(AGENPARL) – ven 24 febbraio 2023 OFFICIAL USE
Dear Colleagues,
Many of you inquired about our upcoming working paper focused on current trade flows between Russia, the Caucasus and Central Asia.
This paper titled “The Eurasian roundabout: Trade flows into Russia through the Caucasus and Central Asia” is [out](https://www.ebrd.com/publications/working-papers/the-eurasian-roundabout) now.
It has been co-authored by Beata Javorcik, Maxim Chupilkin, and Alexander Plekhanov. Please find it attached and an accompanying press release below.
With kind regards,
EBRD External Communications
[]EBRD press release
EBRD analyses trade flows between Russia, Caucasus and Central Asia
Some intermediated trade may be used to circumvent sanctions but its scale is limited
FOR IMMEDIATE RELEASE
http://www.ebrd.com/
http://www.ebrd.com/news
https://twitter.com/ebrdhttps://www.facebook.com/ebrdhqhttps://www.linkedin.com/company/ebrdhttps://www.instagram.com/ebrd_official/https://www.youtube.com/user/ebrdtv
Date: 24/02/2023
Contact: Anton Usov
Twitter: [@AntonUsov1892](https://twitter.com/AntonUsov1892)
Tags: @EBRD
EBRD looks at trade patterns between Russia and its neighbours in the aftermath of introduced sanctions
·EBRD economists analyse trade flows between Western countries, Russia and its neighbours
·Dramatic drop in direct exports to Russia from the West, indirect exports intermediated via Caucasus and Central Asia
·Some intermediated trade may be used to circumvent sanctions, though its scale is limited
Intermediated trade through a number of countries neighbouring [Russia](https://www.ebrd.com/russia.html) may be used to circumvent economic sanctions but on a limited scale according to a newly published EBRD working paper [The Eurasian roundabout: Trade flows into Russia through the Caucasus and Central Asia](https://www.ebrd.com/publications/working-papers/the-eurasian-roundabout).
The paper documents the dramatic reduction of exports from the [EU](https://www.ebrd.com/who-we-are/structure-and-management/shareholders/european-union.html), [UK](https://www.ebrd.com/who-we-are/structure-and-management/shareholders/uk.html) and [US](https://www.ebrd.com/who-we-are/structure-and-management/shareholders/usa.html) to Russia following the introduction of economic sanctions in March 2022 in the aftermath of Russia’s invasion of [Ukraine](https://www.ebrd.com/ukraine.html).
At the same time, it notes the increase of exports to [Armenia](https://www.ebrd.com/armenia.html), [Kazakhstan](https://www.ebrd.com/kazakhstan.html) and the [Kyrgyz Republic](https://www.ebrd.com/kyrgyz-republic.html) (CCA3 – all members of the [Eurasian Economic Union](http://www.eaeunion.org/?lang=en)). The study also documents the increase of trade flows between the CCA3 states and Russia.
“The war has caused significant changes in the regional trade patterns. Direct exports from Western democracies to Russia have dropped dramatically. We saw the emergence of intermediated trade – an increase in Western exports to Central Asian and the Caucasus countries accompanied by an increase in exports from these countries to Russia.”
“This intermediated trade, however, is only a fraction of what was previously exported to Russia directly. This situation created an opening for other suppliers, such as Türkiye and China, both of which have substantially increased their sales to Russia,” explained[Beata Javorcik](https://www.ebrd.com/who-we-are/senior-management/beata-javorcik), the EBRD’s Chief Economist, and one of the new paper’s authors.
The paper argues that trade flows consistent with rerouting of trade to Russia were more pronounced for product groups where goods are at least partially subject to sanctions and for goods that are similar to sanctioned ones.
At the same time it notes that the increase in exports of sanctioned goods to CCA3 countries represents a small fraction of the reduction in direct exports of sanctioned goods to Russia (around 5 per cent) but can be large for specific product groups.
The patterns of trade summarised in the EBRD paper rely on the official export data but do not include any illicit activities such as smuggling of sanctioned or non-sanctioned goods.
OFFICIAL USE
To learn more about EBRD classifications, visit http://www.ebrd.com/ic
______________________________________________________________

