Be nini, Fede ico; Juá ez, Le icia; Ga cía-Lembe gman, Ezequiel
Wo king Pape
The consequences o non- a i ade ba ie s: Theo y and
e idence om impo licenses in A gen ina
IDB Wo king Pape Se ies, No. IDB-WP-1629
P o ided in Coope a ion wi h:
In e -Ame ican De elopmen Bank (IDB), Washing on, DC
Sugges ed Ci a ion: Be nini, Fede ico; Juá ez, Le icia; Ga cía-Lembe gman, Ezequiel (2024) : The
consequences o non- a i ade ba ie s: Theo y and e idence om impo licenses in A gen ina,
IDB Wo king Pape Se ies, No. IDB-WP-1629, In e -Ame ican De elopmen Bank (IDB), Washing on,
DC,
h ps://doi.o g/10.18235/0013271
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/315906
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h ps://c ea i ecommons.o g/licenses/by/3.0/igo/
The Consequences o Non- a i T ade
Ba ie s:
Theo y and E idence om Impo Licenses in A gen ina
Fede ico Be nini
Le icia Juá ez
Ezequiel Ga cía-Lembe gman
WORKING PAPER No IDB-WP-1629
In e -
A
me ican De elopmen Bank
Depa men o Resea ch and Chie Economis
No embe 2024
* Uni e sidad de San And és
** In e -Ame ican De elopmen Bank
*** Pon i icia Uni e sidad Ca ólica de Chile
The Consequences o Non- a i T ade
Ba ie s:
Theo y and E idence om Impo Licenses in A gen ina
Fede ico Be nini*
Le icia Juá ez**
Ezequiel Ga cía-Lembe gman***
In e -
A
me ican De elopmen Bank
Depa men o Resea ch and Chie Economis
No embe 2024
Ca aloging-in-Publica ion da a p o ided by he
In e -Ame ican De elopmen Bank
Felipe He e a Lib a y
Be nini, Fede ico.
The consequences o non- a i ade ba ie s: heo y and e idence om
impo licenses in A gen ina / Fede ico Be nini, Le icia Jua ez, Ezequiel Ga cía-
Lembe gman.
p. cm. — (IDB Wo king Pape Se ies ; 1629)
Includes bibliog aphical e e ences.
1. In e na ional ade-A gen ina. 2. Ta i -A gen ina. 3. Impo quo as-
A gen ina. 4. Non- a i ade ba ie s-A gen ina. I. Juá ez, Le icia. II. Ga cía-
Lembe gman, Ezequiel. III. In e -Ame ican De elopmen Bank. Depa men
o Resea ch and Chie Economis . IV. Ti le. V. Se ies.
IDB-WP-1629
h p://www.iadb.o g
Copy igh © 2024 In e -Ame ican De elopmen Bank ("IDB"). This wo k is subjec o a C ea i e
Commons license CC BY 3.0 IGO (h ps://c ea i ecommons.o g/licenses/by/3.0/igo/legalcode). The
e ms and condi ions indica ed in he URL link mus be me and he espec i e ecogni ion mus be
g an ed o he IDB.
Fu he o sec ion 8 o he abo e license, any media ion ela ing o dispu es a ising unde such license
shall be conduc ed in acco dance wi h he WIPO Media ion Rules. Any dispu e ela ed o he use o
he wo ks o he IDB ha canno be se led amicably shall be submi ed o a bi a ion pu suan o he
Uni ed Na ions Commission on In e na ional T ade Law (UNCITRAL) ules. The use o he IDB's name
o any pu pose o he han o a ibu ion, and he use o IDB's logo shall be subjec o a sepa a e
w i en license ag eemen be ween he IDB and he use and is no au ho ized as pa o his license.
No e ha he URL link includes e ms and condi ions ha a e an in eg al pa o his license.
The opinions exp essed in his wo k a e hose o he au ho s and do no necessa ily e lec he iews o
he In e -Ame ican De elopmen Bank, i s Boa d o Di ec o s, o he coun ies hey ep esen .
Abs ac *
As WTO egula ions limi ed a i s, non- a i ba ie s, such as impo licenses (NAILs), be-
came essen ial ade policy ools. This pape examines how NAILs impac downs eam i ms in
A gen ina. Using a no el da ase and he s agge ed in oduc ion o NAILs be ween 2005-2011
o iden i ica ion, we analyze hei causal e ec s on i ms’ impo s and he subsequen e ec on
expo s and employmen . Resul s indica e ha NAILs educe i ms’ impo s, inducing mo e ex-
posed i ms o educe expo s and employmen . A ade model wi h oligopolis ic compe i ion
sugges s ha i ms’ ma ke powe can mode a e he impac o NAILs in highly concen a ed
ma ke s.
JEL classi ica ions: D43, F13, F14, F42, F68, L1
Keywo ds: Non- ade ba ie s, T ade policy, Ma ke powe
∗We hank Facundo Albo noz, Fe nando Al a ez, Benjamin Fabe , Ch is ophe Campos, Ma cela Esla a, Thibaul
Fally, Ana Fe nandes, Te esa Fo , Cecile Gaube , Al a o Ga cia-Ma in, Ma co Gonzalez-Na a o, Daniel Haanwinckel,
Juan Ca los Hallak, Luciana Ju enal, Da id Kohn, Ke em Kosa , Nuno Limao, Isabela Manelici, Ma hieu Pedemon e,
Alejand o Riaño, And es Rod iguez-Cla e, Da io To a olo, Mau icio Ula e, Jose Vazquez, Roman Za a e. Fi s e sion:
Ma ch 2018
Be nini Uni e sidad de San And és, Buenos Ai es, A gen ina. Email: [email p o ec ed]
Ga cía-Lembe gman ns i u o de Economia. Pon i ia Uni e sidad Ca olica de Chile. Email:
eglembe gman@be keley.edu
Iua ez: n e -Ame ican De elopmen Bank. Email: [email p o ec ed]
1 In oduc ion
Since he es ablishmen o he Wo ld T ade O ganiza ion (WTO) du ing he U uguay Round in
1994, coun ies ha e collec i ely commi ed o lowe impo a i s. Acco ding o he Wo ld Bank,
his ini ia i e has led o a s eep decline in global a e age a i s, which ell om 8.6% in 1994 o jus
2.6% by 2017.1As WTO egula ions ende ed a i s less iable, ade policy landscape expe ienced
a signi ican ans o ma ion. Non- a i ba ie s (NTBs) o impo s ha e su ged and p oli e a ed,
becoming a cen al ins umen in coun ies’ ade policy (Beghin e al. 2015). The e o e, unde -
s anding hei e ec s is c ucial, especially since escala ing geopoli ical ensions and o he global
challenges ha e b ough ade policy back in o he spo ligh in ecen yea s.
Analyzing he consequences o NTBs has been challenging. NTBs a e di icul o quan i y, and he
lack o exogenous a ia ion u he hampe s esea che s’ abili y o assess hei causal e ec s. As
a esul , we know li le abou NTBs and hei consequences. A sys em o non-au oma ic impo
licenses (NAILs) imposed by A gen ina o e s a unique se ing o o e coming hese challenges
and analyze he e ec s o non- a i ade ba ie s.
This pape in es iga es he impac o Non-Au oma ic Impo Licenses (NAILs), a ype o non- a i
ba ie , on he expo and employmen dynamics o downs eam A gen inian i ms ha ely on
impo ed inpu s. Using comp ehensi e i m-le el da a, we cons uc a no el da ase ca ego izing
p oduc s a ec ed by NAILs annually om 2005 o 2011. By employing an e en -s udy design,
we p o ide causal es ima es o he e ec s o NAILs. We in eg a e hese indings in o a model
o impo e s and expo e s ha inco po a es oligopolis ic compe i ion in expo ma ke s. This
analysis quan i ies he ole o NAILs on in e media e inpu s in shaping i m beha io , while also
highligh ing how i m ma ke powe and o e all ma ke concen a ion can media e he e ec s o
hese ba ie s.
Be ween 2005 and 2011, he A gen ine go e nmen olled ou a sys em o Non-Au oma ic Im-
po Licenses (NAILs), which equi ed ha ce ain p oduc s ob ain app o al om a public o icial
be o e being impo ed – a p ocess ha could delay app o al by up o wo mon hs and could be e-
used by he o icial. In p ac ice, NAILs ope a ed as a non- a i ade ba ie , aising i ms’ impo
cos s. This policy is ideal o analyze he impac o impo es ic ions on i m pe o mance o se -
e al easons. Fi s ly, he s akes we e high: by 2011, NAILs a ec ed almos 600 p oduc lines, which
accoun ed o 17% o i ms’ impo s o in e media e inpu s and a ec ed 37% o manu ac u ing
i ms, ma king his as one o he la ges non- a i ba ie s policy globally. Second, a unique aspec
o he policy was ha p oduc s we e phased in o he NAILs sys em a di e en pe iods, wi hou
any appa en sys ema ic app oach, culmina ing in including all p oduc s by 2012. The s agge ed
inclusion o p oduc s in he NAILs sys em p o ides an ideal empi ical amewo k allowing o
causal iden i ica ion o he e ec s o non- a i ba ie s on i m dynamics.2
We c ea e a no el da ase ha combines h ee da ase s spanning om 2003 o 2011. Fi s , we use
he uni e se o A gen ine expo e s and impo e s, de ailing hei expo and impo ansac ions
1We ake he a e age o e ec i ely applied a es weigh ed by he p oduc impo sha es co esponding o each
pa ne coun y.
