d. the trade effects of non-tariff measures and services measures · non-tariff measures and...

World TrAde reporT 2012

134

This section discusses the trade effects of non-tariff measures and services measures in general before focusing on technical barriers to trade (TBT), sanitary and phytosanitary (SPS) measures and domestic regulation in services. It also examines whether regulatory harmonization and/or mutual recognition help to reduce the trade-hindering effects caused by the diversity of TBT/SPS measures and domestic regulation in services.

d. The trade effects of non-tariff measures and services measures

II – TRADE AND PUBLIC POLICIES: A CLOSER LOOk AT NON-TARIFF MEASURES IN THE 21ST CENTURY

135

D. TH

E TR

AD

E E

FFEC

TS O

F

NO

N-TA

RIFF M

EA

SU

RE

S

A

ND

SE

RV

ICE

S M

EA

SU

RE

S

Contents 1 EstimatingthetradeeffectsofNTMsandservicesmeasures 136

2 DisentanglingtradeeffectsofTBT/SPSmeasuresanddomesticregulationinservices 143

3 Harmonizationandmutualrecognition 149

4 Conclusions 152

AppendixD.1 157

Some key facts and findings

• The contribution of non-tariff measures to overall trade

restrictiveness is significant, and in some estimates NTMs

are far more trade restrictive than tariffs.

• TBT/SPSmeasureshavepositivetradeeffectsformore

technologically advanced sectors, but negative effects

in agricultural sectors.

• ThereisevidencethatTBT/SPSmeasureshaveanegativeeffect

on export market diversification.

• ThenegativeeffectsontradecausedbythediversityofTBT/SPS

measures and domestic regulation in services are mitigated

by the harmonization and mutual recognition of these measures.


136

This section examines the trade impact of non-tariffmeasures. Unlike tariffs, NTMs often vary acrosscountriesandsectors,so“advalorem”equivalentsarecalculatedforNTMsinordertomakethiscomparison.Evidence is then presented on the trade effects oftechnical barriers to trade (TBT) and sanitary andphytosanitary (SPS) measures in goods and ofequivalentdomesticregulationmeasuresinservices.1

The rationale for focusingon thesemeasures is that,independent of their policy objectives, economictheory offers a mixed picture – both negative andpositive – of how these measures affect the volumeand direction of trade. For example, standards andtechnical regulations can raise producer costs –because compliance is more expensive – but reduceconsumercosts–becauseproductqualityinformationis more readily available. Trade will increase or falldependingonwhether thepositiveeffectondemandisgreaterthanthenegativeeffectonsupply.

Inordertohighlightthedifferencesbetweennon-tariffmeasures and tariffs, this section also attempts todisentangle the trade effects of these measures byfocusing on: (a) the specific channel through whichtrade isaffected (thevolumeof tradeor thedecisionto export); (b) their specific impact across countries,sectorsandfirms;and(c)whether themeasure itself,orthewayitisapplied,constitutesthemainrestrictionto trade. This section also considers the degree towhich the harmonization or mutual recognition ofTBT/SPS measures and domestic regulation inserviceshelpstoreduceanytrade-inhibitingeffects.

1. EstimatingthetradeeffectsofNTMsandservicesmeasures

A number of studies attempt to quantify the effect ofnon-tariff measures on international trade. Averagingacrosscountries,theyfindthatNTMsarealmosttwiceas trade restrictive as tariffs. They also find that, inseveralcountries,NTMsactuallycontributemuchmorethan tariffs to theoverall levelof trade restrictiveness.Theseresults,however,arebasedonNTMsdatawhichhave not been updated for about ten years. Given thedeclineintariffratessincethen,therelativecontributionofNTMstooveralltraderestrictivenessislikelytohaveincreased, perhaps making them even more importantthantariffsinmostcountries.

Furthermore, evidence suggests that the relativecontribution of non-tariff measures to the overall levelofprotectionincreaseswiththelevelofGDPpercapita.Thetrade literaturealsofindsthatNTMs inagricultureappear to be more restrictive and widespread thanthose in the manufacturing sector. In the case ofservices,whilerestrictionstotradearegenerallyhigherindevelopingcountriesthaninOECDcountries,theydonot appear to be systematically associated with acountry’s level of development. The cross-country

pattern of restrictiveness of services measures variesacross services sectors. It is worth noting that themethods developed in the literature to estimate thesetradeeffectssufferfromanumberoflimitationswhichcanbe traced, inpart, toa lackof transparency in theuse of NTMs. In addition, they do not address thepotentialimpactofglobalsupplychains.

(a) MagnitudeofNTMsasrestrictionstotrade

Earlier sections of the Report have highlighted thatnon-tariff measures can take many different forms –quotas, taxes, subsidies, technical regulations etc. Inorder to facilitate a comparison between the tradeeffects of these different NTMs, studies analyse theimpact of NTMs on international trade by estimatingan“ad-valoremtariffequivalent(AVE)”,i.e.thelevelofanad-valoremtariffthatwouldhaveanequallytrade-restrictingeffectastheNTMinquestion.Thisenablesacomparisontobemadewithtariffs,andisimportantforanyanalysisof thewelfare implicationsof varioustradepolicymeasures. Inthetradeliterature,theAVEof different NTMs is computed using one of twoapproaches – the “price gap” or the “econometrics-basedmethod”(SeeBoxD.1).

(i) Do NTMs matter?

Usingdatafor91countries,Keeetal.(2009)evaluatethetradeimpactofnon-tariffmeasureseconometricallyforeachof4,575six-digitcategoriesoftheHarmonizedSystem (HS) of classifying goods where at least onecountryimposeswhattheycategorizeaseithera“coreNTM” (defined as including price control measures,quantitative restrictions, monopolistic measures, anti-dumping and countervailing measures and technicalregulations)or“agriculturaldomesticsupport”.2

Theyestimate theaverageAVEofcoreNTMsfor theentire sample at 12 per cent. When weighted byimports,thisnumberfallsto10percent.Thenumbersare much higher – 45 per cent and 32 per centrespectively – if the averages are calculated only fortariff lines affected by core NTMs.3 In contrast, thesimple and import-weighted averages of AVEs ofagricultural domestic support are much smaller(generallybelow1percent).Accordingtotheauthors,this is because a small number of products areaffected by agricultural domestic support in mostcountries. The importance of NTMs is reinforced byavailable firm survey evidence. For example, a recentsurvey on non-tariff trade costs between Arabcountries revealed an average AVE of 6 per cent(HoekmanandZarrouk,2009).

Estimatesofthetradeimpactofnon-tariffmeasuresarelargely consistent with the AVEs computed. HoekmanandNicita(2011)findthatreducingtheAVEofNTMsbyhalf, from around 10 per cent to 5 per cent, would


137

D. TH

E TR

AD

E E

FFEC

TS O

F

NO

N-TA

RIFF M

EA

SU

RE

S

A

ND

SE

RV

ICE

S M

EA

SU

RE

S

BoxD.1: Methodology used for estimating the AVE of NTMs

Price gap method

Non-tariffmeasures increase thepricepaidbyconsumers.4Thebasic strategyof the “pricegap”methodinvolvesacomparisonofpricesbeforeandaftertheNTMmark-up,wherethisdifferenceisexpressedasatariffequivalent.Makingthiscomparison,however, isnotstraightforward.ManyfactorsunrelatedtoNTMsalsoaffectcostsandpricesatdifferentpointsinthesupplychain.Forinstance,the“free-on-board”(f.o.b.)priceatthepointofexportincludesthecostoftransporttothepointofexportaswellasthecostsofloadingthe goods, while the “cost-insurance-freight” (c.i.f.) price also includes the cost of international transportandinsurance.Furthermore,thepriceafterborderproceduresincludesanytariffschargedontheproduct.Finally,wholesaleand retailprices include internal transportcostsanddistributionmargins.These factorsmust be removed from the observed price difference before the mark-up can be attributed to non-tariffmeasures(Ferrantino,2006).

However,differentNTMsoccuratdifferentpointsinthesupplychain,whichmeansthatthepriceimpactofaparticularNTMcanonlybeidentifiedbycomparingtwopricesattherelevantstagesintheproductionanddistributionprocess.Forexample,customsproceduresaffectthedifferencebetweenthec.i.f.priceandthelanded duty-paid price. In sum, it is possible but not straightforward to measure and compare therestrictivenessofdifferenttypesofNTMs(Ferrantino,2012).

Econometrics-based method

An alternative to the direct “price gap” method described above is to estimate the impact of non-tariffmeasuresoneitherpriceorquantity(tradeflows)usingeconometricmodels.Estimatingthe“quantityimpact”isparticularlyusefulbecausedataontradeflowsaremoreeasilyavailableatadisaggregatedlevel.Moreover,whentheNTMisabsolutelyprohibitive,nopricesareobserved,orwhentheproductishighlydifferentiated,pricesarenotparticularlyinformative(Ferrantino,2012).

Inmuchof thetrade literature, theAVEsofnon-tariffmeasuresareestimatedthrough“gravityequations”.TheseareeconometricmodelsoftradewhichacquiretheirnamefromthesimilaritiestoNewton’stheoryofgravitation.Theypredict that thevalueof tradebetweenany twocountrieswillbepositively relatedto thesizeoftheireconomiesand inverselyrelatedtothedistance(andothermeasuresoftradecosts)betweenthem.InordertoestimatetheeffectofpoliciessuchastariffsandNTMsontrade,gravityequationsincludemeasures,whichcapturethesepolicyfactors,asexplanatoryvariables.

In order to estimate the effect of policies such as tariffs and NTMs on trade, gravity equations include measures, which capture these policy factors, as explanatory variables. 𝑙𝑙𝑙𝑙 𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉 = 𝑎𝑎 + 𝑏𝑏!𝑙𝑙𝑙𝑙 1 + 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 + 𝑏𝑏!𝑁𝑁𝑁𝑁𝑁𝑁 + 𝑐𝑐𝑐𝑐 where “X” is a set of variables that may also affect trade flows. It typically where“X”isasetofvariablesthatmayalsoaffecttradeflows.IttypicallyincludesGDP,distanceandothertradecosts.Whenprecisedataare lacking,thepresenceofNTMsiscapturedbyadummyvariable,whichassumesavalueofonewhentheNTMinquestionappliesandzerootherwise.

Thegravitymodeloftradeenablesanestimationofthepredictedvalueoftradebetweenacountrypairwithandwithoutthenon-tariffmeasures.TheeffectoftheNTMontradeisestimatedasthedifferencebetweenthetwovalues.Asimilarcalculationcanbemadefortheeffectofatariffcomparedwithnotariff.TheAVEoftheNTMcanthenbederivedbycomparingthesetwopredicteddifferences.Morespecifically,theAVEoftheNTMisatariffthathasthesameeffectonthevalueoftrade.

Thetradeliteraturereferstotheaboveasthe“directapproach”.Thereisalsoan“indirectapproach”whichcomparesactualtradeflowstothetradeflowspredictedbyahypotheticalfrictionlessbenchmarkscenario.ThedeviationofactualfrompredictedtradeflowsistakentobeindicativeoftheimpactofNTMsbecausespecific explanatory variables measuring NTMs are not included in the estimated equation. This “indirectapproach” isparticularlyuseful ifdirectmeasuresof traderestrictionsaresparseor imprecise,as isoftenthecaseforNTMs(ChenandNovy,2012).

increase tradeby2 to3percent.TheroleofNTMs inreducing trade is further highlighted by the followingexamplescitedinAndriamananjaraetal.(2004).Fortheapparelsector,pricesintheUnitedStates,theEuropeanUnion and Canada were 15 per cent, 66 per cent and25percenthigher,respectively,duetothepresenceof

NTMs. In South-East Asia, South Asia and Japan,paper products were 67 per cent, 119 per cent and199percentmoreexpensiverespectivelyduetoNTMs,whileNTMsonleathershoesraisedtheirpricesinJapanby 39 per cent and in Mexico/Central America by80percent.


138

In the agricultural sector, non-tariff measures onvegetable oils and fats increased their prices inMexicoby30percent, inSouthEastAsiaby49percentandinSouthAfricaby90percent,accordingtoAndriamananjara et al. (2004). Analysing bilateralindustry-specific trade flows for countries in theEuropean Union, Chen and Novy (2011) find thatamong the different NTMs, TBT measures are themost important factor. An analysis of the tradeeffects of TBT/SPS measures, in particular, ispresentedinSectionD.2.

The resultsdescribedabovehighlight the importanceofnon-tariffmeasuresinanabsolutesense.Butwhatdo thedata reveal about the significanceofNTMs inrestricting trade relative to tariffs? Kee et al. (2009)findthatfor55percentof tariff lines intheirsamplesubjecttocoreNTMs,theAVEofthesecoreNTMsishigherthanthetariff.Similarly,in36percentoftarifflinessubjecttodomesticagriculturalsupport,theAVEof domestic agricultural support is higher than thetariff. Furthermore, aggregating core NTMs anddomestic agricultural support across all tariff linesunderconsiderationinanoverall traderestrictivenessindex, Kee et al. (2009) find that NTMs – averagingacross countries – almost double the level of traderestrictivenessimposedbytariffs.Infact,inabouthalfof the countries in the sample, the contribution ofNTMs to the overall level of trade restrictiveness ismuchhigherthanthecontributionoftariffs.

Usingtwoindicesoftraderestrictivenessthatestimatehow trade policies affect a country’s imports – thetarifftraderestrictivenessindex(TTRI)andtheoveralltrade restrictiveness index (OTRI), where the latterincludes the effect of both tariffs and non-tariffmeasures – Hoekman and Nicita (2011) find that,averagingacrosscountries,a10percentreductionintheTTRIincreasestradevolumesbyalittlemorethan2percent,whiletheremovalofNTMsincreasestradeby an additional 1.8 per cent.5 This discussionillustrates that NTMs are an important restriction ontrade, even more important than tariffs in severalcountries. Measuring restrictiveness faced byexporters in all destination markets, Hoekman andNicita(2008)comparethemarketaccessversionsoftheTTRIandtheOTRItoshowthattheAVEofNTMsisgenerallymuchhigherthanexistingtariffs.6

In a recent report, UNCTAD (2012) argues that non-tariff measures contribute much more than tariffs tooverall trade restrictiveness. Inparticular, it finds thatNTMscontributemorethantwiceasmuchastariffstooverallmarketaccesstraderestrictivenes.7Thisresultmust be viewed with caution because unlike thestudies described above (which compare NTMs andtariff data in 2001), the UNCTAD report compares2001NTMdatawith2010tariffdata–aperiodoverwhich tariffs have fallen. Hence, the contribution ofNTMs tooverall trade restrictiveness is likely tohaveincreased,assumingthatNTMsdidnotdeclineduring

thesameperiodand that the trade-restricting impactofNTMsdidnotfallbymorethanthatoftariffs.