Testo Allegato: 1Introduction
1
blocwithRussia,theEurasianEconomicUnion.
2
Goodsexportedtotheseeconomiescouldpotentially
beshippedtoRussiawithminimumchecks(akintoshipments,say,withintheEU).Wealsoconsider
2
Belarus,theremainingmember,wasalsosubjecttoeconomicsanctions.
3
OuranalysistreatsexportsfromtheEUandtheUKasasingleaggregate.
3
ChinaandTurkiyetoRussiaincreasedrelativetothepre-sanctionstrendfromaroundthemid-summer
of2022.AdditionalexportstendedtobemorepronouncedforsanctionedgoodsinthecaseofChina,
whileinthecaseofTurkiyepatternsinthecompositionoftradearelessclear-cut.ForChina,thedata
alsorevealasignicantincreaseinexportsofnewproductgroupstoRussia.Overall,ittookabouttwo
tofourmonthsforthesenewsupplyroutestoemerge.
Ourpapercontributestoseveralstrandsoftheeconomicliterature.First,wecontributetothe
literatureonintermediated(or”entrepot”)tradebyshowingthattraderoutedviathirdcountriescan
beusedtoevadeeconomicsanctions.Suchtradehasbeenearliershowntofacilitateevasionoftaris
2.2Bilateraltradedata
SincetheimpositionofsanctionsRussiahassuspendedpublicationoftradestatisticsandanumberof
otherkeyeconomicindicators,suchasthebreakdownofgovernmentexpenditure.Thusour
4
https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:02014R0833-20221007from=ENtocId108,for
instance,AnnexIIandAnnexVII-B
5
https://trade.ec.europa.eu/doclib/docs/2016/february/tradoc
1
54240
:pdf
6
Thelistofproductgroupspartiallycoveredbythesanctionsisaneclecticmixof1,983HS6codes
combining:weapons(HS9301),semi-conductormedia(852352),enginesandpumps(8412,8413),
containers(860900),aircraftandparts(88),ammonia(281420),steelpipesforoilpipelines(730411),
navigationinstruments(9014),skisuits(611220),andothers.
Wealsoidentifyproductlinesthatarenotnecessarilycoveredbysanctionsbutaresimilartogoods
thatarepartiallyundersanctions(i.e.,thosenestedinthesameHS4code).Thosegoodsareofinterest
sincegoodscoveredbysanctionsmaybeintentionallymisclassiedincustomsdeclarationstoavoid
6
Forinstance,Bevan(2005),discusseshowarmscanbedeclaredassportsarmsforcustomspurposes.Theliterature
hasdocumentedproductmisclassicationbeingfrequentlyusedtoevadeimportduties.Seeforinstance,FismanandWei
(2004),JavorcikandNarciso(2008)andJavorcikandNarciso(2017).
7
August2022).Tradeinallproductcategorieswentup.Atthesametime,TurkishexportstoCCA4
sawastratosphericincrease,albeitfromasmallbaselineinabsoluteterms.Thisincreasewasmost
3Empiricalspecications
3.1Intensivemarginoftrade
Ourtriple-dierencedspecicationforasingleexportingcountrytakesthefollowingform:
Log
(
Exports
pit
)=

1
Post
t

Sanction
p

Russia
i
+

2
Post
t

Sanction
p

CCA
3
i
+
+

3
Post
t

Sanction
p

Georgia
i
+

pi
+

pt
+

it
+

pit
(1)
wherethedependentvariableisdenedasthelogarithmofthevalueofexportsofHS6product
p
exportedinmonth
t
totheimportingcountry
i
.Thecoecientsofinterestarethoseoninteraction
7
{adropinexportstoRussiaandanincreaseinexportstoArmeniaortheKyrgyzRepublic,oran
overalldeclineinexportsofaparticularproductinaparticularyear{butnotdierencesinbilateral
tradeinspecicproductswithinagivenmonth.
3.2Extensivemargin:Probabilityoftradetakingplace
Halformoreofobservationsonbilateralmonthlytradeatthishighlevelofdisaggregationarezeros.
Withthisinmind,itisalsousefultolookattheextensivemarginoftrade{theprobabilitythat
exportsinagivenmontharepositiveforagivenpairoftradepartnersandagivenHS6productcode.
Specicationssimilartothosefortheintensivemarginoftradeareestimatedasalinearprobability
model:
P
(
Exports
pit
`
0)=