2Ou analysis concludes in 2012 o wo p ima y easons. Ini ially, no all p oduc s we e a ec ed by he policy
be o e his da e, enabling us o u ilize he s agge ed inclusion o p oduc s as a means o iden i ica ion. Fu he mo e, in
No embe 2011, he A gen ine go e nmen imposed signi ican es ic ions on dolla pu chases, a policy change likely
o ha e in luenced impo s and expo s, he eby complica ing he iden i ica ion o non- a i ba ie e ec s a e 2012.
2
a he i m-p oduc -coun y (des ina ion/o igin) le el om o icial cus oms da a. Second, we in-
co po a e i m-le el employmen da a om A gen ina’s in e nal e enue se ices (AFIP). Finally,
we digi ized a se ies o go e nmen dec ees o cons uc a no el da ase ha sys ema ically docu-
men he annual imposi ion o NAILs on impo ed p oduc s a he 8-digi a i line.
Ou empi ical s a egy exploi s exogenous a ia ion in he iming o p oduc en ies in o he NAILs
sys em om 2005 o 2011, combined wi h da a on each i m’s impo sha e o a ec ed p oduc s
p io o policy implemen a ion. The unde lying idea is ha i ms impo ing in e media e inpu s
ha la e became subjec o NAILs aced g ea e exposu e o he policy, inc easing hei p oduc ion
cos s. We use his i m-le el exposu e as a shock o p oduc ion cos s, examining how downs eam
i ms espond in e ms o impo s, expo s, and employmen . This s udy p o ides he i s causal
e idence on how non- a i ba ie s on inpu s impac hese h ee key economic dimensions.
Ou i s inding is ha he exposu e o his policy signi ican ly a ec ed i ms’ impo ac i i ies.
Fi ms wi h 30% o hei impo s a ec ed by NAILs (a e age o all exposed i ms) educe hei o al
impo s by 46%. This pa e n p o es ha NAILs we e e ec i e as a non- a i impo ba ie , which
migh ha e signi ican ly a ec ed i ms’ p oduc ion cos s.
Once we ha e es ablished ha NAILs e ec i ely educe impo s, we in es iga e hei impac on
i ms’ expo s and employmen . Ou analysis p o ides he i s causal e idence ha non- a i ba -
ie s o impo s, ep esen ed by non-au oma ic impo licenses (NAILs), dec ease i m expo s and
employmen . Speci ically, an exposu e o 30% o i ms’ impo s o NAIL leads o a 18% educ ion
in expo s and a 3% educ ion in employmen . Fi ms mo e a ec ed by NAILs also educe numbe
o expo ed p oduc and des ina ions and inc ease he likelihood o exi expo ma ke s and lea e
ope a ions. In pa icula , expo s o di e en ia ed p oduc s and o OECD des ina ions a e he mos
a ec ed by his policy. In ligh o ou model, his indica es ha i ms ha used impo ed inpu s
a ec ed by he policy ace an inc ease in hei p oduc ion cos s, ende ing hem less compe i i e.
We hen explo e how i ms’ eac ions o NAILs di e ac oss expo ma ke s, depending on hei
ela i e impo ance in each ma ke as indica ed by hei ma ke sha e. In eg a ing hese ind-
ings wi h ou s uc u al model o e s new insigh s in o i ms’ ma ke powe in in e na ional ade
and how ma ke concen a ion can media e he o e all impac o ade policies. We ind ha he
nega i e e ec o NAILs on expo s is smalle in ma ke s whe e he i m is ela i ely la ge . To
s eng hen ou iden i ica ion s a egy and alida e ou indings on he e ogeneous esponses, we
compa e he eac ions o NAILs o mul i-des ina ion i ms ac oss hei a ious ma ke s. Consid-
e ing ha mo e han 95% o A gen ina’s o al expo s a e explained by i ms ha expo o many
ma ke s, unde s anding hei beha io is also ele an in o he con ex s. A med wi h he s uc-
u e o he model, we de elop a me hodology ha equi es a la ge exogenous cos shock o ensu e
enough a iabili y o including i m-yea ixed e ec s and being able o compa e esponses ac oss
di e en des ina ions. NAILs can p o ide such a shock. We ind ha a i m’s esponses in expo
ma ke s o i m-le el exposu e o NAILs a y by i s ma ke sha e in each des ina ion. A i m e-
duces less i s expo s and main ains p ices mo e s able in ma ke s whe e he i m’s ma ke sha e
is ela i ely highe . This implies ha e en he same i m esponds di e en ly in di e en ma ke s
depending on i s ma ke powe in each ma ke .
Explaining he na u e o he obse ed beha io is a he co e o his pape . The e o e, o guide
he empi ical analysis and quan i y he e ec s o NAILs, we de elop a model o expo ing and
3
impo ing ha inco po a es a iable ma kups. On he demand side, he amewo k inco po a es
a iable ma kups o a s anda d model o he e ogeneous i ms, closely ollowing he analysis in
A keson and Bu s ein (2008).3On he supply side, we assume ha i ms d aw co e p oduc i i y
and combine impo ed in e media e inpu s in a CES p oduc ion unc ion. We u he assume ha
inpu ma ke s a e pe ec ly compe i i e as i is s anda d in he impo ing li e a u e.
We demons a e ha ou empi ical esul s e eal new aspec s o ma ke s uc u e in in e na ional
ade. They a e consis en wi h a model o oligopolis ic compe i ion in expo ma ke s, cha ac-
e ized by a iable ma kups a he i m-by-des ina ion le el. In esponse o cos shocks induced
by non- a i ba ie s o impo s, expo e s s a egically adjus hei ma kups mo e signi ican ly in
ma ke s whe e hey ha e a la ge ma ke sha e. This s a egy allows expo e s o mi iga e some o
he shock’s impac by educing hei ma kups, he eby main aining mo e s able p ices and quan i-
ies in ma ke s wi h g ea e ma ke powe . This, in u n, ha e impica ion ega ding he agg ega e
e ec s o non- a i ba ie s o ade on expo s in a con ex o a iable ma kups.
Ou pape con ibu es o h ee s ands o he li e a u e. Fi s , ou pape ela es o he pape s ha
s udies he e ec s o ade policies (Albo noz e al. 2021,Ami i and Konings 2007,Ami i e al.
2019,Bas 2012,Cole and Eckel 2018,De Loecke e al. 2016,Fajgelbaum e al. 2020,Feng e al. 2017,
Flaaen e al. 2020,Flach and G ä 2020,Goldbe g e al. 2010,Romalis 2007). Ou pape is he i s
o analyze he causal impac o non- a i ade ba ie s ha es ic he quan i y o goods ha can
be impo ed, such as impo quo as, impo licenses o impo bans, on downs eam i ms expo s
and employmen . Ou pa icula ocus is on non-au oma ic impo licenses. 4
On his g ound, ou pape ela es o wo concu en pape s. A kin e al. (2024) analyzes he e -
ec o a simila policy o disc e iona y impo licenses in A gen ina 2013-2015 on impo p ices.
Ou s udy complemen s hei wo k in wo ways. Fi s , while A kin e al. (2024) ocuses solely on
he di ec e ec o impo licenses on impo p ices, his is he i s pape o demons a e ha an
impo an aspec o such policies is ha hey can also a ec downs eam i ms’ p oduc ion, em-
ploymen , and expo dynamics by inc easing i ms’ impo ed inpu cos s. Secondly, by u ilizing
he s agge ed implemen a ion o Non-Au oma ic Impo Licenses (NAILs) be ween 2005 and 2011,
and no ing ha no all p oduc s we e included in he sys em a he same ime, we can mo e accu-
a ely es ima e he causal impac s o impo licenses on i m-le el ou comes.5Mo e simila o ou
wo k, Ghose e al. (2023) s udy a ban o e ilize s impo s in S i Lanka. While hei pape ocuses
on a pa icula inpu and e ec s on he ag icul u al sec o , we analyze a la ge -scale ade policy
in ol ing mo e han 600 p oduc s and di ec ly a ec ing a hi d o i ms in he manu ac u ing sec-
o . We also ex end he analysis o he e ec on he labo ma ke s. While some o he li e a u e
has ocused on he e ec o ade policies on labo ma ke s (Au o e al. 2013,Caliendo e al. 2019,
Dix-Ca nei o 2014,Gu ko a e al. 2023), o ou knowledge, we a e he i s o s udy he impac o
non- a i ade ba ie s ( h ough hei e ec on impo s) on employmen .
Ou pape is also ela ed o Fon agné e al. (2015) and Fon agné and O e ice (2018), who examines
how echnical ba ie s o ade imposed by he des ina ion coun y a ec i ms’ expo s o ha
3The main conclusions ega ding a iable ma kups hold in a wide class o models o ade ha ha e been used in e-
cen pape s. Howe e , he di ec ion o he elas ici y o ma kups wi h espec o he i m’s ma ke sha e is model-speci ic.