In fact, using product-level analysis, a study by Hennand Mcdonald (2011) finds that while trade flows fellby5percentasaresultofbordermeasures,suchastariffs, implementedduring the recentfinancial crisis,theyfellby7percentasaresultofbehind-the-bordermeasures (i.e. non-tariff measures). Even within thecategory of border measures, the authors find thattariffsandothertraditionaltradepolicymeasureshavehad a relatively small impact on trade flows, whereasNTMs such as anti-dumping duties have had asubstantialeffect.

(ii) NTMs: variation across countries and sectors

Keeetal. (2009)findthatthevariation intheAVEsofnon-tariff measures across countries is large. Forexample, thesimpleaverageAVEofcoreNTMsvariesfromalmost0to51percent,andfrom0to39percentwhenimport-weighted.TheAVEsfordomesticsupportaregenerallybelow1percent.Thecountrieswith thehighestaverageAVEofcoreNTMsareall low-incomeAfrican countries, including Algeria, Côte d’Ivoire,Morocco,Nigeria,Tanzania,andSudan.Severalmiddle-incomecountries,suchasBrazil,Malaysia,MexicoandUruguay,alsohaverelativelyhighAVEsofcoreNTMs.8The countries with the highest AVEs of agriculturaldomesticsupportareEUmembers.9

AccordingtoKeeetal.(2009),whenconsideringbothcore non-tariff measures and agricultural domesticsupport, the AVEs of NTMs increases with GDP percapita,althoughsomemiddle-incomecountriesseemto have the highest AVEs of NTMs. However,Figure D.1 shows that there is no discerniblerelationshipbetweentheAVEofNTMsandthelevelofGDPpercapitaacrosscountries.Thisisconfirmedbyregression analysis which shows that the associationbetween the AVE of NTMs and the level of GDPper capita is not statistically significantly differentfromzero.10

At the same time, Hoekman and Nicita (2008) findthat tariffs arenegatively associatedwitha country’slevel of income per capita. This evidence, combinedwith the result in Figure D.1, suggests that thecontributionofNTMstotheoverall levelofprotectionis likely to increase with the level of GDPper capita,i.e. as countries become richer, the traderestrictiveness of NTMs relative to tariffs increases.The findings of UNCTAD (2012), which show thatNTMs are relatively more restrictive in high- andmiddle-incomecountriessupportthisinterpretation.

TheworkbyKeeetal. (2009)also reports significantvariation in the AVEs of non-tariff measures acrosstarifflines,amountingtoanaveragelevelof27percentforagriculturalproductscomparedwith10percentfor


139

D. TH

E TR

AD

E E

FFEC

TS O

F

NO

N-TA

RIFF M

EA

SU

RE

S

A

ND

SE

RV

ICE

S M

EA

SU

RE

Smanufactured goods. The greater trade-restrictingimpact of NTMs for agricultural goods relative tomanufactured products is reinforced by the results ofHoekman and Nicita (2008). They also show that therestrictiveness of NTMs for agricultural trade isespeciallyimportantindevelopedeconomies.

However, using data for 2001 to estimate the tradeeffect of non-tariff measures on prices directly in aneconometricmodel,Andriamananjaraetal.(2004)findalmost no statistically significant impact for theagriculturalsector.11Theauthorsexplainthatthismaybe attributable to the definition of NTMs used in thestudy,whichincludesimportquotas,prohibitions,non-automatic licensing, voluntary export restraints,environmental standards and SPS measures, butexcludes tariff-rate quotas. The latter are likely to bethe economically binding constraints on agriculturaltrade.12

Andriamananjaraetal. (2004) identifyapparelas thesector with the largest number of significant NTMs.They estimate a simple average AVE of NTMs of73 per cent across countries. The correspondingestimateinKeeetal.(2009)is39percent.Thehigherorder of magnitude in Andriamananjara et al. (2004)may be explained by the fact that they excludeproducts for which they found a very small impact ofNTMsonimportsordomesticprices.Andriamananjaraetal.(2004)identifypaperproducts,leatherproducts,and vegetable oils and fats as other sectors withmultiplesignificantNTMs.

(b) Methodologicallimitations:Aproblemoftransparency

The previous section outlined the existing empiricalliterature which quantifies the impact of non-tariffmeasures on trade by estimating an ad-valoremequivalent.ItshouldbenotedthattheuseofAVEs–and

thechoicetomodeltheeffectsofNTMsasanegativetax for subsidies, and as a tariff for trade-restrictingNTMs – can be misleading at times. For example, theequivalencebetweentariffsandquotasbreaksdowninthe presence of market uncertainty. Furthermore, theAVEofNTMsdoesnotcaptureanyrelevantfixedcosts,suchasthoseassociatedwithmeetingcertaintechnicalregulations.Beyondtheselimitations,quantificationisachallenging exercise. The methods developed in theliteraturesufferfromanumberoflimitations.

(i) Price gap method

Acomparisonoftwopricestoinferthetradeeffectofa non-tariff measure is indicative of the lack oftransparency associated with the use of NTMs.Unfortunately, given insufficient data on differentprices, even theestimationof apricegap is far fromstraightforward.

The appropriate prices to compare when measuringthepricegapattributabletomostnon-tariffmeasuresare the invoice (c.i.f.) price of the imported good andthe price of the domestic alternative (Deardorff andStern, 1998). However, in reality, the observabledomesticpriceofagoodtypicallydoesnotdistinguishbetweendomesticproductsandimports.Itmeansthatthe actual comparison is between the invoice (c.i.f.)priceandthepriceofthegoodinthedomesticmarket,whether produced at home or imported. This isproblematicfortworeasons.

First,atacertain levelofaggregation,goodsthatareimported into a country are seldom identical to “like”goods produced domestically. The two may be poorsubstitutes for each other – for example, because ofqualitydifferences.Secondly,evenifthedomesticandimported good are perfect substitutes, the price gapmay be suppressed to the extent that the imports ofthe same good from other countries are subject to anon-tariffmeasure.

An additional issue relates to the choice of domesticpricestouseincomputingthepricegap.Manystudiesuseretailpricedatasimplybecausetheyareeasiertoobserve than prices at other stages of the supplychain. Retail price data contain transport, wholesaleand retail margins. Although these can potentially beseparatedout,theyintroduceconsiderableuncertaintyin the identification of the NTM mark-up.13 It is alsodifficulttonetoutthepriceincreaseduetoconsumers’willingnesstopayforhigherquality.

Furthermore, once a price gap is calculated for aparticular good in a particular market, it provides asingle measure of the trade effect of non-tariffmeasures.Sowhenthereisasingle,transparentNTM,the tariff equivalent reflects the effect of that policy.However,inthecaseofmultipleNTMs,thesinglepricegaportariffequivalentreflectsthecumulativeeffectsofallNTMsthatarepresentinthemarket.Thismakes

FigureD.1:AVEs of NTMs and economic development

5 6 7 8

Log of GDP per capita

Log

of A

VE

of N

TMs

9 10 114

-4

-3

-2

-1

0

-5

-6

-7

-8

Source:Author’sestimatesbasedondatafromKeeetal.(2009)


140

it difficult to ascertain the percentage of the priceincrease that is attributable to each of the separateNTMs. It may be that there is one NTM which, whenremoved, eliminates most of the distortion. If so, theprice gap would largely reflect the effect of thisparticularNTM.

Conversely, it may also be true that the removal of anon-tariffmeasuredoesnotpermitmarketaccess. Inthiscase,the“true”tariffequivalentofasinglepolicychangemay in factbe zeroevenwhen themeasuredtariffequivalentofallNTMsjointlymaybequitelarge(Ferrantino,2012).Finally,thepricegapmethodisonlysuitable for analysing NTMs of a single importingcountry for a few products of particular interest. Thedata requirements to address NTMs across multiplecountriesandproductscanbeunmanageable.

(ii) Econometrics-based method

Anotableadvantageofeconometricanalysis,relativetothe “pricegap”method, is that itcanbeused tostudythetradeeffectsofmultiplenon-tariffmeasuresacrossmultiple industries and countries simultaneously. Inaddition, therelativeabundanceofdataontradeflowsmakes it particularly attractive for analytical purposes.However, the econometrics-based methods havecertainshortcomingsaswell.

First, given the lack of transparency, observing non-tariffmeasuresprecisely isdifficult.Hence,adummyvariablewhichequalsoneifthemeasureispresentisunlikelytocaptureseveralNTMs.UsingthedifferencebetweenactualandpredictedimportsasameasureofNTMs is also problematic because it may capturefactorsotherthantradepolicies.

Secondly, like the “price gap” method, this approachcannot disentangle the individual effects of a singlenon-tariffmeasurewhenmultipleNTMsarepresentinamarket.Inmanycases,however,onlyoneNTM–orasmallnumberofNTMs–isappliedtoanygivengood.Cross-countryvariationintheapplicationofNTMscanthen potentially be used to disentangle their tradeeffects (Carrère and De Melo, 2009). Thirdly, theresultsobtainedarelikelytobesensitivetothedetailsoftheeconometrictechniquesused.

(iii) Global supply chains

Themeasurementexercisesdiscussedintheprevioussub-section do not explicitly address the advent ofinternational production networks. They assume alinear supply chain in which a single good is movedfrom place to place without being transformed.However, with the location of different stages ofproduction in different countries, it takes many morecross-bordertransactionstoprovideasingleunitofafinal good than before. This is particularly true formanufactured goods with multiple components, suchaselectronicsandmotorvehicles.

Consider the global supply chain of producing acomputer disk drive as discussed in Hiratsuka (2005)and Baldwin (2008). The disk drive is assembled inThailand,whichactsasthehubofthesupplynetwork,using 43 components from ten other countries inaddition to 11 components produced in Thailand.Hence,thereareatleasttenmovesacrossinternationalborders,andperhapsmore,dependingontheextenttowhich shipments can be bundled. Furthermore, sincethe disk drive will be shipped to the location of finalcomputerassembly(e.g.China),wheretheothermajorcomputer components are gathered, the number ofcross-bordermovesmultipliesevenfurther.

Importantly,inaglobalsupplychainthatrequiressemi-finished goods to move back and forth acrossinternational borders more than once, the effects ofnon-tariff measures (and other trade costs) arecompounded.Thisimpliesthattheeffectofamarginalincrease in trade costs is much larger than would bethecaseiftherewereasingleinternationaltransaction.Box D.2 illustrates this argument with a numericalexample. In addition, the price increase at each stepwould include not only the monetary costs of movingalong thesupplychain,but thecostsassociatedwiththewaitingtimeaswell(Ferrantino,2012).

(c) Servicesmeasures

Themethodologyemployedtoassessthetradeimpactof services measures follows that used in goods. Inaddition,thetradeliteraturealsodevelopsanapproachbased on the construction of Services TradeRestrictivenessIndices(STRIs).14Anumberofstudiesuse these indices to estimate the price effects ofservicesmeasures(controllingforallrelevantindustryand economy-wide determinants of economicperformance of firms) for several services sectorsacross a large sample of countries (McGuire, 2008;FrancoisandHoekman,2010).

(i) Empirical estimates

For a sample of 78 countries across four servicessectors,Walsh(2006)findsanaveragetariffequivalentof72percentforservicesmeasures.15Analysingdatafor 11 services sectors16 across 63 countries, Guillin(2011) finds a much lower average tariff equivalent ofaround40percent.Acomparisonof theseestimates,however, is not very meaningful because differentstudies use different data samples and differentparametersintheeconometricspecification.

Ingeneral,itappearsthatrestrictionstoservicestradeare higher in developing countries than in OECDcountries (Walsh, 2006; Francois et al., 2003;Fontagné et al., 2010). At the same time, traderestrictions in services do not appear to besystematically associated with a country’s level ofdevelopment.Forexample, theworkof theAustralianProductivity Commission shows that some OECD


141

D. TH

E TR

AD

E E

FFEC

TS O

F

NO

N-TA

RIFF M

EA

SU

RE

S

A

ND

SE

RV

ICE

S M

EA

SU

RE

S

countries have restrictions comparable with theaveragesprevailinginmajordevelopingeconomies.

Furthermore, Gootiiz and Mattoo (2009b) find thatalthoughhigh-incomecountriesarequiteopenoverall,there is much more variation in the restrictiveness ofservices trade in developing countries. The authorsshow that some low-income countries in Asia andAfrica are relatively open. So too are some middle-incomecountriesinLatinAmerica,AfricaandEasternEurope. In contrast, some of the most restrictiveservices measures are found in the fast-growingeconomiesofAsiaaswellasintheMiddleEast.Otherstudies also find the emerging economies in Asia tohaverelativelyprotectionistservicesmeasures(Walsh,2006;Park,2002;McGuire,2008).

It appears that variations in the restrictiveness ofservices measures across countries may depend ontheparticularsectorunderconsideration.Forinstance,Indonesia’s tariff equivalent in business servicesappears to be lower than that in more developedcountries, such as Japan and the Republic of Korea,but higher in construction services (Park, 2002;Guillin, 2011). Similarly, analysing members of AsiaPacific Economic Cooperation (APEC) in 1997,McGuire (2008) found that while the United Stateswas among the least restricted markets intelecommunications services, it was among the mosthighly restricted in maritime services. At the sametime, middle-income economies in South Americawerefoundtohaverelativelyhighrestrictivenessindexscoresforfinancialservices,butwereamongtheleastrestrictedmarkets indistribution, telecommunicationsandprofessionalservices.

According to a set of studies, averaging acrosscountries, transport and business services appear tobe the most open sectors, with an average tariffequivalentof21percentand28percentrespectivelyfor services measures. The most protected isconstructionservices,withanaveragetariffequivalentof 58 per cent (Park, 2002; Fontagné et al., 2010;Guillin, 2011). In a different study, however, foreigndirect investment (an important mode of trade inservices) in transport services is among the mostrestricted, while that in construction services is theleast restricted (UNCTAD, 2006). The contradictoryresultssuggestthattheaccuracyandreliabilityoftheaforementioned estimates of the restrictiveness ofservices measures may be questionable. This lack ofprecision and consistency may be attributable to anumberofmethodologicallimitations.

