question, they are able toconcludethattheelasticityoftradetodistanceisabout0.9.Thismeansthat,onaverage,a10percentincreasein distance between trading partners lowers bilateraltradebyabout9percent.Onthesecondissue,theyfinda great deal of variation in the estimated impact ofdistance from the studies. They attribute the largevariation to differences in data sets, econometricmethods and, most important of all, the time period ofthe data used in the estimation. They find that thedistance effect decreased slightly between 1870 and1950andthenbegantoriseagain.

One problem with most gravity estimates involvingdistanceisthatsomefactorswhichvarywithdistancemay not be fully taken into account. For example,tastes, cultural characteristics and information costsmayvarysystematicallywithdistancesothattradewilldecrease with distance even if transportation iscostless(Feyrer,2009;Allen,2012).UsingtheclosureoftheSuezCanalasanaturalexperimenttotakeinto

account theseother factors,Feyrer (2009)estimatesan elasticity of trade with respect to distance ofbetween 0.2 and 0.5, which is half that found in thegravitymodelestimatesreviewedbyDisdierandHead.

Irrespective of the magnitude of the distance effect,why does it persist as an impediment to trade? First,technologicalprogressmayhavebeen less importantin reducing transportation costs than has beenassumed. Secondly, changes in the composition oftrademightbebiasedtowardgoodswithhighdistancecosts. Related to this hypothesis, as was discussedearlier, the influenceof timeontrade is increasingsothat distance may be serving as a proxy for theincreasedtimesensitivityoftrade.

Thirdly,HillberryandHummels(2008)pointoutthatasubstantialpartof trade is intra-industryexchangeofintermediate inputs for assembled outputs. If inputsarehighlyspecific toaproducer, theremayonlybeavery small possibility of substitution from othersources.Thus,industrialimportdemandswillbemuchmoresensitivetotradecosts–asproxiedbydistance– and firms would rather respond to highertransportationcostsbyrelocatingclosetothesourceoftheinputs.Somesenseofthiseffectcanbegleanedfrom Figure C.49, which is taken from Hillberry andHummels (2008). It uses US manufacturers’ (bothorigin and destination) location at the ZIP code levelandproducts identifiedat a verydisaggregated level.Thefigureshowsthevalueofshipmentsdroppingveryrapidly over distance, particularly within the first200 miles, testifying to the large negative impact ofdistanceontradeinintermediategoods.

(iii) Infrastructure

Theamountandqualityoftransportationinfrastructureinsource,destinationandtransitcountrieshaveamajorimpactontransportationcosts.Thefollowingdiscussionfocuses in particular on the effect of investments inroadandportinfrastructure(SectionC.2discusseshowinvestmentsintransportationinfrastructurecanleadtotheemergenceofnewplayersintrade).

Although there has been a huge surge of studiesdocumenting the importance of transportationinfrastructure,theirfindingsarequitesimilar–thereisa critical role that infrastructure plays in reducingtransportationcostsandenablingtrade.Unfortunately,almostallofthesestudiesestimatethebenefitsfrominvestingintransportationinfrastructurebasedontheperspective of a single country. This is too narrow aview since a country’s infrastructure investment alsoreducesthetransportationcostsincurredbyitstradingpartners and neighbouring countries that arelandlocked,allowingthemtobenefitaswell.Thus,themagnitude of the benefits presented in the tradeliterature probably understates the overall gains fromexpandinginvestmentsintransportationinfrastructure.


185

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

Limao and Venables (2001) rank countries using aninfrastructureindexdesignedtomeasurethecostsoftravel in and through a country. They estimate that acountrywhoseroad infrastructurequalityplaced itonthe 75th percentile globally, i.e. three-quarters of thewaydown the list, had transportationcosts thatwere12percentagepointsgreaterthanthemediancountry.As a consequence, its trade was on average 28 percent lower than the median country. For landlockedcountries,animprovementinownandtransitcountries’road infrastructure couldovercomemore thanhalf ofthe disadvantage associated with being landlocked.ApplyingtheirestimatestoSub-SaharanAfrica,LimaoandVenablesconcludethattransportationcoststhereare higher and trade volumes lower than would bepredicted given the economic characteristics of thecountries–incomes,distance,etc.Theythenattributemuchofthisresulttothepoorstateoftransportationinfrastructureinthecontinent.

A laterstudybyFreundandRocha(2010)onAfricanexportsshowsthatuncertaintyinroadtransporthadanegativeandsignificanteffectonacountry’sabilitytoexport. Their results point to improvements in roadsystems – especially infrastructure, security andpoliciesthat improvecompetition in trucking–askeytostimulatingAfrica’sexports.Blyde(2010)andVolpeet al. (2012) look at the impact of increasinginvestmentsinroadinfrastructureinacoupleofLatinAmericancountries–ColombiaandPeru.

Blyde (2010) first establishes that lower domestictransportation costs in Colombia can significantlyimprove the prospects of exporting. He finds thatregions within the country with lower transportationcosts (regions in the 25th percentile) export around2.3 times more than regions with higher transportcosts (regions in the 75th percentile) once otherfactors are taken into account. He then simulates areduction in transport costs that would arise if thecondition of all the roads identified as “bad” and“regular”bythenationalroadauthoritywereimprovedto “good”. He concludes that this simulated

improvement in road conditions decreases averagetransport costs by about 12 per cent and boostsaverageexportsbyaround9percent.

Volpe et al. (2012) assess the effects of new roadsconstructed in Peru between 2003 and 2010 onPeruvian firms’ exports. The authors conclude thatexporters whose routes were shortened due to theconstructionofnewroadshadexportsthatwereabouttwo-thirds higher than exporters whose route lengthremainedthesame.Overall,theadditionalinvestmentsin road infrastructure meant Peruvian exports were20percenthigherin2010thantheywouldhavebeenwithoutthenewroads.

ThestudiesbyClarketal.(2004)andAbeandWilson(2009) look at the relationship between portinfrastructureand transportationcosts.139Clarketal.(2004)usedataonallUSimportstransportedbysea.Theyconstructanindexofportefficiencyusingsurveymeasures drawn from the World Economic Forum’sGlobal Competitiveness Report,whichdepends,amongother determinants, on the general condition of thecountry’s infrastructure:themoreefficientacountry’sport, the higher its score in this index. Clark et al.estimate that a countrywhich improves its ranking inport efficiency from the 25th to the 75th percentilereduces shipping costs by 12 per cent; this, in turn,implies an increase in bilateral trade of around25percent.140

The study by Abe and Wilson (2009) focuses on thegrowingproblemofportcongestioninEastAsia.ThishasworsenednotonlybecauseoftherapidgrowthinEastAsia’stradebutalsobecausemuchofthattradeis seaborne. Port congestion leads to bottlenecks,which significantly increase the cost of transportinggoodstoandfromEastAsia.Theiranalysissuggeststhat expanding facilities in East Asian ports so as tocut congestion by 10 per cent could decreasetransportationcostsbyupto3percent.

(iv) Market competition

The transportation sector is a service industry whoseefficiencywilldepend,inpart,ontheexistingregulatoryregime and the extent of competition. There are anumber of reasons for a lack of competition in thetransportation sector, including natural monopolies,marketaccessbarriers thatprevent foreignfirms fromentering and competing, and the cartelization oftransportation service providers. In some cases,governmentsmayevenallowpractices,e.g.price-fixing,thatwouldotherwisebeillegalunderanti-trustlaws.

Because the literature on competition in thetransportation sector is immense, the followingdiscussion focuses on a specific sector – maritimetransportation – to illustrate the scope for morecompetitiontoreducetransportationcostsandexpandtrade volumes (see WTO Secretariat Notes

FigureC.49:Estimated relationship between shipments of intermediates and distance

2.6

0

20

0

50

0

1,0

00

2,0

00

3,0

00

204.8

Num

ber o

f shi

pmen

ts

Miles

Source:HillberryandHummels(2008).


186

S/C/W/315 and S/WPDR/W/48 for a fullerdiscussion of competition issues in maritimetransport).141

The maritime transportation market is usuallysubdivided into two: the tramp (or bulk) market andliner market. Tramp ships have no fixed route orscheduleandcanbecharteredforaperiodoftimeorfor a specific voyage. In contrast, liner companiesoperate vessels between fixed ports on a stricttimetable(UNCTAD,2010a).

In the tramp market, the carrier chases the cargo(Brooks, 2011) and price competition is generallyconsidered intense (Clarkson Research Studies,2004). In this unrestricted market, freight rates arevolatile because capital costs are high and supplydifficulttoadjustintheshortrun(Brooks,2011).Shipscostmillionsofdollars;ittakesyearstobuildoneandtheoperatinglifetimeofshipsiscountedindecades.

In the liner market, companies typically organizethemselvesintoaconsortium,withaviewtoprovidinga joint transportation service. In stark contrast to thetrampmarket,operatorsinthelinermarkethavebeenexempt fromnationalanti-trust lawssince the turnofthe20thcentury.Partofthereasonforthisexemptionwas the desire to reduce price volatility. If operatorscanfixpricesandiftheycolludetomaximizeindustryprofits, prices will be higher – set at a mark-up tomarginal cost. The size of the mark-up will varyinverselywiththeelasticityofdemandofthegoodthatis transported, i.e. the more inelastic is the finaldemand, the higher the mark-up. Thus, while thisexemption from anti-trust laws may reduce pricevolatility,itwillbeatthecostofhigherfreightchargesandlowertradevolumes.

In 2010, the European Union removed the anti-trustexemption on price fixing although operationalcooperation among consortia members, such assharingspaceontheirrespectivevessels,continuedtobe exempted.142 Liner members are expected tomarketandpricetheirservicesindividually.TherewasasimilarlegislativeeffortmadeintheUSCongressin2010 to remove the liners’ exemption from US anti-trustlawsbutthebillwasnotpassed.

Beyond government policies, one of the reasonsdiscussedaboveforlackofcompetitionmaysimplybethe existence of a natural monopoly. Hummels et al.(2009)havearguedthattheremaybesuchanelementoperating in the case of developing countries. First,the volume of their trade – particularly of small,developing countries – is tiny compared with thecapacity of modern container ships. Secondly, theremay be substantial economies of scope in offeringtransport services over a network of ports. As aconsequence, itmaybedifficult tosustainmore thanone or two operators to service shipping routes tosomedevelopingcountries.Asimilarpatternhasbeen

highlighted in a recent report by the United NationsConference on Trade and Development on maritimetransport(UNCTAD,2010).Between2004and2011,the average number of liner companies dropped bynearly 23 per cent while the size of the largest shipdeployednearlydoubled.

A trend featuring increasing containership sizes andcarrying capacities and declining competition withinthe industry has continued for several years. Thislimited competition means developing countries payhigher transportation costs and have lower tradevolumes. To estimate these effects, the paper byHummels et al. (2009) studies freight costs for theUnited States and a number of Latin Americancountries.TheyestimatethatshippingpricesonLatinAmericanimportsare,onaverage,30percenthigherthanshippingpricesonUSimportsandthatone-thirdof thisdifference isexplainedby thesmallnumberofcarriers serving Latin American importers. They alsocalculate that eliminating market power in shippingwould increase Latin American import volumes byabout15.2percent.

(v) Technological change

Innovations in the transportation sector can have animportant role in bringing down transportation costs(see Section C.3 for a discussion about the linkbetween technological change and trade morebroadly). Notable examples of innovation include thedevelopment of the jet engine and the adoption ofcontainerizationinmaritimetransportation,whichalsoincreasedtheefficiencyofmulti-modaltransport.

AsGordon(1990)observes, the introductionof the jetaircraftinthe1950screatedprofoundqualitychangesin both performance characteristics and operatingefficiency of commercial aircraft. Compared with thepiston-drivenplaneswhichitdisplaced,jetaircraftsarefaster and have lower maintenance and fuel costs.Adjusting for these improvements in performance,Gordon (1990) estimates that the real price of jetaircraftsfellatarateof12.8percentto16.6percentperyearduring1958-72whentheybegantobewidelyadopted. The reduction in quality-adjusted aircraftprices appears to have been transmitted to airtransportationcharges.Usingtheaveragerevenueperton-kilometreshippedasameasureofairtransportationcost, Hummels (2007) estimates that costs fell morethan ten times over the 50-year period since theintroductionofthejetaircraft(seeFigureC.50).

At its simplest, a container is nothing more than ametal box of standardized dimensions. Yet, this boxenabled the unprecedented expansion of world tradeinthesecondhalfofthe20thcenturyandcontributedto the rise of just-in-time manufacturing and globalsupply chains (Levinson, 2006). The value of thecontainerlaynotintheproductitselfbutinthesystemoftransportation involvingcontainerships, trucksand


187

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

freight trains built to handle container cargo (thusfacilitating multi-modal transport) and automatedhandlingthatgrewaroundit.Itseffectontradewassoprofound that it is tempting to assume that maritimetransportationcostsmusthavefallendramaticallyasaconsequence of the widespread adoption ofcontainers.

However, Hummels (2007) finds no strong empiricalsupportforthispresumption.AscanbeseeninFigureC.51,whilerealtramppricesdeclinedoverthis50yearspan,reallinerpriceswerealmostatthesamelevelin2003astheywerein1955althoughtherehavebeenmarkedfluctuations.

There are several explanations for this apparentdiscrepancy. Levinson (2006) observes that mosthistorical data on freight cost cover only the oceanvoyagebetweentwoportsanddonotincludethetotaldoor-to-doorcostofashipment.Thesetotalcostsaremore likely to have declined with the adoption ofcontainerizedtradeaslargeefficiencygainsforinlandtransport and loading and unloading have beenrealized. Hummels (2007) suggests anotherexplanation is that the available price indices do notadequately capture the quality improvement madepossiblebycontainerization.Containershipsarefasterandquicker in termsofcargohandling.Asdiscussedearlier,thisquickerturn-aroundisabsolutelyessentialto today’s just-in-time inventory systems and globalsupply chains. Thus, even if these prices have notdeclined,thefactthatgoodscanbemovedmuchmorequicklythaninthepastmeanstherehasbeenineffectareductionincostfortraders.

Finally, given that fuels costs are a significantcomponent of the cost of operating shipsor airplanes(see discussion below on fuel costs), current R&Deffortsarebeingdirectedatimprovingfuelefficiencyofthese vessels. In the case of a jet aircraft, the threeprincipal areas of concentration are reducing aircraft

weight, improving aerodynamic design to lessen lift-dependent and friction drag, and enhancing engineperformancetocutfuelburnperunitofdeliveredthrust.

Aircraft weight is being reduced through greater useof advanced alloys and composite materials and byreplacing hydraulic control systems with lightweightelectrical ones. Aircrafts are being designed so thateffectivewingspanextensionismaximized,laminarorstreamlineflowisbettermaintainedandwettedareas(areas of the aircraft in contact with the airflow) areminimized. The thermal, transmission and propulsiveefficienciesofnewerjetenginesaresuperiortothoseof previous generations. These improvements havemade new aircraft such as Boeing’s 787 airlinerconsume about 40 per cent less fuel per passengerthantheir1970scounterparts.143

(vi) Trade facilitation

Trade facilitation examines how procedures andcontrolsgoverningthemovementofgoodsacrossandwithinnationalborderscanbeimprovedandsimplifiedto reduce delays and costs. While it is not part oftransportationper se,thegrowingprominenceoftime-sensitivetradeandglobalsupplychainsincreasestheimportance of border or customs-related costs, andhence of trade facilitation. The potential reduction intrade costs from improvements in trade facilitationappearssignificant.ThisiswhyitisamajorpartoftheWTO’s Doha Round negotiations, where the subjecthasmaderapidprogress(seeBoxC.11).

FigureC.50:Air revenue per ton-kilometre, 1955-2003

FigureC.51:Real tramp and liner price indices, 1955-2003

0

200

400

600

800

1,000

1,200

19

55

19

60

19

65

19

70

19

80

19

85

19

90

19

95

20

00

20

03

19

75

19

55

19

60

19

65

19

70

19

75

19

80

19

85

19

90

19

95

20

00

20

03

0

50

100

150

200

250

300

350

400

500

Liner price index Tramp price index

450

Source:Hummels(2007).

Note:Indexisnormalizedto100fortheyear2000andbasedonrevenuesexpressedinyear2000US$.

Source:Hummels(2007).

Note:IndicesaredeflatedusingaUSGDPdeflatorforthetramppriceandaGermanGDPdeflatorforthelinerprice.Furthermore,indicesarenormalizedto100fortheyear2000.


188

Many studies use measures of trade logistics, such astheWorldBank’sDoingBusinessindicatorsorLogisticsPerformance Index.The latter isbasedonaworldwidesurveyoffreightforwardersandexpresscarriersonthelogistics performance of the countries in which theyoperate and those with which they trade. It assessesperformance in six major areas: efficiency of theclearanceprocessbybordercontrolagencies;qualityoftrade- and transport-related infrastructure; ease of

arranging competitively priced shipments; competenceandqualityoflogisticsservices;abilitytotrackandtraceconsignments; and frequency with which shipmentsreachtheconsigneewithinthescheduleddeliverytime.In 2012, Singapore, Hong Kong (China) and Finlandwereinthetopthreeplacesintheindex(seeArvisetal.,2012). OECD countries generally ranked high on theindex,whilecountriesatthebottomweretypicallyleastdeveloped,landlockedorfromSub-SaharanAfrica.

BoxC.11: Trade facilitation negotiations in the WTO

Startingovertwoyearsafter the launchoftheDohaRound,144thenegotiationsontradefacilitation intheWTOhavenowbecomeoneof themostadvancedcomponentsof theDohaDevelopmentAgenda (DDA).IntensediscussionsamongWTOmembershavebrought thetalkstoastagewherethecontentsofanewagreementareclearlytakingshape.

Afirstdraftoftheprospectiveagreementwastabledin2009andhassubsequentlybeenrefined.Followingdiscussions, members have been able to reduce the number of open questions by more than two-thirds.Structured into twomainpillars, theDraftAgreement (TN/TF/W/165and revisions)proposesa seriesoftradefacilitationmeasures (Articles1-15)anda related implementationmechanism(usually referred toasspecialanddifferentialprovisionsfordevelopingcountries).

TheproposedfacilitationdisciplinesarelargelybasedonthreeexistingprovisionsoftheGATT1994:ArticleV (freedom of transit), Article VIII (fees and formalities connected with importation and exportation) andArticle X (publication and administration of trade regulations). An additional, non-GATT based segmentseekstoenhancecooperationbetweencustomsandotherrelevanttradefacilitationauthorities.