1
Post
t

Sanction
p

Russia
i
+

2
Post
t

Sanction
p

CCA
3
i
+
+

3
Post
t

Sanction
p

Georgia
i
+

pi
+

pt
+

it
+

pit
(2)
wherethedependentvariabletakesavalueofoneforpositivetradevaluesandzerootherwise.
3.3Extensiveandintensivemargin
Tocombinetheestimatesontheextensiveandtheintensivemarginsoftrade,weusetheinverse
hyperbolicsinetransformationofthevaluesoftrade,
log
(
Exp
+
p
Exp
2
+1)(seeMacKinnonand
Magee(1990)).Thistransformationapproximatesthelogarithmictransformationforlargetrade
volumeswhileassigningthevalueofzerotozerotraderatherthandiscardingzeroobservations.
4Results
4.1EUexportsofpartiallyorfullysanctionedproducts
WebeginoureconometricanalysisbyconsideringtheEU/UKexports(EU-27plusUK)toRussia,
CCA3,Georgia,Azerbaijan,Mongoliaandtherestoftheworld.Wefocusonthedierentialpatternsof
Theresults,presentedinColumn1ofTable4,suggestthatafterMarch2022,theEU/UKexportsto
Russiaoffullyorpartiallysanctionedgoodswere74percentlowercomparedtoexportsofothergoods
andtowhatcouldhavebeenexpectedbasedonpatternsobservedpriortoMarch2022.Thisdierence
isstatisticallysignicantatthe1percentlevelandcomesontopoftheoveralldropinEUexportsto
Russia(subsumedintherespectivexedeect).
Atthesametime,fromMarch2022onwardstheEU/UKexportsoffullyorpartiallysanctionedgoods
toCCA3were22percenthigherthanexportsofothergoods.Againtheestimatedcoecientis
statisticallysignicantatthe1percentlevel.Thecorrespondingcoecientontheinteractiontermfor
exportstoGeorgiaissmallandnegative,suggestinga-6.5percentrelativedeclineinexportsofthese
productstoGeorgia,withthestatisticalsignicanceofthecoecientreachingonlythe10percentlevel.
Similarpatternsareobservedontheextensivemargin.Forinstance,forsanctionedgoodsthe
probabilityofobservingEU/UKexportstoRussiadeclinedbyadditional8.3percentagepointswhile
theprobabilityofEU/UKexportstoCCA3increasedbyadditional1.8percentagepoints(seeColumn
2).Thisisasizableeect,consideringtheunconditionalprobabilityofobservinganon-zerotrade
oftrade(seeColumn3).Coecientstendtobelargerre
ectingthecombinedeectsontheextensive
andintensivemargins.InthecaseofGeorgia,theestimatesinColumns2and3donotreach
conventionalsignicancelevels.
TheseverystrikingpatternisconsistentwithascenariounderwhichWesternrmscutdirecttrading
relationshipswithRussiabutscaleuptradewithentitieslocatedintheCCA3countries.Itispossible
thattheseCCA3intermediariesthenre-exportproductstoRussia,notnecessarilywiththeknowledge
oftheWesternexporter.
4.2EUexportsofgoodssimilartosanctionedgoods
Asmentionedearlier,theexistingliteraturehasdocumentedfrequentandextensivemisclassicationof
importedproductsandfoundevidenceconsistentwithsuchmisclassicationbeingusedtoevadeimport
9
Recalltheexampleofx-rayapparatusfordentistryandsimilarmedicaluses(HS902213)notbeingcoveredbytheEU
sanctionsonRussia,whilex-rayapparatusfornon-medicaluses(HS902219)beingsubjecttosanctions.
11
TheresultsarepresentedinthebottompanelofTable4.Theyarequalitativelysimilartothosefor
sanctionedproducts,thoughtheestimatestendtobesomewhatsmallerinmagnitude.Forinstance,the
EUexportsofsimilar-to-sanctionedgoodstoRussiadeclinedmoresubstantiallythanthoseofother
goods,althoughtheobserveddierential,ofaround15percent,issignicantlysmallerthanthatforthe
fullyorpartiallysanctionedgoods(around80percentinthesespecications).Moreover,theresults
revealadditionalincreaseinexportsofsimilargoodsfromEUtoCCA3ofaround15percent,
statisticallysignicantatthe1percentlevel.
ThepatternobservedwithrespecttoexportstoRussiamayhavetodowiththeambiguityofsanctions
andcompliancecomplexity(theburdenofproofthatshipmentisdistinctfromasimilarshipmentthat
wouldbesubjecttosanctions).Theresultsmayalsore
ectvoluntarywithdrawalbyrmssupplying
4.4ExportsfromChina
Next,ouranalysisfocusesonChina’sexports,whichdisplayamarkedlydierentpatternfromthoseof
theEU/UKandUSexports.TheresultsinthetoppanelofTable5indicatethatChineseshipmentsof
fullyorpartiallysanctionedgoodstoRussia,CCA3andGeorgiaincreasedbyadditional8to15percent
comparedwithtradeinothergoods,withthesedierencesbeingstatisticallysignicantatthe1
percentlevel(exceptforonecoecientforGeorgia).Thesemagnitudesincreaseto12to18percent
whenweaccountforgoodssimilartosanctionedproductsinthebottompanelofthetable.Largeand
statisticallysignicanteectsarealsoobservedontheextensivemarginoftrade,re
ectingdiverse
natureofproductionandexportcapacityofChina’seconomy.
Strikingly,thereisnoindicationofChinaincreasingexportsofgoodssimilartosanctionedproducts
morethanexportsofotherproducts.Noneoftherelevantinteractiontermsinthebottompanelofthe
tableisstatisticallysignicantatconventionallevels(withtheexceptionoftheinteractiontermsof
Georgiaintheextensivemarginanalysis).
