Seck, Abdoulaye; Fall, Foun y Alassane; Aida a, Khadidia ou
A icle
Regional in eg a ion and g ow h spillo e s: Is A ica an
economic space?
Jou nal o A ican T ade
P o ided in Coope a ion wi h:
A ican Expo -Impo Bank (A eximbank), Cai o
Sugges ed Ci a ion: Seck, Abdoulaye; Fall, Foun y Alassane; Aida a, Khadidia ou (2020) : Regional
in eg a ion and g ow h spillo e s: Is A ica an economic space?, Jou nal o A ican T ade, ISSN
2214-8523, A lan is P ess, Pa is, Vol. 7, Iss. 1/2, pp. 37-44,
h ps://doi.o g/10.2991/ja .k.200724.001
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/267527
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-nc/4.0/
Resea ch A icle
Regional In eg a ion and G ow h Spillo e s: Is A ica an
Economic Space?
Abdoulaye Seck*, , Foun y Alassane Fall, Khadidia ou Aida a
Economics Depa men , Cheikh An a Diop Uni e si y, Daka , Senegal
1. INTRODUCTION
The ques o egional in eg a ion has always been a key componen o economic g ow h and de elopmen s a egies in A ica. The expan-
sion o goods ma ke s as a esul o he emo al o ade ba ie s has he po en ial o o e come he many cons ain s associa ed wi h small
na ional ma ke s and p o ide oppo uni ies o echnology ans e s, inc eased p oduc i i y, income gene a ion, as well as job c ea ion.
The li e a u e on he ade-g ow h nexus wi h a ocus on he A ican con ex has shown ha inc eased ade ends o posi i ely impac eco-
nomic g ow h (see, e.g., Fosu, 1990; Musila and Yiheyis, 2015; Zahonogo, 2016). To he ex en ha ade openness is a esul o disc imina o y
ade policies, he li e a u e also ends o lend suppo o he iew ha g ea e ade ies h ough egional in eg a ion lead o economic g ow h.
Fo ins ance, Vam akidis (1998) has indica ed ha egional in eg a ion, a o m o disc imina o y openness, ma e s o economic g ow h,
as “coun ies wi h open, la ge, and mo e de eloped neighbo ing economies g ow as e han hose wi h closed, smalle , and less de eloped
neighbo ing economies” (p. 1). Addi ional e idence on he po en ially posi i e impac o egional ade on g ow h om bo h heo e ical and
empi ical pe spec i es includes he wo k o Dion (2004), who shows ha a educ ion in ansac ion cos s as a esul o he phasing ou o ade
ba ie s can lead o as e g ow h and con e gence in membe coun ies, and Bong and P ema a ne (2018), who ind ha egional in eg a ion
has a signi ican e ec on economic g ow h o Sou heas Asian coun ies, as does e Velde (2011) o a la ge sample o coun ies.
These a ious empi ical assessmen s o he g ow h impac o egional in eg a ion end o all sho o cap u ing he egional spillo e e ec s
o en associa ed wi h s onge economic ies. T ade lows can indeed ac as a channel h ough which economic g ow h in one coun y
bene i s g ow h in i s ading pa ne s. The la e will ha e he oppo uni y o mee an inc easing demand o hei domes ic goods, hence
an expansion in domes ic ou pu and expo s. On he impo side, g ow h in neighbo ing coun ies o en spells a g ea e a ailabili y o
impo ed goods o he domes ic economy, ei he o inal consump ion o domes ic p oduc ion. The same can apply o he case o G oss
Domes ic P oduc (GDP) con ac ion in one coun y ha may also impai g ow h in he ade-dependen neighbo s, in e ms o educed
expo ma ke s and p oduc ion as well as dec eased impo s. T ade in he end ends o ie he o e all economic dynamics o ading pa ne s.
T ade can also ie na ions’ g ow h ajec o ies. I se es as a ehicle o knowledge lows and echnology ans e . Impo ing knowledge-
o echnology-embodied inpu s (mos ly machine y o equipmen ) has been shown o bene i domes ic g ow h. Fo ins ance, Coe and
Helpman (1995) and Kelle (2004) p o ide e idence, in he case o de eloped economies, ha domes ic p oduc i i y is posi i ely ela ed o
ARTICLE INFO
A icle His o y
Recei ed 06 Janua y 2020
Accep ed 15 July 2020
Keywo ds
Regional in eg a ion
ade
g ow h spillo e s
A ica
spa ial dynamic panel
JEL Codes
C23
F15
ABSTRACT
Ma ke in eg a ion is o en associa ed wi h economic g ow h spillo e s, o he ex en ha s eng hened ade linkages would
make economic g ow h in one coun y an oppo uni y o g ow h o i s pa ne s. This pape uses a spa ial dynamic panel da a
app oach o analyze he ex en o which A ica and i s majo egional g oupings ha e been able o gene a e sha ed g ow h. The
esul s indica e ha he con inen as a whole is indeed a spa ially in eg a ed economic zone. Al hough o a weak p opo ion
compa ed o ad anced economies, he con empo aneous spa ial g ow h e ec is associa ed wi h s ong g ow h con e gence and
a la ge a e o e u n o physical capi al. The esul s also sugges ha space has mo e o do wi h bila e al ade han geog aphical
dis ance, and poin o a g ea deal o he e ogenei y: he posi i e g ow h spillo e e ec in egional g oupings such as Sou h
A ican De elopmen Au ho i y, Common Ma ke o Eas e n and Sou he n A ican, and Economic Communi y o Wes
A ican S a es, which happen o egis e la ge in a egional ade sha es, con as wi h he absence o such an e ec in Economic
Communi y o Cen al A ican S a es and A ab Magh eb Union, wo blocks which ha e achie ed li le in e ms o economic
in eg a ion. These esul s should be a welcome addi ion o he deba e o e he oppo uni y o he p oposed con inen -wide ee
ade a ea and i s abili y o os e inclusi e and sus ainable economic g ow h and de elopmen o A ican economies.
© 2020 A ican Expo -Impo Bank. Publishing se ices by A lan is P ess In e na ional B.V.
This is an open access a icle dis ibu ed unde he CC BY-NC 4.0 license (h p://c ea i ecommons.o g/licenses/by-nc/4.0/).
*Co esponding au ho . Email: [email protected]
Pee e iew unde esponsibili y o he A ican Expo -Impo Bank
Jou nal o A ican T ade
Vol. 7(1-2); Decembe (2020), pp. 37–44
DOI: h ps://doi.o g/10.2991/ja .k.200724.001; ISSN 2214-8515; eISSN 2214-8523
h ps://www.a lan is-p ess.com/jou nals/ja
38 A. Seck e al. / Jou nal o A ican T ade 7(1-2) 37–44
impo s om pa ne s ha spend mo e on Resea ch and De elopmen (R&D). Seck (2012) p o ides simila e idence o g ow h-enhancing
echnology di usion h ough ade be ween de eloped and de eloping economies (mo e so han o eign di ec in es men ).
