Board of Governors of the Federal Reserve System 
International Finance Discussion Papers 
Number 1026 
August 2011 
The Revealed Competitiveness of U.S. Exports 
Massimo Del Gatto 
Filippo di Mauro  
Joseph Gruber 
Benjamin R. Mandel 
NOTE: International Finance Discussion Papers are preliminary materials circulated to stimulate 
discussion and critical comment. References to International Finance Discussion Papers (other 
than an acknowledgement that the writer has had access to unpublished material) should be 
cleared with the author or authors. Recent IFDPs are available on the Web at 
www.federalreserve.gov/pubs/ifdp/. This paper can be downloaded without charge from Social 
Science Research Network electronic library at www.ssrn.com. 
Change font size pdf document - Compress reduce PDF size in C#.net, ASP.NET, MVC, Ajax, WinForms, WPF
C# Code & .NET API to Compress & Decompress PDF Document
pdf form change font size; adjusting page size in pdf
Change font size pdf document - VB.NET PDF File Compress Library: Compress reduce PDF size in vb.net, ASP.NET, MVC, Ajax, WinForms, WPF
VB.NET PDF Document Compression and Decompression Control SDK
acrobat compress pdf; pdf custom paper size
The Revealed Competitiveness of U.S. Exports* 
Massimo Del Gatto (G.d'Annunzio University and CRENoS) 
Filippo di Mauro (European Central Bank) 
Joseph Gruber (Federal Reserve Board) 
Benjamin R. Mandel (Federal Reserve Board) 
Abstract 
The U.S. share of world merchandise exports has declined sharply over the last decade. Using 
data at the level of detailed industries, this paper analyzes the decline in U.S. share against the 
backdrop of alternative measures of the competitiveness of the U.S. economy.  We document the 
following facts: (i) only a few industries contributed to the decline in any meaningful way, (ii) a 
large part of the drop was driven by the changing size of U.S. export industries and not the size 
of U.S. sales within those industries, (iii) in a gravity framework, the majority of the decline in 
the U.S. export share within industries was due to the declining U.S. share of world income, and 
(iv) in a computed structural measure of firm productivity, average U.S. export productivity has 
generally maintained its high level versus other countries over time.  Overall, our analysis 
suggests that the dismal performance of the U.S. market share is not a sufficient statistic for 
competitiveness. 
Keywords:  Trade competitiveness, gravity model, firm productivity 
JEL classification: F14, F17 
* Division of International Finance, Board of Governors of the Federal Reserve System, Washington, D.C. 20551 
U.S.A. m.delgatto@unich.itFilippo.diMauro@ecb.intJoseph.W.Gruber@frb.govBenjamin.R.Mandel@frb.gov.  
The authors thank Brian Andrew for excellent research assistance. The views in this paper are solely the 
responsibility of the authors and should not be interpreted as reflecting the views of the Board of Governors of the 
Federal Reserve System, any other person associated with the Federal Reserve System, or the European Central 
Bank. 
C# PDF insert text Library: insert text into PDF content in C#.net
Powerful .NET PDF edit control allows modify existing scanned PDF text. Ability to change text font, color, size and location and output a new PDF document.
pdf page size; pdf text box font size
C# PDF Annotate Library: Draw, edit PDF annotation, markups in C#.
Able to edit and change PDF annotation properties such as font size or color. Abilities to draw markups on PDF document or stamp on PDF file.
pdf edit text size; pdf file compression
TheRevealedCompetitivenessofU.S.Exports
MassimoDelGatto
FilippodiMauro
JosephGruber
BenjaminR.Mandel
Thisversion:March18,2011
G.d’AnnunzioUniversityandCRENoS.m.delgatto@unich.it.
EuropeanCentralBank. Filippo.diMauro@frb.gov
FederalReserveBoard.Joseph.W.Gruber@frb.gov.
FederalReserveBoard.Benjamin.R.Mandel@frb.gov.
1 Introduction
TheU.S. share of worldmerchandise exports s has declinedsharplyover the lastdecade. . Using
dataat thelevelof detailedindustries,thispaperanalyzesthedeclineinU.S.shareagainstthe
backdropofalternativemeasuresofthecompetitivenessoftheU.S.economy.
Usualsuspects for agivencountry’sdeclineinexportsharemightinclude: : unfavorablerela-
tivepricemovements,crowdingoutfromtheproliferationoflow-costexportersfromdeveloping
countries,unevenreductions intradecostsandbarriersaroundtheworld,orpossiblythedete-
rioriatingproductivityofexportingrmscomparedtoforeignrivals. Disentanglingthesefactors
presentsseveralcomplications. First,relativepricesareonlyweaklycorrelatedwithU.S.market
share,andarethusnotveryhelpfulinexplainingitsrecentdynamics. Thisisevidencedbythe
acceleratingdropinshareduringthe2000’samidstadeclineinthevalueofthebroadrealdollar.