4.5ExportsfromTurkiye
ThepatternsforTurkiyearetheleastclear-cut.Whenwecomparetradeinthefullyorpartially
sanctionedproductstootherproductsinthetoppanelofTable5,weseenodierentialtrajectoryfor
thesetwotypesofproductswhendestinedtoRussia.However,weseeanincreaseinexportsoffullyor
partiallysanctionedgoodstoCCA3(withtwoofthreecoecientsbeingstatisticallysignicant).
Thepicturechangeswhenweadditionallyconsiderproductssimilartosanctionedproducts(seethe
bottompanel).Accordingtothosespecications,exportsfromTurkiyetoRussiaincreasedsignicantly
morerapidlyforfullyorpartiallysanctionedproductsandproductssimilartothesanctionedones,with
thelargestdierential(of35percent)observedforsimilarproducts.Comparedwithnon-similargoods,
tradeinsanctionedgoodsincreased,byadditional21percent.Atthesametime,exportsoffullyor
partiallysanctionedgoodstoCCA3alsoincreasedbyanadditional20percent.Unlikeinthecaseof
China’sexports,theseeectsareonlyobservedattheintensivemargin,whereTurkiye-Russia(or
Turkiye-CCA3)exportrelationshipsalreadyexisted.Ingeneral,theresultsforTurkiyearelessrobust
intermsofstatisticalsignicance.
13
4.6Eventstudyanalysis
Themostly-unanticipatednatureofthewarandtheclear-cuttimingofsanctionsmeansthatour
beproducedondemandforaparticularcustomer)anddurable.Thispatternmaybeexplainedbythe
willingnessofwell-oconsumersinRussiatopayahighpremiumforthesetypesofgoods,apremium
thatcanpayfortheworkofintermediariesinvolved.FordomesticrmsusingEUproductsasinputs,it
maybehardertopassonadditionalcoststoconsumers;thesameistrueforimportersofless
dierentiatednalgoodswheretheproducer(andhencethecountryoforigin)tendstomatterless.
Likewise,genericgoodsmaybeeasiertosourcefromelsewherethanspecicgoodswhilenon-durable
goods(thosethatexpiremorequickly)maynotlendthemselveseasilytocomplexindirectroutings.
Incontrast,increaseddirectexportsfromChinatoRussiadisproportionatelyconcernintermediateand
capitalgoodsasrmsinRussialookedforalternativesupplychains.Forinstance,exportsof
heavy-dutytrucksfromChinatoRussiaincreasedmarkedlyin2022.
4.8Discussion:Themagnitudeofintermediatedtrade
Intotal,theincreaseinexportsoffullyorpartiallysanctionedgoodsfromtheEU,theUKandtheUS
toCCA4amountedtoaroundUS
$
1billionovertheperiodMarch-August2022.Thisamountstomore
thanthreequartersoftheoverallexportsfromthesecountriestoCCA4.However,itcorrespondsto
only4percentofthedeclineinthedirectexportsoffullyorpartiallysanctionedgoodsfromthese
economiestoRussiaoverthesameperiod(seeTable3;theratioisslightlyhigher,at5to6percent,in
theMay-Augustperiod).
Theextentofreroutingappearstobehigherforparticularproductgroups.Forinstance,theincreasein
exportsofpumpsandcompressors(HS841480)toCCA4representsmorethan50percentofthe
correspondingreductioninexportstoRussia.Forinternal-combustionvehicles(870324),thisratiois
around60percent.Forthisproductline,theEU/UKexportstoCCA4averagedUS
$
113millionin
May-August2022,comparedwithanaverageofUS
$
10millioninthesamemonthsof2017-21,while
thecorrespondingEU/UKexportstoRussiadroppedbyUS
$
178million.Increasedvolumeswere
observed,inparticular,forexportstoKazakhstanandtheKyrgyzRepublic.Forwashingmachines
parts(845090)thesubstitutionratioexceeds70percent;forpaintsandvarnishes(320820)itexceeds
90percent.
Foranumberofspeciclines(forinstance,automaticdataprocessingmachines,orcomputers
(847130),track-layingtractors(870130)andcombineharvesters(843351))theratioexceeds100per
cent.Forexample,theEU/UKexportstoCCA4ofHS847130(computers)increasedfromUS
$
11
milliontoUS
$
51million(drivenbyincreasesinexportstoKazakhstanandArmenia)whilethe
correspondingexportstoRussiadeclinedbyUS
$
22million.
15
Productlineswiththissubstitutionratio”inexcessof50percentareadiversemixcoveringmorethan
130HS6productcodes.
WhilethetotaladditionaltradefromselectedadvancedeconomiesviaArmenia,Kazakhstanandthe
KyrgyzRepublic(insanctionedandnon-sanctionedgoods)isasmallpercentageofthedeclineindirect
exportstoRussia,theamountsinvolvedarelargefortheintermediaryeconomies.IntheKyrgyz
RepublicandArmenia,theywereequivalentto4to6percentoftheirGDP(annualized),risingto
valuesinexcessof15percentofGDPifadditionalimportsfromChinaandTurkiyeareincluded.
Thelogisticsandintermediationservicesassociatedwithincreasedexportswouldhavemadeasizable
contributiontotheseeconomies’GDPandcapitalin
ows.Thecostofservicesiseectivelybilledto
end-consumersinRussiawhofacesignicantlyhigherpricesforimportedgoods(seeHinzand
Monastyrenko(2022)forevidenceofpriceincreasesfollowingtheearlierroundofsanctionsand
counter-sanctionsinRussia).Asaresult,thecurrenciesofeconomiesprovidingintermediaryservices
andanintermediaryinCCA4andthatintermediaryandanentityintheEUortheUK.Indeed,capital