Instructed by the negotiating mandate145 to “clarify and improve” the above-mentioned GATT disciplines,members have proposed a series of measures to strengthen the current regulatory regime. Commonunderlyingobjectiveshavebeen thesimplificationofborder crossingprocedures, the reductionof relatedbureaucraticobstaclesandthecreationofamoretransparentandpredictabletradingenvironment.Withtheexistingrulesnothavingbeenrevisedformorethan65years,governmentsagreedontheneedtoupdatealegal framework that reflects the trading realitiesof the1940sasopposed to thoseof today’sglobalized,supply-chaindominatedworld.

Theexpectedbenefitsofanewagreementarewidelydocumentedandcovervariousaspectsofthecross-bordermovementofgoods.AccordingtoarecentOECDstudybyMoïséetal. (2011),thetradefacilitationmeasuresnegotiated in theWTOhave thepotential to reduce total tradecostsbyalmost10percent forOECD countries alone.146 It has also been shown that successfully implemented facilitation programmesincreasecustomsproductivity, improve taxcollectionandattract foreigndirect investment.There isalsoapositive impact on government revenue, with several countries having more than doubled their customsproceedsafterintroducingtradefacilitationreforms(OECD,2009).

Thebenefitshavebeenshowntobeparticularlysignificantfordevelopingeconomiesandleast-developedcountries (LDCs). Research has found that up to two-thirds of the total gains from trade facilitation areobtainedbythedevelopingworld(OECD,2009).

Manydevelopingcountriessufferfromborderproceduresthatarelessefficientthanthoseoftheirdevelopedcounterparts,indicatingagreatpotentialforimprovement.Thecostsofimportinggoodshavebeenfoundtobeabout20percenthigherforlow-incomecountriesthanfortheirmiddle-incomecompetitors,andanother20percenthigherwhencomparedwithhigh-incomeeconomies(HoekmanandNicita,2010).

The recent analysis by Hoekman and Nicita (2010), which is based on the World Bank’s Doing BusinessIndicators,concludesthattheWTOdiscussionsontradefacilitationare“perhapsofgreatestrelevancetolow-incomecountriesfromatradeexpansionperspective…”.147Thestudyfindsthateven“takingrelativelylimitedactionstofacilitatetradecanboostthetradeexpansioneffectsoftheDohaRoundbyafactoroftwo,threeormore”.148Italsosuggeststhat“pursuitoftradefacilitationisparticularlyimportantforlower-incomecountries,especiallyLDCs,thatotherwisewillnotbenefitfromtheDohamarketaccessnegotiations–becausetheyhaveduty-free,quota-freeaccesstomajormarketsandwillnotbeaskedtoreformtheirowntradepolicies”.149


189

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

Withthehelpofthesetwoindices,Portugal-PerezandWilson (2009) show that progress in trade logisticscanboostthetradevolumesofAfricancountries.Evenrelatively small improvements in these indices thatmoveacountrytowardsthescoresofwell-performingAfrican countries have large impacts on trade andwelfare. This result is consistent with other findingsabout the huge trade benefits from trade facilitationfordifferentsetsofcountries(seeWilsonetal.,2003;2005; 2008; Christ and Ferrantino 2011; Márquez-Ramosetal.,2012).

OtherauthorshaveinvestigatedtheeffectoftheWTO’sAidforTradeinitiativeonthecostsoftradingandfoundthatithassignificantlyreducedthosecosts(Königeretal.,2011;Wilsonetal.,2006).ThisindicateshowAidforTradecanbeeffective inhelpingdevelopingcountriesimprove their trade logistics capabilities. It shows thatevensimplereforms,whoseimplementationisrelativelyeasyandcost-efficient,canhavesignificanteffectsoninternational trade. Such reforms include theintroduction of a single window, whereby all customsdocuments have to be submitted to only onegovernmentalauthority,orswitching frompaper-basedtoelectronicsubmissionofdocuments.

Ahmad(2010)describesthereformeffortsinPakistanand shows that a set of improvements that includesthetwoaforementionedchangesreducedtheshareofshipmentstakinglongerthanonedaytoclearcustomsattheportofKarachifrom96percentto7percent.Asdiscussedabove,thecostsrelatedtotimedelaysatcustomsandtheresultinguncertaintycanrepresentaconsiderableshareofthetotalcostofaproduct.

The extensive literature on trade facilitationemphasizestheimportanceofimprovingthereliabilityandtransparencyofcustomsandotherauthorities,ofincreasing competition in service providers in theproximityofbordersandofprovidingadvance rulingsso as to reduce the uncertainty faced by traders.150

Furthermore, most studies agree on theinterdependency of the reforms, highlighting that anintegratedapproachmagnifiesthebenefitsandyieldslastingimprovements.151Solvingonelogisticalproblemwithin customs or inland transport may not producemajorbenefitsunlessotherbottlenecksaretackledaswell.

(vii) Fuel costs

Higher fuel costs increase transportation costs.Studies by Mirza and Zitouna (2010) and UNCTAD(2010b)findthattheelasticityoftransportcostswithrespect to fuel prices is between 0.09 and close tounitydependingon thecountries, timeframes,modesoftransportationandproductsthatarestudied.Thisisquite a wide range since it means that a 1 per centincrease in fuel costs increases transportation costsbybetween0.09percentand1percent.

Thereareanumberofexplanationsforthiswiderangeofestimates,includinghowhigherfuelchargesaffectvariousmodesoftransportdifferently.152However,thesensitivityoftransportationcoststochangesinenergypricesappearstohavebeenheightenedby long-termimprovements in transportation logistics, such asreduced loading times through containerization. As aresult of the reduction in these non-energycomponents of transportation costs, fuel costs nowaccount for almost one half of total freight costs(RubinandTal,2008).

Rising energy prices adversely affect some transportmodes more than others. Moreira et al. (2008) showthat the increase in energy prices in the last decaderaised air transportation costs relative to maritimecosts for several Latin American countries and theUnited States. Over this period, the Chilean andUSmodalshareforairtransportdecreasedoratbestremainedconstant.153Thisisincontrasttothesecondhalfofthe20thcenturywherethetrendhadbeenformoreandmoreproductstobemovedbyairtransport.

Energycostsalsoinfluencethecompositionoftradedgoodsastheyarelikelytohaveamoreadverseimpacton goods with low value-to-weight ratios. For theseproducts,soaringenergycostscanquicklywipeoutacomparative advantage based on differences in thecostsoflabour,particularlywheremarginsarenarrow.This can put pressure on those global supply chainswhich depend heavily on differences in labour costsacross countries. This difference in value-to-weightratios is reflected in producers’ choice of transportsincelightandhighlyvaluablegoodsaremorelikelytobetransportedbyair(seeBoxC.12).

An analysis of maritime transportation confirms thatgoods with low value-to-weight ratios are likely tosuffer more from higher energy costs. A study byUNCTAD (2010) estimates the elasticity of maritimefreightrateswithrespecttooilpricestobelowerforcontainerized products (0.19 to 0.36) than forproductssuchasironore(upto1.0),whichisabulkyand low value-to-weight good. The study also findsthat these elasticities seem to increase at times ofsharplyrisingenergyprices,whichcouldbeexplainedby the increased volatility and uncertainty thatshipping companies have to deal with under suchcircumstances.

High oil prices can also prompt trading partnerslocated further away to divert trade towardsneighbouring regions (see the discussion on theregionalization of trade in Section B2(d)). In otherwords, soaring oil prices can act as tariff surchargesdifferentiatedbyorigindependingon theproximityoftheexportertotheimportingcountry.

Mirza and Zitouna (2010) introduce a theoreticalmodel, where transport costs have a fixed and avariable component, with energy prices being part of


190

BoxC.12: Shares of modes of transport

Themostimportantmodesoftransportusedforinternationaltradeareseaandairtransportation.However,railandroadtransportareofparticularimportancefortradewithneighbouringcountriesorwithinregionalclusters,suchastheEuropeanUnionorNorthAmerica.Martínez-ZarzosoandSuárez-Burguet(2005)findforcertainLatinAmericancountriesinthelate1990sthatairtransportationisusedfor12to25percentoftotalimportsintermsofvalue.Incontrast,seabornetransportaccountsfor45to70percentofthevalueofimports.ThispatterncanbeobservedinTableC.17,wheremodalsharesofEuropean,AmericanandChileanimportsarecompared.

Dependingon theproductcategoryand thespecificbreakdownofacountry’s imports, thevalueshareofmaritimetransportliesbetween45and95percent.Intermsofitsweight,mostinternationaltradeiscarriedbymaritimetransport.UNCTAD(2010a)findsthissharetobeover80percent.Moreover,Hummels(2007)showsthatairtransportaccountsforlessthan1percentofworldtradeinweight.

However, air transport has been gaining in importance as recent growth rates of ton-miles have beensignificantlyhigherthanforanyothermodeoftransport.Moreover,unlikeagriculturalgoodsandfuelsandminingproducts,high-valuegoodssuchasmanufacturedgoods–consumerelectronicsspecifically–aretransportedprimarilybyair.Theshareofairtransportbecomesfarmoresignificantifworldtradeismeasuredintermsof itsvalueinsteadof itsweight.TableC.17showsthat intermsofvalue,29percentoftheEU’smanufactured imports were transported by air but in terms of weight these same goods only account for1.3percentoftheEU’smanufacturedimports.

TableC.17:Shares of modes of transport for imports in value and weight, 2011 (percentage)

Shares in terms of value

EU 2011 Sea Air Rail RoadOther modes of transport

Totalmerchandiseproducts 55.62 18.80 1.16 12.95 11.47



Manufacturedgoods 45.29 29.02 0.95 19.05 5.69

Shares in terms of quantity

EU 2011 Sea Air Rail RoadOther modes of transport






Chile 2011 Sea Air Rail RoadOther modes of transport






US 2011 Sea Air Rail RoadOther modes of transport

Totalmerchandiseproducts 52.51 22.41 - - 25.08

Agriculturalproducts 62.11 3.56 - - 34.33

Fuelsandminingproducts 77.25 1.58 - - 21.18

Manufacturedgoods 45.69 29.64 - - 24.67

Source:Global Trade Atlas,maintainedbyGTIS(GlobalTradeInformationServices).

Note:Modalshares(inpercent)areconstructedbyusingdataonimports.FortheEU,onlyexternal imports(fromoutsidetheEU)areused.Duetodatalimitations,“Othermodesoftransport”forUSincludesrail,roadandothermodesoftransport.ForEUandChile“Othermodesoftransport”isanaggregationoverallremainingmodessuchasinlandwaterways,pipelinesormaildeliveries.


191

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

the latter component. One prediction of theirtheoretical model is that with increasing fuel costs,international tradewillbecomemoreregionalizedandproducts will be sourced more locally. Rather thanthroughtrade,distantmarketsmaythenbeservedbyway of foreign affiliates or licensing arrangements.However,theempiricalevidenceisfarfromconclusive.

Contrary to the model prediction, Mirza and Zitouna(2010) find that the difference in the elasticity oftransportcostswithrespecttoenergypricesbetweencountries close to the importer compared with onesthatarelocatedfarawayfromtheimportingmarketisonly very small.154 They show that in the US market,MexicanandCanadianexportersdonotseemtohaveoutperformed other exporters when the price for oiland other energy sources increased before thefinancialcrisis.RubinandTal(2008)findtheoppositeresult toMirzaandZitouna, showing thatduringpastoilcrisestheshareofUSnon-oilimportsfromEuropeandAsiadroppedwhiletheshareof importsfromtheAmericaswentup.TheycalculatethatatanoilpriceofUS$200,importsfromEastAsiawouldbeequivalentto an additional 15 per cent tariff on comparableimportsfromMexico.

(c) Conclusions

Thefuturescenariooftransportationcostswilldependon how different determinants – distance to marketsand transportation routes, infrastructure, tradefacilitation, competition and regulation, transportationtechnology,andfuelcosts–arelikelytodevelop.

Section C.4 notes the IEA’s baseline prediction of along-termincrease intherealpriceofenergyofabout16 per cent. Although no similarly precise projectionsfor the other determinants of transportation costs areavailable,itispossible,basedontheextensiveliterature,to arrive at estimates of how improvements in theseareas could reduce transport costs (see Table C.18).These estimates are used to perform a series of“thought experiments” to assess whether such costreductions are likely to offset the expected rise inenergyprices.

The estimates referred to in Table C.18 come fromdifferent studies and employ different countries andtime periods. While the estimates are statisticallysignificant, they are still subject to estimation error.Moreover, they do not distinguish which mode oftransportisinvolved.Nordotheyincludetheimpactoftechnological change. Although technologicaladvancescanbekey to reducing transportationcost,there are no available estimates of how additionalinvestments in R&D will translate into reductions intransportation costs. Despite these caveats, theexercisecanberevealing.

Based on the work by Mirza and Zitouna (2010) andUNCTAD (2010b), there is a lot of variation in theestimated elasticities of transportation costs withrespecttofuelprices.Ifwetakethemaximumoftheirestimates–anelasticityofone–a16percentriseinenergy prices will translate to a 16 per cent rise intransportation costs. Rubin and Tal (2008) estimatethat fuel costs represent about half of transportation

TableC.18:Estimates of potential changes to transportation costs

DeterminantsEstimated impact on transportation cost

Sources Remarks

Fuelcost Increasetransportationcostbybetween8%and16%

MirzaandZitouna(2010)

UNCTAD(2010),RubinandTal(2008)

Future scenario:

Energycostsriseby16%

Infrastructure Decreasetransportationcostbyupto12%

LimaoandVenables(2001)

Blyde(2010)

Assumed improvement in infrastructure:

Countriesmakeinvestmentsintransportationinfrastructurethatimprovetheirrankingfromthe75thtothe25thpercentile.

Tradefacilitation Decreasetransportationcostby10%

Decreaselowincomecountries’tradecostsby20%

Moïséetal.(2011)

HoekmanandNicita(2010)

Assumed improvement in trade facilitation:

ImplementtradefacilitationmeasuresbeingnegotiatedintheDohaRound.

Improvelow-incomecountries’tradefacilitationscoretothelevelofmiddle-incomecountries.

Competition Decreasetransportationcostbyupto10%

Hummelsetal.(2009) Assumed increase in degree of competition:

Increasenumberofcarriersservingdevelopingcountrymarkets.


192

costs.IfwetakeRubinandTal’sestimateasthelowermost projection, transportation costs will go up by8percentasa resultofa16percent rise inenergyprices.

The estimates in Table C.18 suggest that there isample scope for improvements in trade facilitation,investments in transportation infrastructure, andintroducing more competition in transportationservicestooffsethigherenergypricesinthefuture.

The estimates from Limao and Venables (2001) andBlyde (2010) suggest that poor countries whichimprove their transportation infrastructure sufficientlyto raise them from the 75th percentile to the25th percentile can expect to lower transportationcosts by about 12 per cent. The Moïsé et al. (2011)study gives an estimated reduction in trade costs ofabout 10 per cent if the trade facilitation measuresbeingnegotiated in theDohaRoundcometofruition.The study by Hoekman and Nicita (2010) suggeststhat the cost of importing for low-income countriescould be reduced by 20 per cent if their borderprocedures were at a comparable level to that ofmiddle-income countries. Based on the study byHummelsetal. (2009), there isalsoa largepotentialto be tapped from increasing competition intransportation routes serving developing countries.Their transportation costs can be cut by as much as10 per cent, which can either wholly or significantlyoffsettheeffectofhigherfuelcosts.

These “thought experiments” underscore theimportanceofpursuinganumberofpolicyinitiativesatboth thenationalandmultilateral levels.They includeimproving the quantity and quality of transportationinfrastructure, successfully concluding the DohaRound of negotiations and introducing morecompetition in routes that serve poor countries.Although the predicted cost impact of technologicalchangeisnotincludedinthetable,giventheinherentdifficulty of predicting future innovation, it is likely tobeapowerfulforceforcostreduction.

Ifnosignificantprogressismadeonthesefronts,theexpectedriseinenergypricesmaywelltranslateintoa long-run rise in transportation costs. Theconsequencewill beslower tradegrowth.Theremaybe more regionalization of trade as highertransportationcostspenalize tradewithmoredistantcountries. There will be a shift in the composition oftrade,whichwillfavourhigh-qualitygoodsandgoodswithhighervalue-to-weightratios.Theshareoftime-sensitivegoodsintradewillfall.Theextensivemarginofinternationaltrade–thequantityofgoodstraded–will be affected adversely. Furthermore, there mightbe a move away from trade in merchandise goodstowardstradeinservices,technologyandideassincethis would entail far less transportation costs(Hummels,2009).

Among the major trends identified in Section B wasthe emergence of new players in international tradeandtheriseofglobalsupplychains.Whilenottheonlyexplanation,areductionintradecostshasbeenakeydriverofthesetrends.

On the basis of various measures of transportationcosts and logistics performance, least-developedcountriesandcountries inSub-SaharanAfrica tend tofareworst,whilepoor,landlockedcountriesfaceuniqueobstacles.Notonlydohigher transportationcostsandlonger delivery time reduce these countries’ overallvolumeoftrade,theymakeitdifficultforthemtobreakintonewmarketsandparticipateinglobalsupplychains.

Their situation can be alleviated through improvedtrade facilitation, introducing greater competition andby making sizeable investments in transportationinfrastructure.Giventhelikelihoodofrisingfuelcostsin the future, there is someurgency in reformingandmodernizing these countries’ transportationinfrastructure and regulatory systems. The pay-offsfrom infrastructure investments appear large andshould justify commitment of more resources on acost-benefit basis. Because the trade partners ofthesecountrieswillalsoseebenefitsfromlowertradecosts, it is in their interest to provide assistancethrough the Aid for Trade initiative, for example.Beyondthis,theremaybeagoodreasontore-examinethe subject of competition policy in the future as theavailable evidence suggests market power intransportation services has been particularlyburdensometoanumberofdevelopingcountries.

6. Institutions

This section studies the relationship betweeninternational trade and the institutional framework.Two broad questions are addressed: How doinstitutions shape international trade relations? Andhowdoestradeaffectinstitutions?Thekeyobservationin this section is that, in the long run, there exists adual relationship between these two variables (in thelanguage of economists, they are endogenous). Putsimply, institutions shape and are shaped byinternational commerce. Understanding thisrelationshipcanhelpshedsomelightonthefutureofinternationaltradeandthemultilateraltradingsystem.

What are institutions? Economists have developed anotion of institutions that incorporates practices andrelationshipsaswellasorganizations.AsNorth(1990)explains, “institutions are the rules of the game in asociety or, more formally, are the humanly devisedconstraints that shape human interaction” (North,1990). In economics, therefore, institutions are thedeepframeworks,suchassocialnorms,ordinarylaws,politicalregimesorinternationaltreaties,withinwhichpolicies – including trade policies – are determinedandeconomicexchangesarestructured.


