Se ić, Ma ina Basa ac; Vučko ić, Valen ina; Andabaka, Ana
A icle
Examining he e ec s o ade libe alisa ion using a g a i y
model app oach
Economics: The Open-Access, Open-Assessmen Jou nal
P o ided in Coope a ion wi h:
De G uy e B ill
Sugges ed Ci a ion: Se ić, Ma ina Basa ac; Vučko ić, Valen ina; Andabaka, Ana (2024) : Examining
he e ec s o ade libe alisa ion using a g a i y model app oach, Economics: The Open-Access,
Open-Assessmen Jou nal, ISSN 1864-6042, De G uy e , Be lin, Vol. 18, Iss. 1, pp. 1-12,
h ps://doi.o g/10.1515/econ-2022-0061
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/306077
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/4.0/
Resea ch A icle
Ma ina Basa ac Se ić*, Valen ina Vučko ić, and Ana Andabaka
Examining he Effec s o T ade Libe alisa ion
Using a G a i y Model App oach
h ps://doi.o g/10.1515/econ-2022-0061
ecei ed July 01, 2023; accep ed Decembe 09, 2023
Abs ac : The main objec i e o his a icle is o s udy he
impac o ade libe alisa ion in C oa ia as one o he fi s
s uc u al e o ms being implemen ed, co e ing he pe iod
om 2000 o 2021. The panel da a model was specified
using a wo-s ep sys em gene alised me hod o momen
es ima o . The ob ained esul s show ha ade libe alisa-
ion measu ed h ough he specific indices encompassing a
b oade se o bo h a iffand non- a iffba ie s, size o he
ade sec o , eedom o o eign exchange ma ke , and
capi al con ols posi i ely impac ed expo pe o mance.
The esul s also showed a nega i e and s a is ically signifi-
can effec o emo eness on ade, confi ming ha geog a-
phical dis ance is an essen ial indica o o anspo a ion
cos s. The coefficien o Schengen accession was posi i e
and s a is ically significan , indica ing ha he Schengen
ag eemen has boos ed expo s and ha we can expec
he same o C oa ia, especially in he con ex o simul a-
neous accession o he Eu ozone and he Schengen a ea on
1 Janua y 2023.
Keywo ds: in e na ional ade, Eu opean Union, Schengen,
C oa ia, GMM
JEL Classifica ion: F43, F53, F62
1 In oduc ion
Wi h globalisa ion being in he mids o a ans o ma ion,
uelled by he cu en poli ical and economic ensions, we a e
inc easingly wi nessing p ocesses o g owing p o ec ionism
and a consequen dis up ion o bypass o mul ila e al ins i u-
ions (Eu opean Commission (EC), 2021). All hese p ocesses,
s eng hened by apid echnological change and econfigu a-
ion o global alue chains (GVCs), will u he impac he
in e na ional ade ou look (Uni ed Na ions (UN), 2021).
O e he las decade, he Eu opean Union (EU) has
adjus ed i s ade policies o balance be ween p o ec-
ionism and globalisa ion. I has also engaged in nume ous
p e e en ial ade alks. I has o e 70 bila e al ade
acco ds, and many o he discussions a e s ill open (Ülgen,
e al., 2022).
Al hough he cos s o open ade a e mo e easily
de ec ed han gains, a weal h o e idence demons a es
how ade libe alisa ion has suppo ed he ise in p os-
pe i y o he g owing global popula ion due o a consis en
inc ease in p oduc i i y (Ce na , 2019). Gnangnon (2018)
iden ified six po en ial channels h ough which ade libe -
alisa ion could induce highe economic g ow h, including
imp o ing wel a e, p omo ing o eign di ec in es men ,
educing ade cos s, mode a ing fluc ua ions in ade and
p o iding ade s wi h s able income, g ea e coope a ion,
and gene a ing highe public e enue. F eeman e al. (2022)
ound ha mos EU membe s benefi significan ly om
ade despi e some diffe ences be ween indi idual coun-
ies. Cen al and Eas e n Eu opean coun ies (CEECs) end
o see significan benefi s, while coun ies such as I aly and
G eece may see mo e limi ed benefi s. I is wo h no ing ha
hese benefi s end o inc ease o e ime as economic in e-
g a ion educes ade cos s and he EU in e nal ma ke
g ows. This is pa icula ly ue o CEECs, which ha e
expe ienced globalisa ion and economic ansi ion while
in eg a ing in o he EU and unde going nume ous s uc u al
e o ms.
In his s udy, we con ibu e o his deba e by exam-
ining he impac o ade libe alisa ion be ween he EU and
C oa ia, he las coun y o join he EU, affec ing i s deg ee o
in eg a ion in o GVCs. The Wo ld Bank (2018) has epo ed
ha C oa ia has unde gone economic and li ing s anda d
imp o emen s since gaining independence. O e a ew dec-
ades, C oa ia has success ully es ablished a libe al democ acy,
emb aced a ma ke economy, and achie ed uppe -middle-
income coun y s a us. A commi men o a ious in e na ional
* Co esponding au ho : Ma ina Basa ac Se ić,Economic Resea ch
Di ision, C oa ian Academy o Sciences and A s, An e Ko ačića 5, 10000,
Zag eb, C oa ia, e-mail: [email p o ec ed]
Valen ina Vučko ić, Ana Andabaka: Facul y o Economics and
Business, Uni e si y o Zag eb, T g J. F. Kennedyja 6, 10000, Zag eb,
C oa ia, e-mail: [email p o ec ed], [email p o ec ed]
ORCID: Ma ina Basa ac Se ić0000-0003-4182-8405; Valen ina Vučko ić
0000-0002-5438-0665; Ana Andabaka 0000-0003-2596-0561
Economics 2024; 18: 20220061
Open Access. © 2024 he au ho (s), published by De G uy e . This wo k is licensed unde he C ea i e Commons A ibu ion 4.0 In e na ional License.
ade ag eemen s, egional pa ne ships, and domes ic
e o ms ma ked C oa ia’s jou ney owa ds g ea e eco-
nomic openness. Mo e p ecisely, C oa ia success ully imple-
men ed ade libe alisa ion in phases, wi h i s membe ship
in in e na ional o ganisa ions and associa ions significan ly
shaping i s in e na ional ade pa e ns. A e becoming a
Wo ld T ade O ganiza ion (WTO) membe in 2000 and a
Cen al Eu opean F ee T ade Ag eemen (CEFTA) membe
in 2002, C oa ian expo s and impo s significan ly inc eased
in goods and se ices (see Figu es A1 and A2 in Annex).