Second, inmanyinstances,andparticularly forinternationalcomparisons,trade costsandrm
productivityarediculttomeasuredirectly.
1
Andthird,exportsharesmayadditionallyre ect
the idiosyncraticcompositionof the U.S. export bundle, which h may y have little to do withthe
abilityofU.S.exporterswithinagivenindustrytocompete.
To tackle e these e issues, , Section n 2 begins by y decomposing the decline inshare into detailed
industry groups; wend d that t only a a fewof these industries s contributed d to the e decline inany
meaningfulway. Moreover,alargepartofthedropwasdrivenbythechangingsizeofU.S.export
industriesandnotthesizeofU.S.saleswithinthoseindustries. ThismeansthatU.S.exporters
arespecializedinindustriesthathappentohavebeengrowingrelativelyslowlyasashareofworld
trade. TheseobservationsoerourrstsuggestionthatthefallinaggregateU.S.sharehaslittle
todowiththeunderlyingproductivityofU.S.exportingrms.
The authors s thank Brian n Andrew for r excellent research assistance. . The e views in n this s paper are solely y the
responsibilityoftheauthorsandshouldnotbeinterpretedasre ectingtheviewsoftheBoardofGovernorsofthe
FederalReserve System, anyother person associated with theFederal ReserveSystem,or theEuropean Central
Bank.
1
Measures of aggregate e tfp, , comparable e across countries, , are e usually obtained d as the residual l component t of
GDPgrowth that cannot be explained bythegrowth of production inputs. . Oneof f the drawbacks of the growth
accountingapproachisthattheroleofthesectoralcompositionofoutputisruledoutbyassumption. Byassuming
that GDPisproducedbyasinglesector, onecannotdisentangletfpdierences (across countries) duetosectoral
specializationfromtfpdierences duetootherfactors.
1
C# PDF Sticky Note Library: add, delete, update PDF note in C#.net
Able to change font size in PDF comment box. Note is a necessary feature in PDF annotation, which bring users quick and efficient working with PDF Document.
pdf compression settings; pdf file size limit
C# PDF Convert to Word SDK: Convert PDF to Word library in C#.net
target PDF document, keeps the elements (like images, tables and chats) of original PDF file and maintains the original text style (including font, size, color
best way to compress pdf; best online pdf compressor
Wethenpresenttwomeasuresoftradecompetitivenesswhich,insofarastheyareinferredfrom
actualtrade ows,werefer toas revealed d competitiveness. . The e rst measure, inSection3,is
derivedasaresidualfromastandardgravityequation. Theobjectiveoftheexerciseistopurge
bilateraltrade owsoftheeectofnationalincomeandgeography,whereintheresidualcontains
informationabout the relative productivity andunmeasuredtradecosts of exporters. . We e nd
thatthemajorityofthedeclineintheU.S.exportshareis infact duetothedecliningshareof
U.S.incomeintheworld. Theresidual,whichweviewasa‘purer’measureofcompetitiveness,
isdecliningbutnotasdramatically.
Our second approach, , in n Section 4, is derived from a structural model and builds on the
multi-country, multi-sector r versionof Melitz-Ottaviano(2008).
2
Inthatframework, the overall
competitivenessofacountryinagivensectoristheoutcomeofaprocessofrmselectiondriven
by:(1)thedegreeof’accessibility’(i.e. tradecosts)ofthecountry y andthesizeofits domestic
market, as s well l as (2) the exogenous s ability of the country y to o generate low cost rms, which
depends onstructuralandtechnological l factors. . Weextendpreviousempiricalapplications s of
thatmodelbyusingricherproduct leveldetail,andadditionallyemployaninnovativeapproach
byNovy(2009)tocomputecompetitivenessindicatorswhicharecomparableovertime. Consistent
withour gravity residualexercisewendthat,notwithstandingsignicant heterogeneity across
sectors,U.S.exportproductivityhasgenerallymaintaineditshighlevelversusothercountriesover
time. Overall,ouranalysissuggeststhatthedismalperformanceoftheU.S.marketshareisnot
asucientstatisticforcompetitiveness.
2 ThestateofU.S.exportshare
From1980to2009theU.S.shareofworldexportsfellbyalmostonethird,decliningfromabout
11percenttojustover8percentofworldexports. Inthissectionweexaminethedeclineinthe
U.S.shareusingNBER-UNbilateraltradedatafromFeenstra,Lipsey,Deng,MaandMo(2005).