Taking accoun o hese g ow h spillo e e ec s is mo e in line wi h he unde lying mo i e ha go e ns egional in eg a ion. Th ough eco-
nomic in eg a ion, membe coun ies aim o os e con e gen economic ajec o ies ha ensu e sha ed p ospe i y. The expec ed mu ual
bene i s a e o en unde es ima ed by a ypical eg ession analysis ha ela es g ow h o ade.
The e is an eme ging empi ical li e a u e ha s i es o cap u e hese spillo e e ec s, bo owing om spa ial economics. The spillo e
e ec s mean ha , h ough ade linkages, g ow h in one coun y will bene i g ow h in i s ading pa ne s. Empi ical e idence ends o
p o ide ample e idence o g ow h spillo e s h ough ade. Fo ins ance, Ho e al. (2013) ha e shown ha in he con ex o O ganisa ion o
Economic Co-ope a ion and De elopmen (OECD) coun ies, he g ow h spillo e e ec is signi ican as a esul o bila e al ade, al hough
i u ns ou no o be s ongly pe sis en . Igno ing his e ec has been shown o unde es ima e he g ow h con e gence among hese
coun ies. Addi ional e idence o spa ial dependence o na ional economic g ow h includes ha p o ided by Ho e al. (2018) o OECD
coun ies, Sei e al. (2017) o he Middle Eas and No h A ica a ea, Yu (2012) o he US s a es, Tian e al. (2010) o Chinese p o inces,
and E u and Koch (2007) o a la ge se o coun ies, o name a ew.
This esea ch adds o his bu geoning empi ical li e a u e on ade- ela ed spa ial g ow h spillo e by ocusing on he A ican con ex , which
o e s a ue pic u e o a “spaghe i bowl.”1 In e ec , he con inen is made up o eigh majo Regional Economic Communi ies (RECs), and
each one o he 54 coun ies is a membe o a leas one REC.2 These egional g oupings end o di e in many espec s, o ins ance in e ms
o economic o popula ion size, age, and mos ly, main objec i es and ac ual le el o implemen a ion. All o hese end o ansla e in o a ious
le els o ade and economic p oximi y among membe coun ies. Mo e speci ically, mos RECs aim o achie e ull economic union [such
as he Economic Communi y o Wes A ican S a es (ECOWAS), o he Sou he n A ican De elopmen Communi y (SADC)], whe eas
o he s ocus mo e on he ealiza ion o a common ma ke [such as he Common Ma ke o Eas e n and Sou he n A ican (COMESA)].
These a ious objec i es, coupled wi h di e en and slow implemen a ion p ocesses, a e key easons o expec di e en ial g ow h spillo e s.
Taking s ock o he he e ogeneous egional in eg a ion expe iences, in e ms o why some may ha e gene a ed signi ican economic g ow h
ex e nali ies, whe eas o he s ha e ailed o do so, could in o m he design o an e ec i e con inen -wide in eg a ion scheme ha would
p omo e inclusi e and sus ainable de elopmen , which so a has been elusi e o a hos o A ican coun ies.
This s udy aims o assess he egional g ow h po en ial o economic in eg a ion in A ica. Mo e speci ically, i aims o (1) de e mine he
ex en o economic g ow h spillo e h ough egional ade ac oss he con inen , and (2) analyze how a ious speci ici ies o he egional
in eg a ion p ocesses explain he likely he e ogenei y in he abili y o he egional communi ies o gene a e g ow h spillo e s.
The pape de elops a spa ial dynamic panel da a model o cap u e any spa ial dependence o na ional g ow h ajec o ies h ough ade link-
ages. The spa ial weigh ma ix is cons uc ed using bila e al ade lows (and, al e na i ely, he mo e adi ional geog aphical dis ance). The
spa ial dimension is hen added o a s anda d Solow g ow h amewo k ha accoun s o addi ional insigh s om he new g ow h heo y.
The model speci ica ion includes bo h a spa ial au o eg essi e e m and a spa ial ime lag e m.
The esul s sugges s ong e idence o a posi i e g ow h spillo e e ec ac oss he con inen as a whole, and ade u ns ou o be mo e
conduci e o hese spillo e gains han geog aphic p oximi y. Howe e , he spa ial dependence appea s o be less s ong han he one ound
in de eloped coun ies. Mo eo e , accoun ing o he spa ial dependence leads o a g ea e economic con e gence. In addi ion, egional
g oupings such as SADC, ECOWAS, and COMESA appea o be d i ing he esul s, as opposed o he A ab Magh eb Union (AMU) and
Economic Communi y o Cen al A ican S a es (ECCAS), owing o he la ge sha e o in a egional ade, as well as hei signi ican e o
o p omo e p oduc i e in eg a ion ough he de elopmen o egional alue chains, inancial and mac oeconomic in eg a ion ha ensu es
easie con e ibili y o na ional cu encies, egional in as uc u e, as well as ee mo emen o people. All o hese di e en ial aspec s could
se e as key policy di ec ions o u he in eg a e he whole con inen and ensu e inclusi e economic g ow h and de elopmen .
The es o he pape un olds as ollows. Sec ion 2 discusses he li e a u e. Sec ion 3 in oduces he me hodology and he da a. Sec ion 4
epo s and discusses he esul s. Sec ion 5 concludes.
2. LITERATURE REVIEW
Regional in eg a ion has been shown o po en ially gene a e signi ican economic gains. Bu un il ela i ely ecen ly, he li e a u e had
ocused on how a coun y’s ade expansion can bene i i s own economic g ow h ajec o y. Such an expansion can be a esul o alling
a i and non a i ba ie s in he wake o ee ade a angemen s. Mos o he ade heo ies sugges indeed ha in e na ional ade can
lead o inc eased e iciency, economic g ow h, and income gene a ion by allowing g ea e capi al accumula ion, indus ial s uc u e upg ad-
ing, echnological p og ess, and ins i u ional ad ancemen .
Mo e speci ically, a la ge segmen o he empi ical li e a u e has indica ed ha ade has a signi ican g ow h po en ial. Ea ly con ibu ions
ha e shown ha ou wa d-o ien ed de eloping coun ies end o enjoy apid economic g ow h han hei inwa d-o ien ed coun e pa s
1Te m used by Bhagwa i (1995) o e e o he p oli e a ion o p e e en ial ade ag eemen s (PTAs).
2In ac , only 11 coun ies a e membe s o only one REC, whe eas se en a e membe s o h ee RECs (Sou ce: UNECA, h ps://www.uneca.o g/o ia/pages/highligh s-%E2%80%93-a ica-
egional-in eg a ion-index- epo -2016, accessed on Ma ch 1, 2019).