See, o ins ance, A kolakis and Mo lacco (2017) o a e iew o di e en ways o inco po a ing a iable ma kups.
4Nici a and Gou don (2013) shows ha non-au oma ic licenses a e he mos used measu e o con ol impo quan i-
ies and hey a e specially implemen ed in de eloping coun ies.
5Pos -2012, all p oduc s became subjec o impo licenses.
4
des ina ion. In con as , we ocus on he e ec s o non- a i ba ie s o i ms based in he imple-
men ing coun y. The e o e, while hei mechanism is ela ed o condi ions a he des ina ion, ou
mechanism is ela ed o how non- a i ba ie s aise he cos s o impo ed inpu s a he o igin,
impac ing i ms’ p oduc ion cos s and expo ing po en ial.
Second, ou pape is also ela ed o he li e a u e ha s udies he di e en ma gins o adjus men
o i ms o ade policy, iewed as a cos shock (De Loecke e al. 2016). We documen a p e i-
ously unexplo ed dimension o i m he e ogenei y. We highligh he impo ance o he elas ici y
o ma kups o a gi en i m, ac oss i s expo des ina ions. P e ious pape s ha e documen ed ha
i ms cha ges di e en p ices ac oss des ina ions (Mano a and Zhang (2012)). Howe e , hese pa-
pe s ha e no analyzed how hese p ices espond o shocks speci ic o he i m. We show ha i ms
adjus no only p oduc scope and o al expo olumes, bu also hei ma kups ac oss des ina ions.
In making decisions, mul i-des ina ion i ms op imally decide o adjus mo e hei ma kups o cos
shocks in ma ke s whe e hey ha e highe ma ke sha es. As mos o he ade lows a e concen-
a ed in a ew i ms ha expo o many ma ke s, his ma gin o adjus men could po en ially
be impo an o es ima e wel a e gains om ade. In addi ion, his may a ec he dis ibu ion o
gains om unila e al ade libe aliza ion in o eign coun ies.
Thi d, ou pape con ibu es o a g owing li e a u e ha s udies he e ogeneous esponses o i ms
o shocks bu in he con ex o exchange a e mo emen s and incomple e exchange a e pass-
h ough. 6Mo e simila o ou s is Ami i e al. (2015), which decomposes he exchange a e pass-
h ough in o he ole o i ms ma ginal cos s, impo in ensi y, and ma ke powe o a i m in a
gi en ma ke and do analyze adjus men s o i ms depending on hei ma ke sha e. We inno a e
by exploi ing an impo cos s shock (supply shock) ha le us iden i y he ma kup elas ici y and
how i depends on ma ke sha e o he i m in di e en ma ke s while holding cons an demand
shocks. By compa ing he same i m ac oss des ina ions, ou es ima e can be in e p e ed as a mo e
accu a e es ima e o he supe -elas ici y o ma kup. Mo e b oadly, by ex ending ou esul s o he
ma ke le el, we con ibu e o he g owing li e a u e on ma ke concen a ion and pass- h ough o
shocks (Ami i and Heise (2024), Bu s ein e al. (2020), Jua ez (2024), Rubens (2023)).
The emainde o he pape is o ganized as ollows. Sec ion 1 in oduces he da a and highligh s
key pa e ns ha in o m ou heo e ical and empi ical app oach. Sec ion 2 p o ides a de ailed
desc ip ion o he da a and his o ical con ex . In Sec ion 3, we in oduce he heo e ical model.
Sec ion 4 ou lines he empi ical s a egy, discusses he policy we exploi , and explains ou iden-
i ica ion assump ions. Sec ion 5 p esen s he main esul s a he i m-le el. Sec ion 6 examines
he ole o ma ke powe and ma ke concen a ion in shaping he e ec s o he policy. Sec ion 7
concludes.
6Fo ins ance, Be man e al. (2012) ind ha highe pe o mance i ms end o abso b exchange a e mo emen s
in hei ma kups so ha hei a e age p ices in he o eign ma ke a e less sensi i e. Ami i e al. (2016) also show he
exis ence o a iable ma kups in he domes ic ma ke and analyze he ole o s a egic complemen a i y. Howe e , hese
pape s do no analyze di e en ial esponses in o eign ma ke s and don’ ake a s and on whe he a i m adjus men
depends on cha ac e is ics speci ic o he i m-des ina ion.
5
3.2 Impo Decision and uni cos s
We conside a s anda d amewo k o impo beha io whe e i ms’ impo decisions a e he so-
lu ion o a maximiza ion p oblem. The impo beha io o he i m, along wi h i s p oduc i i y
d aw, de e mines i s uni cos s. Since o eign supplie s can be mo e e icien a p oducing some o
he in e media e a ie ies, i ms may be willing o demand impo ed inpu s o educe he uni cos
o p oduc ion. A measu e No inal-good p oduce s each p oduce a single di e en ia ed p oduc .
Fi ms a e cha ac e ized by a he e ogeneous a ibu e ϕ ha is in e p e ed as co e p oduc i i y. In
he same way as in Meli z (2003), his pa ame e is exogenously d awn om a p obabili y dis i-
bu ion g(ϕ)and e ealed o he i ms once hey s a o p oduce. The p oduc ion unc ion akes
he ollowing CES o m:
Q=q(z) = ϕ"∑
(z )θ−1
θ#(θ/θ−1)
whe e z deno es he amoun o impo s o p oduc a ie y (i em psou ced om ma ke j) and
θ>1 is he elas ici y o subs i u ion o inpu s. Fo he momen , we will no ocus on he sou ce
ma ke . Le ’s assume he e is only one ma ke om which he i m can sou ce inpu s. Hence,
=p oduc om ha ma ke .20 Impo ing a ie y in ol es a ixed cos (κm), which, in his
sec ion, we assume is common ac oss i ms and sou ces. We u he assume ha i ms ake inpu
p ices, adjus ed by quali y, as gi en. They a e de e mined by cha ac e is ics speci ic o he o igin-
p oduc , A (i.e, quali y, echnology, and wages in coun y j o p oducing p oduc p), and bila e al
ade cos s speci ic o he i m- a ie y (τi ):
P =τi
A
3.3 Fi m Impo Beha io in Equilib ium
In his subsec ion, we b ie ly analyze he i m’s beha io in equilib ium. We de ine a sou cing
s a egy Ωas he se o inpu a ie ies , so he i m impo s posi i e amoun s o hese a ie ies.
Fi s , we will ocus on he i ms’ decisions, condi ional on he sou cing s a egy Ω.
3.3.1 Op imal amoun o impo s condi ional on sou cing s a egy
To ob ain he numbe o impo s o a a ie y , he i m minimizes i s cos unc ion, which is subjec
o i s p oduc ion unc ion.
The op imal quan i ies o a ie y a e gi en by,
z∗
(ϕ,Ω,Q)≡a g min
z
∑
∈Ω
p z s. Q=ϕ"∑
∈Ω
(z )θ−1
θ#(θ/θ−1)
. (3.2)
20 This leads o he same p edic ion as An as e al. (2017), whe e he gains om a ie y come om he p oduc i i y
d aws o o eigne s, which ollow a F éche dis ibu ion unc ion simila o ha p oposed by Ea on and Ko um.
12
A e sol ing, we ge he ollowing exp ession,
z (ϕ,Ω,Q) = Q
ϕ1
p θ
"∑
( )∈Ω1
p θ−1#θ/θ−1∀ ∈Ω, (3.3)
which co esponds o he ollowing impo s alue,
p z (ϕ,Ω,Q) = Q
ϕ1
p θ−1
∑
∈Ω1
p θ−1θ/θ−1∀ ∈Ω, (3.4)
A e sol ing o he in ensi e ma gin o impo s o any a ie y co esponding o he i m sou cing
s a egy (Equa ion 3.4), ob aining he minimum uni cos unc ion o a gi en s a egy is s aigh -
o wa d;
ci=h(Ω)
ϕ=1
ϕ"∑
∈Ω1
p θ−1#−1
θ−1
=1
ϕ"∑
∈ΩA
τi θ−1#−1
θ−1
=1
ϕ[Φi]−1
θ−1, (3.5)
whe e h(Ω)is he pa o he uni cos gi en by inpu s. We de ine he sou cing capabili y o a i m
as,
Φi="∑
∈ΩA
τi θ−1#.
The e o e, he o al amoun o impo s o in e media e goods o i m iis gi en by,
Mi(Ω) = Qi
ϕ"∑
∈ΩA
τi θ−1#−1
θ−1
, (3.6)
and he expendi u e sha e o i m ion impo ed a ie y is gi en by,:
mi (Ω) = A
τi θ−1
∑
∈ΩA
τi θ−1∀ ∈Ω;
mi (Ω) = 0∀ 6∈ Ω
By Shepa d’s Lemma:
∂logci
∂logτi
=mi (3.7)
13
No e ha he model p edic s ha he ba ie o impo has a highe impac on cos s 21, he la ge
he sha e o he i m’s expendi u e on he inpu a ec ed by he ba ie . In ou empi ical sec ion,
we use his o cons uc ou i m-le el shock.