(ii) Methodological limitations

In analysing the trade-restricting effect of servicesmeasures,anestimatedAVEmusttakeintoaccountthepossiblesubstitutionbetweendifferentmodesofsupplywhen one particular mode is affected. For instance,there may be a switch from mode 3 trade (a foreigncompanysettingupsubsidiariesorbranchestoprovideservicesinanothercountry)tomode2trade(consumersorfirmsmakinguseofaservice inanothercountry) inhigher education services as a result of restrictiveservices measures affecting the former (Dee, 2010).Such intermodal substitution is likely in the case ofinsurance services as well (from mode 3 to mode 1,servicessuppliedfromonecountrytoanother).

In order to derive a meaningful AVE, other policyinterventionsthataffect thetrade-restricting impactof

BoxD.2: Cumulation of trade costs in a global supply chain

Supposethatthetotalvalue-addednecessarytoproduceaproductisequaltoone.Theproductisproducedinstages in“n”countries,eachofwhichadds (1/n) to the totalvalueof theproduct.Afterproduction, theproductisexportedtoafinaldestination,sothatitismoved“n”timesaltogether.Letthecostofanon-tariffmeasureonmoving theproduct fromonecountry to another equal “t” onanad-valorembasis.Hence, ateachstage, thecost “t” ischargedon theentirevalueof theproductproducedup to thatpoint, includingprevious trade costs. The total cost of the product (produced in n stages) when delivered to the finalconsumerisrepresentedbyc(n),sothat:

Box D.2: Cumulation of trade costs in a global supply chain Suppose that the total value-added necessary to produce a product is equal to one. The product is produced in stages in “n” countries, each of which adds (1/n) to the total value of the product. After production, the product is exported to a final destination, so that it is moved “n” times altogether. Let the cost of a non-tariff measure on moving the product from one country to another equal “t” on an ad-valorem basis. Hence, at each stage, the cost “t” is charged on the entire value of the product produced up to that point, including previous trade costs. The total cost of the product (produced in n stages) when delivered to the final consumer is represented by c(n), so that:

𝑐𝑐 1 = (1 + 𝑡𝑡) 𝑐𝑐 2 = !

!1 + 𝑡𝑡 ! + !

!(1 + 𝑡𝑡)

𝑐𝑐 3 = !!(1 + 𝑡𝑡)! + !

!(1 + 𝑡𝑡)! + !

!(1 + 𝑡𝑡)

( )in

it

nnc +=∑

=

11)(1

1

𝑐𝑐 𝑛𝑛 =1𝑛𝑛1 + 𝑡𝑡 !

!

!!!

SupposethattheAVEofanNTMateachstageis10percent,i.e.“t”=0.1andc(1)=1.1.Astheglobalsupplychainisfragmentedfurther,tradecostscompoundfairlyquickly:c(5)=1.343(anAVEof34.3percent)andc(10)=1.753 (anAVEof75.3percent).Moreover,marginal increases in tradecostsarecompounded.Forinstance,iftheAVEofNTMs“t”increasesfrom0.1to0.2,adoublingateachstageofthesupplychain,tradecostsalongthesupplychainmorethandouble,withmorecompoundingformorefragmentedsupplychains:c(5)=1.786andc(10)=3.115.


142

aservicesmeasurealsoneedtobetakenintoaccount.For example, in the case of international air services,firmsmay respond toownership limits imposedby thewithholding clauses in air services agreements(affecting mode 3 trade) by negotiating code-sharingarrangements.Moreover, ifmode3 is thepredominantmode of trade (as it is for telecommunications, forexample),highfixedcostsofmarketentry/establishmentwouldnotevenbecapturedbytheconceptofa“tariffequivalent”.

In addition, the methodological limitations associatedwithanalysingthetradeeffectsofnon-tariffmeasuresarealsoapplicabletoservicesmeasures.Forexample,giventhelackoftransparency,itisdifficulttoobserveprecisely different services measures. Attributing thedifference between actual and predicted imports(derivedfromaneconometricestimation)totheimpactof services measures highlights this problem.Furthermore,theremaybemultiplerestraintsontradein services, and it may not be clear which are

economicallybindingandwhicharenot.RepresentingtheseNTMsasanAVEcanthusbemisleadingforthisreasonaswell.Theuseofsubjectivecriteriatoweighthe relative importance of diverse measures whenconstructingSTRIsalsoillustratesthemethodologicaldifficulties involved in estimating the price effects ofservicesmeasures.

Finally, AVEs of services measures calculated usingservices trade flows do not take into account theindirecteffects that thesemeasureshaveon trade ingoods.Sucheffectsarelikelytobestrongbecauseofthe complementarities between goods and services(see Box D.3). For example, a services measure thatrestricts trade and competition in transport andlogistics services has a negative impact onmerchandise trade. However, this is not taken intoconsideration when AVEs of services measures arecalculated using services trade flows only. The rolethatservicestradeplaysinglobalsupplychainsmakesthisanimportantproblem(seeSectionB.3).17

BoxD.3: Complementarities between trade in services and trade in goods

Evidencesuggeststhatexportcompetitivenessinmanufacturingsectors,suchasmachinery,motorvehicles,chemicalsandelectricequipment,ispositivelyassociatedwithinwardforeigndirectinvestmentandimportsof business services (Francois and Woerz, 2008) and negatively affected by regulations that hinder suchtrade(Nordås,2010).Suchcomplementaritybetweentradeinservicesandtradeingoodsmaybeexplainedbyvariousmechanisms.

Afirstmechanism isconstitutedby transportand logistics links.Transportand travelservicesaccount forabouthalfofcross-bordertradeinservicesandarethemostimportantdirectservicesinputtointernationaltrade in goods. For instance, Yeung et al. (2012) find that Chinese manufacturing firms that make use ofthird-partylogisticssuppliers(largelyfromHongKong,China)tendtoperformbetterinexportmarketsthanfirmsthatdologisticsin-houseorpurchasethemlocally.Evidencealsosuggeststhatmeasuresthatrestricttrade and competition in transport and logistics services have a negative impact on merchandise tradeperformance. Market power in the shipping industry, for example, raises trade costs, particularly fordevelopingcountries(Hummelsetal.,2009).

Secondly, goods and services are often bundled in final markets. After-sales services, for instance, areimportantforahostofdurablegoodssuchascars.Aviationengines,printers,vendingmachines,andotherequipmentarealsoincreasinglyrentedorleasedwithaservicescontract.Anotherrecenttrendistoconsidergoodsmainlyasaservicesplatform.Mobiletelephones,forinstance,areoftensoldforanominalamountonthe condition that customers sign up for a fixed-period service contract. When goods and services arecomplementaryorbundled,servicesmeasuresstronglyaffectthetradedgoodinquestionaswell(Lodefalk,2010).EvidencesuggeststhatmanufacturingfirmsinSwedenandtheUnitedKingdom(andalsominingandoilcompanies intheUnitedKingdom)arevigoroustraders inservices,andthattheservicesshareoftheirtotalrevenuehasincreasedovertime(Lodefalk,2010;BreinlichandCriscuolo,2011).

Thirdly,thecomplementaritybetweentradeingoodsandtradeinservicesisincreasedfurtherbytheroleofintermediaries(retailersandwholesalers)ininternationaltrade.18Bernardetal.(2010)findthat35percentofUSexportersarewholesalers,accountingfor10percentofthevalueofUSexports.Similarly,morethan25percentofItalianexportersareintermediaries,accountingfor10percentofthevalueofItalianexports.19Intermediaries, such as leading multinational retailers tend to source their products directly frommanufacturersorfarmers,andtypicallyhaveacentralizedsourcingunitservicingallsalesoutlets,globallyorregionally.Hence,theytendtocontributetoincreasedtradeinconsumergoodsbetweentheirhomecountryandthehostcountriesoftheiraffiliates.20

Marketconcentrationinasectorcomprisingintermediariesmayalsoaffectmerchandisetrade.Forexample,intheeventoftradeopening,retailerswithsignificantmarketpowermayfailtopassreducedtradecostsontoconsumers(FrancoisandWooton,2010).Atthesametime,regulatoryheterogeneity(suchasdifferencesinproduct standards, labellingand recycling requirements)may imposeconsiderablecostson retailersbyrequiringthemtomodifyproductsforeachdestination.21


143

D. TH

E TR

AD

E E

FFEC

TS O

F

NO

N-TA

RIFF M

EA

SU

RE

S

A

ND

SE

RV

ICE

S M

EA

SU

RE

S

2. DisentanglingtradeeffectsofTBT/SPSmeasuresanddomesticregulationinservices

This section focuses on TBT/SPS measures andequivalentdomesticregulationinservices,andreviewswhatweknowabouttheireffectsontradeflows.Onereason to focus on this sub-set of measures is thateconomic theory does not provide simple predictionsas to their trade effects. Assessing their effects istherefore an empirical issue. In contrast, economictheoryprovidesclearguidanceastothetradeeffectsof other non-tariff measures – for example, importquotas reduce imports, export duties reduce exports,whileexportsubsidiesincreaseexports.

Another specific characteristic of these measures isthat they are commonly regarded as having animportant fixed-cost component, which significantlydifferentiatesthemfromtariffs.Forexample,toadapta product to new technical requirements may requirean initial investment independent of the level ofexports.Thepresenceofafixedcosttoenteramarketmay,however,haveeffectsontradethataredifferentfromatariff,andthisaspectdeservesattention.

Inparticular,thissub-sectionexamineshowTBT/SPSmeasures and domestic regulation in services affectthe volumeof tradeand thedecisionsaboutwhetherto export to a certain market. This sub-section alsoconsiders whether these measures affect developinganddevelopedcountriesdifferentlyandwhethertheseeffectsdifferby sectorandfirm.Wherepossible, theimpactofthesemeasuresontradeinindustrieswherethe production process is fragmented is highlighted.Finally,anattemptismadetodistinguishbetweentheimpactofthemeasuresthemselvesandtheimpactofthewayinwhichmeasuresareimplemented.

Economictheoryandassociatedempiricalresearch,ingeneral, do not distinguish between mandatory andnon-mandatory TBT/SPS measures, and the termstandardisoftenusedtodenoteboth.Intheabsenceofatheoreticalpredictionastoadifferentimpactofamandatory versus a non-mandatory measure – evenwhenusingdatabases thatcoveronlynon-mandatorystandards or only mandatory ones – the results areinterpretedmoregenerallyasthe impactofTBT/SPSmeasures.

Empirical economic literature clearly distinguishes,however, between national or country-specificstandards (standards that are different from those inanothercountry)andsharedstandards(standardsthatare identical or equivalent between two countries,including international and regional standards). ThedistinctionismadetodisentangletheimpactontradeofharmonizationofTBT/SPSmeasures.Thereviewoftheliteratureinthissectionfollowsthisapproach.

As far as services are concerned, the economicliterature generally looks at overall indexes of therestrictiveness of domestic regulation – and includesmeasuresthatgobeyondthefocusofthisreport.Thefollowing review of the relevant studies mainlyhighlights an important gap in the existing empiricalliterature.

(a) Overalleffectontrade

WhenexploringtheeffectsofTBT/SPSmeasuresontrade, one would ideally like empirical evidence todistinguishamongdifferenttypesofmeasures.Thisisbecause TBT/SPS measures affect trade throughdifferentchannels.

For example, the introduction of product safetyregulationwill increaseproductioncostsbutcanalsoserveasanimportantqualitysignal,therebyhelpingtopromote the competitiveness of those products thatmeet stringent standards. Product safety regulationsalso increase trust in the quality of foreign products,thus reducing transaction costs and fostering trade.Whether these effects will translate into higherimports or export depends on the effect of themeasureontherelativecostsofdomesticandforeignproducts,andonthewillingnessofconsumers topayhigherpricesforsaferproducts.

Asafurtherexample,considerthecaseofcompatibilitystandards. Innetworkindustries,wherethevalueofaproduct increaseswith thenumberofconsumersandcomplementary goods, compatibility standards arelikelytoincreasetrade.Withoutsuchstandards,thesemarkets may oversupply varieties and the networksizesmayremaintoosmall.Standardsinthesemarketsare generally voluntary and can help consumersacquire information about preferences abroad, andhelp producers to coordinate their activities moreefficiently. This general prediction needs to bequalified, however, since compatibility standards canalsoreflectanti-competitivebehaviour.

Except for environmental and food safety regulation,the existing trade literature does not distinguishamong different types of measures (for example,whethertheyaddressasafetyorcompatibilityconcern,orwhethertheydefinethecharacteristicsofaproductor a testing procedure). Rather, the literature hastendedtorelyonanindexofstandardizationactivities– usually the number of standards or the number oftechnicalmeasuresmaintainedbyacountry.ThefocushasthenbeenontherelationshipbetweenthisbroadmeasureofTBT/SPSmeasuresandtradeflows,oronthecost-raisingimpactofthesemeasures.

Notwithstanding these limitations, the existingempirical literature finds that, at the aggregate level,TBT/SPSmeasuresmaynotbeassociatedwithlowertrade. For example, in a pioneer study on therelationship between standards and aggregate trade


144

performance,Swannetal.(1996)foundthatstandardspromoted trade. They estimated that a 10 per centincrease in the number of country-specific standards(as opposed to “shared” standards)22 increased UKimportsfromtherestoftheworldby3.3percentandexportsby2.3percent.Withadifferentspecificationof themodel,but thesamedataset,TempleandUrga(1997) found an insignificant effect of standards ontrade. Although their findings differed, both studieschallenged the predominant view that nationalstandardsrestricttrade.

Literature that looks at licensing and qualificationrequirementsandproceduresandtechnicalstandardsinservicesisverylimited.Itwouldappearthattheonlystudy that attempts to measure the effects of suchdomesticregulation isKoxandNordås(2007). In thefirstpartoftheirstudy,theauthorsuseareconstructedProduct Market Regulation (PMR) index23 based ontheselectedindicatorsthatintheirview“comeclosestto covering the types of regulation mentioned in[GeneralAgreementonTradeinServices]ArticleVI.4”,that is, domestic regulation as defined in this report.WhiletheestimatedtradeeffectofthisreconstructedPMRonoverallservicestrade(coveringmodes1and2 and mode 4, individuals travelling from their owncountry tosupplyservices inanother) isnegative, theestimatedcoefficienton“licencesandpermitssystem”(that is mostly closely related to domestic regulationasofGATSArticleVI.4)ispositive,thoughsmall.