A. Seck e al. / Jou nal o A ican T ade 7(1-2) 37–44 39
(Balassa, 1986; Dolla , 1992), in addi ion o alle ia ing po e y and educing inequali ies (Dolla and K aay, 2004). In ac , in e na ional
ade p o ides a coun y wi h inc eased supply o capi al and in e media e inpu s, which a e no a ailable in he domes ic ma ke s. As a
esul , p oduc i i y o he manu ac u ing sec o can ise (Lee, 1995).
In e na ional ade is also a channel h ough which o eign echnology and knowledge di use o he domes ic economy, which hen ans-
la es in o subs an ial agg ega e p oduc i i y gains and economic g ow h. By impo ing echnology-embodied goods, such as machine y
and equipmen , he domes ic economy acqui es ad anced echnology o he ex en ha i de elops i s abso p ion capabili ies (Coe and
Helpman, 1995; Kelle , 2004; Seck, 2012).
On he expo side, a g ea e pa icipa ion in o eign ma ke s, whe e domes ic i ms ace in e na ional compe i ion, can also lead o an
inc ease in p oduc i i y (Wagne , 2007). Expo s a e also associa ed wi h a lea ning-by-doing e ec h ough which i ms can copy om
mo e ad anced o eign compe i o s and subsequen ly gain in e ms o inc eased e iciency (K aay, 1999).
The e is also ample e idence ha cas s some doub on his posi i e g ow h e ec o ade and openness, as a esul o ins ance o educed
a i s. As a as he pos -wa pe iod is conce ned, au ho s such as Vam akidis (2002),Yanikkaya (2003), and DeJong and Ripoll (2006) ha e
ound a posi i e co ela ion be ween a e age a i le el ( educed ade) and long- e m economic g ow h.3 This is he case when go e n-
men ’s a i a ge s indus ies subjec o ex e nali ies, as sugges ed by K ugman’s (1987) model o lea ning-by-doing and G ossman and
Helpman’s (1993) endogenous g ow h model wi h R&D ex e nali ies.4
These unse led esul s on whe he mo e ade leads o economic g ow h ha e ex ended o he ecen empi ical li e a u e mos ly ocused
on A ica. Fo ins ance, Musila and Yiheyis (2015) ha e indica ed in he case o Kenya ha mo e ade does no , a bes , lead o signi ican
g ow h gains. Mo eo e , hey ha e ound ade-policy-induced openness o ha e nega i ely and signi ican ly a ec ed in es men and he
a e o economic g ow h. By con as , Zahonogo (2016) has shown ha ade has a bene icial impac on g ow h up o a h eshold le el o
openness, abo e which he e ec hen declines. To he ex en ha A ican coun ies end o ade less wi h each o he and wi h he es o
he wo ld (below he c i ical h esholds), inc easing ade can he e o e be good o economic g ow h.
The g ow h spillo e associa ed wi h inc eased ade, howe e , ends o be igno ed in his li e a u e, he eby unde es ima ing he economic
g ow h e ec o inc eased ade ies, as a esul o , o ins ance, egional in eg a ion. A ela i ely ecen li e a u e sough o di ec ly cap u e
he spillo e e ec by bo owing om geog aphy economics. I builds on ea lie a emp s ha ela e a coun y’s g ow h o he a e age
g ow h o i s ading pa ne s. Such an app oach ends o be no sa is ac o y, as i does no accoun o he he e ogenei y o he ade ies
ac oss pa ne s. Ins ead, spa ial app oaches ela e a coun y’s g ow h o indi idual g ow h o i s neighbo s. The links a e ob ained h ough
a weigh ing ma ix cons uc ed by using bila e al ade lows.
This app oach is be e able o e eal he ade po en ial o egional in eg a ion. The spa ial na u e o he economic ela ionships wi hin he
egional g ouping is synonymous wi h some synch oniza ion o he g ow h ajec o ies.
Impo an con ibu ions o his li e a u e include hose p esen ed by Ho e al. (2013). They de eloped an ex ended amewo k o he Solow
g ow h model by conside ing he spillo e e ec due o bila e al ade in he con ex o OECD coun ies. They showed ha ade is indeed
associa ed wi h signi ican g ow h spillo e s, al hough he e ec was no pe sis ence. Accoun ing o his g ow h spillo e e ec also mean
ha he a e con e gence was highe han o he wise. Fu he mo e, Ho e al. (2018) p o ided simila e idence in he con ex o ad anced
coun ies. Addi ional con ibu ions wi h a ocus on de eloping coun ies include hose p o ided by Tian e al. (2010) o China and Sei
e al. (2017) o A ab coun ies in he Middle Eas and No h A ica.
This pape b ings o his li e a u e he unique con ex o he A ican con inen , which is a place whe e egional in eg a ion has always been
a signi ican pa o he economic p ocesses. This has esul ed in o a dense web o economic in eg a ion a angemen s. Despi e his s ong
appe i e o economic in eg a ion, A ica is also a place ha ades less wi h i sel . The compa a i e app oach ha looks a he egional
speci ics will be able o shed ligh on he a ious mechanisms ha unde lie he p ocess o spa ially gene a ed g ow h spillo e s associa ed
wi h egional in eg a ion.
3. METHODOLOGY AND DATA
The s a ing poin is he Solow g ow h model wi h a Cobb–Douglas p oduc ion YK
i i i
=
()
-
a
a
AL 1, 0 < a < 1, wi h he assump ions o exog-
enous g ow h a e o he labo -augmen ing echnological p og ess g and o labo (o popula ion) n. I can be de i ed using he ollowing
s eady-s a e equilib ium o he g ow h a e, widely used in he g ow h empi ics li e a u e (see, e.g., Mankiw e al., 1992; Islam, 1995; and
Ba o, 1996):
Dyeyeng es
i
c
i
c
i
c
=- --
-
-
++ +-
-
-
--
()ln ()
ln()
()
11
1
1
1
0
a
a
da
a
ln ii
c
i
eAg +- +
-
() ,1 0
ln (1)
3On he opposi e side o he li e a u e on a i and long- e m g ow h, Edwa ds (1992, 1998), Clemens and Williamson (2004), and Nunn and T e le (2010) p o ide e idence o a nega i e
co ela ion.
4See Nunn and T e le (2010) o a de ailed discussion.
40 A. Seck e al. / Jou nal o A ican T ade 7(1-2) 37–44
whe e y ep esen s eal GDP pe capi a, A deno es echnology, d is he dep ecia ion a e o he capi al s ock (assumed o be cons an ac oss
coun ies), c is he con e gen a e ac oss coun ies, and si is he sa ing a e (o en p oxied by he in es men a e). I is gene ally assumed
ha g + d = 5%, as in he wo k o Mankiw e al. (1992), Islam (1995), o Ho e al. (2013).