As showninFigure 1, , the UnitedStates’share of worldmerchandise exports rose slightly
from1986to1999,increasingfrom about 10
1
2
to12
1
2
percent of worldexports,beforefalling4
percentagepointsbetween1999and2009. Thebilateraltradedatarunthrough2004and,inFigure
2,weobservethateveryindustrygroupatSITC1-digitaggregationregisteredadecreaseoverthe
periodfrom 1984to 2004, withmany of the larger changes occuringinthe early 2000’s. . The
largestdeclinesinsharewererecordedamongthebasicmaterialscategories(SITC0through4),
whichaccountforapproximately25percentofU.S.exports,andinmachineryandtransportation
equipment(SITC7),whichaccountforalmosthalfofU.S.exportsales. Itisinterestingtonote
thatthetimingofthedeclineinU.S.sharediersoverSITCcategories. Thefallinbasicmaterial
sharesisgradualandpersistent,whiledeclineinmachinery&transportationequipmentisabrupt,
primarilyoccurringafter1999.
Thedeclineinmarket shareis machineryandtransportationequipmentisparticularnotable
giventheimportanceofthissectorforoverallU.S.exports.ThefallintheU.S.shareofmachinery
andtransportationequipmentisexaminedfurtherinTable1,whichbreaksthecategoryintoSITC
2-digitsubcategories. AlthoughthedeclineinU.S.shareisapparentacrossmost2-digitmachinery
categories,thefallintheU.S.shareofocemachineandcomputerexportsisparticularlystriking,
withU.S.shareofworldexportsfallingfromathirdofthetotaltojustunderonetenth. Aswith
overallexports,thereissomedispersioninthetimingofthedeclineinsharesacrosssubcategories.
Whereasthefallincomputersissteepandsteadyovertheentireperiod,inmostothercategories
of machinerythe U.S.wasabletomaintainor expandexport sharethrough1999beforeshares
plummetedsharply.
2
ThatmodelwasrstbroughttothedatabyDel Gatto,MionandOttaviano(2006)andfurther developedby
Ottaviano,TaglionianddiMauro(2009).
2
C# PDF Field Edit Library: insert, delete, update pdf form field
A best C#.NET PDF document SDK library for PDF form field Able to add text field to specified PDF file position in C# Support to change font size in PDF form.
best pdf compression tool; pdf compress
C# PDF: Use C# Code to Add Watermark to PDF Document
into your C#.NET class application, developers can easily add a transparent watermark with desired font color, size and position onto target PDF document page.
can pdf files be compressed; change file size of pdf document
Amoremeaningfulwayofdecomposingthedeclineintheaggregateexportshareistocompute
the appropriately weightedcontributionfrom disaggregate categories of goods. . The e change in
aggregateexportsharecanbeexpressedasthesumofchangesacrossproductcategories(i)asa
ratioofthechangeinworldexports:
X
US
X
WORLD
=
X
i
X
i
US
X
WORLD
Figure3depictsthecontributionstothechangeinaggregateexportshareforeach1digitSITC
codeovertheperiodfrom1984to2004. Food&liveanimalsprovidedthelargest t contribution
tothedeclineinshare,accountingforalmostonefourthoftheaggregatedecline. Almostaslarge
werethecontributionsofmachinery&transportationandcrudematerials,alsoeachcontributing
aboutonefourthtotheoveralldeclineinshare. Theimportanceofrawmaterialsforthedecline
inU.S.shareraisesanoteofcautionininterpretingaggregateexportsharestatistics.Commodity
pricesfellovermosttheperiodunderconsideration,andsincetheexportsoftheUnitedStatesare
relativelycommodityintensive,sodidtheU.S.shareofworldexports.
Theimportance of commodities is further illustratedinFigure 4, which h depicts s thetop10
contributorstotheaggregatedeclineamong4-digitSITCcodes. Cornandsoybeanscontributea
combinedonesixthoftheoveralldecline.However,the4-digitdataalsorevealsthatanumberof
categoriesofmanufacturedgoodsalsocontributedtothedecline,includingmotorvehiclepartsand
digitalprocessingunits(computers).Thetakeawaymessageisthatatruemeasureofdevelopments
inU.S.competitivenessismorelikelytobefoundbylookingatU.S.exportperformancewithin
relativelynarrowlydenedcategories.
TheimportanceoffoodsfortheexplainingtheoveralldeclineinU.S.shareissomewhatsur-
prisinggivenfoods relativelysmallshareinU.S.exports and,asshowninFigure2,thelackof
anabnormallylargefallintheU.S.shareoffoodspecicexports. However,itisimportantto
notethatthecontributionofeachindividualcategorytothefallintheU.S.aggregateshareoc-
curs alongbothanintensiveandanextensivemargin. . ThedeclineintheU.S.aggregateshare
re ects bothanintensive decline inmarket share withineachcategory, , as s well l as s anextensive
declinestemmingfromchangesinthesizeofeachcategoryinworldexports. Forinstance,corn
(SITC0440)contributestothedeclineinU.S.aggregatesharebothastheU.S.capturesasmaller
proportionof the corn-specicexport market and d also as s corn’s shareof overall l worldexports
declines.