Goods expo s ha e been one o he pas pe iod’smos sig-
nifican g ow h and economic eco e y sou ces. C oa ia has
managed o inc ease i s sha e o global expo s in g oss
domes ic p oduc (GDP), hus confi ming he impo ance
o joining he EU o be e in eg a ion in o Eu opean and
GVCs. Howe e , due o i s p e ailing eliance on he ou ism
sec o , which accoun s o up o 70% o se ices expo s and
he con ibu ion o which o he C oa ian economy is a ound
17%, pos - ansi ion C oa ia emains a small open economy
wi h limi ed p oduc i i y gains (Wo ld Bank, 2018). Despi e
he a ou able de elopmen s, he C oa ian expo sec o
emains small compa ed o o he CEECs (e.g. 20% o GDP
ela i e o a ound 80% in Slo akia), so i s abili y o influence
o e all p oduc i i y and g ow h is s ill limi ed. In addi ion,
he echnological sophis ica ion o expo s emained ela-
i ely low (Official Gaze e, 13/2021). When i comes o he
mos impo an ade pa ne s, he op fi e coun ies (mo e
han 50% o o al ade) o which C oa ia expo ed in 2020
a e Ge many, I aly, Slo enia, Bosnia and He zego ina, and
Hunga y, while he op fi e coun ies om which C oa ia
impo s goods a e Ge many, I aly, Slo enia, Aus ia, and
Hunga y. Finally, acco ding o he Indus ial De elopmen
Repo 2022 (Uni ed Na ions Indus ial De elopmen
O ganiza ion (UNIDO), 2021), C oa ia s ill belongs o he
g oup o de eloping and eme ging indus ialised coun ies
when analysing he le el o indus ialisa ion. In his con ex ,
his s udy emba ks on a comp ehensi e analysis o he
effec s o ade libe alisa ion on ade pe o mance in
C oa ia, wi h he p ima y goal o in es iga ing how ade
libe alisa ion has affec ed expo pe o mance as one o
he coun y’s compe i i eness indica o s. In he spi i o
Spo nbe ge (2022), his s udy con ibu es o he “cos o
non-Eu ope”by e ospec i ely e alua ing he libe alisa-
ion and in eg a ion o EU ade. Ano he con ibu ion o
ou a icle is in using panel da a (o e ime bila e al ade
da a) and he sys em gene alised me hod o momen s
(GMM) model es ima ion. To cla i y he effec o ade
libe alisa ion in C oa ia, we examine he dynamics o
me chandise expo s and impo s om 2000 o 2021 using
da a om EU membe s and he Uni ed Kingdom (UK). The
g a i y model is applied o e a longe ime se ies as
compa ed o p e ious s udies o accoun o he ime a e
he EU accession and cap u es only he ade be ween EU
membe coun ies and C oa ia. Fu he mo e, we wan o
shed ligh on he limi a ions o ade libe alisa ion effec s
on economic g ow h gi en he specific ea u eso his
small open economy. Mo eo e , mos s udies do no
make a clea dis inc ion be ween indica o s ha show
he ou comes o openness (e.g. olume o ade, coun y’s
size, popula ion, and echnological ad ancemen ) and
hose ha eflec ade policy openness (e.g. ade policy
asa esul o in en ionalpolicyac ions ha makea
coun y mo e open o ade) (McCulloch e al., 2001).
1
In
his a icle, we y o b idge he gap be ween ade policy
openness and ou come openness by using he eedom o
ade in e na ionally, a sub-componen o he F ase Ins i-
u e’s Economic F eedom Index. The esul s show ha
ade libe alisa ion posi i ely impac ed bo h expo s and
impo s. The impo s o goods a e p edominan ly d i en
by he expo s o se ices, mainly om he ou ism sec o ,
on which he economy elies hea ily. Conside ing ha he
inc ease in impo s does no s em p ima ily om manu-
ac u ing, he expec ed boos in ade om he accession o
he Schengen a ea indica es he need o ackle he p oblem
o high impo dependence and low echnological sophis-
ica ion o expo s.
The emainde o his a icle is s uc u ed as ollows.
Sec ion 2 gi es an o e iew o exis ing li e a u e on g a i y
model applica ion in ade analysis and he li e a u e on
he effec s o ade libe alisa ion in EU coun ies. Sec ion 3
desc ibes he da a used, he me hod applied, and he ea-
sons behind choosing a sys em GMM es ima o . Sec ion 4
con ains he esul s o he econome ic analysis and hei
in e p e a ion, ollowed by a discussion in Sec ion 5. Finally,
Sec ion 6 concludes and p esen s some limi a ions and
po en ial a enues o u u e esea ch.
2 Li e a u e Re iew
The g a i y model o ade is a highly in ui i e, s uc u al
model wi h sound heo e ical ounda ions, a ealis ic gen-
e al equilib ium se ing ha simul aneously accoun s o
mul iple coun ies, and a e y flexible s uc u e ha can be
in eg a ed in o a b oad class o mo e gene al equilib ium
1The main diffe ence be ween hese wo ypes is ha e en i a
coun y does no ha e an open ade policy, i may s ill be a la ge
ading na ion due o i s na u al ad an ages (McCulloch e al., 2001;
Singh, 2018).