To elax he assump ion o closed economy and allow o coun ies o in e ac h ough ade, we conside he ollowing equa ion, aug-
men ed wi h spa ial e ms:
ln ln ln lnyy wy wy ng
i i j
N
ij j j
N
ij i i
=+ ++++
-==
-
åå
gl b
111
11
ln(ddb bbmhe)()++++ +++
23 4
1ln ln In lnIns si i i i i (2)
In i is he in la ion a e, an indica ion o a mac oeconomic s abili y, and Ins i is he ins i u ional quali y, which “ e lec s pe cep ions o he
abili y o he go e nmen o o mula e and implemen sound policies and egula ions ha pe mi and p omo e p i a e sec o de elopmen .”5
To he ex en ha bo h economic s abili y and e ec i e and sound go e nance a ec he pace o echnological p og ess, hey can d i e as e
economic g ow h. mi and h a e coun y and ime ixed e ec s, and ei is he e o e m. By conside ing a 2-yea in e al o e he pe iod o
s udy (2000–2016), he common con e gence a e is hen gi en by c = −ln g /2.6 The spa ial weigh ma ix W, whose elemen s a e wij, allows
handle s o cap u e bo h con empo aneous j
N
ij j
wy
=
å
()
1ln and lagged j
N
ij i
wy
=-
å
()
11
ln g ow h spillo e s o coun y i om i s ading
pa ne s j. The bila e al ade (sum o expo s and impo s), a e a e aged o e he en i e pe iod.7
As a i s app oxima ion, we ollow he s anda d li e a u e ha uses spa ial econome ics o add ess a ious economic issues and cons uc
W using geog aphical dis ance. This allows us o see whe he spa ial in e dependence ac oss coun ies occu s “na u ally,” as i may ela e
o an exogenous p oximi y. The elemen s a e hen ob ained om wij = exp(−dij), wi h dij he dis ance be ween coun ies i and j, o mo e
p ecisely hei capi al ci ies (see Kelle , 2004; E u and Koch, 2007; and Ho e al., 2013, o applica ions in he li e a u e).
We nex conside bila e al lows ij, and he elemen s o he spa ial weigh ma ix a e simply wij = ij, a e aged o e he pe iod ( ime-
in a ian ). This ade- ela ed ma ix could p o ide oom o al e ing he economic p oximi y h ough mo e o less ade. Ho e al. (2013)
also conside ed bo h de ini ions o p oximi y, and showed ha bo h a e associa ed wi h spa ial co ela ion, wi h a g ea e spillo e e ec
h ough he ade channel.
The es ima ion s a egy, based on a quasi-maximum likelihood es ima e app oach, consis s o i s unning he eg essions o all A ican
coun ies, and hen sepa a ely o he majo egional blocs, namely, COMESA, AMU, ECCAS, ECOWAS, and SADC.8 These egional
communi ies span he whole con inen . The emaining egional communi ies end o ei he o e lap [e.g., Wes A ican Economic and
Mone a y Union wi h espec o ECOWAS, and In e go e nmen al Au ho i y on De elopmen and he Eas A ican Communi y (EAC)
o COMESA—wi h he excep ion o Somalia, Sou h Sudan, and Tanzania], o oo la ge and he e ogeneous [Communi y o Sahel–Saha an
S a es wi h 24 membe coun ies]. The pooling o all he coun ies will ell how, on a e age, one coun y’s g ow h is bene icial o he neigh-
bo ing coun ies. The sepa a e eg essions will help e eal he speci ici ies ha make a gi en in eg a ion p ocess mo e p one o gene a e
g ow h spillo e han he es .
Table 1 o e s some desc ip i e s a is ics o he da a. I appea s ha he con inen is ading li le wi h i sel , despi e he dense web o ee
ade ag eemen s. Ne e heless, he da a clea ly indica e posi i e ends o all egional g oupings. In e ec , hey a e imp o ing hei ade
ies, as he sha e o in a egional ade is ising. SADC and COMESA ha e seen he la ges inc ease, wi h 8.4% and 5.9% poin s, espec i ely.
On he opposi e end, egions wi h he smalles inc ease (AMU and ECCAS) u n ou o be hose whe e in a egional ade is he lowes .
5Sou ce: Wo ld Bank (h p://in o.wo ldbank.o g/go e nance/wgi/#home, accessed on Ma ch 4, 2019).
6A mo e common choice in he li e a u e is a 5-yea in e al. Bu gi en ou ime span (2000–2016) and he size o ou egional g oupings (which go o as low as i e membe s o AMU),
he closes we can ge in sa is ying he assump ion ha he numbe o indi iduals N (membe s a es) is always g ea e han he ime dimension T in ou panel on one hand, while ha ing
su icien ly la ge numbe o o al obse a ions on he o he hand, is by choosing a 2-yea window.
7Fo many coun y pai s, howe e , he pe iod is sho e han 2000–2016, because o ade da a una ailabili y.
8Fo a de ailed discussion o he quasi-maximum likelihood es ima e app oach and i s applica ion using S a a, see Belo i e al. (2016).
Table 1 | Desc ip i e s a is icsa,b
A ica COMESA AMU ECCAS EC OWAS SADC
GDP p.c. 2016 (cons an 2010$) 2693.6 2818.8 3383.9 3402.5 1093.6 3864.3
GDP p.c. g ow h, 2000–2016 (%) 2.1 1.5 1.7 2.2 1.6 2.4
In a egional ade, 2016 (%) 12.6 12.4 3.4 4.2 10.8 19.2
In a egional ade, change 2000–2016 1.4 5.9 0.8 1.1 1.9 8.4
A ea (million sq. km) 30.8 12.0 5.8 6.5 5.1 10.0
# o membe coun ies 54 19 5 11 15 15
A ea/membe 0.57 0.63 1.16 0.59 0.34 0.67
Popula ion g ow h, 2000–2016 (%) 2.5 2.3 1.9 3.0 2.7 2.2
In es men a e, 2000–2016 (% GDP) 22.5 18.7 27.0 22.9 20.6 22.3
In la ion (GDP de la o %, 2016, and change om 2000) 10.1 (–79) 5.0 (176) 1.9 (–7) 4.1 (–320) 3.8 (–8) 8.7 (–236)
Regula ion (2016, and change) 3.43 (–0.22) 3.31 (–0.08) 3.04 (–0.79) 3.07 (0.25) 3.76 (–0.37) 4.09 (–0.25)
aThe sco es o he egula ion quali y, ini ially anging om –2.5 (weak) o +2.5 (s ong), ha e been escaled o 0–10. bSou ce: Au ho s’ calcula ions, om Wo ld Bank, Wo ld
De elopmen Indica o s (WDI) and Wo ld Go e nance Indica o s (WGI), Uni ed Na ions Commodi y T ade (COMTRADE) Da abase and UNECA.