In otherwords, burdensome licensingproceduresarefound to increase services trade. One possibleexplanation is that restrictive licensing proceduresinduce intermodal substitution between export andforeign direct investment (FDI). The finding that thestringency of the “licences and permits system”indicator reduces inward and outward FDI supportsthis view. In the second part of the study, Kox andNordås (2007) use banking regulatory indexesdeveloped by the World Bank (Barth et al., 2008).They show that regulation aiming at ensuringappropriate standards (such as accounting standardsand financial statement transparency) is positivelyassociatedwithcross-bordertradeandFDIinfinancialservices.24

(b) Differencesacrosssectorsandcountries

StudiesbasedondisaggregatedtradedatashowthattheeffectofTBT/SPSmeasuresdependsonthetypeofsector.Oneofthesestudies isbyMoenius(2004).Using a gravity model25 to assess the impact ofnational standards on trade for a dataset covering471 sectors at the four-digit Standard InternationalTradeclassification(SITC) levelandbilateraltradefor12 developed countries, he finds that import-specificstandards have a negative impact on imports inthe non-manufacturing sectors (namely, food,beverages, crude materials and mineral fuels), but

haveapositiveimpactonimportsinthemanufacturingsector (including oils, chemicals, manufacturing andmachinery).

Moenius’s interpretation of the results is thatstandards, by providing exporters with valuableinformation about market preferences, reducetransactioncostseveniftheyimposeadaptationcosts.In more differentiated sectors, such as certainmanufacturing sectors (for example, high-technologysectors), information costs may be higher. Therefore,informationcosts’reducingeffectoutweighadaptationcosts’increasingeffectandtradeincreases.

Moenius’s (2004) conclusions are supported byseveral studies. For example, Blind (2001) finds apositiveandsignificanteffectofstandardsontradein“instruments for measurement and testing”, as doesMoenius (2006) for “electrical products”. Usinginformation on the measures notified under the SPSand TBT agreements, Disdier et al. (2008b) find anoverallnegative impactofSPSandTBTmeasuresontradeinagriculturalproducts.

Focusing on notified TBT/SPS environment-relatedmeasures(ERM)(seeBoxD.4),Fontagnéetal. (2005)also tend to find a positive effect of ERM onmanufacturing trade, but a negativeeffect on trade infreshandprocessedfood.Morerecently,LiandBeghin(2012)performananalysisof27papersthatusegravityequationstoestimatetheeffectofTBT/SPSmeasuresontrade.Theyfindthatestimatesofthetradeeffectsofthesemeasuresonagricultureand food industriesarelesslikelytobepositivethaninothersectors.

In linewiththegeneralfindingofanegativeeffectofTBT/SPSmeasureson trade in agricultural products,thetradeliteraturethatusesmaximumresidual levels(MRLs)ofpesticidesasanindicatorofthestringencyofSPSmeasuresconsistentlyfindsnegativeeffectsofMRLson imports.Otsukietal. (2001)findanegativeeffect of the EU standard on aflatoxin on Africanexports. In particular, they estimate that moving fromthe Codex Alimentarius standard, established by theUNFoodandAgricultureOrganizationand theWorldHealth Organization, to the more stringent uniformEuropean Commission standard decreases Africanexportsofcereals,driedfruits,andnutstoEuropebyUS$ 670 million. Wilson and Otsuki (2004) find asimilar effect for MRLs on chlorpyrifos on bananasexportsfromLatinAmerica,AsiaandAfricatoOECDcountries.

Chenetal.(2008)findanegativeeffectofregulationsontheutilizationofpesticidesandmedicatedfishfeedon Chinese exports of fresh vegetables, fish andaquatic products between 1992 and 2004. Inparticular,theyfindthata10percentstrictermeasurein the level of pesticides (medicated fish feed)decreases vegetable (fish and aquatic product)exportsby2.8(2.7)percent.


145

D. TH

E TR

AD

E E

FFEC

TS O

F

NO

N-TA

RIFF M

EA

SU

RE

S

A

ND

SE

RV

ICE

S M

EA

SU

RE

S

Several studies show that any negative effects ofTBT/SPSmeasuresontradeareconcentratedmainlyindeveloping-countryexportstodevelopedcountries.Incontrast,exportsfromdevelopedcountriestootherdeveloped countries are not significantly impeded bythesemeasures.26

For example, focusing on SPS measures, Anders andCaswell (2009) find substantially different effectsbetween developed and developing countries. Theyestimatethetradeimpactofmandatory“hazardanalysisandcriticalcontrolpoints”(HACCP)27requirementsforseafood products in the United States between 1990and 2004. US seafood imports across all exporterswere reduced. SPS measures caused a loss in tradevalueofbetweenUS$11.4milliontoUS$30.6million.The impact on developing countries as a groupamountedtoanexportvaluereductionof0.9percentunderHACCPstandards,whiledevelopedcountriesasagroupgainedfromthemeasure.

However, there is wide variation across developingcountries. Anders and Caswell (2009) find that largerseafood exporters gained trade shares with the UnitedStates, while smaller exporters lost ground. Developingcountries were found among both the gaining and thelosinggroup.ThetradeimpactofSPSmeasuresappearsto depend in part on the size of the exporter. Similarly,

examining the trade effects of notified SPS and TBTmeasures adopted by the United States, the EuropeanUnion,Japan,Canada,AustraliaandSwitzerland,Disdieret al. (2008a) find an overall negative effect on totalexportsfromAfrican,CaribbeanandPacific(ACP),LatinAmerican and Asian countries. While ACP countryexports appear to have been significantly negativelyaffected by such measures, the impact on Asiancountriesisnotstatisticallysignificant.

Empiricalresearchondomesticservicesregulationhasnotexaminedwhetherthesemeasureshaveadifferentimpactondevelopedanddevelopingcountries.This ismainly due to lack of data on services measures fordeveloping countries. As regards differences acrosssectors,theabove-mentionedstudybyKoxandNordås(2007) finds that regulation matters more for “otherbusinessservices”(includinglegalservices,accounting,architecture and engineering) than for “total services”(asmeasuredbytotaltradethroughmodes1,2and4).Thisisconsistentwiththeimportantrolethatbusinessservicesplay inproductionchainsandhowamarginalincreaseintradecostscanhaveamagnifiedimpactonoverall trade costs when the production process isfragmented across countries (see Section D.1 andBoxD.2).

BoxD.4: Environment-related measures

One of the basic concerns with environmental regulation is that, in a world where countries differ in thestringency of their environmental regulations and industries differ in their pollution intensities, pollution-intensivefirmswilllocateproductioninlessregulatedcountries.Therefore,pollution-intensiveproductswillbeexportedbylessregulatedcountriesandimportedbycountrieswithmorestringentregulation.

IntheirsurveyontheeffectofenvironmentalregulationsonUSmanufacturing,Jaffeetal.(1995)concludedthat therewas littleempiricalevidencethatdifferences inenvironmental regulationsaffected internationaltradeandinvestmentflows.

Morerecentstudieshaveattemptedtoexplainthisfinding,examiningmoredisaggregateddataandtreatingsamplevariationsmorecarefully.Thegeneralfindingisthattheimpactofenvironmentalregulationontradechangesbycountryandsector.Forexample,Ederingtonetal.(2005)arguethatenvironmentalregulationshave strongereffectson thepatternof tradebetweendevelopedanddeveloping economies thanamongdevelopedcountries.

Using data for 21 OECD countries and a gravity model of trade augmented with an indicator of strictenvironmentalregulation,vanBeersandvandenBergh(1997)findthatstrictenvironmentalregulationdoesnot increase imports.However,while theydonotfind thatenvironmental regulations inpollution-intensivesectorshaveasignificantoveralleffectonexports,theydofindthatthesemeasureshaveasignificantandnegative effect for those pollution-intensive sectors that are resource based (being less geographicallymobile).Thefindingthatstricterenvironmentalstandardshaveanegativeimpactonexportsfrompollution-intensiveindustriesisalsoconfirmedinthestudybyOtsukietal.(2001).

Focusing on environment-related measures notified under the SPS and TBT agreements, Fontagné et al.(2005)findthatfortradeinfreshandprocessedfood,thesemeasurestendtorestricttradefromdevelopingcountries and least-developed countries (LDCs). However, exports from developed countries are notrestricted.On theotherhand, for themajority ofmanufacturedproducts, theseenvironmental regulationshave either no significant effect or a positive effect, and this result applies to countries at all stagesofdevelopment.


146

(c) Volumeoftradeandexportmarketsdiversification

TheeconomicliteratureexaminesTBT/SPSmeasuresin goods and equivalent domestic regulation inservices as possible fixed costs of entry in a market(Baldwin, 2000 and 2005, and Deardorff and Stern,2008;KoxandLejour,2005)–thatis,aninitialcosttobepaidtoaccessamarket.Forexample,alargeinitialinvestmentmayberequiredforafirmtocomplywithacertainforeignstandard,butoncethenewtechnologyisacquiredtheremaybenoadditionalvariablecosts.28Similarly,aqualificationorcertificationrequirementforservice-providingpersonnelmayinvolveaninitialfixedcostofobtaining thequalificationor certification,butnoadditionalvariablecosts.

AsdiscussedinSectionB,assumingtheexistenceoffixedcoststoenteracertainmarket,modelsoftradewith heterogeneous firms show that only the mostproductive firms in an industry will export. As tradecosts are lowered, high-productivity exporting firmsexpand. The most productive firms enter exportmarkets,whilelow-productivityfirmsshrinkorexitthemarket.Inthesemodels,thevolumeoftradebetweentwo countries changes both because incumbentexportingfirmsexpandtheirtrade(thusincreasingtheso-called intensivemarginoftrade)andbecausenewfirms enter the foreign market (thus increasing theextensivemarginoftrade).29

Relatively little is known about how TBT/SPSmeasuresaffectindividualfirmsand,inparticular,theirexport decisions. In order to shed light on this issue,thefollowinganalysisstudiesfirms’decisiontoexportto a market and the volume of their exports.30 Theadvantageofusingfirm-leveldataisthatitallowsustodistinguishbetween thenumberof varietiesexportedbyfirms,thenumberofexportingfirms,andthevalueofexportsbyfirms.

Tomeasurethestringencyofregulatorymeasures,thestudy uses the database on specific trade concernsraised by WTO members in the SPS and TBTcommittees.31 While databases typically used32 tocapturetheimpactofTBT/SPSmeasuresincludebothmeasuresthatrestricttradeandthosethatdonot,thisdatabasecontainsinformationonlyonthosemeasuresperceivedtobeapotentialobstacletotrade.Acountrywouldnotraiseaconcernifitdidnotseethatmeasureasanobstacletotrade.33

Drawing on French firms’ custom data34 from 1995to 2005, the study uses a gravity model of trade toevaluate theeffectofSPSandTBTmeasures raisedas specific tradeconcerns onexport performancebyfirms.Thefirms’exportsareassumedtobedeterminedby demand-side factors (such as income), supplyfactors (such as sectoral productivity), trade costs(such as distance) and by an additional variableindicatingthestringencyofSPSandTBTmeasures.35

Although further research is needed to test therobustness of results, preliminary findings show thatTBT/SPS measures raised as concerns in WTOcommittees are associated with a fall in trade. Inparticular,TBT/SPSmeasuresraisedasspecifictradeconcerns appear to reduce the value of exports. Theeffectonthenumberofexportingfirmsisstatisticallynot significant, but the sign of the coefficient isnegative (results of the estimations are reported inAppendixTablesD.1andD.2).

OtherstudiesalsofindthatTBT/SPSmeasureshaveanegative effect on export market diversification. In astudy (not at firm level) focusing on textile, clothingand footwearexports,Shepherd (2007)shows thata10 per cent increase in the total number of EUTBT/SPS measures is associated with about a6percentdecrease in theproductvarietyofexports(measured as the number of six-digit HS productsunderatwo-digitHSsector)totheEU.

UsingdatafromaWorldBankTBTssurvey,Chenetal.(2006) also find that TBT/SPS measures impededeveloping-country exporters’ entry into developedmarkets.Inparticular,Chenetal.(2006)estimatethatthese measures reduce the likelihood of firmsexporting to more than three markets by 7 per cent.The study, however, is based on a sample of only619 firms located in 17 developing countries. Themeasure of a technical barrier to trade is based onfirms answering “yes” to the question “Have quality/performancestandardsimpactedyourabilitytoexportproducts?” In other words, this study finds that firmsthat claim to find TBT/SPS measures an obstacle totradealsotendtoexporttofewermarkets.36

There is also some evidence that the effects ofTBT/SPS measures on export-market diversificationchanges depending on the type of firms. Standardsandtechnicalregulations(ifnotharmonized)appeartobe particularly harmful to trade for firms that importinputs. In fact, outsourcingfirmsappear less likely todiversify their export markets than firms that do notoutsource. The underlying reason may be that, wheninputs are produced, their ultimate destination isunknown and thus they may not meet the technicalrequirements imposed in the market of the finalproduct(Chenetal.,2006).

In addition, TBT/SPS measures appear to negativelyaffect market entry even more for small firms.Focusing on the electronics sector, Reyes (2011)examines the response of US manufacturing firms tothe harmonization of EU product standards withinternational norms. He finds that harmonizationincreases the entry of firms, and that the effect isstronger forUSfirms thatalreadyexport todevelopedcountriesbutnottotheEuropeanUnion.Asexpected,these firms are on average smaller than firms alreadyexportingtotheEuropeanUnion.FocusingonSenegal,Maertens and Swinnen (2009) show that vegetable


147

D. TH

E TR

AD

E E

FFEC

TS O

F

NO

N-TA

RIFF M

EA

SU

RE

S

A

ND

SE

RV

ICE

S M

EA

SU

RE

S

exports have risen sharply despite increasing sanitaryrequirements, resulting in important income gains andpoverty reduction. However, tightening food regulationalsoinducedashiftintheprofileofexportersfromsmallfarmerstolarge-scaleintegratedestateproduction.

Overall, firm-level studies show a negative effect ofTBT/SPS measures on trade, both through a lowervolume of trade per firm and reduced market entry.This result may be partly explained by the type ofvariable used in some of these studies for TBT/SPSmeasures, which tend to capture only restrictivemeasures. In addition, some evidence points toTBT/SPSmeasuresbeingparticularlytraderestrictivefor small firms and outsourcing firms. However, moreresearch is needed to understand how these resultsvaryacrosssectorsandfirms.