Oneestablishedmethodofassessingtheimportanceofcompositionforchangesintradeshares
is constantmarketshareanalysis(seeECB(2005)foradetaileddescription).
3
Constantmarket
share analysis separates changes inaggregate market shareintotwo components, acommodity
eectandcompetitivenesseectdenedasfollows:4
X
i
US
X
WORLD
=
X
i
X
i
US
X
i
WORLD
:
X
i
WORLD
X
WORLD
|
{z
}
+
X
i
X
i
US
X
i
WORLD
:
X
i
WORLD
X
WORLD
|
{z
}
CommodityEect
CompetitivenessEect
Thecommodityeectmeasures theeectofcompositiononthechangeintheaggregateexport
share,byweightingthechangeinthecompositionofworldexportsbytheinitialcompositionof
3Constantmarketshareanalysisisbesetbyanumberofwelldocumentedtheoreticalproblems(seeRichardson
(1971)foranoverview). However,theapproachremainsillustrativeandsimpletoimplementevenifinterpretation
iscomplicatedbyrelativepricechangesandotherissues.
4Theconstant market shareapproachoftenincludes anadditional\market eect"related tothegeographical
pattern of trade. . For r easeof exposition wehavefocused only on the commodity eect, in asense wrappingthe
market eect intoourmeasurement of thecompetitiveness eect. . With h decliningtradecosts it is likelythat the
marketeecthasbecomealess pronounceddeterminateofaggregatesharein anycase.
3
Generate Barcodes in Web Image Viewer| Online Tutorials
Select "Generate" to process barcode generation; Change Barcode Properties. Select "Font" to choose human-readable text font style, color, size and effects;
pdf font size change; pdf page size dimensions
VB.NET Image: Visual Basic .NET Guide to Draw Text on Image in .
Please note that you can change some of the example, you can adjust the text font, font size, font type (regular LoadImage) Dim DrawFont As New Font("Arial", 16
change file size of pdf; change paper size in pdf
the U.S. export bundle. . The e competitiveness eect measures the portionof the changeinthe
aggregatesharethatisduetochangesinthewithincategoryshareofU.S.exports.
Figure 5 5 decomposes the contribution of each 1-digit SITC export category to the change
in the aggregate export share (the blue bars) into components due to commodity (the green
bars)andcompetitiveness(theredbars)eectsoverthe1984to2006period. Thelargenegative
contributionsoffoodandliveanimalsandcrudematerialslargelyre ectthedecliningimportance
ofthesegoodsinworldexports(signiedbynegativecommodityeects),althoughU.S.exports
alsosueredanegativecompetitivenesseectineachcase.Incontrastthenegativecontributionto
theaggregaterecordedbythemachineryandtransportationsectoriscompletelyduetoadecline
inU.S.competitiveness,asthesectorhas greatlyincreasedits weightintheworldexports over
thetimeframeunderconsideration.
Insummary,interpretingthedeclineintheU.S.exportshareiscomplicatedbycompositional
eects. Theprimary y driversofthedeclineinaggregateU.S.sharewere rawcommodities,with
negativecontributionsthatlargelyderivedfromtheirdecliningweightintheworldexportbasket.
Thatsaid,theU.S.didexperiencealargedeclineintheshareinthemachineryandtransportation
sector, whichwas not t re ected in the e compositionof f U.S. . exports but rather r declines s within
detailedsub-categories. Heretheevidence e of afall inU.S.competitiveness ismore compelling.
Againstthisbackground,thefollowingsectionsfocusonU.S.exportperformancewithinindustries
andattmepttoidentifyitsdrivers. Wesuggesttwoalternativeempiricalmethodologiestoparse
out anarrowerdenitionof competitiveness: : exporterproductivity y purgedofgeographicaland
relativemarket size considerations. . Thisstrategyistermed,"revealedcompetitiveness,"which
derivesfromthefactthatitisinferredfromobservabletrade ows.
3 ReducedFormRevealedCompetitiveness
OnepossibleexplanationforthedeclineinU.S.exportshareissimplythattheU.S.nowaccounts
for a a smaller r share of globaloutput. . As s China a andotheremerging economies s expandrapidly
andbecomemoreintegratedintothe globaleconomy,it isnaturalthattheU.S.share of world
exportswouldfallwithoutnecessarilyindicatinganydeclineintheproductivityofU.S.exporters.