3.4 P ice se ing
Gi en a sou cing s a egy, wi h i s co esponding uni cos ci(Ω,ϕ), sol ing o op imal p ice in
ma ke kis s anda d:
Pik =σik
σik −1ci(Ω,ϕ)(3.8)
PROPOSITION 2. Holding cons an he sec o al p ice Pk, he elas ici y o p ice wi h espec o a a i o
inpu o i m i is gi en by,
dlog Pik
dlog τi
=1
1+Γik
mi
Wi h Γik −Mik
logPik ep esen ing he nega i e o he elas ici y o ma kup wi h espec o p ices. Recall ha Γik
is inc easing in ma ke sha e o he i m in des ina ion k.
P oo . See p oo in Appendix C.3.
No e ha he model p edic s ha p ices (and he e o e expo s) will eac less in ma ke s whe e he
i m is ela i ely la ge . In sec ion 6we come back o his esul o de i e p edic ions abou how
ma ke powe and a iable ma kups media e he e ec s o non- a i ade ba ie s.
We hold cons an Pk, as we do so h oughou he empi ical sec ion by including sec o -yea FE in
e e y speci ica ion. I he ma kup is cons an , hen he e ec o a a i on an in e media e inpu
on p ice is equi alen o he ini ial sha e o he inpu ha he i m was using mi . In con as , wi h
a iable ma kups, we expec ha he impac is lowe o la ge i ms ha ha e a highe Γ. This
will be a key ea u e o explain he di e en ial e ec s o (lack o ) access o in e media e inpu s on
expo s depending on he ela i e posi ion o he i m in he ma ke .
3.5 Re enues in equilib ium
Re enues o i m iin ma ke ka e gi en by:
Rik =1
Mρ−1
ik
ϕρ−1
hρ−1
i
Pρ−η
kDk, (3.9)
and o al e enues o a i m a e gi en by, 22
21In wha ollows, we omi he a gumen Ω, as we will no de i e conclusions on he ex ensi e ma gin o impo s.
22No e ha when we ex end he model o allow o en y and exi in o impo and expo , lowe cos s h ough highe
inpu s may impac esul s.
14
Ri=ϕρ−1
hρ−1
i
∑
k
1
Mρ−1
ik
Pρ−η
kDk, (3.10)
3.6 P edic ions
The model gene a es wo se s o p edic ions ha will guide ou empi ical sec ion. The i s se o e-
sul s is i m-des ina ion speci ic. We es ablish he di ec e ec o inc eased ade ba ie s o a gi en
inpu on he i m’s expo s in each ma ke k. This p oposi ion p edic s he expec ed esponses o a
mul i-des ina ion i m in i s di e en ma ke s, depending on a iable ma kups and cha ac e is ics
o he i m-des ina ion. The second se o esul s a e a he i m le el. These p edic ions show
how ade ba ie s a ec o al expo e enues and o al impo s and guide he es ima ion o he
elas ici y o expo s o impo s a he i m le el.
We i s analyze he e ec s a he i m le el.
PROPOSITION 3(Fi m le el p edic ions).
A. (E ec on o al expo s) The e ec on o al expo s is nega i e and dec easing in he size o he i m.
∂log Ri
∂log τi
= (1−ρ)∑
k
Rik
Ri1
1+Γik
mi <0 (3.11)
B. (E ec on o al impo s) P o ided ρ>1, impo s a e weakly dec easing in he ade cos s o impo -
ing a ie y (τi ). In addi ion, he nega i e e ec is s onge , he highe he sha e o i m’s impo s
co esponding o :
∂log Mi
∂log τi
=−mi "ρ∑
k
Qik
Qk
1
1+Γik
−1#≤0 (3.12)
∂log Mi
∂(log τi ∂mi )=−"ρ∑
k
Qik
Qk
1
1+Γik
−1#≤0 (3.13)
C. (Elas ici y o expo s wi h espec o impo s) The o al amoun o expo s o a i m a e inc easing
on he amoun o impo s o he i m. Tha is,
EXM =
∂log Ri
∂log τi
∂log Mi
log τi
=∂log Ri
∂log Mi
=(1−ρ)∑kRik
Rih1
1+Γik i
1−ρh∑kQik
Qk
1
1+Γik i>0 (3.14)
P oo . See p oo in Appendix C.4.
We hen es ablish he e ec o impo cos shocks on expo e enues in a gi en ma ke k.
PROPOSITION 4(Fi m-des ina ion esponses).
A. P o ided ρ>1, e enues in ma ke k a e weakly dec easing in he cos s o impo ing a ie y (τi ). In
15
addi ion, he e ec is la ge (mo e nega i e), he highe is mi :
∂log Rik
∂log τi
= (1−ρ)1
1+Γik
mi ≤0 (3.15)
∂log Rik
∂log τi ∂mi
= (1−ρ)1
1+Γik ≤0 (3.16)
B. The e ec o inc easing impo cos s on expo s o ma ke k is weakly dec easing in he elas ici y o
ma kup Γik (i is s ic ly dec easing i ma kups a e no cons an ):
∂log Rik
∂(log τi ∂mi )∂Γik
≥0 (3.17)
C. P o ided §=∂log Γik
∂log Sik
>0, hen he absolu e alue o he elas ici y o expo s o ma ke k wi h espec o
impo cos s is weakly dec easing on he size o he i m Sik. I is dec easing i ma kups a e no cons an :
∂log Rik
∂(log τi ∂mi )∂Sik
≥0 (3.18)
P oo . P oo s a e s aigh - o wa d om he inspec ion o equa ions abo e. See appendix.
In he nex sec ions, we explo e he p edic ions o he p oposi ions o he model. In Sec ion 4
we examine p edic ions o P oposi ion 6which es ablishes esul s a he i m-le el. In sec ion 6,
we hen e alua e empi ically he p edic ions o P oposi ion 4 ha a e ela ed o he di e en ial
esponses o i ms ac oss ma ke s, depending on hei ela i e size and ma ke powe .
4 Empi ical S a egy
In his sec ion, we pu oge he he model in ui ions wi h a supply shock o impo cos s o speci ic
p oduc s (i.e.: τi ), combined wi h in o ma ion on he sha e o impo s o he p oduc s o a i m
mi . On his g ound, we exploi exogenous a iabili y in impo cos s due o he A gen ine go e n-
men ’s imposi ion o non- a i ba ie s on speci ic p oduc s be ween 2003 and 2011. On his basis,
we exploi exogenous a ia ion in impo cos s due o he A gen ine go e nmen ’s imposi ion o
non- a i ba ie s on speci ic p oduc s be ween 2005 and 2011. Fi s , we explain he me hodology
used o compu e i ms’ exposu e o he policy. Nex , we demons a e ha he policy was e ec i e
in educing impo s. Finally, we show ha he policy also impac ed i ms’ expo s, and ha he
p e- ea men pa allel ends assump ion holds.
4.1 Me hodology
We use he NAILs o cons uc a cos shock o a i m. In pa icula , o cons uc a ime- a ying
i m-le el a iable ha p oxies a i m’s exposu e o impo ba ie s, we p oceed as ollows: we
use he impo baske o he i m in he pe iod 2003-2007 (be o e he la ge inc ease in he p oduc s
included in his policy) and calcula e he sha e o he i m’s expendi u e on impo ed inpu s ha
co esponds o each p oduc (mi ). Then, holding his sha e cons an o e ime, we mul iply i
16
by an indica o ha akes a alue o 1 in hose yea s when he p oduc is a ec ed by he NAILs.
Then, we sum ac oss p oduc s o a gi en i m. Fo mally, we de ine a i m’s exposu e o NAILs in
ime as,
NAILexposu ei =∑
mi NAIL , (4.1)
whe e mi ep esen s he sha e o expendi u e on impo ed inpu in he pe iod 2003-2005 and
NAIL is an indica o ha akes alue 1 i he p oduc is a ec ed by NAILs in pe iod .
In ui i ely, guided by P oposi ion 6.B., we assume ha a i m is mo e exposed o he impo shock,
he highe he ini ial sha e o expendi u e ha co esponded o he a ec ed p oduc in he pe iod
be o e he policy ook place.
4.2 Rele ance o he policy and iden i ying assump ion
4.2.1 E ec i eness o he NAILs in educing impo s
Be o e mo ing o he pape ’s main esul s, we i s explo e whe he he NAILs e ec i ely educed
impo s o i ems ha we e added o he lis . To do so, we use agg ega e da a o pe o m an e en
s udy a he p oduc le el o analyze i being added o NAILs, educes impo s o an i em a he
8-digi s a i le el. Fo mally,
log(Impo s ) =
12
∑
j=−27
βj1[Qua e sSinceNAILs =j] + α +γ +u , (4.2)
whe e he nega i e alues co espond o qua e s be o e p oduc en e ed he NAILs lis . We
ocus on pa ame e β ha ep esen s he impac o he inco po a ion o NAIL on p oduc s’ impo s.
Figu e 3plo s he coe icien s β.23 We do no obse e sys ema ic di e ences in he yea s be o e he
p oduc was added o he NAIL sys em. As expec ed, he NAILs wo k as an impo an ba ie o
ade, especially since he second qua e a e he p oduc was included in he policy.24 We ind
ha impo s o a p oduc ha is added o he NAILs lis decline by 50% he i s yea ela i e o i s
coun e ac ual.