owsfromRussiatoCCA4economiesincreasedmarkedlyinthelatermonthsof2022(althoughthose
increasesmixtradepayments,capital
ight,transfersaccompanyingemigrationandotherpayments).
PartoftheobservedincreasesinimportdemandfromCCA4economiesmayberelatedtootherfactors,
includingthearrivalofasubstantialnumberofmigrantsfromRussiawhocrossedtheborderin2022.
Wefurtherconductplacebotestsbylookingatsimilarpatternsforadditionalimporters.Wedonot
ndsignicantpatternsforotherneighbouringeconomies(AzerbaijanorMongolia).
References
Afesorgbor,SylvanusKwaku(2019)`Theimpactofeconomicsanctionsoninternationaltrade:Howdo
threatenedsanctionscomparewithimposedsanctions?’
EuropeanJournalofPoliticalEconomy
56,11{26
(2017)`WTOaccessionandtarievasion.’
JournalofDevelopmentEconomics
125,59{71
Kaempfer,WilliamH.,andAntonD.Lowenberg(1988)`Thetheoryofinternationaleconomic
sanctions:Apublicchoiceapproach.’
TheAmericanEconomicReview
78(4),786{793
MacKinnon,JamesG.,andLonnieMagee(1990)`Transformingthedependentvariableinregression
models.’
InternationalEconomicReview
31(2),315{339
Peeva,Aleksandra(2019)`DidsanctionshelpPutin?’DiscussionPapers2019/7,FreeUniversityBerlin,
SchoolofBusinessEconomics
Sonnenfeld,Jerey,StevenTian,StevenZaslavsky,YashBhansali,andRyanVakil(2022)`Itpaysfor
Figure1:ExportstoRussia,byexporterandtypeofgoods
Sources:Authors’calculationsbasedonUNComtradeandChinaCustomsAdministration.
Note:Basedontradereportedbyexporters.TradeinnominalUSdollarsisadjustedforUSin
ation.
EUtotalisinclusiveoftheUK.SanctionedreferstoHS6productlineswhereEUsanctionsapplyat
leastpartially.SimilargoodsarethosenotsanctionedbytheEUbutwithinthesameHS4assanctioned
products.CCA4isArmenia,Georgia,Kazakhstan,KyrgyzR.
22
Figure2:ShareofexportstoRussiaandCCA3intotalexports
Sources:Authors’calculations.
Sources:Authors’calculations.
Note:Basedontradereportedbyexporters.TradeinnominalUSdollarsisadjustedforUSin
ation..
23
Figure4:Exportsofsanctionedvsnon-sanctionedgoods
Sources:Authors’calculations.
Table1:
Descriptivestatistics
EU/UKexportstoRussiaChinaexportstoRussia
V
ariablesMeanMedianSt.dev.MinMaxMeanMedianSt.dev.MinMax
Trade,log11.8512.132.840.0120.7911.9012.212.590.0020.15
Trade,hyperbolic10.1212.005.570.0021.488.7411.416.200.0020.85
Trade,non-zero0.811.000.390.001.000.691.000.460.001.00
EU/UKexportstoCCA3ChinaexportstoCCA3
V
ariablesMeanMedianSt.dev.MinMaxMeanMedianSt.dev.MinMax
Trade,log8.858.972.700.0018.7910.2110.442.720.0019.14
Trade,hyperbolic4.040.005.030.0019.494.130.005.550.0019.84
Trade,non-zero0.420.000.490.001.000.380.000.490.001.00
EU/UKexportstorestoftheworldChinaexportstorestoftheworld
V
ariablesMeanMedianSt.dev.MinMaxMeanMedianSt.dev.MinMax
Trade,log14.9215.222.62-6.9323.1115.2015.582.720.0023.72
Trade,hyperbolic15.2415.843.520.0023.8114.4515.955.250.0024.42
Trade,non-zero0.981.000.150.001.000.911.000.290.001.00
USexportstoRussiaTurkiyeexportstoRussia
V
ariablesMeanMedianSt.dev.MinMaxMeanMedianSt.dev.MinMax
Trade,log10.9010.771.887.8219.869.659.992.750.0018.37