193

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

Institutions can be formal or informal. Formalinstitutions are those that are consciously created byagents and that impose clear and visible constraints.Informal institutions are conventions and codes ofbehaviour. Formal institutions can be further sub-divided into political and economic institutions. Theformer impose constraints on government activitieswhereas the latter set rules that directly affect therelationship between economic agents. This sectionexaminesthetwotypesofformalinstitutions,politicaland economic, and then focuses on informalinstitutions,broadlydefinedasculture.

Formal and informal institutions shape and areshapedbyinternationaltrade.Institutionaldifferencescreatetransactioncoststhatmakeitmoredifficulttotradebuttheycanalsoformthebasisofcomparativeadvantage in certain sectors or production tasks.More directly, institutions determine how trade andtrade-relatedpoliciesaresetandnegotiated,leadingto a more or to a less open trading environment. Inthissense,institutionsareclearlyashapingfactoroftrade. At the same time, economic integration is animportantdeterminantofinstitutionaldevelopment,inthe political, economic and cultural spheres. Whilethese dynamic effects are likely to be slow tomaterialize, they feed back into trade relations overthelongerterm.

Part(a)looksatpoliticalinstitutions,suchastheformof government. Part (b) focuses on economicinstitutions, such as the quality of the regulatorysystem. Part (c) examines cultural norms, such asthose embedded in social values. Trade agreementsare both political and economic institutions, in thatthey commit national policy makers and affecteconomic actors. For this reason, trade agreementsarediscussedthroughouttheentiresection.Eachsub-sectionbeginswithsomeevidenceontherelationshipbetweentradeandinstitutions.Thegoalisnottohavea thorough empirical analysis but rather to highlightsomefactsandcorrelationsthatcanthenbeanalysedinlightofeconomictheory.

(a) Politicalinstitutions

Political institutions shape economic interactions intwoways:first,theyimposeconstraintsongovernmentactivities;secondly,theyinfluencethesetofeconomicinstitutions that societies adopt. The economicliterature has tended to focus, in particular, on theimpactoftheformofgovernmentandpoliticalbordersoninternationaltrade.Formofgovernment,definedbythe extent of accountability, legitimacy, transparencyand choice in a political system, may impact tradeindirectlythrougheconomicdevelopmentordirectlybyaltering policy-makers’ incentives to set trade policy.Similarly, political borders impact trade flows directlybyincreasingtradecostsandindirectlybyfragmentingtheinternationalpoliticalsystem.

(i) Form of government

Democratic forms of government have been on therise over the last half-century (Murtin and Wacziarg,2012; Acemoglu, 2012), as has world trade. Much ofthis research uses data from the Polity IV Project todefine and measure the form of government. ThePolity scheme captures key qualities of politicalinstitutions and processes, including executiverecruitment, constraints on executive action, andpolitical competition. Individual ratings are combinedinto a single measure of regime governance – the“Polity score”–ona21-point scale ranging from -10(fully institutionalized autocracy) to +10 (fullyinstitutionalized democracy). The measure examinesconcomitant qualities of democratic and autocraticauthorityingoverninginstitutions,ratherthandiscreetand mutually exclusive forms of governance. Thisperspective results in a spectrum of governingauthority that spans “autocracies” (-10 to -6), mixedauthority regimes or “anocracies” (-5 to +5), and“democracies”(+6to+10).155

FigureC.52showsthatthecorrelationbetweenmoredemocratic formsofgovernment,asmeasuredby the“Polity score”, and trade (measuredas the total tradetoGDPratio)isstrongandpositivebetween1962and2010.

A positive correlation between the value of importsand exports and a more democratic form ofgovernmentcanalsobeseenusingacross-sectionofcountries in2010(seeFigureC.53).Whilenotshownin this report, a similarpicture isevident fordifferentyears from 1962 onwards. However, if the ratiobetweentotal tradeandGDPisused, rather thanthevalueofimportsandexports,thecorrelation(whilestillpositive) appears to be weaker; possibly suggestingthat richer countries are both more democratic andmoreopentotrade.

The main conclusion from these figures is thatcountrieswhichtrademoretendtobemoredemocraticon average, but this relationship is weak and notsupported by a considerable number of individualcountry observations. These correlations also do notshow that particular forms of government are ashaping factorof tradeas theoppositecouldalsobetrue–tradeisashapingfactorofthechoiceofpoliticalsystems. Moreover, both trade openness and thechoice of a particular form of government could bedrivenbyathirdcommonfactor,suchasdevelopmentlevels. Economic analysis sheds some light on thedeterminantsofthisrelationship.

A number of studies argue that more democraticregimes tend to have more liberal trade policies. Themechanism occurs through several channels. Oneargument is that less democratic governments aremoreeasily “captured”byspecial interestgroups thatbenefitfromtheeconomicrentsassociatedwithtrade


194

FigureC.52:Form of government and trade openness, 1962-2010

-3

-2

-1

0

1

2

3

5

4

0

100

200

300

400

500

600

19

62

Ave

rage

Pol

ity S

core

Wor

ld to

tal t

rade

-GD

P ra

tio

19

64

19

66

19

68

19

70

19

72

19

74

19

76

19

78

19

80

19

82

19

84

19

86

19

88

19

90

19

92

19

94

19

96

19

98

20

00

20

02

20

04

20

06

20

10

20

08

Form of government index Trade openness

Sources:Authors’calculationbasedondatafromthePolityIVProject(http://www.systemicpeace.org/polity/polity4.htm),IMFandCOMTRADE.

Note:FormofgovernmentasproxiedbythePolityScoreindex.Itrangesfrom-10to10.Seetextformoredetails.

barriers. Another argument is that democraticgovernments are more likely to enter into tradeagreements to signal to voters their commitment toopenandstabletradepolicies.

Mansfield and Milner (2010) provide empiricalevidence that the probability of a country signing afree trade agreement increases with its level ofdemocracy. In addition, Mansfield et al. (2000) showthat pairs of democratic countries establish lowertrade barriers compared with pairs of countries thatincludeanautocracy.Yu(2010)considersdemocracyin a standard gravity model and finds evidenceconsistentwith thehypothesis that,onaverage,moredemocracy isassociatedwith increasedtrade.Finally,the empirical results in Eichengreen and Leblang(2008) confirm that the relationship between tradeandtheformofgovernmentrunsinbothdirections.

Arelatedissueishowthetransitionfromoneformofgovernment to another affects trade policy. FromFigureC.54, itappears that theempirical relationshipbetween trade policy and the form of government isnot linear: the countries with the lowest and highestpolityscoresonaverageapply lowertariffscomparedtocountrieswithan intermediatepolityscore.156Thisfact may suggest that the transition towards moredemocratic regimes could lead to an initial surge inprotectionism.

O’Rourke(2007)arguesthatthetransitionfrommoreautocratic to more democratic regimes implies atransferofpowerfromasmallrulingelitetothewiderpopulation. As a consequence, trade policies willchangeaccordingtothepreferencesofthemajority.Ina standard Heckscher-Ohlin framework, one would

expect that more open trade policies should beobserved in countries where the majority of workersgain from trade opening. Conversely, in countrieswhere workers stand to lose from trade opening, ademocratictransitionmightbeexpectedtoleadtoanincreaseintradebarriers.Theevidencefromasampleofdevelopedanddevelopingcountriesbetween1870and1914confirmsthisbasictheoreticalinsight.Whilethisfindingmighthelptoexplainwhyan intermediatelevel of democracy generates higher protectionism(see Figure C.54), it does not explain why furthermoves towards democracy result in lowerprotectionism. The transition to democracy has beenthesubjectofaheateddebateamongsocialscientistsinrecentyears.Thereissomeanecdotalevidencethatdemocratic reforms lead to an initial deterioration ofeconomicpolicyandresultinpooreconomicoutcomesandinstability(atleastintheshortrun).Theevidencein Rodrik and Wacziarg (2005), however, appears toreject the notion that nascent democraciessystematicallyunder-performmoreautocraticregimesandmoreestablisheddemocracies.157

Another possible explanation for the weak positivecorrelationbetweentradeandmoredemocraticformsof governments observed in Figures C.52 and C.53runs in the opposite direction. A number of studiesshowthattheeffectoftradeontheformofgovernmentis influenced by the changes in relative wealth andpower among social groups. Acemoglu and Robinson(2006)provideatheorythatexplainshowglobalizationaffects the transition to and consolidation of moredemocratic regimes. The mechanism through whichtrade influences the political regime is a change infactorpricestriggeredbytradeopening.Theyobservethatpoorercountriesaretypically lessdemocratic(or


195

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

more easily subject to authoritarian coups) andrelatively more abundant in labour. Trade openingleads to an increase in wages in poorer countries,leading to convergence in incomes and a decline in“class conflict”. This, in turn, acts on the politicalstructurebecauseasmallerincomegapbetweenrichand poor decreases the demand for highlyredistributivepolicies, thusmakingmoreparticipatoryformsofgovernmentlesscostlyfortheelites.

Puga and Trefler (2012) analyse the effects ofincreasing trade in medieval Venice. They argue thatlong-distancetradeallowedmerchants toaccumulatewealth and to impose constraints on the executive,eventually triggering a switch from a monarchy to amore liberal political system. Both studies, however,point out that the causal relationship between tradeand forms of government is ultimately a question ofdegree. In the case of medieval Venice, the class of

merchants that imposed constraints on the absolutepower of the executive later used their resources toblock political competition by demanding hereditaryparliamentary participation. The cross-countryevidenceontheimpactoftradeopennessontheformof government is not conclusive. Rigobon and Rodrik(2005)andMilnerandMukherjee(2009)findthattherelationship between trade openness and moredemocratic formsofgovernment iseithernegativeorweak, inparticular fordevelopingcountries.However,López-Córdova and Meissner (2008) find that, whilenorelationshipexistsintheshortrun,apositiveimpactof tradeonmore representative formsofgovernmentcanbedetectedinthelongrun.

A separate argument is that trade-related institutionsmaydecreasetheopportunityforrentseeking.LiuandOrnelas (2012) analyse the role played by preferentialtrade agreements (PTAs) in shaping domestic political

FigureC.53:Form of government and imports/exports, 2010

Imports

10

12

-10 -5 0 5

14

16

18

20

22

Log

of im

port

s

Polity Score of importer

10

Exports

10

12

-10 -5 0 5

14

16

18

20

22

Log

of e

xpor

ts

Polity Score of exporter

10

Developed countries Developing countries LDCs Linear fit


Source:Authors’calculationbasedondatafromthePolityIVProject(http://www.systemicpeace.org/polity/polity4.htm)andCOMTRADE.

Note:ImportsandexportsmeasuredasthenaturallogarithmofthevalueoftradeinthousandUS$.FormofgovernmentasproxiedbythePolityScoreindex.Itrangesfrom-10to10.Seetextformoredetails.


196

FigureC.54:Form of government and average tariffs, 2010

-10 -8 -6 2 4 6

Wei

ghte

d av

erag

e of

eff

ectiv

ely

appl

ied

tarif

fs

Polity Score

-4 -2 0

0

8 10

2

4

6

8

10

12

14

16

18

Source:Authors’calculationbasedondatafromthePolityIVProject(http://www.systemicpeace.org/polity/polity4.htm)andTRAINS.

Note:FormofgovernmentasproxiedbythePolityScoreindex.Itrangesfrom-10to10.Seetextformoredetails.

institutions through this mechanism. In particular, theyshowthatparticipationinPTAsworksasacommitmentwhichhelpsgovernments to resist futurepressures forprotectionism from lobby groups. Therefore, organizedgroupsseekingpoliticalpowerforeconomicgainhavealowerincentivetodosoifacountryis“tied”byaPTA.Inunstabledemocracies,thegovernmenthasaparticularlystrong incentive to sign PTAs and thus weaken theposition of organized groups trying to displace theincumbent. The authors find empirical support for apositive correlation between participation in PTAs andthelongevityofmorerepresentativeregimes.

(ii) Political borders

Internationalpoliticscanhavean impacton trade inanumber of ways. The breakdown of internationalrelationsbetweenthetwoworldwars,forexample,wasassociatedwithadramaticfallintradeflowsduringthisperiod.Likewise, tradeflowsandcommercial interestscanhelp toshape theevolutionof theworld’spoliticalmap, as shown by the age of European colonialismbeginning in the1500s (O’RourkeandFindlay,2007).The following sub-section focuses on how tradeinteracts with the Westphalian system, the politicalorder that was born in Europe with the Treaty ofWestphalia in 1648 and that gradually extended toencompass most of the world. At the core of theWestphalian system is the sovereign nation statedelimited by clearly defined political borders. Thequestion addressed here is “how does internationaltradeandnationalsovereigntyinteract?”

The number of sovereign countries has dramaticallyincreased over the last century, from 58 in 1904 to196 today, with most of this increase taking place

since the Second World War. One line of researchargues that political fragmentation and a significantexpansion in political borders increase transactioncosts,andthisnegativelyaffectstrade.Thebulkoftheliterature focuses on measurements of the “bordereffect”,whichisfoundtobesizeable.

In a ground-breaking paper, McCallum (1995)investigates the trade effect of the border betweenCanadaand theUnitedStatesusingstandardgravityequation techniques. Even though the two countriesshare a common language, similar legal systems andother characteristics that might render the borderseparatingthemasinconsequential,McCallum(1995)finds that theborder reduces tradebya factorof22.That is, trade between Canadian provinces isestimated to be 2,200 per cent higher than tradebetweenCanadianprovincesandUSstates.

Subsequent work by Anderson and van Wincoop(2003)findsthatthetradeeffectofpoliticalbordersissmaller than the finding of McCallum (1995) but stillsizeable.Specifically, theirestimatessuggestthattheborder separating Canada and the United Statesreduces trade by 44 per cent, while borders amongindustrialized countries more broadly have a negativeimpactontradeofabout30percent.Finally,arecentpaper by Redding and Sturm (2008) examinesGermany’sseparationintotwostatesaftertheSecondWorld War and its reunification in the 1990s todetermine the trade and development impact ofchangingborders.TheyfindthattheimpositionoftheEast-West border had a large negative impact oneconomic activity (for instance, as measured bypopulationgrowth)intownsclosesttothenewborderbyreducingmarketaccess.


197

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

Whilebordershaveanegativeimpactontrade,FigureC.55 shows a positive correlation between tradeopennessandthenumberofsovereigncountriesovertime (and,hence, thenumberofborders).Clearly, thestatistical volume of international trade increases bydefinition when a new sovereign nation is created (afraction of what was measured as internal tradebecomes internationalasa resultof thecreationofanewborder,asdiscussedinSectionB.2(a)).However,therelationshipbetweenthenumberofcountriesandtrade openness is still positive when the latter ismeasuredbytheleveloftheaveragetariff(Alesinaetal., 2000). This positive correlation suggests that anincreaseintradeitselfmayhaveanimpactonpoliticalbordersandthenumberofsovereigncountries.

Trade openness often involves a reshaping ofsovereignty,whilepoliticalbordersmayalsochangeinnature as well as in number. The increasing role ofregionalorganizations,suchastheEuropeanUnionortheAssociationofSoutheastAsianNations(ASEAN),is one example. Expanding membership in the WTO,and its enhanced role in international trade disputesettlement,isanotherexample.FiguresC.56andC.57providesomeinsightsintothis.

Figure C.56 shows a strong positive correlationbetween the number of sovereign countries and thenumber of preferential trade agreements (PTAs). Thelargesurgeinthenumberofsovereigncountriesinthepast 50 years appears to precede the formation ofnewPTAs.Furthermore,FigureC.57indicatesthatthe

nature of these agreements has changed over time,with deeper forms of agreements becoming moreprominent, particularly when countries engage inshared,cross-borderproduction.158

A limited economic literature helps to explain theseapparently conflicting facts (Alesina and Spolaore,2003; Ruta, 2005). The studies find that economicintegrationchangesthecostsandbenefitsofnationalsovereignty, releasing centrifugal and centripetalforces. On the one hand, trade openness promotespoliticalfragmentation. Inaworldoftraderestrictions(at their maximum in a world where countries do nottrade),largenationsenjoyeconomicbenefitsbecausepoliticalbordersdeterminethesizeofthemarketandthe extent of economies of scale. Economic gainscreate incentives for political integration. However,with more open trade, the extent of the market is nolonger restricted by political borders. The economicincentive for political integration wanes, and cultural,linguistic and ethnic groups within countries maychoose to form smaller more homogenous sovereignstates(Alesinaetal.,2000).

On the other hand, trade openness requires deeperforms of institutional integration which createcentripetal forces. Economic theory makes twocompelling arguments that substantiate this point.First, a number of authors argue that markets neednon-market institutions (political, legalandsocial) fortheir proper functioning (Casella, 1996; Padoa-Schioppa, 2001; Rodrik, 2000). These non-market

FigureC.55:Number of countries and trade, 1962-2012

110

120

130

140

150

160

170

210

180

0

100

200

300

400

500200

190

19

62

Num

ber o

f cou

ntrie

s

Tota

l tra

de-G

DP

ratio

19

64

19

66

19

68

19

70

19

72

19

74

19

76

19

78

19

80

19

82

19

84

19

86

19

88

19

90

19

92

19

94

19

96

19

98

20

00

20

02

20

04

20

06

20

10

20

08

20

12

Number of countries Trade openness

Sources:Authors’calculationbasedondatafromCOMTRADEandAlesinaetal.(2000).


198

institutions are essentially public goods that marketsfail to provide. Others make the point that tradeopennessincreasestheimpactoftradepolicyonothercountries, rendering unilateral decision-makinginefficientcomparedwithcooperativedecision-making(BronerandVentura,2011;EpifaniandGancia,2007;Brou and Ruta, 2011). The coexistence of competingcentripetal and centrifugal economic forcescontributestoexplainingthereshapingofsovereignty/politicalbordersdescribedabove.

Anexampleofthedualrelationshipbetweentradeandsovereigntyisthenewmomentumthatdeepeconomicagreements have gained since 1990. As the 2011WorldTradeReport(WTO,2011a)notes,thechangingnature of international trade (and, specifically, therisingimportanceofglobalsupplychains)isrelatedtotheriseofdeeperformsofintegration.Thefirstisbotha cause and a consequence of the latter. Theexpansion of production networks is driving theproliferation of deep agreements that aim at filling agovernance gap in areas, such as competition policy,investmentandproductregulation,whichareessentialfor the smooth functioning of these networks. Forthese same reasons, governments undertakecommitments in these policy domains that oftenimpose constraints on national sovereignty andeffectivelymakethepoliticalbordersmoreporous.Atthesametime,deepagreementsareashapingfactorof foreign investments flows and outsourcing as theinstitutional environment is a determinant of firms’economic decisions. This issue is discussed in moredetailinthenextsub-section.