4.2.2 Iden i ica ion assump ion
A e showing ha impo s o p oduc s added o he NAILs sys em decline, we u n o es ou
iden i ica ion assump ion. Ou main iden i ica ion assump ion is ha he iming in which a p od-
uc en e s he NAILs sys em is no co ela ed wi h changes in he i m’s expo decisions. In o he
wo ds, he e olu ion o expo s in i ms ha we e mo e exposed o NAILs would ha e been simila
o he e olu ion o expo s o i ms less exposed in he absence o he policy. Fo example, i could
be he case ha he go e nmen a ge ed p oduc s used by i ms ha we e p edic ed o expe ience
a decline in expo s.
23We es ic he sample o hose p oduc s ha en e ed a some poin in o he NAILs sys em.
24In he i s mon hs, impo e s could use p e iously app o ed au oma ic licensing o impo s, so NAILs migh e-
qui e some mon hs o e ec i ely a ec impo s.
17
Figu e 3: E en s udy. The impac o Non Au oma ic Impo License on i ms’ impo s (logs).
-2.5 -2 -1.5 -1 -.5 0
Log Impo s
-12 -11 -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8
Qua e s since p oduc equi es non-au oma ic impo license
95% con idence in e al
No es: The igu e shows he e ec s on he log o impo alues up o 8 qua e s a e he imposi ion o he non-au oma ic impo
license and he p e- end om 12 qua e s be o e be ween 2002Q1 and 2010Q4. S anda d e o s a e clus e ed a he 8-digi a i le el.
In o de o indi ec ly es his assump ion, we design an e en s udy. We de ine as an e en a =0
when he yea o which a leas one p oduc o he i m was a ec ed. We hen g aph he e en
s udy o he di e ences in log (expo s) be ween hese g oups. Fo mally, we un he ollowing
equa ion,
log(expo si ) =
2
∑
j=−4
βj1[Yea sSinceExposu eToNAILsi =j] + αi+γs +ui . (4.3)
Figu e 4plo s he coe icien s βjo his eg ession. Reassu ing, we do no obse e any sys ema ic
di e ences in he i ms’ expo s in he yea s be o e he i m became a ec ed by NAILs. This is
sugges i e e idence ha he pa allel end assump ion may hold in ou con ex . In addi ion, he
Figu e p o ides a i s glance a he esul s ha we will show in he nex sec ion: he alue o
expo s is signi ican ly educed a e he i m is exposed o NAILs.
18
Figu e 4: E en s udy. The impac o Nonau oma ic Impo licenses on i ms’ expo s (logs).
-.4 -.2 0 .2
Log Expo s
-4 -3 -2 -1 0 1 2
Yea s Since Fi m a ec ed by i s NAIL
95% con idence in e al
No es: The igu e shows he e ec s on he expo alues o i ms ha we e exposed o non-au oma ic licenses up o 2 yea s a e he
exposi ion and he p e- end om 4 yea s be o e. A i m is classi ied as exposed i a leas one o i s p oduc s impo ed du ing 2003-
2007 was a ec ed by non-au oma ic impo licenses. S anda d e o s a e clus e ed a he i m le el. The eg ession includes ixed
e ec a he i m le el and sec o -yea le el.
5 Resul s
In his sec ion we p esen he main esul s o he pape . Fi s , we documen he e ec o he policy
on impo s and expo s a he i m le el. We iden i y he di ec e ec o NAILs on expo s a he
in ensi e and ex ensi e ma gin, we es ima e he elas ici y o o al expo s and he he e ogeneous
e ec s o NAILs ac oss p oduc and ma ke di e en ia ion. We also show he impac on employ-
men . Then, in Sec ion 6we use he model’s p edic ions o es ima e whe he i is inc easing on a
i m’s ela i e size in he ma ke .
5.1 The e ec o NAILs on impo s and expo s
5.1.1 A e age E ec o NAILs on impo s and expo s
We begin by es ima ing he e ec o NAILs exposu e on i ms’ expo s. Acco ding o ou model,
in oducing impo ba ie s o in e media e inpu s inc eases he ma ginal cos o i ms exposed
o his ba ie and educes hei compe i i eness in o eign ma ke s. To quan i a i ely es his, we
un he ollowing equa ion
YX
is =βNAILexposu eis +γi+γ +γs +µi , (5.1)
whe e YX
is is a se o ou comes measu ing in ensi e and ex ensi e ma gin o expo s, such as log ex-
po s, expo s a us, numbe o p oduc s and numbe o des ina ions. Resul s om he es ima ion
19
o equa ion 5.2 a e epo ed in Table 2.
Table 2: Reduced o m. The e ec o NAILs exposu e on i m’s o al expo s
(1) (2) (3) (4)
log(expo s)i Expo s a usi #P oduc s #Des ina ions
NAILexposu ei -0.3494∗∗∗ -0.0282∗∗∗ -0.3115∗∗ -0.1690∗∗∗
(0.1058) (0.0094) (0.1350) (0.0367)
Obse a ions 162,981 162,981 162,981 162,981
R-squa ed 0.85 0.80 0.93 0.95
Mean dep a iable 4.66 0.38 3.00 1.67
Fi m FE Yes Yes Yes Yes
Sec o -Yea FE Yes Yes Yes Yes
No es: Clus e ed s anda d e o a he i m le el in pa en hesis. *** p<0.01, ** p<0.05, * p<0.1 NAILexposu ei ep e-
sen s he sha e o i ms’ impo s o he pe iod 2003-2005 a ec ed by NAIL in yea Column (1) ou come use he in e se
hype bolic sine ans o ma ion o accoun expo s o all i ms. Column (2) ou come is a dummy a iable ha akes
alues 1 i i ms i expo a yea and 0 o he wise. Columns (3) and (4) ou comes indica e he i ms’ numbe o expo ed
p oduc s and des ina ions.
Exposu e o NAILs signi ican ly educed bo h he in ensi e and ex ensi e ma gins o expo s.
Fi ms wi h 10% exposu e o NAILs expe ienced a 3.49% educ ion in expo olumes compa ed o
una ec ed i ms. The policy also a ec ed he ex ensi e ma gin o expo s. On a e age, i ms wi h
10% exposu e o NAILs saw a educ ion o 2,8 pe cen age poin s (-7.4% ela i e o he uncondi-
ional mean) on he p obabili y o con inuing expo ing.In addi ion, hey dec ease he numbe o
expo ed p oduc s by 0.31 (-33%) and he numbe o expo des ina ions eached by 0.169 (-18%).
Once we ha e demons a ed he educed o m e ec s, we p oceed o he IV es ima ion o he
elas ici y o expo s wi h espec o impo s a he i m le el. As P oposi ion 6.C indica es, his
elas ici y is gi en by:
EXM =
∂log Ri
∂(log τi mi )
∂log Mi
∂(log τi mi )
.
No e ha his is equi alen o di iding he coe icien o he e ec o NAIL exposu e on expo s
( educed o m) by he coe icien om a eg ession o NAIL exposu e on impo s ( i s s age). Thus,
i is equi alen o unning an IV eg ession o impo s on expo s, using NAILs exposu e as he
ins umen o impo s.
Resul s a e epo ed in Table 3. In he second panel we epo he i s s age coe icien , which is
−1.54. Namely, a i m o which 10% o hei inpu s a e a ec ed by he NAILs educes hei o al
impo s by 15%. The i s s age F-s a is ic is 154. In he i s panel we epo he coe icien o he
elas ici y o expo s wi h espec o impo s. We ind ha he elas ici y is 0.23. An inc ease in 10%
o impo s o in e media e inpu s inc eases expo alues by 2.2%. In addi ion, access o impo s
also ha e conside ably e ec s on he ex ensi e ma gin o expo s, as e lec ed by an inc ease in
expo s a us, numbe o p oduc s and numbe o des ina ions.
We ha e shown ha exposu e o non- a i ade ba ie s educed i ms impo capabili ies, which
in u n a ec ed expo s o i ms ha used in ensi ely inpu s a ec ed by he policy. In he nex
sec ion, we in es iga e he e ogeneous e ec s depending on he ype o des ina ions and p oduc s
20
Table 3: Elas ici y o expo s wi h espec o impo s a he i m le el
(1) (2) (3) (4)
log(expo s)i Expo s a usi #P oduc s #Des ina ions
log(impo s)i 0.2266∗∗∗ 0.0183∗∗∗ 0.2020∗∗ 0.1096∗∗∗
(0.0671) (0.0060) (0.0873) (0.0241)
Obse a ions 162,981 162,981 162,981 162,981
Fi m FE Yes Yes Yes Yes
Sec o -Yea FE Yes Yes Yes Yes
Fi s S age
NAILexposu ei -1.5420∗∗∗ -1.5420∗∗∗ -1.5420∗∗∗ -1.5420∗∗∗
(0.1242) (0.1242) (0.1242) (0.1242)
F 154.07 154.07 154.07 154.07
Mean dep a iable 5.09 5.09 5.09 5.09
No es: Clus e ed s anda d e o a i m le el in pa en hesis. *** p<0.01, ** p<0.05, * p<0.1 NAILexposu ei
ep esen s he sha e o i ms’ impo s o he pe iod 2003-2005 a ec ed by NAIL in yea . Column (1) ou -
come use he in e se hype bolic sine ans o ma ion o accoun expo s o all i ms. Column (2) ou come
is a dummy a iable ha ake alues 1 i i ms i expo a yea and 0 o he wise. Columns (3) and (4)
ou comes indica es he i ms’ numbe o expo ed p oduc s and des ina ions.
ha he i m expo .