There isnofirm-levelstudy lookingspecificallyat theeffectsofdomesticregulation(narrowlydefinedasofGATSArticleVI.4)onexport-marketentryforservices.Using aggregate data, Kox and Nordås (2007) findthatthedeterminantsofmarketentryandthevolumeof trade are largely the same. In particular, domesticregulationsaimedatensuringhigherqualitystandardsin financial services (accounting standards andfinancial statement transparency) appear to beassociated with both higher export values andincreased entry. However, existing evidence onservicesistoolimitedtodrawgeneralconclusions.

(d) Doesconformityassessmentmatterforgoodstrade?

Conformity assessment refers to testing, inspectionandcertification,aswellastoasupplier’sdeclarationof conformity.37 Conformity assessment proceduresare necessary for achieving important policyobjectives,suchastheprotectionofconsumers’healthand safety. They can, however, also be unnecessaryobstaclestotradewhentheyareduplicative,inefficientorappliedinadiscriminatorymanner.

Testing, inspection and certifying compliance with acertain TBT/SPS measure entails costs. These costsare necessary because they assure compliance withthe required standard. Yet, they can also be anunnecessaryobstacletotrade,whenforeignprovidersare competent to provide the required level ofassurance in a cost-effective manner, but thiscompetence is not recognized by the importingcountry. Ideally, attestation of conformity would becarried out just once in a cost-effective manner andthen recognized everywhere. Yet, even the existenceof a well-functioning technical infrastructure in manycountries does not automatically lead to singleconformityassessment,thusunnecessarilyincreasingtransactioncosts(seeSectionB.1).38

There are several dimensions of conformityassessment costs. It is not just that the fees for

testing, inspection or certification may beunnecessarily high. Unnecessary costs also arisebecause exporters need to comply with testing andcertification requirements in each of the countries towhich they are exporting. Even if importing countriesrelyoninternationallyharmonizedproductstandards–oracceptanothercountry’sstandardsasequivalent–theymaystillhaveaseparateconformityassessmentrequirement.Thiscansubstantiallyincreasethecostsofexporting,notleastbecauseexportersfacetheriskthatgoodsarerejectedbytheimportingcountryaftershipment.

When conformity assessment requirements differsignificantlyacrosscountries,andtheproceduresareopaque, companies may face additional costsassociated with obtaining the necessary information,and redesigningproducts tomeetdifferentcountries’conformityassessmentstandardsandrequirements.

Inaddition,lengthyconformityassessmentproceduresalso imply additional costs associated with salesrevenues forgone while the product is under review.Forsometime-sensitiveproducts,suchastextilesandclothingandhigh-technologyproductswithashortlifecycle, time delays can have a severe impact onprofitabilityandmarketpenetration.

Conformity assessment costs have not beensystematically quantified. This is because someaspects,suchastheopportunitycostoflostsales,aredifficult to measure. However, the extent to whichconformity assessment costs are perceived asobstacles to trade clearly emerges from severalsurveysandcasestudies(seeBoxD.5).

Little is known about the impact of conformityassessment procedures on trade. Focusing on asampleofdevelopingcountries,astudybyChenetal.(2006) claims that conformity assessment issuessignificantly impede trade. On the basis of firm-levelsurveydata,theyfindthatfirmsanswering“yes”tothequestions “Have testing procedures impacted yourabilitytoexportproducts?”and“Doyouhavedifficultyobtaining information about applicable regulations inthe countries listed?” also have a significantly lowerpropensity to export. They also find that testingproceduresareparticularlyburdensomeforagriculturalfirms.

In all likelihood, the impact of conformity assessmentprocedureson trade variesacross sectors. TheOECD(1999) survey stresses that even the nature ofconformity assessment costs varies by productaccording to their technical characteristics. Terminaltelecommunications equipment and automotivecomponents, forexample, requirean initialapprovalofthe product before it can be exported. In the caseofdairyproducts,eachindividualconsignmentmustbetestedbothprior toexportand/orat theportofentry.Thus conformity assessment procedures are a fixed


148

cost of exporting telecommunications equipment andautomotivecomponentmarkets–borneinadvance.Ontheotherhand,theyareavariablecostfordairyexports.

AstudybySchlueteretal.(2009)looksattradeeffectsof different types of SPS measures imposed on meatproducts. After grouping 21 types of measures in sixclasses, they find that whereas disease-preventionmeasures, tolerance limits for residues andcontaminants, and conformity assessment andinformation requirements increase trade, production-processrequirementsandrequirementsforhandlingofmeatafterslaughteringrestricttrade.

The paper by Fassarella et al. (2011) looks at theimpactofSPSandTBTmeasuresonexportsofpoultrymeatbyBrazilianexporterstothemainworldimportersintheperiod1996to2009.Theyfindthattheimpactof aggregated TBT and SPS measures on Brazilian

poultrymeatexportsare insignificant.However,whenmeasures are disaggregated, conformity assessmenthasanegativeandsignificantimpactonthevolumeofBrazilian poultry meat exports, while packaging andlabelling requirements, and/or disease-preventionmeasures (regionalization or quarantine treatment)haveapositiveandsignificantimpactonthevolumeofBrazilianpoultrymeatexports.

This report attempts to assess the importance ofconformity assessment requirements relative toproduct-characteristicsregulationsonoverallfoodandagricultural trade.39 Relying on the database onspecific trade concerns regarding SPS measuresdescribed in Section C, the analysis distinguishesbetween concerns related to conformity assessment(such as certificate requirements, testing, inspection,andapprovalprocedures)assetoutinAnnexCoftheSPSAgreement,andconcernsrelatedtootherissues

BoxD.5: Reporting of conformity assessment procedures as barriers to trade: selected examples

The fact thatconformityassessmentcostsareperceivedas importantobstacles to tradeclearlyemergesfromseveral surveys. In thebusiness surveyonnon-tariffmeasures conductedby the International TradeCentre (see Section C.2), product certification, product testing and inspection requirements appliedin importing countries represent more than half of all firms’ complaints about TBT/SPS measures in the11developingcountriesanalysed.

CostsofcertificationalsoappearasaprominentobstacletotradeinasurveyontheeffectsofSPS-relatedprivatestandardsconductedby theWTOSecretariat (seeG/SPS/GEN/932/Rev.1).Seventeenoutof the22 respondents included a reference to high certification costs. The survey also notes that developing-countryexportersconsidercompliancewithprivatestandards tobeaprerequisite forexporting toa largenumberofdeveloped-countrymarkets.

Compliancecostsforprivatestandardsarehigh,andtheyaresignificantlyaffectedbythecostofcertification.While the cost of certification varies depending on the sector, the examples provided indicate that theaverageannualcertification feemay reachbetweenUS$2,000andUS$8,000 foraprivatestandard. Inaddition,countriesreportsignificantcostsassociatedwiththetime-consumingprocessofmeetingprivate-standardrequirements,especiallyformicrobiologicalandchemicalanalyses,nottomentionthedifficultyoffindingaccreditedlaboratorieswithadequatedetectiontechniques.Thesecostsrisesignificantlywhentestshavetobeconductedabroad.Overall,thesecostsaredeemedasignificantimpedimenttotradeforsmall-scaleproducersthat,asaconsequence,areexcludedfromproductionchains.

Testingandcertificationcostsalsoappeartobeasignificantobstacletotradeforexportsfromdevelopedcountries.The2011NationalTradeEstimateReportonForeignTradeBarriers (NTEReport)–anannualsurvey carried out by the United States Trade Representative to identify foreign barriers to US exports –offers several examples. For instance, it claims that “Thailand imposes food safety inspection fees in theformof importpermit feesonall shipmentsofuncookedmeat.Currently, imports face feesof5bahtperkilogram(approximately$160perton)forredmeat(beef,buffalomeat,goatmeat,lamb,andpork)andforoffal, and 10 baht per kilogram ($320 per ton) for poultry meat. Fees for domestic meat inspections aremuchlowerandareleviedintheformofaslaughteringorslaughterhousefee.Thefeesare$5pertonfordomesticbeef;$21pertonforpoultry;$16pertonforpork;andzeroforoffal”.

Lengthycertificationprocedurescanalsobethemainobstaclestotrade.Forexample,the2011NTEReportrelatesUSindustryconcernsabout lengthyapprovalproceduresfornewpharmaceuticalproductsinHongKong,China,whichinhibitstheirabilitytomarketproductsonatimelybasis.Similarly,theNTEReportraisesa concern over Paraguay’s “non-automatic import licenses on personal hygiene products, cosmetics,perfumesand toiletries, textilesandclothing, insecticides,agrochemicals,andpoultry.Obtaininga licenserequires review by the Ministry of Industry and Commerce and sometimes by the Ministry of Health. Theprocessisslow,takingupto30daysforgoodsthatrequireahealthcertification.Onceissued,thecertificatesarevalidfor30days.”


149

D. TH

E TR

AD

E E

FFEC

TS O

F

NO

N-TA

RIFF M

EA

SU

RE

S

A

ND

SE

RV

ICE

S M

EA

SU

RE

S

(suchasrequirementsondiseasetreatment,maximumresidual levels, or thegeographical applicationof themeasure).The impactof thesetwotypesofconcernson the probability that firms will export and on thevolumeoftradeisanalysedusingbothasimpledummyfortheexistenceofanSPSmeasureandafrequencymeasure.40

Theanalysissuggests that, ingeneral,SPSmeasuresimposedbyanimportingcountryandraisedasspecifictrade concerns have a negative impact on theprobability that firms will export to the marketconcerned(resultsarereportedinAppendixTableD.3).However, conditional on the probability that firmsexport (that is, forfirmsalready in theexportmarket),the value of exports increases.41 In particular, theresultssuggestthatitisconformityassessment-relatedfactors that have the most negative impact on theprobabilityofenteringamarket,whilemeasuresrelatedtothecharacteristicsoftheproductexplainmostofthepositive impact on the value of trade. Although moreresearch is needed, one possible explanation is thatSPS measures, by enhancing consumers’ trust inimportedproducts, increase trade for thoseexportersthatmanage toovercome thefixedcostofenteringamarket.

In sum, the empirical evidence suggests thatconformityassessmentcosts(mostlyrelatingtotradein food and agricultural products) are an importantobstacletotrade.

3. Harmonizationandmutualrecognition

Thediscussion in theprevioussub-sectionssuggeststhat the use of TBT/SPS measures and domesticregulation in services by the importing country canhave ambiguous effects on trade. In the event thatthey have adverse trade effects, it is imperative toconsider how these harmful trade impacts may bealleviated.

Harmonization and mutual recognition of TBT/SPSmeasuresarecommonlybelievedtobestepstowardsmoreopen trade.However,economic theoryprovidesan ambiguous answer to the question of whetherharmonizationincreasesordecreasestrade,aswellaswhether harmonization has more impact than mutualrecognition on boosting trade (see Box D.6). Thissection, therefore, reviews the empirical evidence ontheseissues.

(a) Isharmonizationtradecreating?

The empirical literature measures the extent ofharmonization of standards in different ways. Somestudies consider a standard as harmonized if itconforms with an international standard published bythe International Organization for Standardization

(ISO), the International Electrotechnical Commission(IEC),theInternationalTelecommunicationUnion(ITU)or similar bodies. Other studies treat standards asharmonizediftheyarecommontoagroupofcountries.

Notwithstanding these differences, a general findingin the literature is thatharmonization increases trade.For example, using the number of bilaterally-sharedstandardsreportedinthestandards-relateddatafromthe Perinorm database, and taking country-specificstandards into account, Moenius (2004) finds thatsharedstandardshaveapositiveandsignificanteffectonbilateraltrade.

Using a gravity model of trade for the period 1995-2002, Clougherty and Grajek (2008) find thatconformity with ISO 9000 in developing countriesappearstoenhanceexportstodevelopedcountries(asimilar effect was estimated in Grajek (2004)). Theauthorsdonot,however,findthatconformitywithISOstandards in developed countries has a significanteffectoneitherexportsorimports.FocusingontradewithintheEuropeanUnion,VancauterenandWeiserbs(2005)findthatharmonizationhasasignificanteffecton a country’s exports.42 In particular, they find thatcountries that have a larger than average share oftrade in sectors covered by the EU harmonizationdirectiveexportmore.Morerecently,usinganindexofvariations in regulation on veterinary drugs andpesticides across countries, Gervais et al. (2011)estimatethatdifferencesinstandardshaveanegativeeffectontradeinpigmeatandbeef.

Harmonization is also found to have a positive effectonthediversificationofexportmarkets(theso-calledextensivemarginoftrade)–that is,onthenumberofexported varieties and export destinations. Albeitlimitedbythelackoffirm-leveldata,Shepherd(2007)isthefirststudytoexploretheimpactofharmonizationat the extensive margin of trade. Focusing on theexportsoftextiles,clothing,andfootwear,hefindsthatharmonizationisassociatedwithhigherexportvariety,mainly for low-income countries’ exports to theEuropeanUnion.

Focusing on the electronics sector, Reyes (2011)examines the response of US manufacturing firms tothe harmonization of EU product standards withinternationalnorms.Theauthorusestheshareofnon-harmonizedstandardsinanindustry43asameasureoftradecostsduetoavarietyofstandards.

Reyes’ study finds that increasing harmonizationincreases US exports to the European Union. Inparticular,thisincreaseisduetomoreUSfirmsenteringtheEUmarket.Exports fromUSfirmsalreadypresentin the EU market before the harmonization decrease.Overall, exports increase. Product standardharmonizationseemstobemoreimportantthantariffsforthepropensitytoexport.Furthermore,newexportingfirms are smaller than those already exporting to the


150

BoxD.6: Harmonization versus mutual recognition

This box explores the possible role of harmonization and mutual recognition of TBT/SPS measures andcompares their advantages and disadvantages. For the purposes of this box, TBT/SPS measuresanddomestic regulation inservicesare treated togetheras “standards”becausetheconclusionsfromthetheoreticalliteratureapplygenerallytogoodsandservicesregulation.

Supposethattwotradepartnersareconfrontedwiththesamemarketfailurebutaddressitwiththeuseofdifferentstandards.Thismeansthatexistingexporterswillhavetobearthecostsofadaptingtheirproductsto the requirements of the destination country or produce goods that meet both standards. The differentstandardsofregulationhaveanegativeeffectonmarketentry–theextensivemarginoftrade–asitactsasafixedmarketentrycost(KoxandLejour,2005).

Now, consider a case in which a firm operating under increasing returns to scale44 serves the domesticmarketandcanpotentiallyexporttothreeforeignmarkets,uponpayingafixed(sunk)marketentrycost.Ifthiscostismarket-specific,thefirmcanonlyrealizemarket-specificeconomiesofscaleineachoftheexportmarkets.Sincethetwocountrieshavethesamemarketfailure,aneffectivesolutionforbothcountrieswouldbetochooseacommonstandardorrecognizeeachother’sstandard.