AsshowninFigure6,thefall intheU.S.shareofglobalexports s of about 4percentagepoints
overthepastdecadecorrespondstoadecreaseintheU.S.shareofglobaloutputofabout31/2
percentage points onaPPP basis. . The e relativelytight correlationbetween n export t shareand
incomeholdstrueformanyothercountriesaswell. AcrosstheG7,France,Italy,andJapanhave
experienceddeclinesinexportsharethatbroadlymatchtheirdecliningshareofworldoutput. On
theotherhand,Germanyhasmorelessmaintainedexportshareevenasitsshareofworldincome
hasdeclined,whileCanadaandtheUKhavesueredmuchsteeperfallosinexportsthanincome.
Inpercentageterms,theexport sharegrowthof Chinahas outpaceditsincome share,andthe
sameholdsforIndia.
Figure 6 stronglysuggests that changes inmarket share may be con ating competitiveness
eectswithincomedynamics. Specically,countrycharacteristicssuchassizemaybein uencing
marketsharebuthavelittletodowiththeunderlyingabilityofacountry’sexporterstocompete.
Tocontrolforsuchcharacteristics,ourrstapproachistonon-parametricallyestimatetrade ows
minus thecontributionofcountry size, , geographyandcertaintradecosts. . Aderivativeofthe
gravityequationisanaturalcandidatetodoso. PreviousstudiessuchasBaierandBergstrand
(2001),andmorerecentlyWhalleyandXin(2009)andNovy(2009),usegravitytodecomposethe
levelsofbilateraltrade owsintocontributionsfromincome,tradecostsorotherwise. Eachnds
thatexporterandimporterincomeplays asubstantial,evendominant,roleinexplainingtrade.
Our approach extends this logic c to o the case of relative e trade performance, , where e the gravity
equationis ‘folded’ ’ by dividingthroughby areferenceexporter. . Inthe e particularcase where
thereferencecountryistheentireworld,thegravityequationconvertsneatlyintoanexpression
4
formarketshareintermsofrelativeexportersize,relativegeographiccharacteristicsandrelative
productivity.
Ourapproachto‘decomposing’theshareintofactorsthathavetodowithcompetitivenessand
thosethatdon’t involvessimply lookingatthetimevariationintheresidualofapanelgravity
estimation. Theintuitionisthatifacountryisincreasinglyoutperformingtheaverageexporter’s
performance(i.e.,acountryexportsmorerelative toits ownsizeandmoretodistant countries
overtime)thenitsresidualwillgrowovertime. Wepositthatthisresidualcontainsinformation
about changes inthe underlyingproductivity of exporters. . Inthis s section, , wedonotapplya
structuralinterpretationtothatproductivity,itismerelycontainedintheresidual. Inthesection
thatfollows,weapplyastructurethatallowsformorespecicinterpretationoftheresidualand,
moreover,isconsistentwiththereducedformgravityequationherein.
Tobeconcrete,deneTlh
s
ascountryl’sexportstocountryhinsectorsinagivenperiodt:
T
lh
st
=D
l
t
D
h
t
r
l
st
r
h
st
lh
s
t
(1)
Equation(1)correspondstoagenericgravitymodel,wherebilateraltradeisafunctionofcountry
size(D),latentcountry-specicmultilateralresistance(r),geographiccharacteristics()andglobal
shocks (). . Exploitingthe e multiplicative form of the equation, we cancelout importer-specic
termsbydividingthroughbytotalexportstocountryhinindustrys.
T
lh
st
P
l
T
lh
st
=
D
l
t
r
l
st
lh
s
P
l
D
l
t
r
l
st
lh
s
(2)
Theintuitionforthisreducedformisthatthechangeinagivenimporter’sincomeormultilateral
resistancewillaectthelevelofthatcountry’simportsbutnothowthenewimportsareallocated
across exporters. . Moreover, , aglobal shock aectingallexporters s willnot aect their relative
performanceandhencethetermscanceloutaswell. Themethodoftakingratiosofthegravity
equationhasthreeostensiblebenets.First,forourpurposeofrelatingtheshareofU.S.exportsto
underlyingproductivitymeasures,equation(2)isexpressedinthecorrectunitsofshareowingto
income,tradecostsandproductivity. Second,thesizeofthedatamatrixusedintheestimationis
reducedbyfoldingintheimporter-specicterms. Third,multilateralresistanceterms(asdened
inAndersonandvanWincoop(2003))associatedwithimporterscancelout,sparingtheneedto
approximatethemusingxedeects.