A. Seck e al. / Jou nal o A ican T ade 7(1-2) 37–44 41
A ab Magh eb Union, in addi ion o eco ding he smalles sha e and slowes pace o in a egional ade, also u ns ou o be ela i ely
la ge in e ms o geog aphical size as i ela es o he numbe o membe coun ies. The igu e o 1.16 million squa e miles (sq.m.) is an
indica ion o he ela i ely long dis ance a good p oduced in a ypical membe coun y has o a el in o de o ge o he na ional bo de
and be aded wi h a membe pa ne . By con as , he g oup ha ades he mos wi h i sel (SADC) is also he place whe e geog aphical
size is ela i ely smalle (0.67 million sq.m., e sus 0.34 million sq.m. o ECOWAS). This is a sugges ion ha size (o dis ance) migh be a
po en ial p edic o o c oss- egional ade.
A a as economic g ow h is conce ned, he associa ion wi h in a egional ade seems less ob ious. In ac , he egion wi h he la ges sha e
o and inc ease in in a egional ade (SADC) is also he one ha has pos ed he la ges GDP pe capi a g ow h a e o e he 16-yea pe iod
(2.4%), bu he second la ges egion, in a- ade-wise (COMESA), has expe ienced he slowes g ow h a e o 1.5%. Fu he mo e, ECCAS,
which is he second less ade-in eg a ed egion, has pos ed he second la ges g ow h a e o 2.2%.
These igu es do no e eal a de ini e, clea end as o he ex en o which domes ic g ow h could bene i om neighbo ing coun ies’
g ow h, ei he h ough ade o h ough geog aphical dis ance, as wi h he associa ion be ween g ow h and in la ion on one hand, and
g ow h and go e nance quali y, on he o he hand. A mo e obus analysis should come up wi h a be e ep esen a ion o spa ial depen-
dency h ough a spa ial weigh ma ix, bu also design an al e na i e way h ough which g ow h spillo e e ec s could ma e ialize. The
amewo k would hen allow g ow h in one coun y o bene i g ow h in neighbo ing coun ies as a esul o g ea e ade (o geog aphical)
p oximi y, while ac ually con olling o ac o s ele an o g ow h and echnological p og ess a he s eady s a e equilib ium.
4. RESULTS AND DISCUSSION
Table 2 shows he esul s o a ious model speci ica ions o he con inen as a whole. Going om a s anda d dynamic panel speci ica ion
(column 1) o he augmen ed one wi h a spa ial au o eg essi e e m (columns 2 and 3), he associa ed coe icien (λ) appea s o be signi -
ican . This sugges s ha esul s om he i s s anda d model ha igno es his ele an spa ial dependence a e likely o be biased, on one
hand, and ha A ica as whole is indeed an e ec i ely in eg a ed economic space o he ex en ha i allows economic g ow h ex e nali ies
ac oss coun ies, on he o he hand. G ow h in one coun y bene i s g ow h in pa ne coun ies, as i is he case o OECD coun ies
(Ho e al., 2013), o OECD and de eloping coun ies (E u and Koch, 2011 and 2007), and o US s a es (Yu and Lee, 2012).
Ou esul s also sugges ha bo h ade and physical dis ance a e conduci e o g ow h spillo e s ac oss A ica. Compa ing be ween hese
wo channels, he spa ial dependence appea s o occu mo e h ough ade han h ough geog aphical dis ance, sugges ing ha neighbo -
hood o p oximi y is mo e ele an as an economic concep han a physical concep , in he A ican con ex . In e ec , he coe icien es ima e
on he spa ial ime lag e m is highe when he spa ial weigh ma ix is cons uc ed by using bila e al ade han physical dis ance. Mo eo e ,
he common con e gen a e o 7.9% is highe when ade- ela ed spa ial weigh is conside ed han weigh cons uc ed h ough physical
dis ance (5.4%), and e en when igno ing spa ial dependence (5.7%). Fu he mo e, he es ima ed elas ici y o ou pu wi h espec o capi al
(a) is also highe a 66.9% o he ade-weigh ed spa ial model.
These esul s appea o be gene ally consis en wi h he li e a u e al hough wi h some no iceable di e ences. The highe con e gence a e
associa ed wi h he inclusion o spa ial dependence in s anda d models is also ound by Ho e al. (2013) o OECD coun ies, al hough ou
Table 2 | Es ima ion esul s o he whole A ican con inen a
Va iables – coe icien s (1) (2) (3) (4) (5)
ln(yi –1) — (g)0.893*** (0.08) 0.898* (0.49) 0.854* (0.48) 0.882*** (0.08) 0.873*** (0.08)
ln(ni + 0.05) — (b1)–0.178* (0.13) –0.182** (0.08) –0.276** (0.14) –0.263*** (0.04) –0.259* (0.14)
ln(si ) — (b2)0.078 (0.08) 0.192** (0.09) 0.296*** (0.08) 0.223** (0.09) 0.079 (0.09)
ln(1 + In i ) — (b3)–0.043** (0.02) –0.039* (0.02) –0.038* (0.02) –0.046** (0.02) –0.047** (0.02)
ln(Ins i ) — (b4)0.068 (0.08) –0.059 (0.08) –0.062 (0.09) 0.035 (0.07) –0.029 (0.08)
l– (–) 0.015* (0.08) 0.237* (0.14) 0.153* (0.09) 0.286** (0.11)
– (–) – (–) – (–) –0.068 (0.16) 0.142 (0.09)
N46 46 46 46 46
T88888
Obs. 368 368 368 368 368
Implied a0.421 0.653 0.669 0.653 0.384
Con e gence a e 0.057 0.054 0.079 0.063 0.068
Spa ial weigh s No Geog. dis . T ade Geog. dis . T ade
Coun y FE Yes Yes Yes Yes Yes
Yea FE Yes Yes Yes Yes Yes
aThe dependen a iable is log o eal GDP pe capi a. The common con e gence a es (c) a e ob ained om g = exp(–c ), wi h = 2 ( he ime in e al), and he elas ici y o ou pu wi h
espec o capi al (a) om ba
ag
2
1
1=
−
−
()
. The sum o he exogenous a e o echnological p og ess and he capi al dep ecia ion a e is assumed o be equal o 5%. Columns (2) and (3)
in oduce he spa ial au o eg essi e e m, whe eas columns (4) and (5) add he spa ial lag e m. To al bila e al ade lows (sum o expo s and impo s) a e used o cons uc he ade-
ela ed spa ial weigh ma ix which is hen ow-s anda dized, as well he dis ance- ela ed ma ix. All eg essions include bo h coun y and ime ixed e ec s. S anda d e o s a e be ween
pa en heses, and signi icance a 1%, 5%, and 10% is deno ed by ***, **, and *, espec i ely. FE, ixed e ec s.