2Ma ina Basa ac Se iće al.
models o s udy he linkages be ween ade and labou
ma ke s, in es men , he en i onmen , e c. (Yo o e al.,
2016). I explains ade flows be ween pai s o coun ies (bila -
e al ade flows) by a iables d awn om impo ing and
expo ing coun ies. In i s mos basic o m, he model assumes
ha ade be ween coun y iand coun y jis p opo ional o
he p oduc o
GDP
i
and
GDP
j
an in e sely ela ed o he
dis ance be ween hem (B aha e al., 2015).
New on’s law o g a i y, applied o in e na ional ade,
s a es ha jus as pa icles a ac each o he in p opo ion
o hei size and p oximi y, coun ies ade in p opo ion
o hei espec i e ma ke size (e.g. GDP) and p oximi y
(Yo o e al., 2016). The fi s applica ions o New on’s law
o g a i y o economics include Linneman (1966), Poy-
honen (1963), Ra ens ein (1885), and Tinbe gen (1962).
Seminal wo k o Ande son (1979) is he fi s o p o ide a
heo e ical economic basis o he g a i y equa ion based
on cons an elas ici y o subs i u ion (CES) p e e ences
and goods ha a e diffe en ia ed by egion o o igin.
Ano he ea ly con ibu ion o g a i y heo y comes om
Be gs and (1989, 1990) and Dea do ff(1998) who ha e p e-
se ed he CES p e e ence s uc u e and added monopolis ic
compe i ion o a Hecksche –Ohlin s uc u e o explain spe-
cialisa ion (Ande son & an Wincoop, 2003).
This was ollowed by he influen ial wo k o Ande son
and an Wincoop (2003) and Ea on and Ko um (2002),
which p o ided he mic oeconomic ounda ions o he
g a i y model and g ea ly impac ed he u he de elop-
men and applica ion o he g a i y model. Mo e p ecisely,
wi h he publica ion o Ande son and an Wincoop (2003)
and Ea on and Ko um (2002), he con en ional wisdom
ha g a i y equa ions lacked mic o- ounda ions was finally
dismissed (Head & Maye , 2014). Since nei he model elied
on impe ec compe i ion o inc easing e u ns, he e was
no longe a eason o belie e ha g a i y equa ions should
only apply o a subse o coun ies o indus ies (Head &
Maye , 2014).
The nex impo an yea was 2008, because h ee
pape s we e published –Chaney (2008), Helpman e al.
(2008), and Meli z and O a iano (2008) –who b ough
oge he ecen wo k on he e ogeneous fi ms wi h he
de e mina ion o bila e al ade flows (Head & Maye ,
2014). Since hen, he g a i y model has been used in
esea ch wi h many augmen a ions.
T adi ionally, g a i y models ha e been based mainly
on in ui i e ideas abou which a iables will likely influ-
ence ade. The exis ing li e a u e con ains nume ous s u-
dies ha examine he impac o ade libe alisa ion on
a ious economic ou comes, wi h esul s a ying depending
on he a iables and coun ies included in he analyses. The
emaining pa o he li e a u e e iew is di ided in o h ee
pa s acco ding o ou esea ch in e es –a icles analysing
ade libe alisa ion in EU coun ies, hen in New Membe
S a es (NMS), and hose ela ed o C oa ia. Fi s , Gnangnon
(2018) analysed 150 coun ies, including EU coun ies, by
using an unbalanced panel da ase comp ising o e he
pe iod 1995–2015 o assess he impac o mul ila e al ade
libe alisa ion on economic g ow h. His esul s sugges a
s ong posi i e effec , wi h uppe -middle-income and high-
income coun ies benefi ing mos om mul ila e al ade
libe alisa ion due o hei supe io ading capabili ies com-
pa ed o low-income and lowe -middle-income coun ies.
Jena and Ba ua (2020) analysed he con e gence dynamics
in he EU. They showed ha lowe -income coun ies a e
ca ching up wi h he iche coun ies by opening up o
in e na ional ade, excluding in a-EU ade, and by
implemen ing g ow h-enhancing go e nmen expendi u e
policies. Thei esul s showed ha ela i ely lowe -income
coun ies such as Romania, C oa ia, La ia, Mal a, and
Cyp us expe ienced a somewha highe impac o ade
and go e nmen expendi u e han highe -income coun-
ies. Spo nbe ge ’s (2022) analysis co e s 43 coun ies
( he EU28, BRIIC coun ies [B azil, Russia, India, Indonesia,
and China]) and 10 OECD membe coun ies (Aus alia,
Canada, Japan, Mexico, No way, Sou h Ko ea, Swi ze land,
Taiwan, Tu key, and he USA) om 1995 o 2014. This
a icle ocuses on ade in manu ac u ed goods, using a
s uc u al g a i y amewo k and a flexible wo-s ep es i-
ma ion app oach, accoun ing o o e wo- hi ds o in e -
na ional goods ade. The esul s e eal a deep in eg a ion
p ocess ha inc eased ade sha es wi hin he EU, i.e. ade
sha es be ween he EU-15 and he CEECs inc eased by
a ound 40% due o EU in eg a ion effec s.
When i comes o analysing ade libe alisa ion o he
NMS
2
o he EU, he li e a u e is no e y ex ensi e.
Namely, Bussiè e e al. (2005) analysed he apid ade
in eg a ion o he CEECs wi h he eu o a ea om 1980 o
2003. Based on he augmen ed g a i y model, he esul s
sugges ha ade in eg a ion be ween mos o he la ges
CEECsand heeu oa eaisal eady ela i ely well
ad anced, while some Bal ic and Sou h Eas e n Eu opean
coun ies s ill ha e significan scope o ade in eg a ion.