5
Denotingthegeometricmeanofagivenvariableby
X=
Q
l
X
1
n
,takinglogs,andallowingfor
amean-zeroperturbation("),wecanrewritetheaboveexpressionas:
6
ln
T
lh
st
P
l
T
lh
st
=ln
1
n
s
+ln
D
l
t
D
t
+ln
lh
s
h
s
+ln
r
l
st
r
st
+"
lh
st
(3)
Thelogofcountryl’s market share indestinationmarket hisapositivefunctionofits rel-
ativeincome,its geographicproximityandits relativeproductivity. . Again,thisspecicationis
isomorphictoastandardgravitymodel,thoughspeciedinrelativeterms. Withtheadditional
assumptionsthatandnareconstantovertime,variationinexportermultilateralresistanceand
productivityisidentiedastheresidualofamodelwithexporterrelativeincomeandcountry-pair
xedeectsontheright-handside. Thatis,theactualmarketsharechangesovertimerelativeto
thechangesinthegravitymodelpredictioncontainsinformationabout theevolutionofrelative
5
Otherexamples ofcancellingout theimporter xed eects in agravity y framework include: Head and Mayer
(2000),Martin,Mayer and Thoenig(2008)andHead,Mayer andRies(2010).
6
Expression(3)imposesseparabilityacrossright-handsideratioswiththeassumptionthatln
P
T =
P
lnT. In
practice,thismayhavetheeectofoverestimatingtheshareofeachexporter(i.e.,sincethesharesasdecomposed
ontheright-handsidewilladduptomorethan1),but littleimpactontherelativesizeof theshares.
5
exporter productivity. . This s is what we will l refertoas s the share-to-modelratio, , or r simplythe
growthinthemodelresidualbyexportingcountry,averagedacrossindustries:
4ln
rl
t
r
t
=
4ln
Tlh
st
P
l
T
lh
st
4ln
\
Tlh
st
P
l
T
lh
st
!
(4)
Theassumptionof atime-invariant  is similar tostandardgravityapproaches usingvariables
suchas distance,commonborderandcommonlanguagethatdon’ttendtochangemuch. . The
implicationof this assumption, however,is that decreases intradecosts dueto changingtrade
policy willalsobecapturedintheresidualterm. . Inour r implementationwe adddummies for
signicant shifts inpolicy(e.g., NAFTA,EMU)as wellas over thecourseofour sampletotry
tocontrolforchangingtradecosts,butnonethelesstheresiduallikelycaptureselementsoffalling
tradecostsinadditiontorelativeproductivity. Assuch,inthissectionwejointlyestimaterelative
performanceduetothesefactors,bothofwhichtintoareasonable,ifbroad,denitionofexport
competitiveness; inthe followingsectionwe usea structural model toparse the residual more
nely.
Theassumptionofaconstantnumberoftradingpartnersperimporter(n)maybelessbenign.
DuetotheseminalworkofFeenstra(1994),therehasbeenmuchstudyoftheincreaseinproduct
varietywithinevennarrowlydenedproductcategories. Weaddressthisempiricallyintwoways.
First,everyspecicationbelowcontainstimexedeectswhichwouldsoakupaseculartrendin
varieties. Secondly,mostspecications s containcountry-pairxedeects or exporter-timexed
eectswhichwouldpickupatleastaportionoftheleveldierencesinnbycountry. Wenotethat
adisproportionate riseinrelativeproductvarietytocertaincountries over timewoulddecrease
thetermln(1=n)andhenceworkagainstthendingofrisingproductivityintheresidual. Forthe
mostprolictradersintermsoftheirnumberoftradingpartners,whichincludestheU.S.,wethus
takeour estimatesoftherisingresidualas anunderestimateofthetruechangeinproductivity
andtradecosts.
3.1 Reducedformrevealedcompetitiveness: : data&specication
The datausedinthe estimationare bilateral trade  ows as describedinthe previous section,
nominalGDPdatafromthePennWorldTablewhichareconvertedintointernationaldollarsat
PPPexchangerates,dummyvariablesforNAFTA,EUandEMItrade ows,thedistancebetween
capital cities, as s wellas commonborder andcommonlanguage dummies. . We e followprevious
studiesbytruncatingthedataat$10,000perannualbilateral owtoavoidpotentialdistortions
fromerrorsofunitsinthedataandimplausiblysmalltradevalues. Weruneachgravityregression
at theSITC4-digitlevelandconstrainourselvestoproductswithover1,000exporter-importer-
yearobservations.Theamountofdatalostduetoconcordanceissuesforincomeanddistancedata
willvarybyspecicationsincetheuseofxedeectsoftenobviatestheuseofthosevariables,but
themostpunitivecutofthedatastillaccountsforover83percentofglobaltradevaluebetween
1980and2004.
OurestimatorisOLSonthelog-linearspecicationof(3). Cognizantofthefactthatthereare
manydierentwaystospecifythatequation,wetryanarrayofvedierentpanelspecications
withvaryingdegreesofcontrolformultilateralresistanceterms. Again,ourobjectiveistocompute
variousindexesofthechangeintheresidual(4)whichwillbeinformativeoftheportionofU.S.
sharedecline notexplainedby gravity controlssuchasincomeandgeography. . The e dierences
amongtheseveregressionsarethetreatmentoftheterms(whichinsomecasesarecountry-pair
xedeectsandinothersarethestandarddistance,borderandlanguagecontrols),themeasure
ofcountry-specicvariablesD,aswellasthesubsetofdatausedfortheestimation.