42 A. Seck e al. / Jou nal o A ican T ade 7(1-2) 37–44
Table 3 | Es ima ion esul s o egional g oupingsa
Va iables – coe icien s COMESA AMU ECCAS EC OWAS SADC
ln(yi –1) — (g)0.687*** (0.04) 0.785 (0.90) 0.892*** (0.24) 0.862*** (0.25) 0.754*** (0.15)
ln(ni + 0.05) — (b1)0.193* (0.11) –0.283 (0.56) –0.031 (0.09) –0.248** (0.11) –0.252*** (0.09)
ln(si ) — (b2)0.040 (0.03) 0.397 (0.20) 0.104** (0.04) 0.294* (0.17) 0.293*** (0.07)
ln(1 + In i ) — (b3)–0.018 (0.02) 0.006 (0.18) –0.023** (0.01) –0.039* (0.02) –0.036* (0.02)
ln(Ins i ) — (b4)0.042 (0.03) –0.24 (0.85) 0.068** (0.03) 0.035 (0.06) 0.052 (0.12)
l0.216* (0.12) 0.193 (0.43) –0.038 (0.18) 0.284*** (0.10) 0.263*** (0.04)
N14 5 9 15 12
T88888
Obs. 112 40 72 120 96
Implied a0.113 0.580 0.491 0.681 0.544
Con e gence a e c0.188 0.121 0.057 0.074 0.141
Spa ial weigh s WT ade T ade T ade T ade T ade
Coun y FE Yes Yes Yes Yes Yes
Yea FE Yes Yes Yes Yes Yes
aThe dependen a iable is log o eal GDP pe capi a. In all eg essions, he elemen s o spa ial weigh ma ix a e o al bila e al impo and expo lows. The eg essions also include bo h
coun y and ime ixed e ec s. S anda d e o s a e be ween pa en heses, and signi icance a 1%, 5%, and 10% is indica ed by ***, **, and *, espec i ely.
es ima ed con e gence a e appea s o be much la ge (7.9% agains 5.2% in he case o OECD coun ies, o he ade-weigh ed spa ial model).
Like Ho e al. (2013), he no ion o neighbo hood is mo e associa ed wi h ade han pu e geog aphical dis ance, as a as g ow h spillo e s a e
conce ned. Howe e , hei es ima ed spa ial au o eg essi e coe icien is mo e han wice as much as ou es ima e (0.589 agains 0.237). This is
an indica ion ha he A ican con inen is less o an in eg a ed economic space han he OECD g oup when i comes o ade- ela ed g ow h
ex e nali ies. This is because hese ad anced economies ade mo e wi h one ano he (65% wi hin he EU and 50% be ween US, Mexico, and
Canada in 2015), and he p oduc ion base (GDP) is much la ge and allows o mo e expo s and impo s, compa ed o A ican economies.9
Fu he mo e, he gene al p emise ha mac oeconomic s abili y is good o economic g ow h is suppo ed by he esul s, as educed in la-
ion is associa ed wi h inc eased eal GDP pe capi a. Ins i u ional quali y, cap u ed by he quali y o go e nmen egula ion, howe e , does
no seem o be signi ican ly co ela ed wi h economic g ow h. This could be an indica ion o e ec i e coping mechanisms in he ace o an
un iendly (business) en i onmen , o a lack o imp o emen in public go e nance (in ac , Table 1 indica es a sligh ly de e io a ing ins i u-
ional quali y du ing he ime pe iod).
When he spa ial ime lag is in oduced (columns 4 and 5), he associa ed coe icien ( ) u ns ou o be insigni ican , ega dless o how he spa ial
weigh ma ix is cons uc ed. Al hough he inclusion o an i ele an spa ial ime lag e m ends no o be associa ed wi h an ob ious e iciency
loss, as sugges ed by Yu (2012), we p e e d opping ha a iable. In he end, g ow h spillo e s happen con empo aneously: domes ic g ow h in
a gi en yea would bene i om g ow h happening he same yea in neighbo ing coun ies, no om g ow h in he pas (2 yea s). We he e o e
a o model speci ica ion (3), Table 2—con empo aneous spa ial au o eg essi e e m, wi h ade- ela ed spa ial weigh ma ix— o see whe he
his b oad con inen al pic u e o spa ially in eg a ed economies may be hiding some he e ogenei y among he componen egional blocs.
As one would ha e expec ed, Table 3 shows ha egional g oupings ha happen o be ading mo e wi hin hemsel es (COMESA, ECOWAS,
and SADC) appea o be d i ing he esul s in Table 2, as hei combined membe coun ies make up mos o he con inen (87%). In e ec ,
only o hese h ee g oupings is he ade- ela ed spa ial dependence signi ican ly posi i e.10 The combina ion o (1) ela i ely high eco-
nomic g ow h, (2) s onge ade ies which ha e inc eased mo e han anywhe e else on he con inen , and (3) a ela i ely mode a e size o
geog aphical space, which sugges s ha aded goods ha e o a el ela i ely smalle dis ances o each na ional bo de s, could explain why
hese en i ies a e mo e in eg a ed economic spaces, inso a as g ow h spillo e s h ough ade a e conce ned.
Acco ding o he A ican Union Commission–A ican De elopmen Bank–Uni ed Na ions Economic Commission o A ica (AU–A DB–
UNECA) measu emen s o egional in eg a ion ac oss A ica, bila e al ade has been he dimension along which A ica as a whole has
made mo e e o , wi h a sco e o 0.54, as o 2016.11 This is a esul o educed cus oms du ies as well as inc eased in a egional ade sha es.
In he opposi e, a eas whe e economic in eg a ion has been lagging ac oss he con inen is inancial and mac oeconomic in eg a ion. The
a e age sco e o 0.38 sugges s ha , as a as ade and g ow h spillo e s a e conce ned, egional con e ibili y o na ional cu ency is s ill an
issue, o i wo sens unce ain y and adds o he al eady high ade ansac ion cos s, he eby educing ade lows.
Whe eas he egional g oupings ha e been making almos simila p og ess on egional in as uc u e (such as in a egional ligh s and
oaming cos s) and p oduc ion in eg a ion (such as ade on in e media e inpu s and ade complemen a i y), hey end o a e di e en ly
on he on s o ee mo emen o people and inancial and mac oeconomic in eg a ion, wi h ECOWAS a ing be e on bo h dimensions.