5.1.2 He e ogeneous E ec s o NAILs on expo s: p oduc and ma ke di e enci a ion
In his sec ion, we explo e whe he non- a i ba ie s on impo s a ec s i ms’ abili y o expo
speci ic ypes o p oduc s o access ce ain ma ke s. To answe his ques ion, we sepa a e i m
expo s depending on he des ina ions and p oduc s ha hey sell. Fi s , we sepa a e i ms o al
expo s in o hose a e ela ed o di e en ia ed goods and hose ha a e ela ed o undi e en ia ed
goods, acco ding o Mic o-D classi ica ion (Be nini e al. 2018). Second, we sepa a e i ms o al
expo s acco ding o hei des ina ions in OECD, Me cosu (A gen ina’s mos impo an egional
ade ag eemen ), and o he coun ies.
Resul s a e p esen ed in Table 4. The coe icien o di e en ia ed expo s wi h espec o NAILs ex-
posu e is 0.34, which is h ee imes highe han he coe icien o non-di e en ia ed expo s. This
inding indica es ha access o impo ed inpu s is pa icula ly c i ical o p oducing di e en ia ed
goods, and non- a i ba ie s a ec expo e s o hese goods ela i ely mo e. Mo eo e , expo s
o OECD coun ies a e 46% mo e sensi i e o NAILs han expo s o Me cosu coun ies. This
sugges s ha NAILs ha e a mo e signi ican impac on i ms expo ing o high-income economies,
whe e compe i ion is mo e in ense and access o impo ed inpu s o educe p oduc ion cos s and
gain compe i i eness is mo e impo an .
These indings a e consis en wi h he ac ha di e en ia ed p oduc s and mo e complex ma ke s
ypically equi e mo e in ensi e use o high quali y in e media e inpu s.
21
ba ie s. In highly concen a ed ma ke s, non- ade ba ie s may impac sec o sales less, as i ms
can abso b he shock by adjus ing hei ma kups. In con as , in sec o s wi h low concen a ion
ma ke s, he e ec on he cos s o he ba ie s has a highe impac on downs eam i ms.
Figu e 5: Agg ega e e ec o exposu e o NAILs a Ma ke -le el and HHI index
-1 -.5 0 .5
Es ima ed Coe icien
0 2000 4000 6000 8000 10000
Ma ke HHI
No es: Ma ke is de ined as he des ina ion-p oduc combina ion using he 8-digi HS le el o p oduc s.
The con idence in e als a he 95% le el. Es ima ed coe icien by concen a ion o he sec o , de ined a
8-digi a i line-des ina ion-yea . Using he mean s anda d e o pe bin, 5% con idence in e als we e
compu ed a ound he bin sca e poin s
7 Conclusion
The imposi ion o Non-Au oma ic Impo Licenses (NAILs) in A gen ina be ween 2005 and 2011
p o ides a unique oppo uni y o s udy he b oade consequences o non- a i ade ba ie s. This
pape in es iga es how hese impo es ic ions a ec ed downs eam i ms, wi h a pa icula o-
cus on hei impac s on impo s, expo s, and employmen . We ind ha NAILs signi ican ly e-
duced i m impo s, leading o subsequen declines in bo h expo s and employmen o i ms ha
ely on hese impo ed inpu s. These indings unde sco e he c i ical ole ha non- a i ba ie s
play in shaping i m beha io and b oade economic ou comes.
Ou analysis u he explo es he ole o i m ma ke powe and ma ke concen a ion in media ing
he e ec s o NAILs. We de elop a heo e ical model wi h oligopolis ic compe i ion in expo
ma ke s, demons a ing ha i ms wi h g ea e ma ke powe in speci ic des ina ions can adjus
hei ma kups in esponse o cos shocks om NAILs. This abili y o abso b shocks by al e ing
ma kups educes he impac on p ices and ou pu , pa icula ly in mo e concen a ed ma ke s.
Consequen ly, he agg ega e e ec s o non- a i ba ie s like NAILs a e une enly dis ibu ed, wi h
i ms in mo e concen a ed ma ke s being be e equipped o manage hese ade es ic ions.
Addi ionally, ou indings highligh he impo ance o unde s anding how i ms, especially hose
ha expo o mul iple ma ke s, se p ices and eac o shocks. In ou sample, oughly 60% o
expo e s se e mo e han one des ina ion, and hese i ms accoun o o e 99% o o al manu-
ac u ing expo s. Unde s anding he beha io o hese mul i-des ina ion expo e s is c ucial o
assessing agg ega e ade lows and he dis ibu ion o wel a e gains om ade. We documen
ha wi hin- i m esponses o NAIL-induced cos shocks a y ac oss des ina ions, wi h i ms ad-
jus ing hei expo e enues less in ma ke s whe e hey hold a la ge ma ke sha e by educing
28
hei ma kups in hose des ina ions.
This he e ogenei y in esponses ac oss des ina ions has signi ican implica ions o he impac o
ade shocks a he agg ega e le el. Ou esul s sugges ha unila e al ade libe aliza ion, which
educes local cos s o A gen ine i ms, would disp opo iona ely bene i iche coun ies whe e
hese i ms ha e a lowe ma ke sha e, as he educ ion in cos s would lead o ela i ely g ea e
p ice educ ions in hose ma ke s. In con as , in poo e coun ies whe e mul i-des ina ion ex-
po e s ha e a highe ma ke sha e, he cos educ ions would be pa ially abso bed in he i ms’
ma kups, limi ing he ex en o he gains. These insigh s a e c ucial o policymake s, who mus
conside he a ied impac s o ade ba ie s ac oss di e en ma ke en i onmen s when designing
and implemen ing ade egula ions.
29
Re e ences
ALBORNOZ, F., I. BRAMBILLA,AND E. ORNELAS (2021): “Fi m expo esponses o a i hikes,” .
ALBORNOZ, F. AND E. GARCIA-LEMBERGMAN (2022): “Impo ing a e expo ing,” .
AMITI, M. AND S. HEISE (2024): “US ma ke concen a ion and impo compe i ion,” Re iew o
Economic S udies, dae045.
AMITI, M., O. ITSKHOKI,AND J. KONINGS (2015): “Impo e s, expo e s, and exchange a e dis-
connec ,” Ame ican Economic Re iew, 104, 1942–78.
——— (2016): “In e na ional shocks and domes ic p ices: how la ge a e s a egic complemen a i-
ies?” Tech. ep., Na ional Bu eau o Economic Resea ch.
AMITI, M. AND J. KONINGS (2007): “T ade libe aliza ion, in e media e inpu s, and p oduc i i y:
E idence om Indonesia,” The Ame ican Economic Re iew, 1611–1638.
AMITI, M., S. J. REDDING,AND D. E. WEINSTEIN (2019): “The impac o he 2018 a i s on p ices
and wel a e,” Jou nal o Economic Pe spec i es, 33, 187–210.
ANTRAS, P., T. C. FORT,AND F. TINTELNOT (2017): “The ma gins o global sou cing: Theo y and
e idence om us i ms,” Ame ican Economic Re iew, 107, 2514–64.
ARKOLAKIS, C. AND M. MORLACCO (2017): “Va iable Demand Elas ici y, Ma kups, and Pass-
Th ough,” .
ATKESON, A. AND A. BURSTEIN (2008): “P icing- o-ma ke , ade cos s, and in e na ional ela i e
p ices,” Ame ican Economic Re iew, 98, 1998–2031.
ATKIN, D., J. BLAUM, P. D. FAJGELBAUM,AND A. OSPITAL (2024): “T ade Ba ie s and Ma ke
Powe : E idence om A gen ina’s Disc e iona y Impo Res ic ions,” Tech. ep., Na ional Bu-
eau o Economic Resea ch.
AUTOR, D. H., D. DORN,AND G. H. HANSON (2013): “The China synd ome: Local labo ma ke
e ec s o impo compe i ion in he Uni ed S a es,” Ame ican economic e iew, 103, 2121–2168.
BAS, M. (2012): “Inpu - ade libe aliza ion and i m expo decisions: E idence om A gen ina,”
Jou nal o De elopmen Economics, 97, 481–493.
BEGHIN, J. C., M. MAERTENS,AND J. SWINNEN (2015): “Non a i measu es and s anda ds in
ade and global alue chains,” Annu. Re . Resou . Econ., 7, 425–450.
BERMAN, N., P. MARTIN,AND T. MAYER (2012): “How do di e en expo e s eac o exchange
a e changes?” The Qua e ly Jou nal o Economics, 127, 437–492.
BERNINI, F., J. GONZÁLEZ, J. C. HALLAK,AND A. VICONDOA (2018): “The Mic o-D classi ica ion:
A new app oach o iden i ying di e en ia ed expo s,” Economía, 18, 59–85.