In addi ion, Papazoglou e al. (2006) used a g a i y model o
o ecas he po en ial impac o he 2004 EU enla gemen
on ade balances and ade flows. The esul s sugges ha
g oss ade c ea ion o he accession coun ies is abou
25% o hei 2003 ade. Simila ly, Hagemeje and Mućk
(2019) e alua ed he effec s o expo s and GVC
2I e e s o membe s a es ha joined he EU a e 2004.
Effec s o T ade Libe alisa ion Using a G a i y Model App oach 3
pa icipa ion on economic g ow h in CEECs om 1995 o
2014. The au ho s showed ha expo s had played a i al
ole in con e ging he CEECs wi h hei ad anced coun e -
pa s. Also, hey showed ha he significan g ow h d i e s
a e GVC pa icipa ion, impo s o echnology, and capi al
deepening.
Se e al au ho s ha e used he g a i y model o ana-
lyse C oa ian ade libe alisa ion. Fo example, Klimczak
(2016) examined he Wes e n Balkan egion (Albania, Bosnia
and He zego ina, C oa ia, Mon eneg o, Se bia, and FYR o
Macedonia) by using panel da a o he 2001–2014 pe iod.
Based on he es ima ion esul s, i appea s ha ade libe al-
isa ion has only had a minimal impac on he alue o
expo s. Demand, a ious supply- ela ed a iables, and o -
eign di ec in es men s a e key ac o s influencing expo s.
In e es ingly, a la ge in e nal ma ke , measu ed by popula-
ion a he han GDP, appea s o limi expo s. Ranilo ić
(2017) analysed he effec s o C oa ian accession o he EU
on me chandise ade using he g a i y model. The esul s
confi med he posi i e effec o he EU accession on ade.
On he o he hand, ee ade ag eemen s wi h non-EU coun-
ies ha e no s a is ically significan and posi i e effec on
C oa ian ade. S ojčiće al. (2018) analysed he ade libe al-
isa ion effec s wi h he EU on changes in he s uc u e and
quali y o expo s om NMS om 1990 o 2015. Thei esul s
showed ha he iming o ade libe alisa ion wi h he EU
affec ed he expo pe o mance e olu ion and he s uc u e
and quali y o expo s om NMS wi h a eco ded inc ease
in he sha e o high echnology-in ensi e indus ies. The
au ho s showed ha he mos ad anced NMS ob ained ull
benefi s o p e e en ial access o EU ma ke s, wi h smalle
effec s eco ded in Slo enia and C oa ia. Jošićand Bašić
(2021) analysed C oa ia’sCEFTAandEUmembe shipeffec s
on ade c ea ion and ade di e sion using he g a i y
model o in e na ional ade. Thei analysis encompassed
180 ading pa ne coun ies om 2000 o 2016. The esul s
showed posi i e effec s o C oa ia–CEFTA in eg a ion e i-
den in a dominan ade c ea ion effec . Con e sely, he
C oa ia–EU in eg a ion exhibi ed a ade di e sion effec
in cases o impo s and expo s and is inconclusi e in
o al ade flows. Ris ano iće al. (2020), using a dynamic
econome ic model, e alua ed he ac o s ha showed
he impac on he o al ade o Se bia and EU membe
s a es in a pe iod om 2001 o 2018. Thei esul s showed
ha he size o he economy and popula ion played an
impo an olein he adeo Se bia.A hesame ime,
he geog aphical dis ance nega i ely affec s he bila e al
ade be ween Se bia and o eign ading pa ne s om
he EU, p o ing he basic assump ion o a g a i y model.
Addi ionally, a common bo de and a common language
also show posi i e effec s on bila e al ade.
3 Resea ch Me hods
In his a icle, we use panel da a (o e ime bila e al ade
da a), which is conside ed o ha e an ad an age o mi i-
ga ing he bias gene a ed by he e ogenei y ac oss coun-
ies. P ecisely, he models a e es ima ed using a wo-s ep
sys em GMM es ima o (A ellano & Bo e , 1995; Blundell &
Bond, 1998) sui able o cases o “la ge nand small ”
(Roodman, 2009) whe e lagged le els a e used as ins u-
men s o cu en diffe ences, and ice e sa (WTO, 2012).
Namely, he GMM is a s a is ical es ima ion echnique used
in econome ics (Hansen, 1982) and o he fields o es ima e
model pa ame e s when he assump ions o o dina y leas
squa es (OLS), fixed-effec s, o andom-effec s models may
no apply o a e inapp op ia e. GMM can be a powe ul
and flexible me hod o es ima ing model pa ame e s unde
ce ain condi ions, making i a be e choice o hese o he
me hods in some cases. In pa icula , OLS, fixed-effec s
models, and andom-effec s models make specificassump-
ions abou he da a, such as linea i y and homoscedas i-
ci y. GMM does no ely on hese assump ions and can
p o ide consis en es ima es, bu i is no wi hou i s chal-
lenges. In addi ion, he lagged dependen a iable on he
igh -hand side o he equa ion (as in ou model) can lead
o a co ela ion be ween ha a iable and he e o e m.
In such cases, he con en ional es ima o s lis ed p e iously
may yield biased and inconsis en es ima es (Heo e al.,
2021).
Nex , he sys em GMM can effec i ely add ess endo-
genei y issues (A ellano & Bo e , 1995), whe e he expla-
na o y a iables a e co ela ed wi h he e o e m. This is
a common p oblem in eg ession analysis, and GMM can
help mi iga e bias caused by endogenei y. Fu he mo e,
using bo h le els and diffe ences o he da a can lead o
g ea e p ecision in pa ame e es ima es. The e o e, GMM
is a e y flexible me hod ha can be adap ed o diffe en
ypes o da a and model specifica ions, ocusing on he
change o a iables o e ime, which is pa icula ly impo -
an in ou model. Hence, i is a na u al choice when
dealing wi h da a ha ha e bo h c oss-sec ional and
ime-se ies dimensions.