The specications are describedinTable 2. . In n specication(i), we e regress s the e exporter’s
shareofglobalsalesineachSITCproductontheexporter’srelativenominalincome(recallthat
6
the importer-specicterms cancelby dividingbya referenceexporter), exporter-importer xed
eects,yearxedeects,anddummiesforNAFTAandEMU. Anactualmeasureofincomeis
usedto control for r thetrends describedinthe previous section. . Theexporter-importerFE E is
astaticmeasureoftradecosts whichwipes outvariationinborder, , distanceandlanguage,and
arguablyincludesmanymoreunmeasured(andunchanging)barrierstotrade.Tocontrolforsome
largepolicychangesduringour samplewhichwedonotviewasendogenoustocompetitiveness,
dummiesforpost-NAFTAandpostEuroyearsareincludedfortheappropriatecountries. Finally,
yearxedeectssoakupseculartrendsinn.
Specication(ii)usesthesameregressorsas(i),butonasubsetofthedatathathasobservations
foratleast20ofthe25yearsinthesample(i.e.,withineachexporter-importer-SITCcell). Itis
informativetoconstrainourselvestothesubsetofbilateraltrade owsthatarebalancedoverthe
courseofthesampleforatleasttworeasons. First,theaverageresultsstatisticsreportedacross
productsmaybeskewedbycompositionalchangesovertimeintheunbalancedpanel. Secondly,
ourlinear-in-logs specicationpotentiallyintroducesselectionbiasbydroppingtheobservations
withzerotrade ows.Onepossiblewaytoassessthesensitivityoftheresultstolooseningthedata
truncationatzerowouldbetotightenitfurther; thatis,anyselectionbiascausedbydropping
zerovalueswouldbeenhancedbydroppingsporadicones.
Specication(iii)usesexporter-yearxedeectsintheplaceofGDP. Sincethesexedeects
alsoapproximatechangestothemultilateralresistancetermsoftheexporter,theymayinfactbe
soakingupsomeoftheinformationoncompetitiveness intendedtobemeasured. . As s such,the
robustnessoftheresultconsistsofasimilarproleofresidualchangesinspecications(i)through
(iii),duetothefollowingtrade-o:inthersttwothereislikelysomeomittedvariablebiassince
implicitindexes of multilateralresistance(asdenedinAndersonandvanWincoop(2003))are
themselvesafunctionofgeographicvariablesincludedintheregression. Ontheotherhand,the
appropriatecontrolformultilateralresistanceremovesfromtheresidualatleastsomeinformation
on the e relative e performance e of exporters. . Specications s (iv) ) and d (v) ) check k the e robustness of
theresultstomorestandardgravityspecications,byunfolding(3)intolevelsandincorporating
conventionalmeasures of statictradecosts. . Specication(vi)uses s aanalternativedatasource
onbilateralinternationaltrade owsaggregatedintobroaderISIC2-digitsectors.7
3.2 Reducedformrevealedcompetitiveness: : results
After controllingfor model l factors inseveralalternativeformulations s of thegravity model, , we
ndthattheU.S.exportshareisonlyinslightdecline. Inourbenchmarkspecication(i),the
majorityoftheroughly20percentdeclineinaggregateU.S.exportshareisexplainedbythemodel
withabouta6percentdeclineintheresidual.
8
Table3shows the estimates ofcontrolvariables for (3) estimatedacross all l products.
9
As
expected,exporterGDPshareispositivelyrelatedtoexportshare,witha1percentdecreasein
relativeincomedecreasingexportsharebyroughly0.4-0.6percent. Thesemagnitudesaresimilar
tothecoecients onGDPinthelevelregressionsandslightlylowerthanthose usingtheISIC
data. Theeect t of NAFTAandtheintroductionoftheeuroarebothpositiveandsignicant,
withcoecients rangingfrom0.4-1.5and0.1-0.5, , respectively. . Measuresofdistance,language
andborderhavetheexpectedsign.
7
Specication(vi)conrmstheconsistencyofthereducedformresultswiththeempiricalexerciseinthefollowing
section. WhilethereducedformregressionsusetheFeentraetal. . (2005)datadescribedabove,themethodologyin
thenextsection additionallyrequiresdataonsectoralintra-nationaltrade,whichnecessitatesusinganalternative
dataset. Thosedataaredescribedbelow.
8ResultsfortheremainingvespecicationscanbefoundinAppendixB.