Trade,hyperbolic3.800.005.550.0020.554.400.005.420.0019.06
Trade,non-zero0.330.000.470.001.000.420.000.490.001.00
USexportstoCCA3TurkiyeexportstoCCA3
V
ariablesMeanMedianSt.dev.MinMaxMeanMedianSt.dev.MinMax
Trade,log10.059.781.604.2218.348.538.612.440.0017.23
Trade,hyperbolic1.440.003.710.0019.032.910.004.500.0017.92
Trade,non-zero0.130.000.340.001.000.320.000.460.001.00
USexportstorestoftheworldTurkiyeexportstorestoftheworld
V
ariablesMeanMedianSt.dev.MinMaxMeanMedianSt.dev.MinMax
Trade,log14.4714.622.467.8223.5012.1512.462.920.0020.67
Trade,hyperbolic7.110.007.750.0024.194.360.006.310.0021.36
Trade,non-zero0.470.000.500.001.000.340.000.470.001.00
Source
:Authors’calculationsbasedonUNComtradeandChinaCustomsAdministration.
Note
:UnitofobservationisbilateralexportsatHS6levelinagivenmonthfromJanuary2017toAugust2022(forChina,toDecember2022).Observationswithzero
tradeinallmonthforagivencountrypairandHS6productgrouparenotincluded.
Table2:
Averagemonthlycross-bordertrade
owsinMay-July2022,inpercentofthe
May-Julyaveragein2017-19
FromCCA4WorldARMGEOKAZKGZBLRRUSAZEMNGTJKUZB
Turkiye143120139133182202197119209203164
EU+UK+US12410018612211518861424013296
EU12292172119114184614439121111
Germany137931969913430952387810188101
UK80932238669144112315547394
US14311244813312425210125311032034
Source
:AuthorscalculationsbasedonUNComtrade.
Note
:Basedontradereportedbyexporters.TradeinnominalUSdollarsisadjustedforUSin
ation.
25
Table3:
Changeinexportvolumes,Mar-Aug2022relativetotheaverageofMar-Aug
2017-21,inUS
$
billion
TotaltradeSanctionedSimilarOther
RussiaCCA4RussiaCCA4RussiaCCA4RussiaCCA4
EU/UK-32.791.14-24.600.76-4.960.06-3.230.31
US-3.240.27-1.740.26-0.340.00-1.160.01
China-0.173.22-1.892.611.860.53-0.140.09
Turkiye1.490.700.560.370.390.160.540.18
Source
:AuthorscalculationsbasedonUNComtradeandChinaCustomsAdministration.
Note
:Basedontradereportedbyexporters.TradeinnominalUSdollarsadjustedforUSin
ation.EUtotalisinclusiveoftheUK.Sanctioned
referstoHS6productlineswhereEUsanctionsapplyatleastpartially.SimilargoodsarethosenotsanctionedbutwithinthesameHS4as
sanctioned.
Table4:
Exportsofsanctionedvsothergoods:EU/UK,US
123456
Exporter
EU/UKUS
Dep.var.Exports
Log0-1HyperbolicLog0-1Hyperbolic
PostxSanctionedxRussia-0.739***-0.0829***-1.612***-0.785***-0.0678***-0.972***
(0.0430)(0.00743)(0.0814)(0.132)(0.0101)(0.120)
PostxSanctionedxCCA30.223***0.0182***0.409***0.1050.005330.175**
(0.0349)(0.00450)(0.0500)(0.0833)(0.00682)(0.0782)
PostxSanctionedxGeorgia-0.0652*-0.004340.01150.110-0.00864-0.0216
(0.0344)(0.00550)(0.0551)(0.141)(0.00803)(0.0908)
Observations1,544,4242,618,2042,618,204228,3621,254,1241,254,124
R
2
0.9110.7310.8650.9400.7230.811
789101112
Exporter
EU/UKUS
Dep.var.Exports
Log0-1HyperbolicLog0-1Hyperbolic
PostxSanctionedxRussia-0.808***-0.0980***-1.834***-0.837***-0.0950***-1.255***
(0.0504)(0.00813)(0.0901)(0.183)(0.0115)(0.136)
PostxSimilarxRussia-0.150***-0.0334***-0.492***-0.0980-0.0546***-0.568***
(0.0523)(0.00870)(0.0942)(0.206)(0.0131)(0.151)