9Mo eo e , ou implied e u n o capi al, p oxied by he elas ici y o ou pu wi h espec o capi al, is much highe (66.9% agains 51.7%), which could be indica i e o he ela i ely sca ce
capi al s ock h oughou he con inen as opposed o he mo e ad anced economies ha make up he OECD.
10The esul s o AMU seem e y imp ecise and ine icien , as sugges ed by he la ge s anda d e o s, owing ce ainly o he smalle sample size as well as o he imbalance be ween he wo
dimensions o he panel s uc u e (T exceeds N). We also a gue ha di e ences in he esul s a e a ansla ion o di e en ia ed in eg a ion p ocesses.
11The o e all index elies on i e dimensions: ade in eg a ion, egional in as uc u e, p oduc i e in eg a ion, ee mo emen o people, and inancial and mac oeconomic in eg a ion.
The sco e anges om 0 (lowes ) o 1 (highes ). The 2016 epo can be ound he e: h p://www. alac.o g/images/docs/9384/a ica- egional-in eg a ion-index- epo -2016.pd (accessed
on Ma ch 1, 2019).
A. Seck e al. / Jou nal o A ican T ade 7(1-2) 37–44 43
The ela ionship be ween g ow h and ins i u ions on one hand, and wi h in la ion on he o he , wi hin he subg oups end o gene ally
e lec he o e all indings o he con inen as a whole, shown in Table 2. In e ec , highe in la ion ends o be associa ed wi h, a bes , no
signi ican change in economic g ow h, whe eas ins i u ional quali y change does no signi ican ly a ec g ow h, excep o ECCAS membe
coun ies whe e he impac is mo e in line wi h he gene al inding in he ele an li e a u e on a posi i e impac o go e nance quali y
imp o emen . This could be because only his egion has seen an imp o emen in he a e age sco e o e he ime pe iod.
O e all, ECOWAS and SADC ha e egis e ed he highes a e age sco es, mos ly when i comes o ee mo emen o people, inancial and mac-
oeconomic in eg a ion, and egional in as uc u e. COMESA has a ed ela i ely well on he dimension o ade and p oduc i e in eg a ion.
The esul s showing hese h ee egional blocs being mo e conduci e o g ow h spillo e h ough ade a e ano he ansla ion o hei ela i ely
g ea e e o o in eg a e membe economies. The esul s a e also an indica ion ha , in addi ion o inc eased ade lows and egional alue
chains, ee mo emen o people, inancial and mac oeconomic in eg a ion, and egional in as uc u e a e key o gene a ing a mo e in eg a ed
economic space ha would allow economic g ow h in any domes ic economy o bene i economic g ow h in i s ading pa ne s.
5. SUMMARY AND CONCLUSION
This pape was conce ned wi h he abili y o egional in eg a ion p ocesses ac oss A ica o gene a e g ow h spillo e h ough ade, especially
in he p ospec o he A ican Con inen al F ee T ade A ea (A CFTA). Using a spa ial dynamic panel da a model, he pape showed ha A ica
is indeed a place whe e na ional economic g ow h bene i s om g ow h in neighbo ing coun ies, mo e so h ough ade linkages han h ough
geog aphical p oximi y. Howe e , he gene a ed spillo e gains appea ed o be o a small magni ude when compa ed o economic spaces o
mo e ad anced coun ies, such as OECD o US s a es. The esul s also indica ed ha his o e all pic u e hides a g ea deal o he e ogenei y
ac oss RECs, wi h SADC, ECOWAS, and COMESA egis e ing signi ican , posi i e spillo e e ec , whe eas g oupings such as ECCAS and
AMU show no signi ican spa ial co ela ion, owing mainly o he di e en ia ed leng h a which hey ha e gone o imp o e bila e al ade lows.
Fo A CFTA ag eemen s o p oduce inclusi e and sus ainable economic g ow h and de elopmen , he spa ial dependence o economic
en i ies needs o be u he s eng hened, so ha na ional g ow h ajec o ies could be ein o cing one ano he , in line wi h he solida i y
spi i ha go e ns his collec i e ini ia i e. To ha end, ade ies ha e o be u he s eng hened, o ins ance, by con inuously educing
ade impedimen s (pa icula ly non a i ba ie s ha end o be le ou o majo a angemen s). Ha monized indus ial policies should
also con ibu e o he de elopmen o egional alue chains, on he one hand, whe eas inancial and mac oeconomic policies should ensu e
less bu densome con e ibili y among he oo many na ional cu encies on he o he hand, wi h he goal o u he educing he high c oss-
bo de ansac ion cos s and inc easing he g ow h-spillo e conduci eness o ade ac oss he con inen .
CONFLICTS OF INTEREST
The au ho s decla e hey ha e no con lic s o in e es .
AUTHORS’ CONTRIBUTION
Abdoulaye Seck is he p incipal au ho , wi h esea ch assis ance by Foun y Fall and Khadidia ou Aida a.
REFERENCES
Balassa B., 1986. The employmen e ec s o ade in manu ac u ed p oduc s be ween de eloped and de eloping coun ies, J. Pol. Mod. 8, 371–390.
Ba o R.J., 1996. De e minan s o economic g ow h: a c oss-coun y empi ical s udy, NBER Wo king Pape No. 5698. A ailable a SSRN: h ps://
ss n.com/abs ac =3422.
Belo i F., Hughes G., Mo a i A.P., 2016. Spa ial panel da a models using S a a, CEIS Wo king Pape No. 373. Re ie ed om h ps://pape s.ss n.
com/sol3/pape s.c m?abs ac _id=2754703.
Bhagwa i J.N., 1995. US ade policy: he in a ua ion wi h FTAs. Discussion Pape Se ies No. 726, Depa men o Economics, Columbia Uni e si y,
New Yo k, N.Y.
Bong A., P ema a ne G., 2018. Regional in eg a ion and economic g ow h in Sou heas Asia, Global Bus. Re . 19, 1403–1415.
Clemens M.A., Williamson J.G., 2004. Why did he a i -g ow h co ela ion change a e 1950? J. Econ. G ow h 9, 5–46.
Coe D.T., Helpman E., 1995. In e na ional R&D spillo e s, Eu . Econ. Re . 39, 859–887.
DeJong D.N., Ripoll M., 2006. Ta i s and g ow h: an empi ical explo a ion o con ingen ela ionships, Re . Econ. S a . 88, 625–640.
Dion D.P., 2004. Regional in eg a ion and economic de elopmen : a heo e ical app oach, Discussion Pape Se ies o SFB/TR 15 Go e nance
and he E iciency o Economic Sys ems 20, F ee Uni e si y o Be lin, Humbold Uni e si y o Be lin, Uni e si y o Bonn, Uni e si y o
Mannheim, Uni e si y o Munich, Ge many.