Harmonization impliesacommondefinitionofboth thepolicyobjectiveand the technical requirements toachieve it, while mutual recognition refers to the reciprocal acceptance of the measures applied in bothcountries.Bothapproachesareconsideredtrade-enhancingastheyproduceeconomiesofscaleandpermitamoreefficientallocationofresources(ChenandMattoo,2008).Takingtheexampleofthefirmdescribedinthepreviousparagraph,ifthefixedcostofentryisthesameforallexportmarkets,asisthecaseundermutual recognition and harmonization, the firm can realize global economies of scale, and realize costsavings.However,eachsolutionaffectstrade inadifferentwayand, ingeneral, it isnotpossibletodefinewhetherharmonizationormutualrecognitionismoretrade-enhancing.

Ingeneral,harmonizationisexpectedtoboosttrademorethanmutualrecognitionforthefollowingreasons.As countries adopt the same standards, products are more homogenous and better substitutes for bothproducersandconsumers than inamutual recognition framework, thus reducinghome-bias– that is, thegeneral preference for domestically-produced goods (World Trade Organization (WTO), 2005b). Commonstandardslowertheinformationcostsfacedbyconsumersandincreasetheirconfidenceaboutthequalityofimportedproducts(Dissanayakaetal.,2001).Thisalsoappliesforbusiness-to-businessrelationships,whereharmonizationenhancescommunicationeffectiveness(Grajek,2004).Theyalsoallowcompatibilitybetweenimportedanddomestically-producedproducts(Baller,2007).

However,itispossiblethatharmonizationcanhaveanegativeimpactontradethatcanbeavoidedthroughmutualrecognition.Harmonizationreducesthenumberofvarietiesinthemarket(forexample,harmonizationtoacertainhigher-qualitystandardremovesfromthemarket lower-qualityproducts thatsomeconsumersmayhavebeenwillingtobuy).Whendemandforforeignproductsisdrivenbyloveforvariety,alowerdegreeof differentiation among products will diminish trade. Moreover, harmonization may generate compliancecoststhatvaryfordifferentcountriesifcertaincountrieslacktheexpertisetotakefullpartinthesettingofinternationalstandardsoriftheylackbargainingpower.Inthiscase,thegainsfromharmonizationwillnotbeequallydistributedamongparticipatingcountries.

In contrast,mutual recognition allowsan equal distribution of gains from removing TBTamong countries.Whenthisapproachisinplace,firmscansellinforeignmarketswithoutbearingthecostofharmonization.Therefore,when love for variety is important for tradeorwhencostsofadaptation toanew (harmonized)technologyarehigh,mutualrecognitionshouldbeexpectedtoboosttrademorethanharmonization.

Harmonization and mutual recognition also take place within regional agreements, with differentconsequencesfortradewithcountriesthatarenotpartoftheagreement(WorldTradeOrganization(WTO),2005b;ChenandMattoo,2008;MattooandSauvé,2003).Ontheonehand,harmonizationdecreasesthecostsoflearningabouttheregulationofeachmemberoftheagreementandavoidstheassociatedcostsofcompliance, thus benefiting producers that are not in the agreement. On the other hand, it can increasecompliance costs for firms outside the agreement, especially for firms in less developed countries, whichoftenlacktheinfrastructureandexpertiserequiredtocomplywithnewregulations(Otsukietal.,2001).Withmutual recognition, external producers can choose to produce according to the standards adopted in thecountrythatbettersuittheirproductionadvantages,implyinglowercosts.


151

D. TH

E TR

AD

E E

FFEC

TS O

F

NO

N-TA

RIFF M

EA

SU

RE

S

A

ND

SE

RV

ICE

S M

EA

SU

RE

S

European Union before harmonization. These resultssuggest that working towards a harmonization ofproduct rules across markets could assist small- andmedium-sizedfirmsinenteringnewexportmarkets.

Economistshavearguedthatdifferencesinregulationacross countries (policy heterogeneity) reduceservicestradeinthesamewaythatitdoesforgoods.AsdiscussedinBoxD.2,KoxandLejour(2005)showthat in a standard monopolistic competition model oftrade, different standards of regulation acrosscountries reduce bilateral trade.45 In support of thistheoretical prediction, empirical evidence shows thatmutualrecognitionorregulatoryharmonizationhaveapositiveeffectontrade.46

De Bruijn et al. (2008) consider the prospectiveeffects of the EU Services Directive, proposed in2004 by the European Commission to reduce theimpedimentstotrade,onbilateraltradeincommercialservices. By combining the changes in regulatorydiversity with the empirical results of the gravityanalysis, theyestimate that total trade of commercialservices within the European Union increases by anaverage of 28 per cent as a result of the ServicesDirective, as approved in 2006. This rises to 44 percent for the original proposal by the EuropeanCommission, which included the country of originprinciple.47 As they argue, such large differencesimplicitly show the economic benefits of mutualrecognitionofregulatorystandards.

In addition, Kalemli-Ozcan et al. (2010) consider theretrospective effects that regulatory harmonizationbased on the EU’s Financial Services Action Plan(FSAP) had on cross-border banking activities. Suchactivities increased significantly among Europeancountries that quickly adopted the financial servicesdirectives of the FSAP. Their results suggest thatlegislative harmonization in financial markets had apositive effect on cross-border banking integrationthat is additional to the generally positive effects ofeuroareamembership.48

(b) Regionalintegration

A growing number of regional/preferential tradeagreementsincludeprovisionsonTBT/SPSmeasures.The analysis of the content of preferential tradeagreements (PTAs) in last year’s report (WTO, 2011)showthatapproximately60percentoftheagreementsincludesuchprovisions.

In particular, mutual recognition of conformityassessmentandharmonizationoftechnicalregulationare among the most common approaches ofintegration in the TBT area. While the objective offosteringmutualrecognitionofconformityassessmenttendstobeafeaturethatoccurswithequalfrequencyacross several types of PTAs, significant differencesexist in terms of their tendency to include

harmonization of technical regulations between EU-type and North American-type agreements. Forexample,whiletheagreementssignedbytheEuropeanUniontypicallyincludeharmonizationprovisions,PTAsinvolving North American countries tend to includemutualrecognitionoftechnicalregulations.

Furthermore,lastyear’sreporthighlightstwofeaturesof PTAs. First, PTAs that harmonize standards arelikelytopresent“hub-and-spoke”characteristics,withthe larger partner representing the hub to whosestandards the spokes conform. Therefore, the reportcautions that this tendency may hinder further tradeopening among major regional groupings. Secondly,“deep”PTAs(thatis,moreambitiousPTAsintermsofthe depth of integration of TBT provisions) are morelikelybetweencountriesathigherandsimilarlevelsofdevelopment. Therefore, the report warns about therisksofmovingtowardsatwo-tieredworldthatwouldfurthermarginalizedevelopingcountries.

Thisyear’sreporttakestheanalysisastepfurtherandlooksattheevidenceofhowharmonizationandmutualrecognition provisions in PTAs affect trade.Harmonization and mutual recognition, when theyoccurattheregionallevel,affectcountriesoutsidetheregion differently. While harmonized standards allowentry into the whole regional market once theharmonized standard is adopted, mutual recognitionmaynotprovideaccesstothirdcountries.Forexample,agreementsinvolvingmutualrecognitionofconformityassessment procedures are likely to have trade-divertingeffects for countriesoutside theagreementif they are subject to strict rules of origin (i.e. laws,regulations and administrative procedures whichdetermineaproduct’scountryoforigin).

Suppose, for example, that following an agreementbetweencountryAandcountryB,onlygoodsmadeincountry A (satisfying specific rules of origin) cancirculate freely in country B after being tested andcertified in A. This privilege does not extend toproducts originating in third countries. Therefore, afirm located in country C will have to pay twice asmuch as a firm located in A (or B) for conformityassessmentinordertoaccessmarketsAandB.Inthecaseofservices,supposethatcountriesAandBhavesignedanagreementprovidingformutualrecognitionofqualificationrequirements.Aservicesproviderfromcountry C willing to serve both A and B markets willhave to pay twice as much to obtain the necessaryqualification requirements. Mutual recognition ofconformity assessment procedures (in the formerexample)orofqualificationrequirements(inthelatterexample)betweenAandBwhenaccompaniedbyruleoforiginthereforeincreasesthecostsforfirmslocatedin thirdcountries relative tofirms located inAandB,thusdivertingtrade.

Very few empirical studies have looked at howSPS/TBT-relatedpoliciesinPTAshaveaffectedtrade


152

both within and outside the region covered by theagreement. Existing studies indicate that regionalagreementsonharmonizationtendtodiverttradeandthat trade diversion affects exports negatively,especially from developing countries. For example,Cadot et al. (2010) show that the existence of PTAsbetweendevelopedanddeveloping countries (North-South agreements) hurts trade between developingcountries(South-Southtrade)andimpedesdevelopingcountries’attemptstodiversifyintonewmarkets.

ChenandMattoo (2008)estimateagravitymodelofbilateral trade of 28 OECD countries and 14 non-OECDcountriesat the three-digitSITCproduct level.Theiranalysisindicatesiftwocountrieshavesignedamutual recognition agreement (MRA) for a certainsector and the number of harmonization directivesbetweenthetwocountriesforaproduct.TheanalysisalsoindicateswhetherMRAsincluderulesoforigin.

ChenandMattoofindthatharmonizationagreementscanincreasetradebetweenparticipatingcountriesbutwillnotnecessarilyincreasetradewithothercountries.In particular, they find that harmonization increasesexports from developed countries outside the region,but it reduces exports from developing countriesoutsidetheregion.MRAstendtoincreasetradewithinthe region. MRAs also increase trade with countriesoutsidetheregioniftheyarenotassociatedwithrulesof origin. However, when the MRAs contain rules oforigin, trade with countries outside the region isnegativelyaffected,especiallyexportsfromdevelopingcountries.

Finally, focusing on two sectors, telecommunicationsequipment and medical devices, Baller (2007)examines the impact of MRAs and harmonizationagreements on bilateral trade among 26 OECDcountries and 22 non-OECD countries.49 Her resultsindicatethatwhileMRAsincreaseboththeprobabilityof entering a new market (the extensive margin oftrade)and thevolumeof trade (the intensivemargin),harmonizationofstandardsortechnicalregulationhasambiguouseffects.LikeChenandMattoo(2008),herfindingssuggestthatregionalharmonizationincreasestradewithdevelopedcountriesbuthinders tradewithdevelopingcountries.

Thereisnoempiricalanalysisthatlooksspecificallyatthe discriminatory effects of MRAs concerningdomestic regulation in services. The few empiricalstudiesontradediversionintheservicessector50usedummy variables indicating the existence of apreferential trade agreement between two givencountries.Suchvariablesdonotallowustodistinguishbetween market access and national treatmentcommitments (i.e. the principle of giving others thesame treatment as one’s own nationals), on the onehand, and mutual recognition of standards andrequirements,ontheotherhand.

As argued by Fink and Jansen (2009), the scope fordiscriminationislikelytobelimitedbytwofactors.OneisthatMRAs tend toapplymostly to restrictions relevantformode4movements,amodeoftradethatevenattheregionallevelhasnotbenefitedfromsignificantlevelsoftrade opening. The other factor is that MRAs tend toapply to only a small number of professional servicessectors,notablyaccounting,architectsandengineering,andonly a fewMRAs featureautomatic recognitionofqualifications(OECD,2003).

Tosumup,evidencesuggeststhatregionalintegrationof TBT/SPS measures has trade-diverting effects,especially to the detriment of developing countries.Thisfinding isconsistentwith theevidencethatdeeppreferentialtradeagreementsintheareaofTBT/SPSmeasures are more likely among countries with ahigher and more similar level of income. This findingalso highlights the risk that regional integration onTBT/SPS measures may lead to a multi-tiered worldwherecertaindevelopingcountriesaremarginalized.

4. Conclusions

The trade literature estimates the degree ofrestrictiveness of non-tariff measures and servicesmeasuresbyestimatingan“ad-valoremtariffequivalent(AVE)”, i.e. the level of an ad-valorem tariff that wouldhaveanequallytrade-restrictingeffectasthemeasuresatissue.TheuseofAVEstomeasurethetradeimpactof NTMs, however, presents conceptual andmethodologicallimitations.Forexample,theequivalenceof tariffs and quotas breaks down in the presence ofmarket uncertainty, or when NTMs take the form offixedmarketentrycosts,suchasthoseassociatedwithmeetingcertaintechnicalrequirements.

AVEs do not adequately capture the trade-restrictiveimpact of certain non-tariff measures when theproduction process is fragmented across countriesbecause they fail to take intoaccount thecumulativeeffect of measures along the production chain.Additionally, in the case of services measures, theestimated AVEs neither account for the possiblesubstitutionbetweendifferentmodesofsupplynorforthe complementarity between trade in services andtradeingoods.

Notwithstanding these limitations, existing empiricalevidence suggests that non-tariff measures andservices measures can significantly restrict trade. Inparticular,NTMscanbeastrade-restrictiveastariffs,and even more so in the case of certain high- andmiddle-income countries. In the case of servicesmeasures, while restrictions to trade are generallyhigher in developing countries than in developedcountries, they do not appear to be systematicallyassociatedwithacountry’slevelofdevelopment.

Acomparativeanalysisoftherolethatvarioustypesofnon-tariff measures play in the overall level of NTM


153

D. TH

E TR

AD

E E

FFEC

TS O

F

NO

N-TA

RIFF M

EA

SU

RE

S

A

ND

SE

RV

ICE

S M

EA

SU

RE

S

restrictivenessdoesnotexist.However,itisclearthattheimpactontradeisnotnecessarilyrestrictiveforallmeasures.TBT/SPSmeasuresdonotunambiguouslyincrease or decrease trade. In general, TBT/SPSmeasures have positive effects for moretechnologicallyadvancedsectors,butnegativeeffectson trade in freshandprocessedgoods.Aseconomictheory suggests, the introduction of a new TBT/SPSmeasure yields a trade-off between higher costs ofadaptation to new requirements for producers andlower information costs for consumers, who can beconfidentaboutthequalityoftheproductinquestion.The prevalence of a positive effect of TBT/SPSmeasures on manufacturing goods may suggest thatinformation costs are more important or adaptationcostslowerinthesesectorsthaninnon-manufacturingsectors.