Ou model employs da a o all EU coun ies ( he UK is
also included
3
). The analysis co e s he 2000–2021 pe iod.
Symbolically, he gene al panel baseline model is specified
as ollows:
=+ + + +
+= =
−
XαβX β β θ
εi N T
ln ln lnGDP lnRem
,1,…,, 1,…,,
i i i i i
ii
1123
(1)
3This s udy does no ake B exi in o accoun .
4Ma ina Basa ac Se iće al.
whe e Nis he numbe o uni s o obse a ion, Tis he
numbe o pe iods,
X
i
s ands o he alue o he depen-
den a iable (in his case, he C oa ian expo and impo )
iin he pe iod , he pa ame e
α
is he cons an ,
β
is he
scala ,
−
X
i
1
is he one-pe iod lagged (1 yea ) dependen
a iable,
θ
is he fixed elemen o andom e o o he
uni o obse a ion, and ε
i
is he e o e m in he model.
The dependen a iable measu es bila e al ade (expo
and impo ) be ween C oa ia and o he EU membe s plus
he UK. The a iables o in e na ional ade, impo s, and
expo s a e defla ed by he Ha monised Index o Consume
P ices. The e a e also independen a iables influencing
he s uc u e o expo s. Fi s , lnGDP is nominal GDP, and
he eason o using nominal, no eal GDP is explained in
Shephe d e al. (2019). Namely, he GDP da a a e effec i ely
defla ed by he mul ila e al esis ance e ms (MRT), which
a e unobse ed p ice indices. Defla ing by some o he ac-
o s, such as a eadily obse able p ice index, is likely o be
misleading (Shephe d e al., 2019). Fu he mo e, a p oxy
“ emo eness”(lnRem) is also included in he analysis.
Mo e specifically, a o mula ha measu es a coun y’s
a e age weigh ed dis ance om i s ading pa ne s,
whe e weigh s a e he pa ne coun ies’sha es o EU
GDP (deno ed by
GDP
E
U
), is used o con ol o he mul i-
la e al esis ance e ms o expo ing and impo ing coun-
ies (WTO, 2012):
=ΣRem dis
.
ij
j
GDP
GDP j
EU
In ou analysis, MRT is con olled in GMM es ima ions
using he a iable emo eness. Remo eness measu es a
coun y’s a e age weigh ed dis ance om i s ading
pa ne coun ies. Namely, he mul ila e al esis ances
bea he in ui i e in e p e a ion ha , all else equal, wo
coun ies will ade mo e wi h each o he he mo e emo e
hey a e om he es o he wo ld (Yo o e al., 2016).
P ope accoun o he mul ila e al esis ances is he c i-
ical diffe ence be ween he nai e s. heo y- ounded
applica ions o he ade g a i y model (Ande son & an
Wincoop, 2003). Some esea che s c i icise a educed-
o m e sion o he cus om ea men om Ande son
and an Wincoop (2003), whe e he mul ila e al esis ance
e ms a e app oxima ed by he so-called “ emo eness
indexes”cons uc ed as unc ions o bila e al dis ance
and GDP (Head & Maye , 2014).
Howe e , ade cos s go beyond a iffs and anspo a-
ion expenses. They can include ac o s such as cus oms
p ocedu es, bo de delays, cul u al diffe ences, and egu-
la o y ba ie s. Remo eness se es as a p oxy o some
o hese non- a iff ade ba ie s because mo e dis an
coun ies a e likely o ace g ea e logis ical and egula o y
challenges when ading wi h each o he . Analys s can
be e accoun o hese hidden ade cos s by including
emo eness in he g a i y model.
In addi ion o emo eness, se e al o he a iables can
be used as mul ila e al esis ance e ms in he g a i y
model o in e na ional ade o cap u e a ious aspec s
o ade cos s and ba ie s. Tha is why we added he
Schengen a iable and an index o economic eedom,
4
in o de o cap u e diffe en dimensions o ade cos s
and ba ie s in a mo e comp ehensi e manne .
The selec ion o a iables in he model is made on he
basis o p e ious esea ch (e.g. B aha e al., 2015;Heo&
Doanh, 2020; Ris ano iće al., 2020; WTO, 2012) and he
specific needs o his esea ch.
5
The dependen one-pe iod-
lagged a iable will be used as an ins umen al a iable.
In he nex s ep, wo ex ended model e sions we e
in oduced and a e he subjec o he analysis. Model (2), in
addi ion o he base a iables, includes he a iable Schengen
en y (Schengen), popula ion (lnpop), and eedom o ade
in e na ionally (ln i) as a p oxy o ade libe alisa ion:
=+ + +
+++++
−
XαβX β β
βββθε
ln ln lnGDP lnRem
Schengen lnpop ln i .
i i i i i
i i i i
1123
456
(2)
The dummy a iable ep esen ing Schengen en y is a
bina y a iable ha akes he alue o 1 i he coun y is in
Schengen and 0 o he wise.
6
In addi ion, he dummy a i-
able assumes a alue o 1 om he yea he coun y en e s
he Schengen a ea. The Schengen Ag eemen ep esen s a
significan achie emen in he ongoing p ocess o deepening
4In an ex ended model ha will be explained la e .