(b) Economicinstitutions

Economic institutions – especially the quality ofregulations and the rule of law – provide a criticalstructure for economic interaction. According to theWorldwide Governance Indicators (2011), regulatoryquality “reflects perceptions of the ability of thegovernmenttoformulateandimplementsoundpoliciesandregulationsthatpermitandpromoteprivatesectordevelopment”. In the same way, the rule of law index“reflects perceptions of the extent to which agentshave confidence in and abide by the rules of society,and in particular the quality of contract enforcement,property rights, thepolice, and the courts, aswell asthelikelihoodofcrimeandviolence”.

Clear,stableandenforceablerulesarefundamentaltointernational trade relations as they limit uncertaintyby creating a framework within which economicexchangetakesplace.Moreover,economicinstitutionsmayshapetradeflowsby influencingthecomparativeadvantage of countries. The following sub-sectionattempts to uncover the determinants of therelationshipbetweentradeandeconomicinstitutions.

(i) Stronger institutions promote trade

How do economic institutions relate to internationaltrade?Theevolutionof theaverage levelsof the ruleoflawandregulatoryqualityacrossallcountriesdoesnotshowaclearpattern.Bothmeasuresofinstitutionalquality decreased in the last half of the 1990s andimproved in the following decade, returning

FigureC.56:Preferential trade agreements (PTAs) and number of sovereign countries, 1950-2011

0

50

100

300

150

250

200

Num

ber o

f agr

eem

ents

Num

ber o

f cou

ntrie

s

19

50

19

52

19

54

19

56

19

58

19

60

19

62

19

64

19

66

19

68

19

70

19

72

19

74

19

76

19

78

19

80

19

82

19

84

19

86

19

90

19

94

19

98

20

02

20

08

20

06

20

04

20

00

19

96

19

92

19

88

20

10

PTAsNumber of countries

80

100

120

140

160

180

200

Sources:Authors’calculationbasedondatafromAlesinaetal.(2000)andWTOSecretariat.


199

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

approximately to the 1990 level by 2010. Over thesametimeperiod,however,worldtradeincreased,withthe exception of the 2008-09 fall after the globalfinancial crisis. However, this lack of any obviouspositive relationship at the aggregate level can bedeceiving. Figure C.58 shows the pattern of exports,ruleof lawand regulatoryquality forbest-performing exporters, i.e. for the sample of countries thatregistered the highest increase in exports between1996 and 2010.159 The growth of exports in thesecountrieswasaccompanied,onaverage,bysignificantimprovements in economic institutions, as measuredby the two indices. Cross-country evidence for 2010reported in Figure C.59 also confirms this positiverelationship between trade openness, regulatoryqualityandruleoflaw.

Whyare thequalityofeconomic institutionsand tradepositively related? As already observed for politicalinstitutions and trade, the relationship runs in bothdirections.Propertyrights,efficientregulationsandtherule of law allow economic actors to establish a traderelationship in which rules and individual positions areclearlyunderstood.Theseinstitutionscreateincentivesto exchange goods and services as they reducetransactioncostsassociatedwithuncertaintyand lackof transparency. Available empirical evidence confirmstheimportanceofthischannel.

Examining the effect of corruption and imperfectcontract enforcement on trade, Anderson andMarcouiller (2002) find that improvements in the

qualityofinstitutionslowersthepriceoftradedgoodsand increases trade flows. They also find thatinstitutional quality can be an alternative explanationforwhyhigh-income,capital-abundantcountriestradedisproportionately with each other rather than withlow-income, labour-abundant countries. Indeed, theyargue that efficient economic institutions in high-income countries lower the transactions costs oftrading with each other relative to trade withdevelopingcountries.

Atthesametime,tradeopennesscanhaveanimpacton economic institutions in various and sometimesconflicting ways. A number of studies point out thateconomic institutionsare inter-linkedwithchanges intheeconomy.Contractenforcementandtheprotectionof property rights, for example, can depend on avarietyof factors, suchasgovernments’ incentives toact and economic actors’ incentives to respect therules. Changes in relative prices brought about byinternational trade are likely to influence theseincentivesandhenceshapeinstitutions(CopelandandTaylor,2009;Anderson,2008).

Consider a country with weak protection of propertyrightsthatisrelativelyabundantinforestryresources.Asthepriceofforestryproductsincreaseswithtradeopenness, poachers may be tempted to extract moreforest products but the government also has anincentivetobettermonitorandmanageanincreasinglyvaluable resource. Copeland and Taylor (2009) offervarious examples of how trade opening had either a

FigureC.57:Trade in intermediate goods and “depth” of preferential trade agreements, 1958-2011 (thousandsofUS$)

Depth Trade

0

1958-1982 1982-1986 1987-1991 1992-1996 1997-2001 2002-2006 2007-2011

0.1

0.2

0.3

0.4

0.5

0.6

1.0

0.7

0.9

0.8

Ave

rage

agr

eem

ents

' dep

th

Ave

rage

trad

e in

par

ts a

nd c

ompo

nent

s

400,000

350,000

300,000

250,000

200,000

150,000

100,000

50,000

0

Sources:WTOSecretariatandOreficeandRocha(2011).

Note:ThedepthofPTAsisconstructedconsideringthetenmostimportantprovisionsusingprincipalcomponentanalysis.SeeOreficeandRocha(2011)formoredetails.


200

positiveornegativeimpactontheeffectiveprotectionofpropertyrights.

Tradecanalsohaveanimpactoneconomicinstitutionsthrough more indirect channels. In the previous sub-section,itwasshownhowtradeopeningcaninfluencetherelativeeconomicpowerofdifferentsocialgroupsand thus formal political institutions that evolve. Aseconomic institutions are established and reformedthrough political processes, changes in politicalinstitutionsandorganizationswillclearlymatter(Greif,2006).160

Acemoglu et al. (2005) suggest that the effect oftradeoneconomicinstitutionsdepends,especially,oninitialpoliticalconditions.Theyargue that thevaryinggrowth patterns among European countries after1500areexplainedbyhowtheincreaseintransatlantictrade affected economic reforms. In countries withmoreopenpoliticalsystems,suchasGreatBritainandthe Netherlands, the increase in Atlantic tradestrengthened and enriched merchant groups whoobtainedimportantreforms,includingbetterprotectionof property rights. This, in turn, paved the way forsustained economic growth. Similar changes ineconomic institutions did not take place in countriesthathadweakerchecksandbalancesonthemonarchy,such as Spain and Portugal. This finding suggestsan important and complex interaction betweentrade openness, political institutions and economicreforms.161

(ii) Institutions create comparative advantage

Economic institutions can also be a source ofcomparativeadvantage.Countrieswithbetterinstitutionsspecializeinindustriesforwhichtheexistenceofstableand reliable institutions ismore important.FigureC.60showsapositivecross-countrycorrelationbetweentheruleoflawandtheshareofexportsinsectorswithhighinstitution-intensity. As discussed in more detail below,the indicator for institutional intensity measures theproportion of intermediate inputs that require relation-specificinvestmentsforeachindustry(Nunn,2007):themore complex the production process required to usetheseintermediateinputs,themoreitisreliantonstronginstitutions.

Economictheoryandempiricalevidenceconfirmthatacountry’s institutional framework, in addition to itstechnological level or relative factor abundance, is asource of comparative advantage. Recent literatureemphasizes in particular the role of cross-countrydifferencesincontractenforcement–andthereforeinthe degree of contract incompleteness – in shapingtradepatterns(Levchenko,2007;Nunn,2007).

Therelativeneedforcontract-dependentinputsvarieswidely across sectors. For example, the automotiveindustry is more institution-dependent than flourmilling.Infact,mostoftheintermediateinputsinvolvedinautomotiveproductionaredesignedforaparticularmodelandcannotbeusedbydifferentcarproducers.

FigureC.58:Exports and the quality of economic institutions in the best-performing exporters, 1996-2010

19.5

20

20.5

21

22 0.3

0.2

0.1

0

-0.2

-0.3

21.5

-0.1

Export (log) Rule of law Regulatory quality

19

96

19

98

20

00

20

02

20

03

20

04

20

05

20

06

20

07

20

08

20

09

20

10

Ave

rage

exp

orts

Ave

rage

leve

l of e

cono

mic

inst

itutio

ns

Sources:Authors’calculationbasedondatafromWGIandCOMTRADE.

Note:AverageexportsaremeasuredasthenaturallogarithmofthevalueofexportsinthousandsofUS$.Ruleoflawandregulatoryqualityareindicesrangingfrom-2.5to2.5.


201

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

In contrast, flour production requires mainly cerealsthat are exchanged and priced in uniform markets.Bettereconomic institutionsreducethe inefficienciesassociatedwithcontractincompleteness,whichinturnwillhaveadisproportionateimpactoncostsinsectorsthat requiremorecontract-dependent inputs, suchasthecarindustry.Oneimplicationisthatcountrieswithbettereconomic institutionsaremore likely tohaveacomparativeadvantageinthesesectors.162

A study by Nunn (2007) offers a convincing test ofwhether countries with stronger contract enforcementspecializeintheproductionofgoodsforwhichrelation-specific investments are most important.163 He showsthat the average contract intensity of exports at thecountry level is positively correlated with differentmeasures of the quality of economic institutions,including judicial efficiency and contract enforcement.At the country-industry level, countries with better

contract enforcement specialize in industries whererelationship-specificinvestmentsaremostimportant.

Thedevelopmentoffinancialinstitutionsalsohelpstodetermine trade patterns. Beck (2002) shows thateconomieswithmoredevelopedfinancialsectorshaveacomparativeadvantage inmanufacturing industries.Examining 65 countries over a 30-year period, heshowsthatfinancialdevelopmentexertsalargeimpacton the levelofbothexportsand the tradebalanceofmanufacturedgoods.

Svaleryd and Vlachos (2005) find that the financialsectorisasourceofcomparativeadvantageconsistentwith the Heckscher-Ohlin-Vanek model. Countrieswith well-functioning financial systems tend tospecialize in industries that are highly dependent onexternal financing. They find that differences infinancialsystemsaremore importantdeterminantsof

FigureC.59:Trade openness and institutional quality, 2010

Regulatory quality

Trad

e op

enne

ss



0

0.5

1

1.5

2

2.5

-2.5 -2 -1.5 -1 -0.5 0

Regulatory quality

21 1.50.5

Rule of law

Trad

e op

enne

ss

0

0.5

1

1.5

2

2.5

-2.5 -2 -1.5 -1 -0.5 0

Rule of law

21 1.50.5

Sources:Authors’calculationbasedondatafromWGIandCOMTRADE.

Note:Ruleoflawandregulatoryqualityareindicesrangingfrom-2.5to2.5.


202

FigureC.60:Rule of law and comparative advantage, 2010 (share)

0

0.05

0 0.4 0.6 0.7 0.8 0.9 1

0.1

0.15

0.25

0.35

Sha

re o

f exp

orts

in h

ighl

y in

stitu

tion-

depe

nden

t ind

ustr

ies

Rule of law

0.2

0.3

0.4

0.45

0.5

0.1 0.2 0.3 0.5

Source:Authors’calculationbasedondatafromWGIandNunn(2007).

Note:Ruleoflawisanindexrangingfrom-2.5to2.5.

specializationamongOECDcountriesthandifferencesin human capital. Weak financial institutions result inlarger transactioncostsandother“financial frictions”.These frictions also affect the volumes of trade bydistorting firms’ production decisions and entry intointernationalmarkets.

Manova (2008b) shows that countries with weakfinancial institutions export fewer varieties to fewerdestinationmarkets, thus registering loweraggregatetrade volumes. These distortions are amplified infinancially vulnerable sectors that need more outsidecapital and that have fewer assets that can becollateralized. Ferguson and Formai (2011) show thatcountries with more developed financial systemsexportdisproportionatelymoreinsectorsthatproducecomplex goods and that have a high propensity forverticalintegration.164

(iii) Institutions and the changing nature of trade

Another important issue is the association betweencountries’ economic institutions and trade policies(including regulations, protection of intellectualpropertyrights,andinvestment).FigureC.61plotsthecorrelationbetweentheruleoflawandaveragetariffsacrosscountriesin2010.Therelationshipisnegative,suggesting that countries that have better contractenforcementalso tend tohave lower tariffs.Asimilarnegativecorrelationcanbefoundbetweenthequalityoftheregulatorysystemandtariffs.

FigureC.62showstherelationshipbetweeneconomicinstitutions and deep preferential trade agreements.

The relationship appears to be less pronounced thanthecorrelationbetweenruleoflawandaveragetariffsbut the figure still shows a positive associationbetween countries’ international commitments andtheirdomesticenforcementcapacity.

What factors could explain this relationship betweentrade policies and economic institutions? Onecompelling argument is that it is shaped by thechangingnatureoftradeandthegrowing importanceofcross-borderproduction(seeSectionB.2(e)).

Theemergenceofopportunitiestoparticipateinglobalsupplychainslowersincentivestoimposetradebarriers.AsnotedbyBaldwin(2010b), ratherthanbuildingtheirown supply chains behind tariff walls over severaldecades,theICTrevolutionallowsdevelopingeconomiestosetupmanufacturingfacilitiesinamatterofmonthsby joining in supply chains. In this context, tariffs andother trade-related policies that were conceived topromoteimportsubstitutionbecomeoutdated.Domesticeconomic institutions, however, interact with thechangingnatureoftradeincomplexways.

First,whetherjoiningsupplychainscanbeasuccessfulstrategy for developingcountries crucially dependsonthe strength of domestic economic institutions. Thereason is that the quality of domestic institutionsdetermines in which country firms choose to offshore(Grossman and Helpman, 2005). In developingcountrieswithstrongercontractenforcement,therewillbe more investment and the costs of producingintermediateinputswillbelowerthanincountrieswithpoorinstitutions.Hence,tariffcuttingandparticipationindeeppreferential tradeagreements ismore likely to


203

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

characterize the first rather than the latter group ofcountries,inlinewiththediscussionabove.

Secondly,thequalityoftheinstitutionalframeworkisanimportant determinant of firms’ choice to integrate a

particularproductionstageortooutsourceit.Considerthecasewhereafirminanadvancedeconomyhastodecide whether to outsource or to integrate theproduction of an intermediate input in a developingcountry. If economic institutions in the developing

FigureC.61:Rule of law and average tariffs, 2010 (percentage)

Wei

ghte

d av

erag

e of

eff

ectiv

ely

appl

ied

tarif

fs

0

5

10

15

20

25

30

-2 -1.5 -1 -0.5 0 0.5 1.5 21

Rule of law


Sources:Authors’calculationbasedondatafromWGIandTRAINS.

Note:Ruleoflawisanindexrangingfrom-2.5to2.5.

FigureC.62:Rule of law and “deep” preferential trade agreements, 2010

-2 -1.5 -1 -0.5 0 0.5 1.5 21

Dep

th


Rule of law

0

0.2

0.4

0.6

0.8

1

1.4

1.2

Sources:Authors’calculationbasedondatafromWGIandWTOSecretariat.

Note:Ruleoflawisanindexrangingfrom-2.5to2.5.ThedepthofPTAsisconstructedconsideringthetenmostimportantprovisionsusingprincipalcomponentanalysis.SeeOreficeandRocha(2011)formoredetails.


204

country are strong, contracts between suppliers ofintermediate goods and the final good producer aremorelikelytobeenforced.Inadditiontoincreasingthelikelihoodofoffshoring,thisimpliesthatthestrengthofeconomicinstitutionsaffectstherelativeprevalenceofFDIorforeignoutsourcing(AntrasandHelpman,2004).

AsrecentevidencebyBernardetal.(2010)shows,betterqualityeconomicinstitutionsareassociatedwithahigherprobabilityofoffshoring.However,furtherstrengtheningof the institutional environment is associated with arelative decrease in FDI. As the authors argue, this ispresumablyrelatedtothegreatereasewithwhicharm’slengthcontractscanbewrittenandenforced.

(c) Culture

Inaddition to the formal institutionsdiscussedabove,informal institutions, such as social norms andconventions, impose constraints on and thereforeinfluence human interactions. The multiple forms ofcodesofbehaviourareoftencapturedinasingleword:culture. This sub-section highlights some basic factsabouthowdifferences incultureacrosscountriesarerelevant to international trade. Simply put, it asksif cultural differences are a shaping factor ofinternationaltrade.

Thequestionofhowinternationaltradepatternsrelatetoculturaldifferencesandhowthisrelationshipevolvesovertimehasnotbeenofinteresttoeconomistsonly.Inhis well-known study The Clash of Civilization and the

Remaking of World Order, political scientist Samuel P.Huntington writes: “In the emerging world, patterns oftrade will be decisively influenced by the patterns ofculture.Businessmenmakedealswithpeopletheycanunderstand and trust; states surrender sovereignty tointernational associations composed of like-mindedstatestheyunderstandandtrust.Therootsofeconomiccooperation are in cultural commonality” (Huntington,1996).Thekeyhypothesisisthatculturalidentitieswillbeamoreprominentdeterminantofthepatternoftradeandoftradeagreementsinthepost-ColdWarworld.

Definingandmeasuringcultureisnoteasy.Religionandlanguagehaveoftenbeenusedasaproxy forculture.However,eachofthesemeasureshassomedrawbacks.For instance, differences in religion as a measure ofculturaldifferenceshasbeencriticizedbecausereligionhasrelativelymorerecentrootsthanthelatter(Guisoetal., 2009). As a result, countries that have substantialcultural differences may share the same religion. Forthis reason, a numberof recent economic studiesusegenetic distance as a proxy for differences in cultureamong countries.165 Genetic distance measures thetime since two populations have shared commonancestors. The assumption is that populations thatshare more recent common ancestors have had lesstime to diverge in a wide range of traits andcharacteristics,suchasimplicitbeliefs,customs,habits,biases and conventions, which are transmitted acrossgenerations(SpolaoreandWacziarg,2009a;2009b).

FigureC.63showsthecorrelationbetweentotaltradeand cultural differences, as measured by genetic

FigureC.63:Total trade and cultural differences, 1980-2011

Log

of to

tal t

rade

0 500 1,000 1,500 2,000 2,500 3,5003,000

-5

0

5

10

15

20

Genetic distance

Country-pairs in 2011 Country-pairs in 1980 Linear fit (country-pairs in 1980)Linear fit (country-pairs in 2011)

Sources:Authors’calculationbasedondatafromSpolaoreandWacziarg(2009a)andCOMTRADE.

Note:Geneticdistanceisusedasaproxyforculturaldifferences.Itmeasureshowtwopopulationsaregenerallyrelatedtoeachother.Thevariableisanindexbetween0and10,000.Itisconstructedasaweightedaverageofaprobability(between0and1),weightedbytheproductofthesharesofthepopulationinpercentagepoints(between0and100)inanytwocountries.Seetextformoredetails.