WhenTBT/SPSmeasureshaveanegativeeffect,theimpact tends to be greatest for developing-countryexports. There is also evidence that TBT/SPSmeasures have a more negative impact on trade infood and agriculture – mainly because of the costsassociatedwithconformityassessmentprocedures.Inaddition, TBT/SPS measures appear to reduce thelikelihoodofexportmarketdiversification.Smallfirms–andfirmsthatoutsourcetheir intermediateinputs–appeartobemostaffectedbyTBT/SPSmeasures.

Harmonization and mutual recognition of standardsare ways in which any negative effects of TBT/SPSmeasurescanbemitigated.Harmonizationisshowntoenhance the presence of small and medium-sizedfirms in export markets. However, if harmonization ormutual recognition occurs within regional tradeagreements, there may be significant trade-divertingeffects on countries outside the agreement. Thisappears to be especially the case for developingcountries. Furthermore, as stressed in last year’sWorldTradeReport,thereisariskofa“lock-in”effect,wherebytheregionalharmonizationofstandardsmayreduce incentives for further trade opening. There isalsoariskofamulti-tieredregulatoryworldemerging,inwhichdevelopingcountriesaremarginalized.

The economics literature on domestic regulationrelatedtoqualificationandlicensingrequirementsandprocedures and technical standards is extremelylimited. Most studies look at a much wider set ofservicesmeasuresandare, therefore,not informativefor this report. In relation to the financial servicessector, the existing literature finds that regulationaimed at ensuring appropriate standards (such asaccounting standards and financial statementtransparency) is positively associated with cross-bordertradeandforeigndirectinvestmentinfinancialservices. As with TBT/SPS measures, there is alsosome evidence that a reduction in policy diversity,carriedoutthroughmutualrecognitionorconvergenceof international standards, has increased servicestrade.

Regardlessoftheirobjective,TBT/SPSmeasuresanddomesticregulationinservicesmayormaynotreducetrade. Negative trade effects, when they exist,generate negative spillovers across countries. Thisprovides a rationale for international cooperation.Harmonizationandmutual recognitionhelp to reducethe undesired negative trade effects of legitimatepublicpolicy.However, bothapproacheshighlight theneed for capacity building to address regulatorychallengesindevelopingcountries.

The costs related to compliance and conformityassessment impinge particularly on developingcountries. This is because they lack the technicalinfrastructure necessary to effectively develop anddesigntechnical regulation,standardsandconformityassessment procedures. Also, they lack thelaboratoriesandaccreditedcertificationbodiestotestandcertifycompliancewithacertainstandard.TheseissuesarethefocusofSectionE.


154

1 Thissectiononlyfocusesondomesticregulationmeasuresrelatingtoqualificationandlicensingrequirementsandprocedures,andtechnicalstandards.ThisnarrowsetofmeasuresistheequivalentinservicesofTBT/SPSmeasuresingoods.

2 TheagriculturalsectormayalsobesubjecttocoreNTMs.

3 ItisworthnotingthattheseAVEswereconstrainedtobetradeimpedingthroughanexponentialtransformationintheestimatedequation.ThistakesawayfromthefactthatNTMsmayactuallyenhancetradeattimes.

4 SeeBoxD.1foradescriptionoftheTTRIandOTRI.

5 Thisassumesperfectinformation.If,forexample,qualitydifferencesbetweenproductsaresignalledbytechnicalregulations,suchNTMscouldlowerpricesandincreasetrade.

6 SeeBoxD.1foradescriptionofthemarketaccessversionsoftheTTRIandOTRI.

7 AsexplainedinBoxD.1,thisisameasureoftherestrictivenessfacedbyexporters.

8 ThisfollowsaWorldBankclassificationofthesecountriesaccordingtodatain2001.

9 Usingthe“pricegap”methodtoestimatetheimpactofNTMsontrade,Bradford(2003)findstheAVEsofNTMstobeofthesameorderofmagnitudeforasampleofdevelopedcountries,therebyreinforcingtheresultsofKeeetal.(2009).However,theformer’sestimatesaredistinctlyhigherbecausethestudyusesthe“pricegap”method–AVEsaremeasuredasthedifferencebetweenimportandretailprices,aftercorrectingfortransportanddistributioncosts,andhenceincludemorepolicyrestrictionsintheirdefinitionofNTMs(Keeetal.,2009).Atthesametime,itispossiblethatforcertainNTMs,quantity-basedeconometricmethodsgivebiasedestimates.InthecaseofTBTandSPSmeasures,forinstance,ifcompliancecostsarepassedontounitvalues,regressingthevalueofimportsonameasureofNTMswillunderestimatetheirtradeimpact.Similarly,ifthereismarketpowerintheimportingcountry,thedomesticpricewillrisebymorewithaquantitativerestriction(QR)thanatariffreducingimportsbythesameamount.Hence,theAVEofaQR,derivedfromaquantity-basedestimation,wouldbeunderestimated.

10 RegressingthenaturallogarithmoftheAVEofNTMsin2001onthelevelofGDPpercapitain2001,wefoundap-valueof0.133.

11 TheestimatedtradeeffectrepresentsthepercentagepremiumonproductsrestrictedbyanNTMinacountryrelativetothepriceofthoseproductsincountrieswithoutNTMs.

12 Thisisdifferentfromtheimplicationof“binding”inalegalsense.ItreferstothefactthatconditionalonpresenceoftariffsandotherNTMs,thetradeeffectofaparticularNTMmaynotbestatisticallysignificantlydifferentfromzero.

13 Eventhelandedduty-paidpricemaycontainwholesaleandretailmarginsbecauseimporters,wholesalers,andretailersmaysharetheNTMrentsamongthemselves,especiallysincelargeretailersareintegratedintotheearlierstagesofthedistributionprocess(Bannister,1994;KrishnaandTan,1992).

14 SeeSectionCforadescriptionofthemethodology.

15 Thefourservicescategoriesaretravel,transport,governmentandcommercialservices.

16 Transport,travel,communicationsservices,construction,insurance,financialservices,royaltiesandlicencefees,computerandinformationservices,otherbusinessservices,governmentservicesandpersonal,culturalandrecreationalservices.

17 Fordevelopedcountries,asmuchasthree-quartersofservicestradeisinintermediateinputs(Miroudotetal.,2009).

18 Manufacturersmaychoosetoexportdirectlyorthroughintermediarieswhomovegoodsthroughwholesaleandretaildistributionnetworks.Ahnetal.(2011)showthattheshareofexportthroughintermediariesispositivelycorrelatedwiththedifficultyofaccessingdestinationmarkets.Thisisbecausewhenbarrierstotradearelarge,relativelysmallandlessproductiveexportersuseintermediariestoexport.

19 AccordingtoBernardetal.(2011),however,therearelargevariationsintheimportanceofintermediariesacrosscountries(andproducts).

20 Multinationalretailersalsotendtosourcetheirprivatelabelsfromdevelopingcountries(Nordås,2008)andtherearecaseswheretheyhaveprovidedthescaleandstabilityofdemandnecessaryfordevelopingcountryfarmerstoinvestinmodernproductiontechnology(DolanandHumphrey,2010).

21 Thetradeeffectsofregulatoryheterogeneity(withafocusonTBT/SPSmeasuresanddomesticregulationinservices)arefurtheranalysedinSectionD.3.

22 Perinormcontainsinformationonallstandardsdevelopedinthe21countriescovered,includinginformationontherelationshipamongstandardsoriginatedindifferentcountries.Thisinformationdefineswhethertwostandardsareidentical,equivalentornon-equivalent,onthebasisofISO/IECGuide21.

23 ThereisalargeliteraturethatstudiestheeffectofregulationinservicesontradeusingProductMarketRegulation(PMR)indicators.SeeforinstanceNicolettiandMirza(2004),Lennonetal.(2009)andSchwellnus(2007).Ingeneralthisliteratureestimatesanegativeeffectofregulationonservicestrade.However,PMRcoversarangeofmeasuresthatgoesbeyonddomesticregulationasofGATSArticleVI.4.Therefore,theyarenottakenintoaccountinthisreview.ThesameissuepertainsalsotootherstudiessuchasNicolettietal.(2003)thatusetheindexofnon-manufacturingregulations(NMR)andKimuraandLee(2006)thatusean“EconomicFreedomoftheWorld”(EFW)indicator.

24 TheAnnexonFinancialServicesintheGATSexplicitlyallowscountriestotakeprudentialmeasurestoprotectinvestorsanddepositorsandtoensuretheintegrityandstabilityofthefinancialsystem.TheanalysisofKoxandNordås(2007)showsthatmostsuchmeasureshaveapositiveeffectonservicestrade.Thiseffectislargerforregulationintheexportingcountrythanforregulationintheimportingcountry.

Endnotes


155

D. TH

E TR

AD

E E

FFEC

TS O

F

NO

N-TA

RIFF M

EA

SU

RE

S

A

ND

SE

RV

ICE

S M

EA

SU

RE

S

25 GravitymodelsareeconometricmodelsoftradewhichacquiretheirnamefromtheirsimilaritytoNewton’stheoryofgravitation.Thegravitymodeloftradepredictsthatthevolumeoftradebetweenanytwocountrieswillbepositivelyrelatedtothesizeoftheireconomies(usuallyGDP)andinverselyrelatedtothedistance(andothermeasuresoftradecosts)betweenthem.

26 See,forexample,OECD(1999);Otsukietal.(2001);WilsonandOtsuki(2004);Gebrehiwetetal.(2007);andDisdieretal.(2008a).

27 HACCPisafoodsafetyandqualitymanagementsystemthatinvolvesmonitoring,verifyingandvalidatingcompliancewithregulatoryrequirementsinallstagesofproductionatalltimes.

28 Fixedcostsareindependentoftheamountproducedorexported,whilevariablecostsincreasewiththelevelofproductionorexports.

29 Forareviewofthetheoreticalliteratureonheterogeneousfirms,seeHelpman(2011)andRedding(2010).

30 DetailsofthisanalysiscanbefoundinFontagnéetal.(2012).

31 Foradescriptionofthisdatabase,seeSectionC.

32 MeasuresnotifiedatWTOorPerinorm.

33 SeeSectionC.1foradiscussiononavailabledatasets.

34 FrenchCustomdatacontainfirm-leveldataonannualshipmentsbyallexportingFrenchfirmsintheperiod1995-2005toallpartnercountriesaroundtheworld.WethankCEPIIforprovidingaccesstothesedata.

35 Theestimatedequationis:

29

20 Multinational retailers also tend to source their private labels from developing countries (Nordås, 2008) and there are cases where they have provided the scale and stability of demand necessary for developing country farmers to invest in modern production technology (Dolan and Humphrey, 2010). 21 The trade effects of regulatory heterogeneity (with a focus on TBT/SPS measures and domestic regulation in services) are further analysed in Section D.3. 22 Perinorm contains information on all standards developed in the 21 countries covered, including information on the relationship among standards originated in different countries. This information defines whether two standards are identical, equivalent or non-equivalent, on the basis of ISO/IEC Guide 21. 23 There is a large literature that studies the effect of regulation in services on trade using Product Market Regulation (PMR) indicators. See for instance Nicoletti and Mirza (2004), Lennon et al. (2009) and Schwellnus (2007). In general this literature estimates a negative effect of regulation on services trade. However, PMR covers a range of measures that goes beyond domestic regulation as of GATS Article VI.4. Therefore, they are not taken into account in this review. The same issue pertains also to other studies such as Nicoletti et al. (2003) that use the index of non-manufacturing regulations (NMR) and Kimura and Lee (2006) that use an “Economic Freedom of the World” (EFW) indicator. 24 The Annex on Financial Services in the GATS explicitly allows countries to take prudential measures to protect investors and depositors and to ensure the integrity and stability of the financial system. The analysis of Kox and Nordås (2007) shows that most such measures have a positive effect on services trade. This effect is larger for regulation in the exporting country than for regulation in the importing country. 25 Gravity models are econometric models of trade which acquire their name from their similarity to Newton’s theory of gravitation. The gravity model of trade predicts that the volume of trade between any two countries will be positively related to the size of their economies (usually GDP) and inversely related to the distance (and other measures of trade costs) between them. 26 See Box D.4. 27 See, for example, OECD, 1999; Otsuki et al., 2001; Wilson and Otsuki, 2004; Gebrehiwet et al., 2007; and Disdier et al., 2008a. 28 HACCP is a food safety and quality management system that involves monitoring, verifying and validating compliance with regulatory requirements in all stages of production at all times. 29 Fixed costs are independent of the amount produced or exported, while variable costs increase with the level of production or exports. 30 For a review of the theoretical literature on heterogeneous firms, see Helpman (2011) and Redding (2010). 31 Details of this analysis can be found in Fontagné et al. (2012). 32 For a description of this database, see Section C. 33 Measures notified at WTO or Perinorm. 34 See Section C.1 for a discussion on available datasets. 35 French Custom data contain firm-level data on annual shipments by all exporting French firms in the period 1995-2005 to all partner countries around the world. We thank CEPII for providing access to these data. 36 The estimated equation is:

y!,!,! = β!STC!,!,! + D! + D! + D! + D!,! + D!,! + ε!,!,!, where subscripts s, ii) the average value exported by firms, (iii) the number of new firms, (iv) the total number of exporters. The explanatory variable STC is: (i) a dummy variable equal to one if a specific trade concern was raised by France against an SPS or a TBT measure to be adopted in an export market, (ii) the frequency ratio of the number of HS4 sectors affected by the measure within each HS2 sector and the number of HS4 sectors in that HS2. Explanatory variables are lagged one year to capture the possibility that the measure related to a specific trade concern can affect trade with a

wheresubscriptss,dandtindicatesector,destinationcountryandyear.yisinturn:(i)theaveragenumberofvarietiesexportedbyfirms,(ii)theaveragevalueexportedbyfirms,(iii)thenumberofnewfirms,(iv)thetotalnumberofexporters.TheexplanatoryvariableSTCis:(i)adummyvariableequaltooneifaspecifictradeconcernwasraisedbyFranceagainstanSPSoraTBTmeasuretobeadoptedinanexportmarket,(ii)thefrequencyratioofthenumberofHS4sectorsaffectedbythemeasurewithineachHS2sectorandthenumberofHS4sectorsinthatHS2.Explanatoryvariablesarelaggedoneyeartocapturethepossibilitythatthemeasurerelatedtoaspecifictradeconcerncanaffecttradewithadelay.Infact,STCsmayrelatetodraftmeasuresnotyetinforce.Fixedeffectsincludedintheregressionaddresstheomittedvariableproblembycontrollingforalldestination-timespecificvariables(suchasincomeandalldemandsidevariablesindestinationcountries)andsector-timespecificaspects(suchassectoralproductivityshocks).