BLAUM, J. (2017): “Impo ing, Expo ing and Agg ega e P oduc i i y in La ge De alua ions,” Un-
published Manusc ip , B own Uni e si y.
30
BLAUM, J., C. LELARGE,AND M. PETERS (2013): “Non-homo he ic impo demand: Fi m p oduc-
i i y and quali y bias,” Unpublished pape .
BOZ, E., E. CERUTTI,AND E. PUGACHEVA (2019): “Dissec ing he global ade slowdown: A new
da abase,” Tech. ep., IMF Wo king Pape , In e na ional Mone a y Fund, Washing on, DC.
BURSTEIN, A., V. M. CARVALHO,AND B. GRASSI (2020): “Bo om-up ma kup luc ua ions,” Tech.
ep., Na ional Bu eau o Economic Resea ch.
CALIENDO, L., M. DVORKIN,AND F. PARRO (2019): “T ade and labo ma ke dynamics: Gene al
equilib ium analysis o he china ade shock,” Econome ica, 87, 741–835.
COLE, M. T. AND C. ECKEL (2018): “Ta i s and ma kups in e ailing,” Jou nal o In e na ional
Economics, 113, 139–153.
DELOECKER, J., P. K. GOLDBERG, A. K. KHANDELWAL,AND N. PAVCNIK (2016): “P ices,
ma kups, and ade e o m,” Econome ica, 84, 445–510.
DIX-CARNEIRO, R. (2014): “T ade libe aliza ion and labo ma ke dynamics,” Econome ica, 82,
825–885.
EDMOND, C., V. MIDRIGAN,AND D. Y. XU(2015): “Compe i ion, ma kups, and he gains om
in e na ional ade,” Ame ican Economic Re iew, 105, 3183–3221.
FAJGELBAUM, P. D., P. K. GOLDBERG, P. J. KENNEDY,AND A. K. KHANDELWAL (2020): “The
e u n o p o ec ionism,” The Qua e ly Jou nal o Economics, 135, 1–55.
FENG, L., Z. LI,AND D. L. SWENSON (2017): “The connec ion be ween impo ed in e media e
inpu s and expo s: E idence om Chinese i ms,” Jou nal o In e na ional Economics, 101, 86–
101.
FLAAEN, A., A. HORTAÇSU,AND F. TINTELNOT (2020): “The p oduc ion eloca ion and p ice
e ec s o US ade policy: he case o washing machines,” Ame ican Economic Re iew, 110, 2103–
27.
FLACH, L. AND F. GRÄF (2020): “The impac o ade ag eemen s on wo ld expo p ices,” Re iew
o In e na ional Economics, 28, 168–208.
FONTAGNÉ, L. AND G. OREFICE (2018): “Le ’s y nex doo : Technical ba ie s o ade and mul i-
des ina ion i ms,” Eu opean Economic Re iew, 101, 643–663.
FONTAGNÉ, L., G. OREFICE, R. PIERMARTINI,AND N. ROCHA (2015): “P oduc s anda ds and
ma gins o ade: Fi m-le el e idence,” Jou nal o in e na ional economics, 97, 29–44.
GHOSE, D., E. FRAGA,AND A. FERNANDES (2023): “Fe ilize Impo Bans, Ag icul u al Expo s,
and Wel a e: E idence om S i Lanka,” Tech. ep., The Wo ld Bank.
GOLDBERG, P. K., A. K. KHANDELWAL, N. PAVCNIK,AND P. TOPALOVA (2010): “Impo ed in-
e media e inpu s and domes ic p oduc g ow h: E idence om India,” The Qua e ly Jou nal o
Economics, 125, 1727–1767.
31
GURKOVA, E., E. HELPMAN,AND O. ITSKHOKI (2023): “T ade Libe aliza ion, Wage Rigidi y, and
Labo Ma ke Dynamics wi h He e ogeneous Fi ms,” .
HALPERN, L., M. KOREN,AND A. SZEIDL (2009): “Impo ed inpu s and p oduc i i y,” Cen e o
Fi ms in he Global Economy (CeFiG) Wo king Pape s, 8, 28.
JUAREZ, L. (2024): “Buye ma ke powe and exchange a e pass- h ough,” A ailable a SSRN
4420344.
MANOVA, K. AND Z. ZHANG (2012): “Expo p ices ac oss i ms and des ina ions,” The Qua e ly
Jou nal o Economics, 127, 379–436.
MELITZ, M. J. (2003): “The impac o ade on in a-indus y ealloca ions and agg ega e indus y
p oduc i i y,” Econome ica, 71, 1695–1725.
NICITA, A. AND J. GOURDON (2013): A p elimina y analysis on newly collec ed da a on non- a i
measu es, UN.
ROMALIS, J. (2007): “NAFTA’s and CUSFTA’s Impac on In e na ional T ade,” The e iew o Eco-
nomics and S a is ics, 89, 416–435.
RUBENS, M. (2023): “Ma ke s uc u e, oligopsony powe , and p oduc i i y,” Ame ican Economic
Re iew, 113, 2382–2410.
32
A Appendix: Da a cons uc ion
A.1 Da a Sou ces
Da a Da a Sou ce No es
A gen inian Expo s Aduanas (2000-2012) Access h ough Minis y o P oduc i e De elopmen
A gen inian Impo s Aduanas (2000-2012) Access h ough Minis y o P oduc i e De elopmen
Dec e e In o ma ion Sec e a y o T ade, A gen ina Minis y o P oduc i e De elopmen
Employmen Fo m 931 Decla a ion Adminis acion Fede al de Ing esos Publicos (AFIP)
Mic o-D Classi ica ion Be nini e al. (2018) Classi ica ion o di e en ia ed expo s
NAILS in A gen ina In oLEG, MECON Cen e o Documen a ion and In o ma ion (CDI)
NAILS Wo ldwide Wo ld T ade O ganiza ion WTO Impo Licensing Po al
A.2 Baseline Sample
In his sec ion, we desc ibe how he da a o he baseline analysis was cons uc ed. We pu oge he
h ee da ase s: (i) AFIP Employmen Da a, (ii) Cus oms impo da a, and (iii) In oLEG dec ees.
Fi s , we ake AFIP Employmen Da a. This da ase includes in o ma ion on employmen and
ac i i y sec o s o he uni e se o i ms in A gen ina (e.g. expo e s, impo e s, domes ic i ms,
e c.) om 2001 o 2019. We keep in o ma ion o he pe iod 2003-2011. To cons uc ou sample
we p oceed wi h some cleaning s eps: (i) keep i ms wi h posi i e employmen (e.g. mo e han
1 employee), (ii) keep i ms wi h in o ma ion on he ac i i y sec o , (iii) keep all i ms ha we e
ac i e in 2007 31 and we e ac i e o a leas 1 yea in ou sample 32.
Second, we add da a om Cus oms con aining he uni e se o impo e s and expo e s in A -
gen ina. The cus oms da ase is a he i m le el and includes in o ma ion on he ade lows o
each i m, des ina ion o o igin, yea , and p oduc a he mos de ailed agg ega ion le el (12-digi
le el, which includes HS 6-digi le el and 6 digi s speci ic o A gen ina). We es ic he sample o
(i) manu ac u ing i ms o a oid ading companies whose impo s a e no in e media e inpu s o
hei p oduc ion and whose expo s a e no p oduced by o he i ms and (ii) i ms ha expo ed
a leas once in 2002-2007. Exclusions include impo s o used goods, p oduc s o igina ing om
p o inces in A gen ina, hose associa ed wi h consignmen expo e u ns, and p oduc s o igina -
ing om A gen ina. Rega ding he expo da abase, i m-le el da a be ween 2000 and 2012 a e
conside ed, excluding non- eexpo ed p oduc s and hose p oduced in A gen ina. P oduc s des-
ined o A gen ina a e also excluded, e aining only newly expo ed i ems.
Thi d, we cons uc ed a unique da abase con aining mon hly da a on (non) a i ba ie s o di -
e en p oduc s imposed in A gen ina du ing he 2002-2011. We acked and digi ized execu i e
dec ees du ing he pe iod o cons uc a da abase lis ing he mon h-yea in which an adminis a i e
ba ie was imposed on each o he p oduc s a (HS-8-Digi ). We ge his in o ma ion om In o-
LEG. In oLEG is a ju idical da abase, whe e he Legisla i e In o ma ion and Documen a ion A ea
o he Cen e o Documen a ion and In o ma ion (CDI) o he Minis y o Economy and Finance
(MECON) co-o dina es he collec ion and upda ing o na ional legisla ion, i s ules o in e p e a-
ion and backg ound.
31No e ha his s ep does no ha e ele an consequences since mos o he i ms being excluded he e a e e y small
and do no impo o expo .
32Resul s emain quali a i ely unchanged i we don’ impose his las es ic ion.
33
The main challenge in cons uc ing p ice and olume indices wi h cus oms da a is he uni alue
bias. Uni alues, de e mined by di iding obse ed alues by quan i ies, do no accu a ely e lec
eal p ices. They can luc ua e e en when he e is no ac ual p ice change due o shi s in composi-
ion. We ollow he me hodology de eloped by Boz e al. (2019) o mi iga e his issue.