205

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

distance.Eachpointinthefigurerepresentsacountry-pair.Theunconditionalcorrelationisnegative,meaningthat on average countries that are more distantculturally trade less with each other compared withcountries that share more similar cultural traits.Perhaps more surprisingly, it also shows that therelationship between trade and culture does not varymuchover time.To theextent thataslightdifferenceexists between international trade in 2011 and in1980,culturaldifferencesappearmorerelevanttodaythan30yearsago.

While Figure C.63 focuses on trade in final goods,Figure C.64 correlates cultural differences acrosscountries(measuredbygeneticdistance)andtradeinintermediate goods. Trade in intermediate goods is asimple(ifnotfullyaccurate)proxyfortherelevanceofcross-border production networks. Also in this case,the correlation is negative and relatively constantacross time, suggesting that cultural differencesrepresent a cost in the development of global supplychains.

Whydoculturaldifferencesappeartonegativelyaffecttrade? Economics provides two overlapping answers.The first is that differences in informal institutions,such as cultural traits, are an implicit barrier to tradeas theycreate transactionand informationcosts.Thelogical implication is that deeper cultural differenceshave a negative impact on trade. If this argument iscorrect,however,oneshouldalsoexpectthatnetworksof people with similar cultural traits, but located in

different countries, should trade more. The availableevidencesupportsthisconclusion.

Inparticular,RauchandTrindade(2002)examinehowethnic networks influence trade volumes. The studyfocuses on international transactions involvingChinesenetworks,thelargesttransnationalnetworkintheworld.TheauthorsfindthattheeffectofChinesenetworksispositiveforallgoods,andthatitisstrongerforbilateraltradeindifferentiatedproducts,forwhichinformation frictions are likely to represent a moreimportantbarrierrelativetoundifferentiatedgoods.166

Thesecond,andrelated,reasonwhyculturaldifferencesnegatively affect international trade is trust. Trust is acrucial component in determining economicrelationships, including trade relationships. Trust isparticularlyimportantinthosesocietieswhereinformalinstitutions, such as social norms, regulate economicexchanges between individuals. Guiso et al. (2009)provideevidence that trust isan importantcomponentin trade relationships.Theyshowthatculturalaspects,measuredbyreligious,geneticandphysicalsimilarities,andbythehistoryofconflicts,affectbilateraltrust(and,hence,trade)betweenEuropeancountries.

However,therelationshipbetweencultureandtradeisprobably more complex than simple cross-countryregressions may suggest. One reason is that over along period of time, trade may shape culturaldifferences.Forexample,trademayactasavehicletoincrease or establish trust between culturally diverse

FigureC.64:Trade in intermediate goods and cultural differences, 1980-2011

Log

of to

tal t

rade

0 500 1,000 1,500 2,000 2,500 3,5003,000

-5

0

5

10

15

20

Genetic distance

Country-pairs in 2011 Country-pairs in 1980 Linear fit (country-pairs in 1980)Linear fit (country-pairs in 2011)

Sources:Authors’calculationbasedondatafromSpolaoreandWacziarg(2009a)andCOMTRADE.

Note:Geneticdistanceisusedasaproxyforculturaldifferences.Itmeasureshowtwopopulationsaregenerallyrelatedtoeachother.Thevariableisanindexbetween0and10,000.Itisconstructedasaweightedaverageofaprobability(between0and1),weightedbytheproductofthesharesofthepopulationinpercentagepoints(between0and100)inanytwocountries.Seetextformoredetails.


206

FigureC.65:Cultural differences and “deep” preferential trade agreements, 1958-2011

0

0.2

0 200 400 600 800 1000 1200

0.4

0.6

1

1.4

1.6A

gree

men

ts' d

epth

Genetic distance

0.8

1.2

Sources:Author’scalculationbasedondatafromSpolaoreandWacziarg(2009a)andWTOSecretariat.

Note:Geneticdistanceisusedasaproxyforculturaldifferences.Itmeasureshowtwopopulationsaregenerallyrelatedtoeachother.Thevariableisanindexbetween0and10,000.Itisconstructedasaweightedaverageofaprobability(between0and1),weightedbytheproductofthesharesofthepopulationinpercentagepoints(between0and100)inanytwocountries.Seetextformoredetails.ThedepthofPTAsisconstructedconsideringthetenmostimportantprovisionsusingprincipalcomponentanalysis.SeeOreficeandRocha(2011)formoredetails.

agents (Tabellini, 2008).167 Moreover, culturaldifferences may not only work as a trading cost butalso as a trading advantage. Like formal institutions,theycanbeasourceofcomparativeadvantage.

Greif (1994) offers a theoretical framework in whichcultural factors determine institutional structures,which in turn have an impact on the economicdevelopmentandtradepatternsofdifferentsocieties.BycomparingtheMaghrebisandtheGenoeserolesinMediterraneantradeduringthe11thand12thcenturies,the author argues that collectivist and individualistculturalbeliefsshapedthe institutionalbackgroundinwhich the two traders’ groups operated, leading todifferentpatternsoftradeandeconomicsuccess.

OnepuzzlingaspectofFiguresC.63andC.64 is thecomparison of the relationship between culturaldifferencesandtradeinfinalandintermediategoods,respectively. To the extent that trade in intermediategoods captures trade in parts and components and,more generally, the importance of cross-borderproduction between pairs of countries, one wouldexpect that cultural differences matter more than intrade in final goods. Cultural differences are likely tobe associated with different formal institutions and,other things being equal, to discourage offshoring.However, while in both charts the relationship isnegative, trade in intermediate goods appears to beless – rather than more – affected by culturaldifferences. At first sight, this finding is at odds withbasiceconomicintuition.

Apossibleexplanationforwhyculturaldifferencesdonotappear toprovidea formidablebarrier to trade inintermediate goods is that states often cooperate toovercome these barriers. Put differently, formalinstitutionssuchasdeeppreferentialtradeagreementsmay“compensate” for the implicit tradecostscreatedby cultural distance and divergent domesticinstitutions.SomeevidencesupportingthispossibilityisprovidedinFigureC.65.

It indicates that deeper agreements are entered intobycountries that,onaverage,havedifferentcultures.Oneexplanationisthatculturallydistantcountriesareless likely to have similar formal institutions in areassuchas intellectualpropertyrightsor investmentthatareessential tothesuccessfuldevelopmentofcross-bordervaluechains.Forthesecountries,deepertradeagreementsserveasasubstituteforpoorordivergentdomestic institutions and may be a necessaryprerequisite for taking advantage of the gains frominternationalproduction.

An intriguing question is whether the Huntingtonhypothesis–whichpredictsthatculturaldiversitywouldmatter more in the post-Cold War era compared withtheColdWarperiod–issupportedbythedata.Arecentstudy offers an affirmative response. Gokmen (2012),using different measures of culture, including religion,ethnicity, language, civilization and genetic distance,findsthatcultureaffectedtrademorefollowingtheendof the Cold War. However, the observation that thedepth of economic agreements between countries is


207

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

positively correlated with their cultural distancecounters Huntington’s view that a shared culture is aprerequisiteforeconomiccooperation.

(d) Conclusion

This section makes two key points. First, theinstitutionalframework isan importantshapingfactorof trade, along with traditional factors such astechnology and endowments. Secondly, therelationshipbetweeninstitutionsandtradeiscomplexbecauseofthetwo-waynatureofhowoneinfluencestheother.

Institutions are a shaping factor of trade. Domesticpoliticalinstitutionsandtheinternationalpoliticalmapaffecthowtradeandtrade-relatedpoliciesaresetandnegotiated. Stronger economic institutions facilitateinternational commerce and influence trade patterns,as they representasourceofcomparativeadvantageand a determinant of firms’ offshoring decisions. Inaddition,differences inculturecancreatetransactioncosts that may limit commercial relationships. In thecoming years, institutions will continue to be animportant shaping factor of trade. This is becauseinstitutionsarelikelytoaffecttheflowsofintermediategoodsinglobalsupplychainsevenmorethanflowsoffinal goods. In light of this, governments are likely topay even more attention to reforming domestic andinternationalinstitutionsinthenearfutureasawayofreducing transaction costs, gaining comparativeadvantage in sectors with higher value-added, andlinkingtointernationalproductionnetworks

Institutions,however,arealsoshapedby internationaltrade.Economicintegrationisassociatedwithchangesindomesticpoliticalinstitutionsandwiththereshapingofsovereignty.Tradealsocreatesincentivestoimprovethe quality of economic institutions. The increasingimportanceofcross-bordersupplychainsisadriverofdeep preferential trade agreements, in part becausegovernments need to address the new cross-borderpolicy spillovers created by the internationalization ofproduction. Finally, trading relationships contribute tobuilding trust between diverse communities and arevectors for cultural influence. Institutions that shapehuman interactions tend to be more persistent thaneconomic forces, such as international trade. Thisinconsistency between the reach of markets and thereach of regulators must be a fundamental policyconcernincomingyears.

WhatdoesthiscomplexrelationshipbetweentradeandinstitutionsimplyfortheWTO?Ontheonehand,certainaspects of this relationship reinforce the multilateraltrading system. As stronger commercial ties createincentivestoadoptmoreefficienteconomicinstitutionsand reinforce trust and cooperation across countries,global trade policy-making may flourish. On the otherhand,thereareimportantreasonsforconcern.

First, while trade openness can encourage domesticpolitical and institutional reform, the transition mayinitially lead to a surge in protectionist incentives.Long-term policy commitments are needed to keepthese self-defeating temptations in check. Secondly,economicintegrationisreshapingsovereigntybutthecentral actors in existing global organizations remainnationstates.Thegrowingnumberofcountriesintheinternational system makes it more difficult tocooperate and to reach meaningful agreements. TheWTOalreadyprovidesaroleforregionalorganizations(notably the European Union as a WTO member) butthisrolecouldbefurtherpromoted.

Thirdly,weakeconomicinstitutionscanbeareasonforinefficientspecializationandtheinabilitytojoinglobalsupply chains, especially for developing countries.Deep preferential trade agreements can be aninstrumenttohelpovercomethesebarriers.Inaddition,aid programmes aimed at promoting trade shouldcontinue to focus on building institutionalinfrastructure. Finally, the proliferation of deeppreferential trade agreements is in part an efficientresponse to the changing nature of trade. However,theriskofsegmentingmarketscanbeanunintendedconsequence of these arrangements. Improvingcoherencebetweenmultilateralandpreferentialtradesystems is necessary to avoid discrimination amongtradingpartners.TheseissueswillbefurthertakenupinSectionE.

7. Conclusions

This section has looked at a number of factors –demography, investment, technology, naturalresources,transportationandinstitutions–andaskedhow each one is likely to shape the future ofinternational trade. These concluding reflections willexamine what implications they hold for individualcountriesorcountrygroups.

Developing countries

One of the biggest stories of the past 20 years hasbeen the successful integration of many developingcountriesintotheglobaleconomyandtheiremergenceas key players in international trade. Developingcountriesarediverseinthequalityoftheirpoliticalandeconomic institutionsbut therearestrong reasons tobelieve that “better” institutions give countries acompetitive advantage and produce better tradeoutcomes.Itisnotclear,however,whetherdevelopingcountrygrowthwillcontinueatthesamerapidpaceortaper off. Improving the quality of institutions wouldprovide developing countries with a way of ensuringthatgrowthcontinues.

ArapidlyageingpopulationmeansthatakeysourceofChina’s comparative advantage – its workforce –which fuelled its rapid economic rise could diminish.


208

At the same time, China is undergoing a process ofrapid capital accumulation. If this momentumcontinues, China could shift the source of itscomparative advantage in the direction of morecapital-intensive exports. A similar process is underway in other developing countries, such as Chile andTurkey,whichhaveseenrapidcapitalaccumulation inrecentyearsandrisingaggregatecapital-labourratios.It is not clear whether the impetus behind China’sdemand fornatural resourceswill recedeor intensify.Given the likely moderation in China’s future growth,there is reason to think that the need for naturalresources will dissipate. At the same time, however,China faces a growing scarcity of certain keyresources, especially water, that is unlikely to abateanytimesoon.

India and countries in the Middle East and Sub-Saharan Africa will continue to see their populationsgrow.Notonlywillmedianagesbelowbutdependencyratios will decline over the next decades. If thesecountriescanstrengthenpublic institutionsandkeepeconomicpoliciesoutward-looking,theycouldbecometheworld’sfastest-growingeconomies.Thehighratesof population growth in the Middle East and Sub-Saharan Africa also offer these countries thepossibility of reducing their dependence on naturalresource exports. However, this in turn depends onsuccessfully providing their growing populations withthenecessaryskills.

InthecaseofmanycountriesinSub-SaharanAfrica,italso requires efforts to reduce the “distance” tomarkets.CountriesinSub-SaharanAfricatendtofareworst on various measures of transportation costs.Sincereducingtransportationcostsanddeliverytimesarepreconditions forsuccessfully integrating into theglobal economy and global supply chains, increasinginvestment in transportation-related infrastructure iscritical.Thiswillinvolveharnessingdomesticresources(public and private) and using the Aid for Tradeinitiative as the lynchpin to mobilize internationalassistance.Theremaybesomescope tousecurrentregional integration efforts to identify and prioritizeregional infrastructureprojects thatcan reduce tradecostsandtousethecurrentDohaRoundnegotiationson trade facilitation as a way of improving customsprocedures and other regulations. African countriescontinue to lagbehind in innovationand inabsorbingtechnology transfers.These toocanbeaddressedbyimprovingthequalityofeducationandtraining.

Developed countries

In recent decades, developed countries have grownmoreslowly thandevelopingcountriesandhaveseentheir share of world trade shrink. The ongoing GreatRecession is likelytoproducea lostdecadeformanyadvancedcountries,particularlythoseinEurope.

Demographics in theformofanageinganddecliningpopulation will confront Japan and a number ofEuropeancountrieswithstrongheadwinds togrowth.Thiswillhaveadverseeffectsontheirfutureshareofglobal trade exacerbating the trends outlined inSection B. Greater openness to migration may helpalleviate these demographic challenges. Thesecountries will need to maintain a highly skilledworkforce,investahighshareofGDPonresearchanddevelopment and promote innovation. Newtechnologies such as robotics and 3D printing maybecome more widespread. Their adoption is likely tovary significantly across sectors: currently they areused, for instance, in construction, aerospace,jewellery and healthcare. More importantly, thesetechnologiesarelikelytoprovedisruptivebyreducingthe importance of low labour costs (provided bydeveloping countries). This can lead manufacturingand its associated supply chains to return (“in-sourcing”)todevelopedcountries.

Compared to Japan and Europe, the United Statesdoes not face as serious a set of demographicchallenges. Its population is expected to continue togrow (although at a slower rate) and it remains moreopentoimmigrants,whonowmakeupaboutasixthofitsworkforce.Immigrantsareparticularlyimportantinagricultureandinformationtechnology,sectorswherethe United States is an export powerhouse.Nevertheless, theUnitedStateswill need toupgradeitsinfrastructureandinvestinitsworkerssotheycancontinue to provide innovation and entrepreneurship.Dependent on oil imports for decades, technologicalimprovements in natural gas extraction now promisetheUnitedStatesenergyself-sufficiencyinthefuture.Sincethiswillleadtolowerenergycosts,itislikelytocould give a substantial competitive boost to UnitedStates’manufacturingexports.

From “fundamentals” to other shaping factors

This assessment of the key factors shaping trade –and how they will play out among various countriesand regions – is incomplete. It does not take intoaccount how trade affects income distribution, alterstherelativepowerofnationsorcreatesspillovers(e.g.environmentaldegradation)thatcertaincountriesmayfind unacceptable. These effects can weaken publicsupportfortradeopennessorpromptgovernmentstoadopt policies that directly or indirectly have anadverse impact on trade. We turn to these issues inSectionDofthisreport.


209

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

1 Therearevariouscausesforthefallinfertility.Technologicalprogressandtheaccumulationofhumanandphysicalcapitalmakelabourmoreproductiveandincreasetheopportunitycostsofraisingchildren(GalorandWeil,1996).Moreover,raisingincomeshiftsthecompositionofdemandtowardsnon-agriculturalgoodsandservices,whicharerelativelymoreintensiveinskilledlabour.Therelatedriseinthereturntoeducationleadstoincreasedinvestmentineducation,furtherincreasingtheopportunitycostofraisingchildren.Furthermore,parentswithhigherincomedevotemoreresourcestoeachchild.Sincethisraisesthecostofeachchild,italsoleadstofewerchildren(Becker,1981).

2 Theyoung-age(old-age)dependencyratioisdefinedastheratioofindividualsbelow15(above65)toworkingagepopulation(15-65).Theoveralldependencyratioisthesumoftheyoung-andold-agedependencyratios.

3 Theslow-downinpopulationgrowthinChinaispartlyduetotheone-childpolicy,introducedin1979.Thispolicyhasalsocontributed,invaryingdegrees,toadeclineinfertility,anincreaseinthesexratio(definedastheproportionofmaletofemalelivebirths)from1.06in1979to1.19in2010(WorldBankData,GenderStatistics)andanincreaseintheold-agedependencyratio(Heskethetal.,2005).

4 TheWorldBank(2012)estimatesthatcloseto42millionjobswillhavetobegeneratedgloballyby2020tocopewiththegrowthinthenumberofolderpeople.One-quarterofthesewillhavetobegeneratedinChina,butbythenthesizeoftheChineselabourforcewillhavestartedtodeclineinabsoluteterms.

5 SeeSectionC.1(b)foranaccountofcurrentandprojectededucationtrendsinselectedcountriesandregions.

6 Theautarkyrelativepriceisthepriceofthecapital-intensivegoodrelativetothepriceofthelabour-intensivegoodthatwouldbeobservedinacounterfactualsituationofnotrade.

7 SeeSectionB.2foranexplanationoftheHeckscher-Ohlintrademodel.ThestudiesbySayan(2005)andNaitoandZhao(2009)alsolookatthedistributionofgainsacrossgenerations.InSayan’s(2005)model,tradebasedondifferencesduetounequalpopulationdynamicsdoesnotnecessarilyleadtowelfaregainsforbothcountries.NaitoandZhao(2009)showthattheoldgenerationintheageingcountry(acountrywithdecliningfertilityrate)gainsfromtrade,butsubsequentgenerationsloseduringthetransitionphase.Acompensationschemeconsistingofcountry-specificlump-sumtaxes(transfers)andsavingssubsidies(taxes)can,however,makefreetradeParetosuperiortoautarky.AnotherpaperonthedifferentialeffectsoftradeliberalizationonoldandyoungwrittenbyGokcekusandTower(1998)arguesthatretireesarecapitalowners.AccordingtotheStolper-Samuelsontheoremtheywillfavourtradeopeningifthecountryhasacomparativeadvantageincapital-intensivegoods(suchastheUnitedStates).