5In ecen yea s, hese models ha e been es ima ed using da a on
in e na ional ade flows only, bu a numbe o ecen s udies s ess
he impo ance o including in a-na ional (i.e. domes ic) ade flows
as well (Campos e al., 2021). Fo ins ance, Yo o (2012) a gued ha he
solu ion o he dis ance puzzle is he applica ion o he app op ia e
measu e o globalisa ion – ha is he inc ease in in e na ional eco-
nomic in eg a ion ela i e o he in eg a ion o in e nal ma ke s. Con-
sequen ly, he au ho emphasised he need o e alua e he effec s o
bila e al dis ance and in e na ional ade cos s ela i e o he effec s
o in e nal dis ance and in e nal ade cos s in o de o cap u e he
effec s o globalisa ion. Howe e , he es ima es o he s anda d g a i y
a iables based on he sample wi h in e na ional and in a-na ional
ade a e s a is ically no diffe en om he co esponding es ima ed
pa ame e s based on he sample wi h in e na ional ade only (Yo o
e al., 2016). Ano he sho coming is ha in e nal ade includes ade
in se ices, whe eas ex e nal ade is la gely limi ed o me chandise
ade (B ei & on Pe e , 2017).
6Today, he Schengen a ea includes mos EU coun ies, wi h he
excep ion o Bulga ia, Cyp us, I eland, and Romania, as well as he
Uni ed Kingdom.
Effec s o T ade Libe alisa ion Using a G a i y Model App oach 5
Eu opean in eg a ion, and i s p ima y aim is o simpli y he
mo emen o goods, se ices, and people wi hin Eu ope’s
in e nal bo de s. The e o e, i is easonable o expec ha
i will affec ade ba ie s. Mo eo e , acco ding o Aussilon
and Le Hi (2016a), he Schengen Ag eemen mus ha e
educed he “economic dis ance”be ween i s membe coun-
ies h ough a ious channels. A icles ha analyse he
impo ance o he Schengen Ag eemen based on he g a i y
model (Aussilloux & Le Hi , 2016a; Chen & No y, 2011; Da is
& Gi , 2014; Felbe may e al., 2016) conclude ha i yields
posi i e effec s on ade. Mo e p ecisely, Da is and Gi
(2014) a gued ha labou mobili y esul ing om Schengen
posi i ely affec s ade by inc easing demand o o eign
goods, imp o ing awa eness o low-cos p oduce s ab oad,
and lowe ing he isks associa ed wi h buying and selling
ou side he coun y. Chen and No y (2011) ound ha c oss-
coun y ade in eg a ion is lowe o hose coun ies ha
joined he EU mos ecen ly and ha e no ye implemen ed
he Schengen Ag eemen ha abolishes physical bo de con-
ols. Aussilloux and Le Hi (2016b) showed ha e-es ab-
lishing pe manen bo de con ols wi hin he Schengen
A ea would dec ease ade be ween Schengen coun ies
by 10–20%. The Schengen a ea’s GDP would be educed by
0.8 poin s, equi alen o mo e han 100 billion eu os. Nex ,
acco ding o Felbe may e al. (2016), Schengen has boos ed
ade by 3% on a e age (equi alen o a d op in a iffsby0.7
pe cen age poin s). Goods ade is mo e obus ly affec ed
han se ices, and pe iphe al coun ies benefi mo e han
cen al ones. Mo e in e es ingly, Spo nbe ge (2022) con-
cluded ha ade in eg a ion has no deepened o he EU-
15, while ade sha es among he newly joined cen al and
eas e n EU membe s doubled.
Fu he mo e, nume ous analyses show ha he effec s
o he popula ion on ade a e ambiguous (Kuma & Ahmed,
2015;Ris ano iće al., 2020). Addi ionally, acco ding o Fi z-
simons e al. (1999), ade ises wi h popula ion, which
indica es ha la ge and weal hy coun ies end o ade
mo e wi h each o he based on a gi en GDP pe capi a.
F eedom o ade in e na ionally is a measu e o
a wide a ie y o es ain s ha affec in e na ional
exchange: a iffs, quo as, hidden adminis a i e es ic-
ions, and con ols on exchange a es and he mo emen
o capi al, and i is o en used as a p oxy o ins i u ional
quali y. I can be defined as eedom o exchange –in i s
b oades sense, buying, selling, making con ac s, and so
on –is essen ial o economic eedom, which is educed
when eedom o exchange does no include businesses
and indi iduals in o he na ions (The F ase Ins i u e,
2022). I is in e es ing o ou analysis since i cap u es
bo h a iffand non- a iffba ie s, he size o he ade
sec o , eedom o he o eign exchange ma ke , and capi al
con ols. Acco ding o Sono a (2014), he economic eedom
o a ading pa ne has a s a is ically significan and posi-
i e effec on he olume o ade be ween he US and i s
ading pa ne s. The nex a ia ion (3) includes he index o
economic eedom (lnecon ) ins ead o eedom o ade
in e na ionally:
=+ + +
++++
+
−
XαβX β β
βββθ
ε
ln ln lnGDP lnRem
Schengen lnpop lnecon
.
i i i i i
i i i
i
1123
456 (3)
The index o economic eedom o He i age Founda ion
7
is used as a gene al measu e o he economic eedom o a
coun y measu ed h ough he lenses o ou ca ego ies o
eedom, including ma ke openness ( ade eedom,
in es men eedom, and financial eedom). We employ
i as a obus ness check and as a subs i u e o he o me
F ase Ins i u e Index because bo h indices gauge eco-
nomic eedom by using a weigh ed a e age o a ious
componen s. Howe e , while he o me p ima ily elies
on quan i a i e a iables, he la e inco po a es quali a-
i e assessmen s ( esea ch ha conside s his index in a
con ex o measu ing ade libe alisa ion includes Gnangnon
(2018), San os-Paulino (2005), and Wall (1999)).