36 Itisuncleartowhatextentaproblemofself-selectionmaybiastheseresults.

37 Inawidersense,italsoincludestheareaofmetrology,whichisanimportantprerequisiteforconformityassessmentandaccreditation(theevaluationofthecompetenceofanyinstitutioninvolvedinconformityassessment).

38 Forthisreason,governmentsencouragecooperationbetweenconformityassessmentbodiesandsometimesareactivelyinvolvedinmutualrecognitionagreements(MRAs).

39 DetailsofthisstudycanbefoundinCrivelliandGröschl(2012).ThestudyusesaHeckmanmodeltoestimatetheresults.Theyestimateaprobitbinarychoicemodeloftheform

Little is known about the impact of conformity assessment procedures on trade. Focusing on a sample of developing countries, a study by Chen et al. (2006) claims that conformity assessment issues significantly impede trade. On the basis of firm-level survey data, they find that firms answering “yes” to the questions “Have testing procedures impacted your ability to export products?” and “Do you have difficulty obtaining information about applicable regulations in the countries listed?” also have a significantly lower propensity to export. They also find that testing procedures are particularly burdensome for agricultural firms. In all likelihood, the impact of conformity assessment procedures on trade varies across sectors. The OECD (1999) survey stresses that even the nature of conformity assessment costs varies by product according to their technical characteristics. Terminal telecommunications equipment and automotive components, for example, require an initial approval of the product before it can be exported. In the case of dairy products, each individual consignment must be tested both prior to export and/or at the port of entry. Thus conformity assessment procedures are a fixed cost of exporting telecommunications equipment and automotive component markets – borne in advance. On the other hand, they are a variable cost for dairy exports. A study by Schlueter et al. (2009) looks at trade effects of different types of SPS measures imposed on meat products. After grouping 21 types of measures in six classes, they find that whereas disease-prevention measures, tolerance limits for residues and contaminants, and conformity assessment and information requirements increase trade, production-process requirements and requirements for handling of meat after slaughtering restrict trade. The paper by Fassarella et al. (2011) looks at the impact of SPS and TBT measures on exports of poultry meat by Brazilian exporters to the main world importers in the period 1996 to 2009. They find that the impact of aggregated TBT and SPS measures on Brazilian poultry meat exports are insignificant. However, when measures are disaggregated, conformity assessment has a negative and significant impact on the volume of Brazilian poultry meat exports, while packaging and labelling requirements, and/or disease-prevention measures (regionalization or quarantine treatment) have a positive and significant impact on the volume of Brazilian poultry meat exports. This report attempts to assess the importance of conformity assessment requirements relative to product-characteristics regulations on overall food and agricultural trade.1 Relying on the database on specific trade concerns regarding SPS measures described in Section C, the analysis distinguishes between concerns related to conformity assessment (such as certificate requirements, testing, inspection, and approval procedures) as set out in Annex C of the SPS Agreement, and concerns related to other issues (such as requirements on disease treatment, maximum residual levels, or the geographical application of the measure). The impact of these two types of concerns on the

1 Details of this study can be found in Crivelli and Gröschl (2012). The study uses a Heckman

model to estimate the results. They estimate a probit binary choice model of the form Pr import!"#$%& > 0 = 𝛟𝛟(α! + α!SPS!" !!! !"# + α!𝐗𝐗𝐢𝐢𝐢𝐢𝐢𝐢 + D! + D! + D! + D!"# + ε!"#$%&),

where ϕ (•) is a standard normal distribution function. And an outcome equation of the form

ln import!"#$%&|import!"#$%& > 0 = α! + α!SPS!" !!! !"# + α!𝐗𝐗𝐢𝐢𝐢𝐢𝐢𝐢 + α!λ(𝛂𝛂) + D! + D! + D! + D!"# + ε!"#$%&,

where D denotes dummy variables and X is a vector of standards gravity control variables and multilateral resistance terms and λ(𝛂𝛂) is the inverse mills ratio.







whereϕ(•)isastandardnormaldistributionfunction,andanoutcomeequationoftheform













whereDdenotesdummyvariablesandXisavectorofstandardsgravitycontrolvariablesandmultilateralresistancetermsandλ(α)istheinversemillsratio.

40 ThisisthecountofthenumberofSPSmeasuresinplaceonHS4productlineswithinanHS2sectordividedbythenumberofproductswithinanHS2sector.

41 ThislastresultisincontrastwiththefindingofFontagnéetal.(2012)discussedabovethatexportsofFrenchfirmsarenegativelyaffectedbyTBT/SPSmeasuresonwhichspecifictradeconcernshavebeenraised.ThismaybeduetothefactthatCrivelliandGröschl(2012)’ssampleincludesdevelopingcountries.Forthesecountries,thepositivedemandeffectsofSPS/TBTmeasuresarelikelytobemorerelevantthanforFrenchexporters.

42 SimilarresultsarefoundinDeFrahanandVancauteren(2006)forfoodproducts.

43 DefinedasthenumberofCENELECstandardsthatarenotidenticaltoanexistingIECstandardoverthetotalnumberofstandardsineachSIC4industry.

44 Aproductiontechnologyischaracterizedbyincreasingreturnstoscalewhenaveragecostsfallasthelevelofproductionincreases.

45 Policyheterogeneityisconsideredasafixedsunkcost.Duetoitsfixedcostnature,policyheterogeneityhastwoeffectsonthelevelofbilateralservicestrade.First,itreducesthenumberofexportingfirms.Secondly,itincreasestheaveragesizeoftheexportingfirms.InthetheoreticalframeworkofKoxandLejour(2005),thefirsteffectdominates.Therefore,thelevelofbilateralexportsisnegativelyrelatedtothedegreeofbilateralpolicyheterogeneity.

46 AsarguedbyFinkandJansen(2009),mutualrecognitioninthecontextofservicescancoverawiderangeofpracticesincludingrecognitionofprudentialregulationsunderfinancialservices(tofacilitatemode3),recognitionofeducationalqualificationswithaviewtoenrolmentinhighereducationorfurthertraining(tofacilitatemode2),aswellasrecognitionofprofessionalqualifications(tofacilitatemode4).

47 The“countryoforiginprinciple”(CoOP)wasakeyelementintheoriginalproposalbytheEuropeanCommission.Accordingtothisprinciple,operatorsprovidingcross-borderservicesintoanothermemberstatewithoutestablishingtherepermanentlywouldberequiredtorespectonlytherulesandregulationsoftheircountryofestablishment,withoutbeingsubjecttoothermemberstates’ruleseachtimetheycrossedaborder.TheCoOPinfactwouldhaveappliedmutualrecognitionofregulatorystandardsbetweenEUmemberstates(withsomelimitations).However,theamendedServicesDirectiveadoptedbytheEuropeanParliamentandtheCouncilattheendof2006excludedtheCoOP,whichhadcomeunderfirebecauseoffearsofsocialdumping.Asfarasdomesticregulationisconcerned,theServicesDirectiveprovidesforthesimplificationofqualificationandlicensingrequirementsandprocedures.


156

48 OtherstudiessuchasKoxandLejour(2005)andKoxandNordås(2007)alsoattempttoestimatehowanynegativeeffectofburdensomeregulationonservicestradecanbereducedthroughharmonizationormutualrecognition.However,theyuseindicatorsofregulatoryheterogeneitybasedonthePMRdata,measuringheterogeneityinamuchwidersetofmeasuresthanjustdomesticregulationcoveredinthisreport.

49 Baller(2007)’sdatabasecontainsinformationoneightMRAsrelevanttomedicaldevicesand14MRAsrelevanttotelecommunicationsequipment.Italsocontainsinformationon22EUharmonizationagreementsand19ASEANharmonizationagreements.

50 ParkandPark(2011)applyagravityregressionanalysistofourmajorservicessectors–financial,business,communicationsandtransportationservices.TheyfindthatthePTAscreateservicestradeamongmembersanddonotdivertservicestradefromnon-members.VanderMarelandShepherd(2011)findevidencethatfromanumberofsectors–transport,communications,businessservices,finance,andtradeservices–PTAsarenotonlytradecreatingbetweenmembercountries,butalsowithrespecttonon-members.FrancoisandHoekman(2010)istheonlystudythatisolatespossibletradediversioneffectsinservices,inparticularwithintheEuropeanUnion.Inthiscase,evidenceoftradediversionisfoundonlyforbusinessandinformaticsandtelecomsservices,wheretheyestimatea13.3percentincreaseintradevolumeswithintheEUrelativetothirdcountries.


157

D. TH

E TR

AD

E E

FFEC

TS O

F

NO

N-TA

RIFF M

EA

SU

RE

S

A

ND

SE

RV

ICE

S M

EA

SU

RE

S

AppendixTableD.1:Effects of SPS measures on export performances by firm

Dependentvariables

Lnn.ofvarietiesexportedbyfirms

Lnn.ofvarietiesexportedbyfirms

Lnexportsvaluebyfirms


Numberofexportingfirms

Numberofentryfirms

OLS OLS OLS OLS Poisson Poisson

(1) (2) (3) (4) (5) (6)

SPSd,s,t-1 -0.130*** -0.725*** 0.065

(0.021) (0.106) (0.314)

SPSFreqd,s,t-1 -0.167*** -0.910*** -0.166

(0.036) (0.197) (0.671)

Observations 86850 86850 86850 86850 86850 86850

R-squared 0.343 0.343 0.425 0.425 - -

Note: The variable SPS denotes a dummy for the existence of a measure (against which a concern was raised) in the sector. The variableSPSFreqisacountoftheconcernsraisednormalizedbythenumberofproducts(HS4)withinanHS2sector.Resultsareobtainedusingone-year lagexplanatoryvariables(aggregateestimationatHS2 level, thesample includesonlyfirmsexportingforat leastfiveyearsduringtheperiod1995-2005).All regressions include time,sector,destinationcountry, time-sectorand time-destinationcountryfixedeffects.Robuststandarderrorsinparentheses.***indicatesasignificancelevelof1percent.

Source:Authors’calculationsusingthedatabasefromFontagnéetal.(2012).

Appendix d.1


158

AppendixTableD.2:Effects of TBT measures on export performances by firm

Dependentvariables

Lnn.ofvarieties

exportedbyfirms

Lnn.ofvarieties

exportedbyfirms



Numberofexportingfirms

Numberofentryfirms

OLS OLS OLS OLS Poisson Poisson

(1) (2) (3) (4) (5) (6)

TBTd,s,t-1 -0.065*** -0.661*** -0.193

(0.018) (0.098) (0.319)

TBTFreqd,s,t-1 -0.062*** -0.876*** -0.217

(0.023) (0.133) (0.503)

Observations 86850 86850 86850 86850 86850 86850

R-squared 0.342 0.342 0.425 0.425 - -

Note: The variable TBT denotes a dummy for the existence of a measure (against which a concern was raised) in the sector. The variableTBTFreqisacountoftheconcernsraisednormalizedbythenumberofproducts(HS4)withinanHS2sector.Resultsareobtainedusingone-year lagexplanatoryvariables(aggregateestimationatHS2 level, thesample includesonlyfirmsexportingforat leastfiveyearsduringtheperiod1995-2005).All regressions include time, sector,destinationcountry, time-sectorand time-destinationcountryfixedeffects.Robuststandarderrorsinparentheses.***indicatesasignificancelevelof1percent.

Source:Authors’calculationsusingthedatabasefromFontagnéetal.(2012).

II – TR

AD

E A

ND

PU

BLIC

PO

LICIE

S: A

CLO

SE

R LO

Ok

AT N

ON

-TAR

IFF ME

AS

UR

ES

IN TH

E 21

ST C

EN

TUR

Y

159

D. THE TRADE EFFECTS OF NON-TARIFF MEASURES AND SERVICES MEASURES

AppendixTableD.3:Impact of SPS measures on agricultural and food trade, 1996-2010

SPSVariable: SPSFreqij(t-1)HS2 SPSij(t-1)HS4

DependentVariable: Pr(importijtHS4) ln(importijtHS4) Pr(importijtHS4) ln(importijtHS4) Pr(importijtHS4) ln(importijtHS4) Pr(importijtHS4) ln(importijtHS4)

(1) (2) (3) (4) (5) (6) (7) (8)

SPSmeasureij(t-1)HS4 -0.160*** 0.641*** -0.144*** 0.661***

(0.06) (0.15) (0.05) (0.14)

SPSConformityij(t-1)HS4 -0.309*** -0.473* -0.270*** -0.406*

(0.08) (0.28) (0.07) (0.23)

SPSCharacteristicij(t-1)HS4 0.019 0.988*** 0.012 0.962***

(0.07) (0.24) (0.06) (0.19)

Controls YES YES YES YES YES YES YES YES

FixedEffects YES YES YES YES YES YES YES YES

Estimatedcorrelation(rho)0.461 0.508 0.460 0.460

(0.01) (0.01) (0.01) (0.01)

Estimatedselection(lambda)1.372 1.091 1.370 1.371

(0.04) (0.04) (0.04) (0.04)

Logpseudolikelihood -7773030 -7772832 -7772958 -9756160

WaldChi2 49855.54 49752.98 49914.95 49838.46

Observations 5,452,530 5,452,530 5,452,530 5,452,530

Note:EstimationmethodistheHeckmanSelectionModel(maximumlikelihood).SPSFreqisacountoftheconcernsraisednormalizedbythenumberofproducts(HS4)withinanHS2sector(resultsusingthesevariablesarereportedincolumns(1)to(4)).SPSdenotesadummyfortheexistenceofameasure(againstwhichaconcernwasraised)inthesector(resultsreportedusingthisvariablearereportedincolumns(5)to(8)).ControlsincludethelogoftheproductofGDPs,thelogoftheproductofpopulations,thelogofdistance,adjacency,commonlanguageandcolonialheritage.Commonreligionistheselectionvariableinthefirststageestimation.Importer,exporter,HS4product,yearfixedeffects,andmultilateralresistance(MR)termsàlaBaierandBergstrand(2009)areincludedinallregressions.Standarderrorsinparentheses.***,*indicateasignificancelevelof1and10percent,respectively.

Source:CrivelliandGröschl(2012).

d. the trade effects of non-tariff measures and services measures · non-tariff measures and...

Documents