8 Standardnationalaccountingshowsthatcapitalflowsandthetradebalancearecloselyrelated.Thecurrentaccountdeficit–theexcessofpayments(M)totherestoftheworldforgoods,services,investmentincome,andunilateraltransfersoverreceipts(X)fromtherestoftheworldforsimilaritems–equals(apartfrommeasurementerrors,whichmaybesubstantial)theexcessofaggregate

expenditure(E)forgoodsandservicesovernationaloutput(Y).Thelatter,inturn,equalstheexcessofinvestment(I)overaggregatesavingsS(thesumofpublicsavingsSG=T–GandprivatesavingsSP).Insymbols:M–X=E–Y=I–S.Thus,acurrentaccountdeficitimpliesanexcessofinvestmentoversavings.Foratextbooktreatment,seeMankiw(2010).OntheinfluenceofdemographicfactorsonlargeandpersistentUnitedStates’tradedeficits,seeCooper(2008)andFerrero(2010).

9 Anothermechanismatworkisthatinageingcountries,dueto“capitaldeepening”(highercapital-labourratios),capitalbecomesmoreproductive,producingcapitalinflows.

10 Helliwell(2004)arguesthat,althoughthereissomeevidencethatcountrieswithhighdependencyratiostendtoimportmorecapitalandruncurrentaccountdeficits,theseeffectsarestrongerfornon-OECDthanforOECDcountries.Inthelatter,hearguesthattheshareofthepopulationaged65ormorehasnostatisticallysignificanteffectonthecurrentaccountbalance.

11 Thelife-cyclehypothesiswasfirstadvancedbyModiglianiandBrumberg(1954).ItiscloselyrelatedtothepermanentincomehypothesisofFriedman(1957),whichpositsthattransitoryincomefluctuationsdonotaffectconsumption,becausethelatterisonlyinfluencedbychangesinpermanentincome.IntheHall(1978)formulationofthelife-cyclepermanentincometheory,individualschoosetheirconsumptionpatternsoastokeeptheexpected(discounted)marginalutilityofconsumptionconstantovertime.

12 AsunderlinedbyAttanasio(1999),theimportanceoftheprecautionarymotiveisultimatelyanempiricalquestion,depending,amongotherfactors,ontheavailabilityofsafetynetsandonthecharacteristicsofindividuals’preferencesandincome.

13 Theimportanceofliquidity/borrowingconstraintsismainlydocumentedindevelopingcountries.

14 InthecaseofFrance,Desvauxetal.(2010)estimatethatby2030maturehouseholds(aged65orabove)andprime-earninghouseholds(aged40to59)willaccountforabout70percentoftotalconsumption.Familyfragmentationisalsoprojectedtoincrease:averagehouseholdsize,equalto2.8in1980,willdeclineto2by2030.

15 Thedropinconsumptionfollowingretirementisa“puzzle”becauseitseemstocontradicttheconsumption-smoothingpredictionofthelifecycle-permanentincomehypothesis.Theempiricalevidence,however,ismixed.UsingUnitedStatespaneldataovertheperiod1980-2000,Aguilaatal.(2011),forinstance,findnoevidenceofadeclineinoverallconsumptionatretirement.InthecaseofItaly,Miniacietal.(2003)presentevidencethatwork-relatedexpensesfallafterretirement,butnon-durableconsumptiondoesnotfall.TheyconcludethattheretirementconsumptionpuzzleisabsentintheItaliancontext.Hurst(2008)arguesthattheobservedevidenceisnotinconsistentwiththelifecyclemodel,oncethisisextendedtoallowforhomeproduction(seealsoLührmann,2010)andhealthshocks(seealsoBanksetal.,1998).

Endnotes


210

16 FollowingMilanovicandYitzhaki(2002),WorldBank(2007)usesanabsolutedefinitionofmiddleclass,whichincludesindividualsearninganannualpercapitaincomebetweentheaverageofBrazilandthatofItaly(i.e.betweenaboutUS$4,000andaboutUS$17,000peryear,in2000PPP).Otherstudiesusewiderincomeintervalstodefinetheglobalmiddleclass,obtaininghigherprojectionsforitssize,bothinabsolutenumbersandasashareoftheglobalpopulation(KharasandGertz,2010;WilsonandDragusanu,2008).

17 Chinaisalreadytheworld’slargestautomarket,with13.6millionvehiclessoldin2009,comparedwiththe10.4millionsoldintheUnitedStates.Moreover,thecountryisalsotheworld’sfirstcellphonemarket,withapproximately700millionsubscribersin2010(KharasandGertz,2010).

18 KharasandGertz(2010)estimatethatin2010Chinesehouseholds’finalconsumptionaccountedfor37percentoftotaloutput,belowtheglobalaverage(61percent)andthepercentageobservedforotheremergingeconomiessuchasVietNam(66percent),Indonesia(63percent),India(53percent)andThailand(51percent).However,Atsmonetal.(2012)highlightthatinrecentyearsChinesehouseholds’consumptionincreased,alsothankstothemeasuresincludedinthecountry’slatestfive-yearplan,whichenhancedsocialsecurityandfinancialintegration.

19 AswillbedetailedinSectionD.1,anotherrecenttrendwhichcharacterizesbothmoredevelopedandlessdevelopedareasistheincreaseininequalitywithincountries.Thistrendisalsoinfluencingdemandpatterns,contributingtotheluxurymarket’sgrowthinmanyeconomies,includingChina(Atsmonetal.,2012;KharasandGertz,2010).

20 Thereare,however,anumberofpolicybarrierstotradeineducation,intheformbothofquantitativerestrictionsonthenumberofforeignsuppliersandofproceduralrequirementsrelatedtorecognitionofqualifications(LimandSaner,2011).

21 AscanbeseeninFigureC.7,fordevelopingcountriestheincreaseineducationlevelswasmainlyachievedthoroughincreasesinprimaryenrolmentrates,whilefordevelopedcountriesitwasachievedmainlythroughincreasesatthesecondaryandtertiarylevels.

22 KCetal.(2010)providepopulationprojectionsbylevelofeducation,ageandsexfor120countries,fortheperiod2005-50.Startingfrombaselinecountry-levelsurveydatafortheyear2000,theyproduceeducationprojectionsforfourdifferentscenarios,amongwhichthemostrealisticoneassumesthatcountries’educationwillevolveaccordingtoaglobalupwardtrend.

23 Inparticular,Romalis(2004)arguesthatcountriesthatareabundantinskilledlabourandcapitaldocapturelargersharesofUSimportsinindustriesthatintensivelyusethosefactors.Inasimilarvein,Chor(2010)showsthatcountrieswhicharemoreskillabundantexhibithighervolumesofbilateralexportsinmoreskill-intensiveindustries.Finally,Kowalski(2011)findsthat,togetherwithphysicalcapitalendowments,thelengthofschooling(aproxyforhumancapitalendowment)isamongthemostimportantvariablesexplainingindustrypatternsoftradeflows.AccordingtoKowalski’sestimates,astandarddeviationincreaseinyearsofschoolingresultsonaverageinabout14-17percentincreaseinexports.Healsoclaimsthatsecondaryandtertiaryeducationhavedifferentimpactsontradepatterns,withcross-countrydifferencesinsecondaryschoolingbeingamoreimportantexplanationofindustrytradeflows.

24 InCostinot’s(2009)model,thequalityofinstitutionscomplementshumancapitalindeterminingcomparativeadvantage.Duetothiscomplementarity,improvementsininstitutionshavelargereffectsincountrieswithmoreeducatedworkers.Similarly,improvementsineducationhavelargereffectsincountrieswithbetterinstitutions.SeeSectionC.6forfurtherdiscussion.

25 Thisisbecauseintheformercountryefficientorganizationofproductionrequiresthematchingofworkerswithsimilartalent.Conversely,inthelattercountryitrequireshiringone,orfewverytalentedindividuals,whoarecomplementedbyseverallessertalentedindividuals.

26 OtherpapersthatdeveloptheideathatworkerheterogeneitymattersforcomparativeadvantageareGrossman(2004)andOhnsorgeandTrefler(2007).Seealsothediscussionon“lumpiness”inSectionC.1(c).

27 AmitiandFreund(2010),forinstance,documentthatsince1992China’sexportsweresubstantiallyreallocatedawayfromapparel,textiles,footwear,andmiscellaneousmanufacturing(includingtoys)andtowardelectricalmachinery,officemachines(whichincludescomputers),andtelecommunications.Thesearepreciselythesectorsthatrelymostheavilyonprocessingtrade.

28 Thelabourforceparticipationrateisdefinedastheratiobetweenlabourforce(employedandunemployedactivelylookingforajob)andpopulationagedover15years.

29 Thefractionofthelabourforcecomposedofwomenincreasedineverycountrywheresignificantreductionsinfertilitywereobserved(SoaresandFalcão,2008).

30 Inthestandardmodeloflaboursupply,ahigherwagerateinducestwoeffectsonlabourmarketparticipation:asubstitutioneffect(theopportunitycostofleisureincreases,thereforeindividualsworkmoreandreduceleisure)andanincomeeffect(higherincomeopportunitiesincreasethedemandforleisure,inducingindividualstoworkless).SeeBlundellandMaCurdy(1999)forareviewofthelaboursupplyliterature.

31 However,asunderlinedbyKlasenandPieters(2012),theU-shapedhypothesishasbeendocumentedmainlybycross-sectionalanalyses,whilestudiesusingpaneldatafindmoremixedresults(seeforinstanceGaddisandKlasen,2011).Itshouldalsobementionedthatinmanydevelopingcountries,especiallyinAsianones,womenincreasetheirparticipationinthelabourmarketinresponsetoadverseeconomicshocks.Thisformofwomen’slabourforceparticipationmaycreatepovertytraps(BhalotraandUmaña-Aponte,2010).

32 TheUnitedStates(notshowninFigureC.9)followsasimilarpatterntotheEuropeanUnion.

33 KlasenandPieters(2012)forinstance,showthathusbands’higherincomereducesfemaleLFPRsinIndia.

34 TheMiddleEastischaracterizedbylowlevelsofLFPRsalsoformales.Indeed,increasinglabourforceparticipationisrecognizedasaprioritybymanygovernmentsintheregion(ILO,2012).

35 Thereisalsosomeevidenceofapositivecorrelationbetweenwomen’sshareinemploymentandaggregateexportsfordevelopingcountriessuchasMauritius,Mexico,Peru,thePhilippinesandSriLanka(Nordås,2003).

36 Comparativeadvantageismeasuredwithrevealedcomparativeadvantageindices.

37 SeeMorrissonandJütting(2005)foranempiricalcontributionusingmeasuresofdiscriminationbasedoninstitutionalconstraints.


211

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

38 Inparticular,theUnitedStateshosted42.8millionmigrants(20percent),followedbytheRussianFederation(12.3million,5.7percent),Germany(10.8million,5percent),theKingdomofSaudiArabiaandCanada(3.4percenteach),France,theUnitedKingdomandSpain(3percenteach),IndiaandUkraine(2.5percenteach).

39 Theaverageshareofinternationalmigrantsinthepopulationofthetenlargesthostswas13.2percentin2010.Inthesameyear,amongthosewithmorethan1millioninhabitants,thehighestproportionofinternationalmigrantswerefoundinQatar(87percentofthepopulation),theUnitedArabEmiratesandtheStateofKuwait(about70percent),JordanandPalestine(about45percent),Singapore,IsraelandHongKong,China(about40percent).

40 InAustralia,theyearlyaverageinflowofmigrantsincreasedfrom154,000between1980and1989to318,000between2000and2008.InCanada,itincreasedfrom126,000between1980and1989to241,000between2000and2009.IntheUnitedStates,itincreasedfrom633,000between1980and1989to1millionbetween2000and2010.Thesedata,aswellasotherdataonmigrantinflowsreportedinthissection,arefromtheUnitedNationsPopulationDivision,InternationalMigrationFlowstoandfromSelectedCountries:The2010Revision,database.

41 Forinstance,theUnitedStates’ImmigrationReformandControlAct(IRCA)of1986acceleratedthecountry’simmigrationflows,regularizing2.7millionimmigrantsbetween1989and1994(UnitedNations,2011a).

42 Forinstance,between2005and2007,inflowsfromPolandtoGermanyregisteredannualnetgainsof43,000arrivals,amountingtoanannualaverageof146,000.

43 TheWorldBankGlobalBilateralMigrationDatabase(GBMD)onlyincludesdataupto2000.Bilateralmigrantstocksin2010areusedbyWorldBank(2011c).However,thesedata,asinanupdateofRathaandShaw(2007),includeasmallersetofcountry-pairswithrespecttotheGBMDdata,andthereforearenotpreciselycomparabletothelatter.Nonetheless,thecalculationofintra-regionalsharesofmigrantstocksfor2010indicatesthattheshareofintra-regionalmigrantsdeclinedfrom2000to2010inAsia,EuropeandtheMiddleEastwhileitincreasedintheotherregions.

44 Otherdeterminantsofcross-bordermigrationareseasonalweatherpatterns,conflictsandnaturaldisasters(RathaandShaw,2007).Concerningthedeterminantsofinternalmigration,earlierstudiesfocusedontheroleofgeographicalincomedifferentialsindetermininginternalmigration(HarrisandTodaro,1970;Todaro,1969).Differently,theso-called“NewEconomicsofLaborMigration”(NELM)underlinedtheroleofmigrationasastrategyundertakenbyhouseholdsinpoorcountriestodiversifyandthusreducerisk(KatzandStark,1986;LucasandStark,1985;RosenzweigandStark,1989;StarkandLevhari,1982).Hoddinott(1994)generalizestheTodaroandNELMapproachesandprovidesevidenceontheimportanceofbothindividual-andhousehold-leveldeterminants.Foracomprehensivereviewoftheinternalmigrationliterature,seeTaylorandMartin(2001).

45 OnRCPsinAfrica,seeIOM(2011).

46 Inparticular,AfricaisthemainsourceregionforFrance(withashareof43percentofthetotalinflowofimmigrantsin2008)andtheCommonwealthisthemainsourceregionfortheUnitedKingdom(withashareof34percentofthetotalinflowofimmigrantsin2009).

47 FigureC.11showstheaveragefordevelopedcountries.Intheperiod2005-10,netmigrationincountrieslikeItaly,PortugalandJapanmorethandoubledthecontributionofnaturalincrease(birthsminusdeaths)topopulationgrowth.Inafurther29countriesorareas,netmigrationcounterbalancedtotallytheexcessofdeathsoverbirths(UnitedNations,2011b).

48 Theliteratureonmigrationandfertilityoffersfourbroadhypothesestoexplaintheobservedpatterns.Thesocializationhypothesis,emphasizingthedifferencesinfertilitybetweenmigrantsandnativesatdestination,positsthat,onceatdestination,migrantsmaintainthefertilitynormstowhichtheywere“socialized”duringtheirchildhood.Studiesmaintainingtheadaptationhypothesisstressthatmigrants’fertility,eventhoughitcandifferfromtheoneofnativesatdestination,tendstoconvergetothatofnativesovertime.Accordingtootheranalyses,however,thesimilaritiesbetweenmigrants’andnatives’fertilitylevelsobservedinsomecontextsarenotduetoadaptation,butinsteadtheyarerelatedtotheselectionofmigrants,i.e.tothefactthatmigrantsareanon-randomsampleofthepopulationatorigin,characterizedbyfertilitylevelsdifferentthanthoseofothernativesatorigin.Finally,accordingtothedisruptionhypothesis,thereductioninfertilityobservedforsomemigrantsatdestinationismainlyduetotheeconomicandpsychologicalcostsassociatedwithrelocation.Dependingonthecontextofanalysisandonthemethodologyused,eachofthesehypothesesfindssomesupportintheliterature,withmorerecentanalysesprovidingrelativelymoresupporttotheadaptationhypothesis.Foracomprehensivereview,seeKulu(2005).

49 Eurostat,Migrationandmigrantpopulationstatisticsdatabase.

50 Atthegloballevel,inspectionoftheUnitedNationsPopulationDivision,WorldMigrantStockdatabaserevealsthat,relativetothetotalpopulation,theyoungareunderrepresentedamonginternationalmigrants,whilethoseofworkingageandoverage65areoverrepresented.

51 TheprojectionsinTableC.2arebasedonthe2008RevisionoftheUnitedNationsPopulationDivision’sWorldPopulationProspects.ThefiguresondependencyratioshouldnotbecomparedwiththeonesinFigureC.4,whicharefromthe2010Revision.

52 Theoverallstabilityofskilledemigrationratesisconfirmedifalongertimeperiodisconsidered.Inparticular,Defoort(2008)analysesemigrationratestoasubsetofsixOECDdestinations(UnitedStates,Canada,Australia,Germany,UnitedKingdomandFrance)foreachfive-yearperiodbetween1975and2000.Theauthorshowsthatoverallemigrationratesarestableovertheperiod,buttheyincreasedincertainregions(especiallyinSub-SaharanAfricaandCentralAmerica)anddecreasedinothers(mainlyintheCaribbeanandNorthernAfrica).Interestingly,inspectionoftheDocquieretal.(2009)datasetrevealsthattheemigrationrateishigheramonghigh-skilledwomenthanamonghigh-skilledmenby17percentonaverage.

53 DatafromtheDocquieretal.(2009)dataset.Beineetal.(2007)pointoutthat,withoutcontrollingforageofentry,high-skillemigrationratesarelikelytobeoverestimated.Thisisbecauseonecouldcountashigh-skillemigrantevenindividualswhomovedalreadyaschildrenandacquiredtheireducationatdestination.However,theirestimatescorrectedforageofentryarehighlycorrelatedwiththeuncorrectedones.


212

54 ThisandthefollowingparagraphdrawextensivelyonthetextbookexpositionofFeenstraandTaylor(2008).

55 Theassumptionofconstantrelativepriceimpliesthatboththesendingandthereceivingcountriesare“small”.

56 Noticethatconstantcapital-labourratiosimplyconstantfactorprices.Therefore,inthelongrun,ashockinfactorendowmentisfullyabsorbedbychangesinthecompositionofoutputthatgoinoppositedirectionsinthetwosectors.Thisisdifferentfromtheshortrun,wherechangesinthecompositionofoutputgointhesamedirectioninthetwosectorsandthereisafallinthereturntothefactorwhoserelativeabundanceincreases(inthecaseoflabourmigration,thewageratefalls).Theeffectofmigrationonwagesisthemostresearchedtopicinthemigrationliterature.AreviewcanbefoundinHanson(2009).