PostxSanctionedxCCA30.294***0.0284***0.538***0.02450.009730.242**
(0.0389)(0.00523)(0.0564)(0.129)(0.00825)(0.0948)
PostxSimilarxCCA30.153***0.0226***0.284***-0.1380.008730.131
(0.0461)(0.00604)(0.0632)(0.140)(0.00932)(0.104)
PostxSanctionedxGeorgia-0.110***-0.003900.002570.0769-0.0181*-0.116
(0.0398)(0.00626)(0.0615)(0.172)(0.00933)(0.106)
PostxSimilarxGeorgia-0.107**0.00119-0.0191-0.0643-0.0197*-0.197
(0.0484)(0.00733)(0.0716)(0.233)(0.0116)(0.131)
Observations1,544,4242,618,2042,618,204228,3621,254,1241,254,124
R
2
0.9110.7310.8650.9400.7230.811
Source
:Authors’calculations.
Note
:StandarderrorsareclusteredattheHS6level.Allregressionsincludemonthximporter,monthxHS6andHS6ximporterxedeects.*,**,***denote
statisticalsignicanceatthe10%,5%and1%levels,respectively.Thedependentvariableisthelogarithmofbilateraltradeinagivenmonthinagivenproductgroup
attheHS6level;0-1variablefortradetakingplaceortheinversehyperbolicsinetransformationoftrade.ImportereconomiescompriseArmenia,Azerbaijan,Georgia,
Kazakhstan,KyrgyzR.,Mongolia,Russia,Turkiyeandtherestoftheworld(aggregated).SanctionedreferstoHS6productlineswhereEUsanctionsapplyatleast
partially.SimilargoodsarethosenotsanctionedbutwithinthesameHS4assanctioned.Incolumns1-6similargoodsarepartof”other”group.
26
TheEurasianroundabout:
Trade
owsintoRussiathroughtheCaucasusandCentralAsia
*
MaximChupilkin
EuropeanBankforReconstructionandDevelopment(EBRD)
BeataJavorcik
UniversityofOxford,EBRDandCEPR
AlexanderPlekhanov
EBRD
February2023;Firstdraft
Abstract
Thispaperdocumentsasubstantialchangeinregionaltradepatternsprecipitatedbythewaron
UkraineandthesubsequentintroductionoftradesanctionsonRussia.Itprovidesevidencesuggestive
ofintermediatedtradevianeighbouringeconomiesbeingusedtocircumventthesanctions.The
analysis,basedonHS6product-leveldataonbilateralmonthlyexports,showsasharpdropindirect
exportsfromtheEUtoRussiafollowingtheintroductionofsanctionsinMarch2022.Atthesame
time,anincreaseintheEUexportstoArmenia,KazakhstanandtheKyrgyzRepublic(CCA3,all
membersofEurasianCustomsUnionalongsideBelarusandRussia)isobserved.Bothpatternsare
particularlypronouncedforproductgroupspartiallyorfullysubjecttotheEUsanctionsaswellas
goodsthataresimilartothesanctionedones.ThedropinEUexportstoRussiawasaround80
percentsteeperforsanctionedgoodsthanforotherproducts,whileexportsofsanctionedgoodsto
CCA3rosebyextra30percentrelativetothoseofothergoods.Theincreaseinexportsofsanctioned
goodstoCCA3representsasmallfractionofthereductionintheirdirectexportstoRussiabutcan
belargeforspecicproductgroups.
Keywords:sanctions,tradediversion,re-exports,intermediatedtrade,Russia
JELClassicationNumber:F14,F15,F51
*
Chupilkinemail:chupilkim@ebrd.com;Javorcikemail:beata.javorcik@economics.ox.ac.uk;Plekhanovemail:
plekhana@ebrd.com.Theviewsarethoseoftheauthorsandshouldnotbeattributedtotheinstitutionstheyarealiated
with.TheauthorsaregratefultoJosephSassoonfortheexcellentresearchassistance.
1
The Eurasian roundabout:

Trade flows into Russia
hrough
the
Caucasus and Central Asia
axim Chupilkin, Beata Javorcik
and Alexander Plekhanov
Abstract
This paper documents a
substantial change in regional trade patterns precipitated by the war on
Ukraine and the subsequent introduction of trade sanctions on Russia. It provides evidence
suggestive
of intermediated trade via neighbouring economies being used to circumvent the
sanctions. The
analysis, based on HS6 product
level data on bilateral monthly exports, shows a
sharp drop in direct
exports from the EU to Russia following the introduction of sanctions in March
2022. At the same
time, an increase in the EU exports to Armenia, Kazakhstan and the Kyrgyz
Republic (CCA3, all
members of Eurasian Customs Union alongside Belarus and Russia) is
observed. Both patterns are
particularly pronounced for product groups partially or fully subject to
the EU sanctions as well as
goods that are similar to the sanctioned ones. The drop in EU exports
to Russia was around 80
per
cent steeper for sanctioned goods than for other products, while
exports of sanctioned goods to
CCA3 rose by extra 30 per
cent relative to those of other goods.
The increase in exports of sanctioned
goods to CCA3 represents a small fraction of the reduction in
their direct exports to Russia but can
be large for specific product groups.

Keywords:
sanctions, trade diversion, re
exports, intermediated trade, Russia
JEL
Classification Number: F14, F15, F51
The authors are grateful to
Ralph de Haas
for valuable comments and
suggestions and to
Joseph Sassoon
for the excellent research
assistance.
The working paper series has been produced to
stimulate debate on economic transition and
development. Views presented are those of the authors and not necessarily of the EBRD.
Working Paper No.
276
Prepared in
February 2023
Table5:
Exportsofsanctionedvssimilarvsothergoods:ChinaandTurkiye
Exporter
Turkiye
Dep.var.Exports
0-1HyperbolicLog
0-1Hyperbolic
ostxSanctionedxRussia
0.151***0.0257***0.370***0.0413-0.01000.0804
(0.0344)(0.00618)(0.0725)(0.0628)(0.00815)(0.0866)
PostxSanctionedxCCA3
0.146***0.0160***0.200***0.184***0.0001830.122**
(0.0350)(0.00450)(0.0538)(0.0521)(0.00596)(0.0574)
PostxSanctionedxGeorgia
0.0784*0.0178***0.212***-0.0129-0.0163**-0.0908
(0.0438)(0.00602)(0.0676)(0.0458)(0.00694)(0.0671)
Observations
1,125,6542,251,8722,251,872575,2481,689,0141,689,014
Exporter
Turkiye
Dep.var.Exports
0-1HyperbolicLog
0-1Hyperbolic
ostxSanctionedxRussia
0.181***0.0299***0.413***0.210***-0.0168*0.174*
(0.0422)(0.00775)(0.0897)(0.0777)(0.00961)(0.0984)
PostxSimilarxRussia
0.05730.008290.08290.354***-0.01510.202*
(0.0445)(0.00790)(0.0925)(0.0828)(0.0107)(0.111)
PostxSanctionedxCCA3
0.151***0.0193***0.223***0.228***-0.009650.0489
(0.0459)(0.00564)(0.0664)(0.0626)(0.00716)(0.0683)
PostxSimilarxCCA3
0.008270.006490.04400.0962-0.0215***-0.156**
(0.0482)(0.00614)(0.0728)(0.0750)(0.00767)(0.0739)
PostxSanctionedxGeorgia
0.119**0.0279***0.258***-0.00534-0.0284***-0.167**
(0.0533)(0.00730)(0.0795)(0.0551)(0.00823)(0.0788)
PostxSimilarxGeorgia
0.07500.0192**0.08830.0159-0.0268***-0.168*
(0.0573)(0.00777)(0.0848)(0.0611)(0.00921)(0.0906)
Observations
1,125,6542,251,8722,251,872575,2481,689,0141,689,014
ce
:Authors’calculations.
:StandarderrorsareclusteredattheHS6level.Allregressionsincludemonthximporter,monthxHS6andHS6ximporterfxedeects.*,**,***denote
statisticalsignifcanceatthe10%,5%and1%levels,respectively.Thedependentvariableisthelogarithmofbilateraltradeinagivenmonthinagivenproductgroup
attheHS6level;0-1variablefortradetakingplaceortheinversehyperbolicsinetransformationoftrade.ImportereconomiescompriseArmenia,Azerbaijan,Georgia,
Kazakhstan,KyrgyzR.,Mongolia,Russia,Turkiyeandtherestoftheworld(aggregated).SanctionedreferstoHS6productlineswhereEUsanctionsapplyatleast
partially.SimilargoodsarethosenotsanctionedbutwithinthesameHS4assanctioned.Incolumns1-6similargoodsarepartof”other”group.