A.3 In oLEG - Cen e o Documen a ion and In o ma ion (CDI)
A page on he In oLEG websi e o a speci ic esolu ion, such as Resoluciøsn 1660/2007, ypically
includes he o icial i le and numbe , he da e o issuance, he main ex de ailing he legal p o i-
sions and egula ions, and he names and posi ions o he signa o ies. I also p o ides in o ma ion
on ela ed legal documen s and amendmen s, he applicabili y and scope o he esolu ion, and
speci ic implemen a ion ins uc ions, including imelines and esponsible au ho i ies.
Figu e 6: Example o NAILs
No es: The igu e shows an example o one o he digi alized dec e es.
Sou ce:In oLEG
34
B Appendix: Empi ical Pa
B.1 B oad economic ca ego ies a ec ed by NAILs
Figu e 7: Impo s wi h NAIL by b oad economic ca ego ies
No es: This g aph co esponds o all coun ies wi h NAILs.
B.2 NAILs by sec o
Figu e 8: A e age i m’s sha e o impo s co esponding o a ec ed inpu s (2011)
No es: Fi m‘s sha e o impo s co esponding o a ec ed inpu s (2011), by sec o HS2. Sou ce: Cen e o Documen a ion and
In o ma ion (CDI) in A gen ina.
35
B.3 Robus ness in e en s udy
Figu e 9: E en s udy. The impac o Nonau oma ic Impo licenses on i ms’ expo s (logs). CLAE
2 digi s
-.4 -.3 -.2 -.1 0 .1
Log Expo s
-4 -3 -2 -1 0 1 2
Yea s Since Fi m a ec ed by i s NAIL
95% con idence in e al
Figu e 10: E en s udy. The impac o Nonau oma ic Impo licenses on i ms’ expo s (logs).
CLAE 6 digi s
B.4 Ma ke Sha e
Dis ibu ion o Ma ke sha e a iable Sisk
36
Table 7: Ma ke Sha e dis ibu ion. Yea 2006
pe cen ile Sisk
p10 0.004
p25 0.038
p50 0.299
p75 2.043
p99 9.633
A e age 4.163
B.5 Expo e s a e also impo e s
We show ha expo e s a e also impo e s. The i s igu e highligh s ha a la ge sha e o expo e s
also impo , wi h his sha e emaining s able bu sligh ly inc easing o abou 61% by 2011. The
second igu e ocuses on a subse o expo e s, showing an e en highe p opo ionâ ˘
Aˇ
Tconsis en ly
a ound 72-73% in la e yea s.
These igu es unde sco e he in e connec ed na u e o expo and impo ac i i ies, sugges ing ha
many i ms ely on impo ed inpu s o p oduc ion. This dual ole as bo h expo e s and impo e s
implies signi ican impac s om ade policies, such as non- a i ba ie s, on i m pe o mance,
esilience o cos shocks, and s a egic adap a ion o egula o y changes.
46
54
49
51
52
48
53
47
55
45
58
42
59
41
59
41
60
40
61
39
0 20 40 60 80 100
pe cen
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Pe cen age o Expo e Fi ms ha also Impo by Yea
Impo s==1 Impo s=0
61
39
64
36
67
33
68
32
69
31
71
29
72
28
72
28
73
27
73
27
0 20 40 60 80 100
pe cen
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Pe cen age o Expo e Fi ms ha also Impo by Yea
Impo s==1 Impo s=0
No es: In he i s pane, a i m is conside ed an impo e i , in he co esponding yea , i makes a leas one impo ope a ion. In he
second panel, a i m is conside ed an impo e i be ween 2002-2012 makes a leas one impo ope a ion.
B.6 O he ends on he s udied pe iod
The g aphs in he appendix p o ide an insigh ul o e iew o exchange a es and ade dynamics
o e ime, speci ically ocusing on he pe iod su ounding he implemen a ion o non-au oma ic
impo licenses (NAILs). Panel A shows he exchange a es (ARS/USD) o e ime, dis inguishing
be ween o mal and in o mal a es. Despi e luc ua ions, he e a e no signi ican changes du ing
he pe iod o NAILs applica ion. Panel B illus a es ade o e ime, depic ing expo s, impo s,
37
Msk ="Nk
∑
i=1
(1−σ−1
ik )Sik#−1
(C.2)
="Nk
∑
i
Sik −
Nk
∑
i
Sikσ−1
ik #−1
(C.3)
="1−
Nk
∑
i1
ρ(1−Sik) + 1
ηSikSik#−1
(C.4)
=1−1
ρ+1
ρ−1
ηHHIsk−1
(C.5)
=1+1
ρ(HHIsk −1)−1
ηHHIsk−1
(C.6)
We can de ine Λs,kas:
Λs,k=∂log Ms,k
∂log ps,k
=−(1
ρ−1
η)∂log HHIs,k
∂log ps,k
h1
ρ(HHIs,k−1)−1
ηHHIs,ki<0
Ms,k=1+1
ρ(HHIs,k−1)−1
ηHHIs,k−1
(C.7)
Taking logs:
log Ms,k=log 1+1
ρ(HHIs,k−1)−1
ηHHIs,k−1!≈log 1
ρ(HHIs,k−1)−1
ηHHIs,k−1!
log Ms,k=−log 1
ρ(HHIs,k−1)−1
ρHHIs,k
Di e en ia ing
∂log Ms,k=−∂log 1
ρ(HHIs,k−1)−1
ηHHIs,k=−(1
ρ−1
η)∂log HHIs,k
∂log ps,k∂log ps,k
h1
ρ(HHIs,k−1)−1
ηHHIs,ki
Λs,k=∂log Ms,k
∂log ps,k
=−(1
ρ−1
η)∂log HHIs,k
∂log ps,k
h1
ρ(HHIs,k−1)−1
ρHHIs,ki<0
C.6.4 Ma ke Le el Ou comes
In his sec ion, we explo e ma ke -le el p edic ions ega ding he he e ogeneous e ec s o ma ke
sha es. Fi s , he e ec on o al expo s is inc easing in he He indahl-Hi schman Index (HHI).
44
Speci ically, he ela ionship is desc ibed by he equa ion
Rsk =1
Mρ−1
s,k
ϕρ−1
hρ−1
s
Pρ−η
kDk, (C.8)
The elas ici y o o al expo s o a i m‘s p ice is gi en by
∂log Rsk
∂log τsk
= (ρ−1)∑
k
Rsk
Rs
1
1+Λsk
ms >0, (C.9)
which is posi i e i ρ>1. The Λsk unc ion inco po a es he HHI index, indica ing ha highe
ma ke concen a ion leads o a g ea e sensi i i y o expo s o p ice changes.
Second, he e ec on o al impo s sugges s ha impo s a e weakly dec easing in he ade cos s
o he impo ing a ie ies, p o ided ρ. The equa ion cap u es his ela ionship.
∂log Ms
∂log τs
=−ms "ρ∑
k
Qsk
Qs
1
1+Λsk
−1#≤0, (C.10)
implying ha as ade cos s inc ease, o al impo s dec ease, e lec ing he sensi i i y o impo
olumes o cos a ia ions.
Finally, he elas ici y o expo s wi h espec o impo s indica es ha he o al amoun o expo s
in a sec o is posi i ely ela ed o he amoun o impo s in ha sec o . This is o malized by he
equa ion.
ΣX,M=
∂log Rsk
∂log τsk
∂log Ms
∂log τs
=∂log Rsk
∂log Ms
=(1−ρ)∑kRsk
Rsh1
1+Λsk i
(1−ρ)h∑kQsk
Qk
1
1+Λsk i>0, (C.11)
indica ing a posi i e ela ionship be ween impo s and expo s, unde sco ing he in e connec ed
na u e o ade dynamics wi hin a sec o . Toge he , hese indings highligh he impo ance o
conside ing ma ke sha es and ade cos s in unde s anding he b oade economic impac s on
expo s and impo s. Below, we can show he o mal p oposi ions:
PROPOSITION 6(Ma ke le el p edic ions).
A. (E ec on o al expo s) E ec o o al expo s is inc easing in HHI.
Rs,k=1
Mρ−1
s,k
ϕρ−1
hρ−1
s
Pρ−η
kDk(C.12)
∂log Rs,k
∂log τs,k
= (ρ−1)∑
k
Rs.k
Rs
1
1+Λs.k
ms, >0 (C.13)
I ρ>1 ha equa ion is posi i e. Inside he Λs,k unc ion is he HHI index.
B. (E ec on o al impo s) P o ided ρ, impo s a e weakly dec easing in he ade cos s o he impo ing
45
a ie ies.
∂log Ms
∂log τs,
=−ms, "ρ∑
k
Qs.k
Qs
1
1+Λs.k
−1#≤0 (C.14)
C. (Elas ici y o expo s wi h espec o impo s) The o al amoun o expo s o a sec o a e inc easing
on he amoun o impo s o he sec o . Tha is,
ΣX,M=
∂log Rs,k
∂log τs,k
∂log Ms
∂log τs,
=∂log Rs,k
∂log Ms
=(1−ρ)∑k
Rs,k
Rsh1
1+Λs.ki
(1−ρ)h∑k
Qs,k
Qk
1
1+Λs.ki>0
46