All a iables no exp essed in sha es a e loga i hmically
ans o med (nominal GDP, eedom o ade in e na ion-
ally, index o economic eedom, and emo eness). Table 1
desc ibes he a iables, hei sou ces, and expec ed signs
ha align wi h economic heo y and p e ious esea ch.
4 Resul s
This sec ion p o ides es ima es ob ained wi hin he sys em
GMM model. Table 2 shows he esul s o he es ima ed
impac o he ade libe alisa ion in C oa ia and he diag-
nos ic es s o dynamic panel da a analysis. We es ima ed
ou models: specifica ion (2) –analyses he influence o
he basic g a i y a iables (nominal GDP and emo eness)
and addi ional a iables, which include Schengen en y
(Schengen), popula ion (lnpop), and eedom o ade in e -
na ionally (ln i) as a p oxy o ade libe alisa ion – o
expo (a) and impo s (b); and specifica ion (3) –which
includes he index o economic eedom (lnecon ) ins ead
o eedom o ade in e na ionally – o expo (a) and
impo (b). The esul s o he ade libe alisa ion in Model
(2) sugges ha he coefficien s a e in line wi h he expec a-
ions o he s udy and consis en wi h he p edic ions o
7h ps://www.he i age.o g/index/abou
6Ma ina Basa ac Se iće al.
he heo e ical model, bo h when analysing expo s om
C oa ia (a) and when analysing impo s o C oa ia (b).
Namely, he coefficien o nominal GDP is posi i e and
s a is ically significan , indica ing a posi i e impac o GDP
g ow h on expo s/impo s in C oa ia a he 1% le el. Mo e-
o e , he esul s show a nega i e and s a is ically significan
influence o geog aphical dis ance.
In Model (3), ins ead o he a iable eedom o ade
in e na ionally, he a iable index o economic eedom
is included in he analysis, which is also one o he ways
o checking obus ness. This addi ional a iable has an
app op ia e sign and is significan . Hence, we confi m
he obus ness o he es ima ed models p esen ed in se -
e al ways; fi s , ega dless o whe he C oa ian expo s o
EU membe s (a) o impo s om EU coun ies o C oa ia (b)
a e analysed, he main conclusion emains. Also, wi h bo h
p oxies o ade libe alisa ion, i.e. bo h indexes o eco-
nomic eedom, he ob ained esul s hold in all models.
I can be concluded ha all analysed a iables in bo h
a ia ions a e s a is ically significan and o app op ia e
signs. The esul s e e ing o impo s a e e y simila o
hose ob ained o expo s. The only excep ion is he
Schengen a iable in Model (3)b, which has he app o-
p ia e sign bu is no s a is ically significan .
The lowe pa o Table 2 p esen s he esul s o he
diagnos ic es s (e.g. A ellano–Bond es s o au oco ela-
ion o second o de [AR2]). The e was no au oco ela ion
be ween esiduals o he second o de in none o he ana-
lysed sys em GMM models, meaning ha he models we e
alid. The alidi y o he ins umen s selec ed o e alu-
a ing he model was es ed wi h a Sa gan es . Hence, based
on he Sa gan es , he hypo hesis ha no co ela ion exis s
be ween he esiduals and he ins umen s was accep ed.
Table 1: Da a desc ip ion and sou ces
Code Va iable Sou ce Expec ed sign
lnexpo Goods, alue o expo s, ee on boa d (FOB), US dolla s IMF –Di ec ion o T ade S a is ics Posi i e
lnimpo Goods, alue o impo s, cos , insu ance, eigh (CIF), US dolla s IMF –Di ec ion o T ade S a is ics Posi i e
lngdp Nominal GDP (cu en p ices; million eu o) Eu os a Posi i e
ln em GDP (cons an 2015 US$) The Wo ld Bank Nega i e
Dis ance h ps://www.dis ancecalcula o .ne /
ln i F eedom o ade in e na ionally F ase Ins i u e Posi i e
lnecon_ ee Index o economic eedom The He i age Founda ion Posi i e
lnpop Popula ion on 1 Janua y by age and sex Eu os a Posi i e
Sou ce: IMF, The Wo ld De elopmen Indica o s, F ase Ins i u e, The He i age Founda ion, Eu os a .
Table 2: Resul s o he dynamic panel models
Value o me chandise
expo s o C oa ia’s
ading pa ne s a
Value o me chandise
impo s om C oa ia’s
ading pa ne s b
Model (2) Model (3) Model (2) Model (3)
Lagged
dependen
a iable
0.6646*** 0.5850*** 0.6631*** 0.5672***
[0.000] [0.000] [0.000] [0.000]
GDP 0.1713*** 0.3130** 0.1356*** 0.1825**
[0.002] [0.012] [0.001] [0.014]
Remo eness −0.4436* −0.6246** −0.5891** −0.8753**
[0.053] [0.046] [0.022] [0.017]
Schengen 0.2408*** 0.2063*** 0.0548** 0.0471
[0.000] [0.009] [0.049] [0.115]
F eedom o
ade
in e na ionally
2.1165*** 2.3472***
[0.000] [0.000]
Popula ion 0.1315* 0.0315 0.1760*** 0.2417***
[0.079] [0.766] [0.000] [0.000]
Index o
economic
eedom
0.6587** 0.5602*
[0.043] [0.075]
C−4.3422* −1.1720 −3.7394 −0.3252
[0.078] [0.609] [0.143] [0.894]
Numbe o
obse a ions
513 539 513 539
Numbe o
coun ies
27 27 27 27
m2 es (p-
alue)
0.1988 0.2076 0.3884 0.3777
Sa gan es (p-
alue)
0.9274 0.9463 0.8608 0.9162
Sou ce: Au ho s’calcula ions.
No es: *, **, *** indica e s a is ical significance a le els o 10, 5, and 1%;
s anda d e o s a e in b acke s.