57 BeverelliandGroppo(2013)analysetherelationshipbetweenskilledimmigrationandthestructureoftradeinskill-intensivesectorsinOECDeconomies.Preliminaryresultsindicatethat–controllingfortherelativeendowmentofskillednativesandcapital–countriesthatarerelativelymoreendowedwithskilledimmigrantscaptureahighershareofworldtradeinskill-intensivesectors.

58 Thisisbecauseoftheassumptionsofsymmetricdemandandtechnologybetweencountries.

59 GastonandNelson(2013)discussvariousothercasesinwhichthereiscomplementaritybetweenmigrationandtrade.Theysuggestthatintra-industrytrade,tradeinthepresenceofeconomiesofscaleandinthepresenceofinternationaldifferencesinthedegreeofimperfectcompetitionallgiverisetosuchcomplementarity.

60 WhileWong(1988)estimatesanindirecttradeutilityfunction,HijzenandWright(2010)treatimportsandimmigrantsasintermediateinputstofinaloutput.Tofindwhetherimmigrantsarequantitycomplementsorsubstitutewithtrade,theyestimate“Rybczynskielasticities”,namelythepercentagechangeinthedemandforimportsduetoapercentagechangeinimmigrants.

61 Thetwochannelscanalsobedenotedas“businessandsocialnetworkeffect”andthe“transplantedhome-biaseffect”(Brattietal.,2012).

62 AreviewcanbefoundinBrattietal.(2012).Theliteratureevolvedfromcross-countrystudiestopanel-dataonesandtorecentcontributionstryingtoestablishacausaleffectofimmigrationontrade.Notonlypermanent,buttemporarymigrationhasalsobeenshowntomatter(JansenandPiermartini,2009).

63 ThelargereffectofmigrantnetworksondifferentiatedratherthanonhomogeneousgoodsfoundbyAleksynskaandPeri(2012)andseveralotherstudiesisinlinewithRauch’s(1999)hypothesisthattrade-relevantinformationconveyedbymigrantnetworksisespeciallyrelevantondifferentiatedgoods.

64 Empiricalevidenceindeedshowsthatinmarketswithanincreaseinlesseducatedimmigrantsthereisalargeproportionofsectorswithahigherintensityofunskilledworkers(CardandLewis,2007)andasloweradoptionofskill-intensivetechniques(Lewis,2005).

65 KerrandLincoln(2010).Theyalsoshowthatin2000,47percentofthePhD-holdersworkinginscienceandtechnologyintheUnitedStateswereforeign-born.

66 Chellarajetal.(2008)findthatlargerenrollmentsofinternationalgraduatestudents,asaproportionoftotalgraduatestudents,resultinasignificantincreasein

patentsawardedtobothuniversityandnon-universityinstitutions,aswellasinincreasesintotalpatentapplications.Themarginalimpactofanotherforeigngraduatestudentisaround0.88patentapplicationsand0.57patentgrantseconomy-wide.HuntandGauthier-Loiselle(2012)findthataonepercentagepointriseintheshareofimmigrantcollegegraduatesinthepopulationincreasespatentspercapitaby9-18percent.Partofthiseffectreflectsthepositivespillovereffects(crowdingin)onnativeinventors(whichmayinturnbeduetocomplementaritiesininnovation).KerrandLincoln(2010)findthatincreasesinH-1BadmissionssubstantiallyincreasedratesofIndianandChineseinventionincitiesthatrelymoreonimmigrantscientists.A10percentgrowthintheH-1Bpopulationcorrespondsto1-4percenthighergrowthinIndianandChineseinventionforeachstandarddeviationincreaseincitydependency.Theyalsofindsomeevidenceforcrowding-ineffects.TurningtostudiesonEUmemberstates,Ozgenetal.(2011)showthattheaverageskilllevelofimmigrantsaffectspatentapplicationsinasampleof170EUregions.Moreover,patentapplicationsarepositivelyaffectedbythediversityoftheimmigrantcommunity.Anincreaseinthefractionalizationindexby0.1fromtheregionalmeanof0.5increasespatentapplicationspermillioninhabitantsbyabout0.2percent.FocusingonFrance,GermanyandtheUnitedKingdom,Venturinietal.(2012)findthathighlyeducatedmigrants,ingeneral,playapositiveroleinpromotinginnovation.Inhigh-technologysectors,inparticular,highlyskilledforeignworkerscontributepositivelytoinnovationwithoutcrowdingoutnatives.

67 SeeHanson(2009)andliteraturecitedtherein.

68 RecentcontributionsalongtheselinesincludeDiMariaandStryszowski(2009)andAzarnert(2012).

69 Empiricalsupportforthebraingainhypothesis,atleastinsomecountriesincludingBrazil,China,IndiaandIndonesia(representingmorethan80percentofthesamplepopulation)isfoundbyBeineetal.(2008;2010).

70 Thereareothermechanismsthroughwhichthemigrationofeducatedindividualscanhavepositiveeffects.First,theremittancessenthomebymigrantsboosttheincomeofthoseleftbehind.Thiscancontributetoinvestmentinthesendingcountry(seeSectionC.2).Remittancesmayalsocompensatetheamountspentoneducatingmigrantsseveraltimesover–asshownbyNyarko(2011)inthecaseofGhana.Secondly,migrantnetworkscanboosttradeinvariousways(seeBoxC.2)andhelpalleviatecapitalconstraintspreventingthedevelopmentofsmallenterprisesinthesourcecountry,asshownbyWoodruffandZenteno(2007)inthecaseofMexico.

71 Conversely,Azarnert(2012)arguesthat,ifprospectivemigrantsforeseethepossibilityoflow-skilledguest-workeremploymentinahigherwageforeigncountry,therelativeattractivenessofskilledemploymentinthehomecountrymightbereduced,withadverseeffectsonhumancapitalformationandanincreaseinfertility.

72 Themechanismisasfollows.Beforethedemographictransition,theurbandeathrateishighduetoinfectiousdiseasesandurbangrowthisonlysustainedbymigration.Whenthedemographictransitionsetsin,theurbandeathratefallsmorerapidlythantheruralone.Theurbannaturalincreasebecomespositiveanditdrivesthegrowthinurbanpopulation.Migrationbecomesagainthemainsourceofurbangrowthtowardstheendofthedemographictransition,when,duetolowfertility,theurbannaturalgrowthrateisverylow(ornegative).


213

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

73 Lumpinessiscloselyrelatedtoagglomeration,definedasthespatialconcentrationofeconomicactivity.Therearethreemaindriversofagglomerationconsideredintheliterature.First,agglomerationhasbeenshowntobedrivenbyfirms’objectivetoshareinputsandfacilities,andtotakeadvantageoflargermarkets.Secondly,agglomerationisalsoguidedbythebenefitsprovidedbybiggerandthickerlabourmarkets,intermsofhigherlaboursupply,bettermatchingbetweenemployersandemployeesandhigherworkerspecialization.Finally,anothermajordriverofagglomerationisfirms’andworkers’objectivetobenefitfromthehigherknowledgeflowscharacterizingbigcities.

74 Asshouldbeclearfromthemaintext,crucialtotheresultthatlumpinessaffectscomparativeadvantageistheviolationoffactorpriceequalization.Thatis,theresultisobtainediffactorendowmentswithinacountryareoutsidethe“coneofdiversification”(thefactorpriceequalizationset)andoneregionfullyspecializes.Factorpriceequalizationwithinthecountrycanbeviolatedifsomefactors(likenaturalresources)areimmobileinthepresenceofdifferencesinthelevelofamenitiesbetweenregions(CourantandDeardorff,1993)andinthepresenceofagglomerationeffectsà laKrugman(1991)–seeBrakmanandvanMarrewijk(2013)foradetailedexplanation.NotethatlumpinesscangiverisetoadirectionoftradecontrarytotheonepredictedbytheHeckscher-Ohlintheorem.

75 InacasestudyusingSpanishdata,Requenaetal.(2008)alsofindsomeevidenceoflumpiness.

76 Foracomprehensivereviewoftheliteraturethatdocumentstheexistenceofagglomerationeconomies,includingproductivitygains,seePuga(2010).Meloetal.(2009)underlinethedistinctionmadeintheliteraturebetweenlocalizationandurbanizationeconomies.Theformerindicategainswhicharemainlyrelatedtoindustrialconcentration,whilethelatterrepresentgainsfromcitysize.Generally,bothhaveanimpactonproductivity,withurbanizationeconomiesbeingrelativelymoreimportantforlightindustriesandknowledge-intensiveservices,suchasfinanceandrealestate.

77 Fortheeffectofinstitutionsondemographicchangesee,forinstance,McNicoll(1980)andBumpass(1990).

78 Intheempiricalanalysis,Doetal.(2012)useinstrumentalvariabletechniques(inparticular,geography-basedinstrumentfortradepatterns)toisolatethecausaleffectofcomparativeadvantageonfertility.

79 Moreover,variablessuchascolonialoriginsandlinguisticproximitycanbothinfluencetradeandimmigration.Ifnotproperlycontrolledfor,theycanconfoundtherelationshipbetweenimmigrantsandtradeflows.

80 Briantetal.(2009)usethestocksofimmigrantsin1875,1982and1990asaninstrumentforthecurrentstockofimmigrants.PeriandRequena-Silvente(2010)andBrattietal.(2012)useanapproachà laAltonjiandCard(1991),wherebythenetinflowofimmigrants(intheformerstudy)orthestockofimmigrants(inthelatterstudy)areimputedbasedonhistoricalimmigrationenclaves.

81 SeeBrülhart(2010)forasurvey.

82 InKrugmanandLivasElizondo(1996),tradeopeningleadstodispersionofeconomicactivitywithinacountry.Inthemodel,therearetwoagglomerationforces,forwardlinkages(becauseofatasteforvarietyandinterregionaltransportcosts,consumersliketolocateclosetoaslargeanumberofproducersaspossible)andbackwardlinkages

(inordertosaveontransportandfixedset-upcosts,monopolisticallycompetitiveproducersseektolocatetheirsingleplantasclosetotheirconsumersaspossible).Thedispersionforceisconstitutedbycongestioncosts.Forlowenoughtradecosts,thecongestionforcecomestodominatethebackwardandforwardlinkages,leadingtodispersionofeconomicactivity.Theimplicationtheydrawisthat“thegiantThirdWorldmetropolisisanunintendedby-productofimport-substitutionpolicies,andwilltendtoshrinkasdevelopingcountriesliberalize”.However,inamodelclosertoKrugman(1991),wheretheintensityofthedispersionforcefallswithtradeopening,MonfortandNicolini(2000)gettheresultthatthelatterinducesinternalagglomeration.

83 Withnochangeinacountry’sterritory,thisimpliesanincreaseinroaddensity.

84 http://www.cio.com/article/123230/South_Africa_Outsourcing_Scorecard

85 http://www.icta.mu/mediaoffice/2007/IPLC_en.htm

86 Authors’calculationsbasedondatafromtheInternationalMonetaryFund.

87 SuchastheHeckscher-Ohlinmodeloftrade.

88 Attimes,theycontractwithfirmsinthehostcountrytoestablishajointventure(Desaietal.,2004).

89 Analysingindustry-leveldatafor91countriesbetween1980and1997,Manova(2008a)showsthatequitymarketliberalizationboostsexportsdisproportionatelymoreinsectorsthatarerelativelymoredependentonexternalfinancing.Thisisindicativeofadirectlinkbetweenportfolioinvestmentinflowsandgreaterdomesticinvestmentinplant,machineryandequipment,which,inturn,increasesthesupplycapacityoffirms.

90 Bothphysical(shares,bonds,property)andhuman(educationandexperience).

91 Ashortcomingofmanyofthesestudiesisthattheyusecombinedprivateandpublicsavingsdata.

92 Authors’calculationsbasedondatafromtheInternationalMonetaryFund.

93 Inthiscontext,amultinationalfirmwillinternalizeitsactivitiesinaforeigncountrythroughFDIiftheinternalizationcostislowerthanthecostassociatedwithestablishinganarm’slengthcontract(BuckleyandCasson,1976).

94 Rapidcapitalmobilityhas“levelledtheplayingfield”forinternationalbusinesstosomeextent.Firmsthatwouldliketotakeadvantageofregulatoryortradepoliciesinaforeigncountrycansimplymoveorsub-contractthroughafirmlocatedthere(Feenstra,1998).

95 SeveralfactorsaffecttherelationshipbetweenR&Dexpenditureandinnovations.Clearly,innovationispartiallytheresultofchance.Therefore,therelationshipbetweenR&Dandinnovationisbynaturestochastic.But,inaddition,R&Dproductivitymaydependonspecificconditions,suchasthequalityoftheeducationsystem.ForadeeperunderstandingoftherelationshipbetweenR&Dandinnovation,seeChapter4ofWorld Intellectual Property Report (WIPO,2011).

96 Forguidelinesonthecollectionanduseofdataoninnovation,seeThe Oslo Manual(OECD,2005).

97 SeeKhanandWunsch-Vincent(2011)foradiscussiononthedifferentmeasuresofpatentsavailable.


214

98 Toaddressthislimitation,JaffeandTrajtenberg(2002)suggestusingthenumberofpatentsweightedbytheircitations.

99 Thebalanceofpaymentsisanimportantsourceofinformationinthisrespect.

100 Hall(2010)hadpreviouslyplottedastandardLorenzcurveofbusinessR&DandGDPfor40economiesfortwoperiods,1999and2005.Duetodataavailability,the37countriesinoursampleincludeArgentina,Australia,Austria,Belgium,Canada,China,CzechRepublic,Denmark,Estonia,Finland,France,Germany,Greece,Hungary,Iceland,Israel,Italy,Japan,RepublicofKorea,Luxembourg,Mexico,Netherlands,NewZealand,Norway,Poland,Portugal,Romania,RussianFederation,Singapore,SlovakRepublic,Slovenia,Spain,Sweden,Switzerland,UnitedKingdomandUnitedStates.

101 Asdiscussedinsub-section(a),thereareseveralmeasuresoftechnologicalinnovation.Asimplecomparisonofthenumberofpatentapplicationsdoesnotcapturethevalueofspecificpatents.Onewaytheeconomicliteratureattemptstoaddressthisconcernisbycountingthenumberofpatentapplicationsinspecificfilingoffices.Wealsoperformtheanalysislookingatthesealternativemeasures.Whilethespecificrankingofacountrymaychange,thefindingthatAsiancountrieshaveemergedamongthemajorinnovatingcountriesisconsistent.

102 R&Dservicescoverthoseservicesassociatedwithresearch(e.g.chemistry,biotechnology,medicalsciences,appliedscienceandtechnologywhichmayberelatedtomachinery,electricity,communications,vessels,aircraft,civilengineering,construction,information,etc.)andexperimentaldevelopmentofnewproductsandprocesses.

103 ThisgroupwasestablishedbyUNSDfollowingarequestbytheUNStatisticalCommissiontotheInter-agencytaskforceonstatisticsofinternationaltradeinservicestodevelopcompilationguidancetoaccompanytheManualonStatisticsofInternationalTradeinServices2010.TheUNexpertgroupincludesallparticipatingagenciestotheinteragencytaskforceaswellasnationalexpertsintradeinservicesstatistics.

104 SeeChapter1ofWorld Intellectual Property Report (WIPO,2011).

105 Notethatforeignsubsidiariesarecountedasresidentswhentheyprovidetheirlocaladdress.

106 OtherstudiesthatexplorethegeographicaldimensionofinternationaltechnologyincludeBottazziandPeri(2003),Branstetter(2001),EatonandKortum(1999),IrwinandKlenow(1994).

107 “Verticallyintegratedcountries”aredefinedasthosecountry-pairswithashareoftradeinintermediategoodsabovethemedian.

108 Dataareavailableathttp://epp.eurostat.ec.europa.eu/portal/page/portal/microdata/cis.

109 SeethearticleinThe Economist(12January2013)“Innovationpessimism:Hastheideamachinebrokendown?”,The Economist(2013).

110 Indevelopingcountries,theroleofgovernmentinresearchismuchmorepronounced.SeeWIPO,(2011).

111 Neweconomicgeographytheoryalsopredictsthatastradecostsfall,productioninitiallybecomesmoreconcentrated,butthenbecomesmoredispersed.Thisisbecauseasconcentrationincreases,forcesthatactagainstagglomerationbecomemoresalient.Forexample,inorder

topersuadeworkerstomoveintothesectors,firmswillhavetopayhigherwages.Thiswilltendtoreducetheincentiveforafurtherexpansionofthesector.Thelevelofaggregationatwhichthisturnintheagglomerationpatternwilloccurwilldependonanumberoffactors.Oneoftheseisthetechnologicalspilloverintensityandthegeographicalextentofknowledgespillovers.Evidencesuggeststhattheadvantagetoclusterisparticularlyimportantinsomeknowledge-intensivesectors(AudretschandFeldman,1996).Thisiscompatiblewiththefactthatknowledge-intensivesectorshaveasubstantialpartoftacitknowledgethatislesseasilytransferableacrosscountries.

112 SeeChart13inthe2008World Trade Report(WTO,2008).

113 LaursenandMeliciani(2010)showthatICTaffectsexportmarketsharesalsoinnon-ICTsectorsandthatsmallopeneconomiesbenefitmorethanothercountriesfromICT-relatedforeignknowledgeflows.

114 Meliciani(2011).

115 Severalfactorsaffecttheabilitytoappropriatereturnsfrominnovation.Theseincludeleadtime,secrecy,complementaryassetsandpatentprotection.Onthebasisofasurveyquestionnaireadministeredto1,478R&DlabsintheUSmanufacturingsectorin1994,Cohenetal.(2000)findthatofthesemechanismspatentstendtobetheleastemphasizedbyfirms.

116 Forexample,severalstudiesshowthatcountries’growthrates(theultimateresultofinnovation)arepositivelyassociatedwiththevolumesoftrade(AlcaláandCiccone,2003;FrankelandRomer,1999;SachsandWarner,1995)andtradeopening(SachsandWarner,1995;andSala-i-Martin,1997).

117 AlsoseeSectionB.1.

118 OtherstudiesthatpointtothesamedirectionincludeCleridesetal.(1998)andVanBiesebroeck(2005)forAfricancountries,andHallward-Driemeieretal.(2002)forEastAsiancountries.

119 WIPO(2011),Chapter2andMaskus(2012)forareview.

120 Thisistheview,forexample,ofJavorcik(2004)whofindsthatweakprotectionofintellectualpropertyrightsdetersFDIintechnology-intensivesectorsthatrelyheavilyonintellectualpropertyrights.

121 SeeSectionC.1aswellas,forexample,AgrawalandOettl(2008),Kerr(2008),Singh(2005).

122 ForareviewoftheexistingempiricalevidenceonFDIandtechnologyspillovers,seeKeller(2010).Asanexampleofrelevantempiricalpapersontheissue,seealsoBlalockandGertler(2008),Javorcik(2004),AitkenandHarrison(1999),DjankovandHoekman(2000),HaddadandHarrison(1993)andKonings(2001).