9
Asmentioned,thegravityresidualsareestimatedattheSITC4-digit levelfor specications(i)-(v) andatthe
ISIC 2-digit level for specication (vi). . In n the table, due tocomputational constraints on such a a large dataset,
wepresent aggregate control variables estimated without product xed eects. . As s such, the coecients can be
interpretedas simpleaveragesacrossSITCproducts,orinthecaseof specication(vi),ISICproducts.
7
AnindexofmarketsharechangesfortheU.S.,alongwithanindexofmodelpredictedvalues,
areshowninFigure7. TheindexineachyearisageometricmeanofsharechangesacrossU.S.
destinationcountriesandproducts,whereeachchangeinshareisweightedbytheSITC-importer
valueintheyear 2000.
10
Despitea wideningofthegapbetweenthetwo indices s intheearly
period,themodelpredictionbroadlyfollows thesharetrend. . Since e there arenot manytime-
varying regressors s inour gravityestimation, , this s result is closely relatedtothe observation n in
Figure6thatU.S.incomeshareandtradesharehavesimilardynamics.
To construct a statistic for r the overall l percent change in n market t share due to o the gravity
residual,theratioofactualtopredictedshareisaveragedacrosstimeperiodsintheearlypartof
thesample(1980-1992)andthelatterpart(1993-2004)andthelog-dierenceofthesetworatiosis
takenforeachexporter-importer-SITCgroup. Theaverageofthosestatisticsacrossdestinations
and4-digit productgroupsis showninFigure8fortheG20plusSingapore,TaiwanandHong
Kong.11 Acrossallproducts,theU.S.isinthemiddleofthepackwithdecreasesinitsresidual
of 6 percent. . This s canbe interpretedas adecrease inU.S.export market share of 6percent
thatisnotaccountedforbythedynamicsofincome,andisnotablysmallerthantheoverallshare
declineofapproximately20percentoverthatperiod. ThissuggeststhatU.S.relativeproductivity
competitiveness,albeitinslightdeclinebythismeasure,didnotdeclinebynearlyasmuchasitsfall
insharemightsuggest. Thisresultisconsistentacrossproductcategories,showninTable4,even
forSITC7(machineryandtransportation)whereU.S.shareperformancewasparticularlygrave,
as wellasforotherspecications showninAppendixB. . For r otherexporters,clearwinnersand
losersemerge. Indonesia,China,IndiaandMexicohadamongthehighestincreasesintheirgravity
residualbyalargemargin,astheirexportgrowthfaroutpacedtheincreaseintheirincomeshares.
Ontheotherhand,certainlargeAsianexportershaddramaticfallsintheirresidualspresumably
duetotheriseofChinaandlargeincreasesinMexicanexportstotheU.S.overthesampleperiod.
European countries andCanada hadmore e moderate changes s in their r export performance and,
withafewexceptions,tendedtolagbehindtherestoftheworld.
Insummary,thisreducedformexercisestronglysupportsthestorythatexporterincomeshares
areanimportantdeterminantoftradeshares. Beyondthat,however,itisdiculttoknowwhether
thegravityresidualre ectstheactualevolutionintheunderlyingproductivityofexportersrather
thanother factors, , such as evolving trade costs. . In n the following sectionwe e take a dierent
approach to identifying relative e cost t competitiveness across s countries s by y modelling the e micro-
foundationsoftradesharesexplicitly.
4 StructuralRevealedCompetitiveness
Inthis sectionwe buildonamulti-country multi-sectorversionof theMelitz-Ottaviano(2008)
modeltoobtaina(computable)structuralequationfortherelativecompetitivenessofacountry.
AfulldescriptionofthereferencemodelisreportedinCorcosetal. (FEEM,2010),althoughits
mainpropertiesaresummarizedinAppendixA.
Themodelyieldsthefollowingexpressionforaggregatebilateraltradefromcountryltocountry
10For the sakeof comparability, the predicted and actualmarket sharechanges areaggregatedover exactlythe
sameSITC-destination pairs. . Theindexofsharechangedoes s notexactlymatch that inFigure1,since:(a) itis
ageometricindex, whereas simplyaddingup shareacross products as in Figure 1is analogous toan arithmetic
mean,and(ii)becausetheindexismatchedineachperiod(i.e.,thetrade owhadtooccurinbothtimetandt-1
forittobeincluded),thecompositionofitemsintheFigure2indexwillbeasubsetofthoseinFigure1. Overall,
the magnitudeof thedropofthegeometricindexseems reasonablyclosetotheaggregatedropandthedynamics
of contractionsintheearly1980’sand2000’s paralleloneanother.
11
Thislistcorrespondswellwiththetoptwentyexportersbysizein1980. Inthetable,thecategory‘OtherEU’
includes:Austria,Belguim,Denmark,Finland,Ireland,Netherlands,PortugalandSweden.
8
Documents you may be interested
Documents you may be interested