Effec s o T ade Libe alisa ion Using a G a i y Model App oach 7
The dependen lagged a iable was s a is ically significan
and had a posi i e algeb aic sign. O e all, based on he
specifica ion es s conduc ed, i can be concluded ha he
es ima es ob ained h ough sys em GMM a e eliable.
5 Discussion
The openness o he C oa ian economy has led o a con-
side able expansion o i s ma ke , esul ing in an imp o e-
men in i s economic pe o mance. The empi ical model
clea ly demons a es ha EU ading pa ne s’GDP g ow h
posi i ely influences C oa ian expo s. The esul s a e
in line wi h he abo e-men ioned s udy pe o med by
Ranilo ić(2017) and confi m he posi i e impac on expo
pe o mance ha was no p e iously ound by D užiće al.
(2011) and Klimczak (2016). This also aligns wi h mac oeco-
nomic heo y, which p edic s ha an inc ease in o eign
incomes inc eases C oa ia’s ne expo s and agg ega e
demand, he eby con ibu ing o GDP g ow h. The ade
libe alisa ion inc eased bo h expo s and impo s, bu
he impo s g ew as e in compa ison wi h expo s be o e
he onse o he global financial c isis. Du ing he obse ed
pe iod, C oa ia eco ded a me chandise ade defici , which
was pa ially offse by a ade su plus in in e na ional se -
ices, excep om 2015 o 2017, when he ne expo s we e
posi i e (CNB, 2022). The C oa ian economy elies hea ily on
he ou ism sec o ha d i es i s high impo dependence,
which weakens i s posi i e impac on he economy. Expo s
o se ices c ea e g ea e demand o he impo o con-
sump ion goods as domes ic p oduc ion is no able o sa is y
he inc easing demand du ing he peak ou is season and o
sa is y p e e ence o o eign-p oduced goods (O sini, 2017).
O sini (2017) ound ha his “leakage effec ”in ou ism e -
enues, which is usually associa ed wi h small-island ou ism
economies, also applies o C oa ia due o he high impo
elas ici y conce ning expo o se ices. To o e come his
issue and educe he “leakage effec ,”he sugges s a less
seasonal pa e n o ou ism, allowing domes ic consump ion
o adjus capaci y o he inc ease in demand, inc easing he
alue o se ices o mo e away om he low pe -capi a
spending, and expanding he ange o se ices wi h lowe
impo con en such as medical and cul u al se ices. The
posi i e esul s achie ed in he ade o goods and se -
ices du ing he pe iod 2015–2017 we e no he esul s o
imp o ed compe i i eness bu he consequence o he
slowe economic eco e y o C oa ia compa ed o o he
EU coun ies, causing he expo s o g ow as e han
he impo s. In he ollowing pe iod, i became ob ious
ha C oa ia could no main ain i s ex e nal balance and
should add ess s uc u al weaknesses in he economy. In
he pe iod 2000–2008, economic g ow h in C oa ia was
d i en by deb capi al inflows mainly om EU coun ies.
Howe e , he in es men s we e no di ec ed owa ds he
adable sec o ; ins ead, hey we e mainly channelled o
financial in e media ion. C oa ia also missed an oppo u-
ni y o in eg a e in o GVC as i was he las membe o join
he EU on 1 July 2013, while mos CEEC benefi ed om
hei membe ship in he EU by a ac ing subs an ial o -
eign di ec in es men , allowing hei as in eg a ion in
GVC, especially au omo i e indus y (O sini, 2017). S ojčić
e al. (2018) also emphasised he impo ance o he iming
o ade libe alisa ion wi h he EU as i affec ed he expo
pe o mance e olu ion and he s uc u e and quali y o
expo s, allowing coun ies ha joined ea lie o eap he
benefi s o p e e en ial access o EU ma ke s and inc ease
he sha e o high echnology-in ensi e indus ies.
D užiće al. (2011) ound ha he EU was insignifican
wi hin he g a i y model, since C oa ia expo ed o (o
impo ed om) he EU coun ies no mo e han o he coun-
ies wi h he same ma ke size and dis ance. Howe e ,
he esul s we e ob ained o he p e-accession pe iod,
using da a o 2008, when he sha e o me chandise ade
wi h he EU was 60%. Ou esul s show ha he EU accession
inc eased o al me chandise ade olume wi h he EU coun-
ies. T ade wi h EU membe s a es in 2021 accoun ed o
abou 70% o o al expo s and abou 77% o o al impo s.
Mo eo e , he s udy unequi ocally e eals ha emo eness
has a nega i e and s a is ically significan impac , indica ing
ha geog aphical dis ance plays a key ole in de e mining
anspo a ion cos s. C oa ia’s la ges ading pa ne s ha e
been Ge many, I aly and Slo enia, wi h Slo enia being he
pa ne coun y o which he mos goods we e expo ed in
2021, while he la ges sha e o goods impo ed in he same
yea came om Ge many (C oa ian Bu eau o S a is ics,
2022), he la ges economy in Eu ope. Mo eo e , hese a e
also he coun ies om which adi ionally he mos ou is s
come o C oa ia, hus ep esen ing also he mos impo an
pa ne s o expo ing C oa ian se ices. The e o e, C oa ia
e u ns pa o i s e enues o he mos impo an ou is
ma ke s by impo ing hei goods. The Schengen en y coe -
ficien was posi i e and also s a is ically significan , as
expec ed. These esul s a e suppo ed in simila s udies,
e.g. B aha e al. (2015) and Ris ano iće al. (2020). Con-
ce ning he o he addi ional a iables –popula ion and
eedom o ade in e na ionally – esul s a e quali a i ely
simila (s a is ically significan and wi h app op ia e sign) o
he baseline model. Howe e , he e a e some diffe ences in
he magni ude o some coefficien s.
8Ma ina Basa ac Se iće al.