123 Seehttp://www.bbc.co.uk/news/business-20639545andhttp://www.dw.de/south-stream-pipeline-construction-begins/a-16435203.

124 Thereareseveralmeasuresthatcanbeusedtomeasurevolatility.ChenandHsu(2012)useamovingaverageofthestandarddeviationofprices,realizedvolatilityandaGARCHmodelasmeasuresofenergypricevolatility.GARCHistheacronymforgeneralizedautoregressiveconditionalheteroskedasticityandreferstoeconometricmodelsthatallowthevarianceofatimeseriestodependonthevolatilitythatwasrealizedintheprecedingperiods.Therefore,itcancapturepotentialclusteringofvolatilityaroundspecificpointsintime.


215

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

125 Sweeneyactuallyusestheterm“depletable”butwetreatitassynonymouswithexhaustible.

126 Seehttp://www.eia.gov/energy_in_brief/about_shale_gas.cfm.

127 Inthecaseofbiofuelsforinstance,ajointreportbytheFAO,IFAD,IMF,OECD,UNCTAD,WFP,theWorldBank,IFPRI,UNHLTFandtheWTO(FAOetal.,2011)observesthatbiofuelsnowaccountforasignificantshareoftheglobaluseofseveralcrops–sugarcane,vegetableoil,coarsegrainsandsugarbeet.Beyond,thepitfallposedbyindustrialpolicy,thereportalsonotestheenvironmentalandsocialproblemsthathavearisenfromthesebiofuelsubsidies.Growingcropsforfuelcanpotentiallyemitmoregreenhousegasesthantheysaveandthesubsidiesthemselvesmayhaveplayedabigroleintheincreaseincommoditypricesin2008,whichwasparticularlyharmfultofood-importingdevelopingcountries(Mitchell,2008).

128 Partofthereasonliesinthedifficultyofdeterminingnaturalresource“abundance”althoughtheremaybescopetouseenvironmentalaccountingmethodshereastheyhavebeenemployedtomeasurenaturalcapital,e.g.forestry,inthevaluationofnationalwealth(PearceandAtkinson,1993;AronssonandLofgren,2010).

129 Theterm“450”comesfromthefactthatclimateresearchersassumethattheconcentrationofgreenhousegasesintheatmosphereshouldnotexceed450partspermillionofcarbon-dioxideequivalenttobeabletohaveareasonablechanceofreachingthe2°Cgoal.

130 SeethediscussioninMilner(1997),Milneretal.(2000)andMilnerandZgovu(2006).

131 Thisisanexampleofwhatiscalledthe“Alchian-Alleneffect”ortheoremafterthetwoeconomists,ArmenAlchianandWilliamR.Allen,whofirstanalysedtheissue.Notethatitfocusesonlyonthepuresubstitutioneffectandignorestheincomeeffectoftheincreaseincost,whichcouldruncountertothesubstitutioneffect.However,toourknowledgethereisnoempiricalevidencetosuggestthattheincomeeffectdominatesthesubstitutioneffect.

132 Strictlyspeaking,theyestimatedthedependenceoffreightchargesonpriceandfoundanelasticityofabout0.125to0.716,withtheir“preferred”specificationbeing0.125.Pureicebergtransportationcostwillhaveproducedanelasticityof1andpurelyadditivetransportationcostwillhavegeneratedanelasticityofzero.The0.125estimateisclosertozero.

133 AsofthethirdweekofOctober2012,ametrictonofironoregoesforUS$120,whileatroyounceofgoldisworthUS$1,700.Thereare32,151troyouncesinametricton.CompareTableC.17foraveragevalue-to-weightratiosfordifferentproductgroupsandmodesoftransportation.

134 Thisisrelatedtowhatistermedthe“O-ring”theoryofproduction(Kremer,1993).TheaccidentthatbefellthespaceshuttleChallengerin1986hasbeenattributedtothefailureofjustoneofitsmanythousandsofcomponents–theO-ring–becausetheverycoldweathermadeittoobrittletowithstandtheexplosivepressureoftheChallenger’srockets.Whenappliedtoglobalsupplychainsandtradeinintermediateinputs,the“O-ring”theorysaysthatadelayinthearrivalofevenoneinputhasacascadingeffectonthewholeproductionprocess,withverycostlyconsequencesforthefirm.

135 Thelistof31landlockeddevelopingcountriescanbefoundat:http://www.un.org/special-rep/ohrlls/lldc/list.htm.

136 C.i.f.referstothepriceinvoicedbyasellerthatincludesinsuranceandallotherchargesuptothenamedportofdestination,whilef.o.b.includesallchargesuptoplacingthegoodsonboardacarrierattheportofdeparture.

137 However,distanceisnotalwaysimmutable.Humanactionandnaturalprocessescanhaveadramaticeffectonit.Forinstance,theopeningoftheSuezandPanamacanalsdramaticallyreducedthemaritimedistancebetweencountries(seethehistoricaldiscussioninSectionBontheeffectoftheseeventsontrade).Moreover,ArcticwarmingmightopenupapolarroutethatwoulddramaticallyshortentheshippingdistancebetweenEuropeandAsia.

138 Ameta-analysisisastatisticalanalysisofacollectionofpriorstudies,whichinthiscaseareestimatesofgravityequations.Thegravityequationseekstoexplainthevolumeoftradebetweenanypairofcountries.SincetheGDPofthetradepartnersandthedistancebetweenthemareusuallyincludedasexplanatoryvariablesintheequation,ithasbeendubbedgravityequationinreferencetotheanalogousroleplayedbymassanddistanceinthetheoryofgravity.

139 ThesourceofdataontransportationcostsisUScustomsdataanditsmeasureof“importcharges”.UScustomsdefinesimportchargesas“…theaggregatecostofallfreight,insurance,andothercharges(excludingUSimportduties)…”Thesecostsreflecttransportationbetweencountriesandexclude,inalmostallcases,inlandtransportation.

140 BlonigenandWilson(2008)builduponthismethodandrefinetheobtainedresultsbyusingvariationinportefficiencyovertime.TheyconcludethattheeffectofportinfrastructureitselfisconsiderablysmallerthansuggestedbyClarketal.(2004),whoseestimatesarearguedtoincludeothercountrycharacteristicsthatarenotdirectlyrelatedtoportefficiency,suchasinlandinfrastructureorexportpolicies.

141 Severalstudiesconfirmthatincreasedcompetitionhasalsoreducedtransportpricesandincreasecargoquantitiesinairfreight.Theimpactofbilateralopenskiesagreementsonrouteofferings,airtransportationpricesandtradevolumesarereferredtoin,amongothers,MiccoandSerebrisky(2006),Zhangetal.(2011)andCristeaandHummels(2011).Regardingairpassengerflows,thestudybyPiermartiniandRuosova(2013)investigatestheimpactofairservicesliberalizationusinginformationon2,300AirServicesAgreementscovering184countries.Theyareabletoidentifyprovisionsintheseagreementsthatareimportantdeterminantsofthedegreeofliberalizationoftheinternationalaviationmarket.Inincreasingorderofliberalization,theseare(i)includemultipledesignations,(ii)freedeterminationofcapacity,(iii)freepricingandcommunityofinterest,and(iv)cabotage.Theythensimulatetheeffectofeachprovisionbeingadoptedbyallcountrypairswhosecurrentairtrafficregulationsdonotincludesuchaprovision.Theypredictthatairpassengertrafficwouldincreaseby0.5,5,9and11percentifallexistingagreementsintroducedmultipledesignation,freedeterminationofcapacity,freepricingandcommunityofinterestandcabotage,respectively.

142 ThisisbasedonRegulation(EC)No906/2009.

143 http://www.economist.com/node/21527035.

144 Thetradefacilitationtalkswerenotaddedtothe2001DohaDevelopmentAgendabeforemid-2004.

145 SetoutinAnnexDoftheGeneralCouncilDecisionof1August2004(WT/L/579).


216

146 Extendingthismethodtonon-OECDcountriesinthefuturewillgiveamorecompletepictureandislikelytoshowthatreductionsforothercountriesareevenlargerasotherstudieslikeOECD(2009)showed.

147 TheAidforTradeinitiativewasmentionedinthesamecontextaswell.

148 HoekmanandNicita(2010),page77ff.

149 Ibid,page78.

150 SeeforexampleArvisetal.(2007),Otsuki(2011),Wilsonetal.(2003)andWilsonetal.(2005).

151 SeeforexampleChristandFerrantino(2011)andArvisetal.(2007).

152 Formaritimetransportation,therearealotofcoststhatareindependentoffuel.Therelativeimportanceoffuelisgreaterforlongervoyages.Furthermore,thereisaneasytechnologicalfixtoadapttorisingfuelpricesasshipscanslowdownandburnlessfuel.Inthecaseofplanes,theyburnalotoffuelontake-offandlanding.Forverylongflights,planeshavetocarrymorefueladdingtotheweightoftheplaneandreducingitsfuelefficiencypercargocarried.Theresultisaquadraticeffectofdistanceinteractingwithfuelprices.Whileplanescanslowdownandburnlessfuel,theyhavealotlessfreedominthisrespectthanships.Finally,risingfuelpricesinducesubstitutionawayfromairplanesbecausefuelpricesrepresentahighershareofoperatingcostsforplanes,sothecostelasticityisgreater.Aswitchfromplanestoshipsforthesamecargowillsharplylowerfreightcharges,whilealsoincurringgreatertimecosts.Thiswillhaveabiggerimpactoncargoesthataretimesensitive.

153 Modalsharesarecalculatedbyusingimportvalues.

154 Theformerelasticityis0.088,whilethelatterisonly0.103.

155 Forfurtherdetails,seehttp://www.systemicpeace.org/polity/polity4.htm.ThissectionproducesarangeofsimplestatisticsusingtradedataandthescoresfromthePolityIVProject.NothinginthissectionimpliesajudgementonthepartoftheWTOofanyparticularformofgovernment.

156 ThebluelineinFigureC.54isasimplelinearfitofthedata.Instead,theredcurverepresentsthebestfitofthedatathatallowsfornon-linearities.

157 Itappearsthattodatenostudyhaslookedinaformalwayattheinverted-UrelationshipbetweenregimetransitionandtradepolicyobservedinFigureC.54.

158 TheconstructionoftheindicatorofthedepthoftradeagreementsfollowsOreficeandRocha(2011).Weconsider100tradeagreementsspanningfrom1958to2011.Thedepthmeasureisconstructedconsideringthetenmostimportantprovisionsinthefactoranalysis,namelytrade-relatedintellectualpropertyrights(TRIPS),IPR,countervailingmeasures,movementsofcapital,publicprocurement,competitionpolicy,anti-dumping,investmentandstateaid.

159 Detaileddatafor“ruleoflaw”and“regulatoryquality”forabroadcross-sectionofcountriesareonlyavailableafter1996(Kaufmannetal.,2010)

160 Abodyofeconomicliteraturefocusesonthecausaleffectsofeconomicandpoliticalreforms(see,amongothers,GiavazziandTabellini(2005)andGiulianoetal.(2012)).

161 ThestudybyAcemogluetal.(2005)openedthewaytoavastliteratureonthepoliticaldeterminantsoftherelationshipbetweenglobalizationandeconomicinstitutions.ImportantrecentcontributionsincludeDalBóandDalBó(2011),DoandLevchenko(2009),Levchenko(forthcoming),Segura-Cayuela(2006),andStefanadis(2010).

162 Costinot(2009)offersanalternativeframeworkinwhichcontractenforcementisacrucialdeterminantofcomparativeadvantage.Inthismodel,betterinstitutions,representedbyahigherprobabilityoftheenforcementofacontract,allowacountrytospecializeintheproductionofmorecomplexgoods.Thesearesectorsthatrequireahighernumberoftasks(suchasresearch,design,assembly)toproduceaunitofthegood.

163 Thestudyintroducesameasurethatquantifiestheimportanceofcontract-dependentinputsintheproductionoffinalgoods.Inparticular,foreachintermediategooditispossibletodeterminewhetheritissoldinanorganizedmarketorifitisreference-pricedinatradepublicationorifitisnoneofthese.Goodsthataremorecontract-dependentarethosethatuseahigherfractionofinputsthatarenotsoldinorganizedmarketsanddonothaveareferenceprice,asthoseinvestmentsaremorelikelytoberelation-specific.

164 Forasurveyofthisliterature,seeWTO(2011b).

165 See,inparticular,Giulianoetal.(2006),Guisoetal.(2009),SpolaoreandWacziarg(2009a;2009b),Gokmen(2012).ThismeasureisbasedontheworkofCavalli-Sforzaetal.(1996).

166 Forabroaderdiscussionofthepro-tradeeffectsofimmigration,seeSectionC.1(c).

167 Arelatedstrandofliteratureanalysestherelationbetweenconflictsandtrade.Inparticular,Rohneretal.(2011)provideatheoryoftradeandconflictwheretradehingesontrustandcooperation.Theyshowthatpoliciesthatfosterinter-ethnictradeincreasetrustbetweensocietiesandreduceconflicts.


217

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

AppendixAppendixTableC.1:BERD in the manufacturing sector (sumover18countries,valuesinPPPUS$million,2005constantprices)

IndustryValue Share

1990-95 average 2005-10 average 1990-95 average 2005-10 average

Fabricatedmetalproducts,machineryandequipment,instrumentsandtransport

85,570 139,638 69.1% 72.7%

Coke,petroleum,nuclearfuel,chemicalsandproducts,rubberandplastics

25,914 37,067 20.9% 19.3%

Basicmetals 4,240 4,052 3.4% 2.1%

Food,beveragesandtobacco 2,594 4,139 2.1% 2.2%

Non-metallicmineralproducts 2,145 2,142 1.7% 1.1%

Wood,paper,printing,publishing 1,404 1,649 1.1% 0.9%

Textiles,furandleather 1,070 1,802 0.9% 0.9%

Furnitureandothermanufacturing

880 1,576 0.7% 0.8%

Manufacturingtotal(sumover18countries)

123,815 192,079

Source:Authors’computations,basedondatafromOECDScience,TechnologyandR&DDatabase.

Note: Forthepurposeofconsistencyandcomparability,aggregationisperformedonlyoncountrieswithBERDdatainallindustrialbreakdownsofthemanufacturingsector,forboththeperiodof1990-95and2005-10.Asaresult,18countriesinthedatabasesatisfythesecriteria,andthey are: Czech Republic, Denmark, Finland, Germany, Hungary, Iceland, Ireland, Italy, Japan, Republic of Korea, Mexico, Norway, Portugal,Singapore,Slovenia,Spain,SwedenandTurkey.


218

AppendixTableC.2:BERD in the services sector (Sumover14countries;valuesinPPPUS$million,2005constantprices)Industry 1990-95 average 2005-10 average Annualized growth

Community,socialandpersonalserviceactivities,etc. 436 728 3.47%

Financialintermediation(includesinsurance) 414 1,465 8.79%

Realestate,rentingandbusinessactivities 3,921 16,088 9.87%

Transport,storageandcommunications 824 1,761 5.19%

Wholesale,retailtradeandmotorvehiclerepair 603 2,337 9.45%

TotalBERDinservicessector(sumof14countries) 5,710 22,294 9.51%

Source:Authors’computations,basedondatafromOECDScience,TechnologyandR&DDatabase.

Note: Forthepurposeofconsistencyandcomparability,aggregationisdoneusingonlycountrieswithdatainallindustriesundertheservicessectorfortheperiodsofboth1990-95and2005-10.Asaresult,the14countriesinthesampleareAustria,Canada,CzechRepublic,Germany,Greece,Hungary,Ireland,Netherlands,Norway,Portugal,Singapore,SlovakRepublic,SpainandTurkey;accordingtotheOECDdatabase,theindustrybreak-downofBERDdoesnotadduptothetotalsectoralBERD.


219

II C. Fu

nd

am

en

tal e

Co

no

mIC

Fa

Cto

rs

aFFe

CtIn

g

Inte

rn

atIo

na

l tra

de

AppendixTableC.3:Number of fixed-telephone subscriptions, mobile-cellular telephone subscriptions and internet users per 100 inhabitants, 2011 (Top30economies)

Fixed-telephone Mobile-cellular telephone Internet users

Monaco 96.40 Macao,China 243.50 Iceland 95.02

ChineseTaipei 72.68 HongKong,China 209.64 Norway 93.97

CaymanIslands 65.63 Panama 203.88 Netherlands 92.30

Germany 63.05 SaudiArabia,Kingdomof 191.24 Sweden 91.00

HongKong,China 61.06 AntiguaandBarbuda 181.64 Luxembourg 90.89

Korea,Republicof 60.90 RussianFederation 179.31 Denmark 90.00

Switzerland 60.82 Suriname 178.88 Finland 89.37

St.Helena 59.65 Oman 168.97 Bermuda 88.34

SanMarino 58.88 CaymanIslands 167.67 Qatar 86.20

Iceland 58.43 Anguilla 166.31 NewZealand 86.00

France 55.92 Finland 166.02 Switzerland 85.20

Malta 54.89 Maldives 165.72 Liechtenstein 85.00

Luxembourg 54.10 Dominica 164.02 Korea,Republicof 83.80

Liechtenstein 53.99 Libya 155.70 Guernsey 83.63

UnitedKingdom 53.24 Austria 154.78 Canada 83.00

Barbados 51.35 Italy 151.84 Germany 83.00

Japan 51.06 Lithuania 151.30 AntiguaandBarbuda 82.00

Greece 49.91 Singapore 149.49 UnitedKingdom 82.00

Sweden 48.72 UnitedArabEmirates 148.62 Andorra 81.00

UnitedStatesofAmerica 47.91 Luxembourg 148.27 FaroeIslands 80.73

Canada 47.86 Seychelles 145.71 Austria 79.80

Australia 46.63 VietNam 143.39 France 79.58

Israel 46.28 Botswana 142.82 Japan 79.53

Ireland 45.22 Kazakhstan 142.55 Australia 79.00

Denmark 45.13 Uruguay 140.75 Belgium 78.00

Andorra 44.57 Bulgaria 140.68 UnitedStates 77.86

Belarus 44.02 Guatemala 140.38 Bahrain,Kingdomof 77.00

Montserrat 43.41 Estonia 138.98 Ireland 76.82

Belgium 43.06 TrinidadandTobago 135.57 Estonia 76.50

Slovenia 42.89 Argentina 134.92 Singapore 75.00

Source:ITU.

C. Fundamental economic factors affecting international trade

Documents

Transcript of C. Fundamental economic factors affecting international trade