Dolla D., 1992. Ou wa d-o ien ed de eloping economies eally do g ow mo e apidly: e idence om 95 LDCs, 1976–1985, Econ. De . Cul .
Change 40, 523–544.
44 A. Seck e al. / Jou nal o A ican T ade 7(1-2) 37–44
Dolla D., K aay A., 2004. T ade, g ow h and po e y, Econ. J. 114, F22–F49.
Edwa ds S., 1992. T ade o ien a ion, dis o ions and g ow h in de eloping coun ies, J. De . Econ. 39, 31–57.
Edwa ds S., 1998. Openness, p oduc i i y and g ow h: wha do we eally know? Econ. J. 108, 383–398.
E u C., Koch W., 2007. G ow h, echnological in e dependence and spa ial ex e nali ies: heo y and e idence, J. Appl. Econ. 22, 1033–1062.
E u C., Koch W., 2011. A con ibu ion o he heo y and empi ics o Schumpe e ian g ow h wi h wo ldwide in e ac ions, J. Econ. G ow h 16,
215–255.
Fosu A.K., 1990. Expo s and economic g ow h: he A ican case, Wo ld De . 18, 831–835.
G ossman G.M., Helpman E., 1993. Inno a ion and G ow h in he Global Economy, MIT P ess, Camb idge, MA, p. 359.
Ho C.Y., Wang W., Yu J., 2013. G ow h spillo e h ough ade: a spa ial dynamic panel da a app oach, Econ. Le . 120, 450–453.
Ho C.Y., Wang W., Yu J., 2018. In e na ional knowledge spillo e h ough ade: a ime- a ying spa ial panel da a app oach, Econ. Le . 162, 30–33.
Islam N., 1995. G ow h empi ics: a panel da a app oach, Q. J. Econ. 110, 1127–1170.
Kelle W., 2004. In e na ional echnology di usion, J. Econ. Li . 42, 752–782.
K aay A., 1999. Expo a ions e pe o mances économiques: e ude d’un panel d’en ep ises chinoises, Re ue d’Economie du Dé eloppemen , 1-2, 183–207.
K ugman P., 1987. The na ow mo ing band, he Du ch disease, and he compe i i e consequences o M s. Tha che : no es on ade in he p esence
o dynamic scale economies, J. De . Econ. 27, 41–55.
Lee J., 1995. Compa a i e ad an age in manu ac u ing as a de e minan o indus ializa ion: The Ko ean case, Wo ld De elop. 23, 1195–1214.
Mankiw N.G., Rome D., Weil D.N., 1992. A con ibu ion o he empi ics o economic g ow h, Q. J. Econ. 107, 407–437.
Musila J.W., Yiheyis Z., 2015. The impac o ade openness on g ow h: he case o Kenya, J. Policy Model 37, 342–354.
Nunn N., T e le D., 2010. The s uc u e o a i s and long- e m g ow h, Am. Econ. J. Mac oecon. 2, 158–194.
Seck A., 2012. In e na ional echnology di usion and economic g ow h: explaining he spillo e bene i s o de eloping coun ies, S uc . Chang.
Econ. Dyn. 23, 437–451.
Sei A.M., Panahi H., Razi D.H., 2017. Regional economic g ow h and spa ial spillo e e ec s in MENA a ea, I an. Econ. Re . 21, 765–787. Re ie ed
om h ps://ideas. epec.o g/a/eu /jou nl/ 21y2017i4p765.h ml.
e Velde D.W., 2011. Regional in eg a ion, g ow h and con e gence, J. Econ. In eg . 26, 1–28.
Tian L., Holly Wang H., Chen Y., 2010. Spa ial ex e nali ies in China egional economic g ow h, China Econ. Re . 21, .
Vam akidis A., 1998. Regional in eg a ion and economic g ow h, Wo ld Bank Econ. Re . 12, 251–270.
Vam akidis A., 2002. How obus is he g ow h-openness connec ion? His o ical e idence. J. Econ. G ow h 7, 57–80.
Wagne J., 2007. Expo s and p oduc i i y: a su ey o he e idence om i m‐le el da a, Wo ld Econ. 30, 60–82.
Yanikkaya H., 2003. T ade openness and economic g ow h: a c oss-coun y empi ical in es iga ion, J. De . Econ. 72, 57–89.
Yu H., 2012. The in luen ial ac o s o China’s egional ene gy in ensi y and i s spa ial linkages: 1988–2007, Ene gy Policy 45, 583–593.
Yu J., Lee L.F., 2012. Con e gence: a spa ial dynamic panel da a app oach, Global J. Econ. 1, 1250006.
Zahonogo P., 2016. T ade and economic g ow h in de eloping coun ies: e idence om sub-Saha an A ica, J. A . T ade 3, 41–56.
ANNEXES
A. Da a Sou ces
Bila e al ade da a a e ob ained om COMTRADE. Geog aphical dis ances a e p o ided by CEPII. Mac oeconomic a iables ( eal GDP
pe capi a, and in es men a e) and popula ion g ow h a e come om he Wo ld Bank’s Wo ld De elopmen Indica o s, and da a on eg-
ula ion quali y a e om he Wo ld Bank’s Wo ld Go e nance Indica o s. A ea size and o he ele an in o ma ion ega ding he con inen
and he egional g oupings a e om UNECA (h ps://www.uneca.o g/o ia/pages/ egional-economic-communi ies).
B. Regional G oupings and Coun y Lis s
COMESA (Common Ma ke o Eas e n and Sou he n A ica): Bu undi, Como os, Democ a ic Republic o Congo, Egyp , E hiopia,
Kenya, Libya, Madagasca , Malawi, Mau i ius, Rwanda, Uganda, Zambia, and Zimbabwe (exclusion o Djibou i, E i ea, Seychelles, Sudan
and Swaziland).
AMU (A ab Magh eb Union): Alge ia, Libya, Mau i ania, Mo occo, and Tunisia.
ECCAS (Economic Communi y o Cen al A ican S a es): Angola, Bu undi, Came oon, Cen al A ican Republic, Congo, Democ a ic
Republic o he Congo, Equa o ial Guinea, Gabon, and Rwanda (exclusion o Chad and Sao Tome and P incipe).
ECOWAS (Economic Communi y o Wes A ican S a es): Benin, Bu kina Faso, Cabo Ve de, Co e d’I oi e, Gambia, Ghana, Guinea,
Guinea Bissau, Libe ia, Mali, Nige , Nige ia, Senegal, Sie a Leone, and Togo.
SADC (Sou he n A ican De elopmen Communi y): Angola, Bo swana, Democ a ic Republic o Congo, Madagasca , Malawi, Mau i ius,
Mozambique, Namibia, Sou h A ica, Tanzania, Zambia, and Zimbabwe (exclusion o Leso ho, Seychelles, and Swaziland).