Djidonou, Robe ; Fos e -McG ego , Neil; Ma hew, Nandi ha
Wo king Pape
Fo mal-in o mal supply chain linkages and i m
p oduc i i y in Sub-Saha an A ica: The ole o human
capi al
UNU-MERIT Wo king Pape s, No. 2025-006
P o ided in Coope a ion wi h:
Maas ich Economic and Social Resea ch Ins i u e on Inno a ion and Technology (UNU-MERIT),
Uni ed Na ions Uni e si y (UNU)
Sugges ed Ci a ion: Djidonou, Robe ; Fos e -McG ego , Neil; Ma hew, Nandi ha (2025) : Fo mal-
in o mal supply chain linkages and i m p oduc i i y in Sub-Saha an A ica: The ole o human
capi al, UNU-MERIT Wo king Pape s, No. 2025-006, Uni ed Na ions Uni e si y (UNU), Maas ich
Economic and Social Resea ch Ins i u e on Inno a ion and Technology (UNU-MERIT), Maas ich ,
h ps://doi.o g/10.53330/JNER2108
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/326935
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-sa/4.0/
#2025-006
Fo mal-In o mal Supply Chain Linkages and Fi m P oduc i i y in
Sub-Saha an A ica: The Role o Human Capi al
Gbenoukpo Robe Djidonou, Neil Fos e -McG ego and Nandi ha Ma hew
Published 13 Feb ua y 2025
DOI: h ps://www.doi.o g/10.53330/JNER2108
Maas ich Economic and social Resea ch ins i u e on Inno a ion and Technology (UNU-MERIT)
email: in o@me i .unu.edu | websi e: h p://www.me i .unu.edu
Boschs aa 24, 6211 AX Maas ich , The Ne he lands
Tel: (31) (43) 388 44 00
UNU-MERIT Wo king Pape s
ISSN 1871-9872
Maas ich Economic and social Resea ch Ins i u e on Inno a ion and Technology
UNU-MERIT
UNU-MERIT Wo king Pape s in end o dissemina e p elimina y esul s o esea ch ca ied
ou a UNU-MERIT o s imula e discussion on he issues aised.
h ps://c ea i ecommons.o g/licenses/by-nc-sa/4.0/
Fo mal-In o mal Supply Chain Linkages and Fi m
P oduc i i y in Sub-Saha an A ica: The Role o
Human Capi al
Gbenoukpo Robe Djidonou1,2,3,∗, Neil Fos e -McG ego 4, and Nandi ha
Ma hew1,5,6
1UNU-MERIT, Uni ed Na ions Uni e si y, Maas ich , Ne he lands
2IFC, In e na ional Finance Co po a ion, Washing on DC, US
3GPI, Global Policy Incuba o , Be lin, Ge many
4ADB, Asian De elopmen Bank, Manila, Philippines
5UNU-CRIS, Uni ed Na ions Uni e si y, B uges, Belgium
6CREF, Cen o Rice che En ico Fe mi, Roma, I aly
∗
∗Co esponding au ho : Gbenoukpo Robe Djidonou, Email: djidonou@me i .unu.edu
1
Abs ac
Mic o, Small, and Medium En e p ises (MSMEs) play a c ucial ole in educing
po e y and inequali y by gene a ing he majo i y o jobs, income, and pa hways o
be e employmen oppo uni ies. Howe e , in o mal en e p ises a e o en cha ac e -
ized by low p oduc i i y and signi ican decen wo k de ici s. In Sub-Saha an A ica,
whe e a la ge sha e o he wo k o ce is engaged in in o mal en e p ises, ansi ioning
o o mali y is essen ial o enhancing p oduc i i y, os e ing economic g ow h, and en-
su ing decen wo k o all. A c i ical pa hway o in o mal i ms o o malize is h ough
p oduc ion and wo ke linkages wi h o mal i ms. Using a sample o 13,626 in o mal
i ms om h ee Sub-Saha an A ican coun ies, his s udy examines he pe o mance
e ec s o in o mal i ms wi h o mal linkages and explo es he media ing ole o human
capi al. We ind ha o mal backwa d linkages—whe e in o mal i ms sou ce inpu s
om o mal i ms—a e signi ican ly mo e common han o he ypes o o mal-in o mal
linkages. Employing he e oskedas ici y-based iden i ica ion, ou indings e eal ha
he p oduc i i y gains om hese linkages a e no au oma ic - highe human capi al is
essen ial o i ms o bene i om knowledge and echnology ans e s. This highligh s
he c i ical ole o abso p i e capaci y in enabling in o mal i ms o le e age knowledge
and echnology ans e ed h ough o mal backwa d linkages, he eby emphasizing he
impo ance o a ge ed capaci y-building in e en ions in os e ing inclusi e economic
g ow h.
Keywo ds: Fo mal-in o mal linkages, supply chains, in o mal i m, i m p oduc i i y, ech-
nology ans e , knowledge ans e , abso p i e capaci y, human capi al, Sub-Saha an A ica.
JEL Codes: J46 ; L14 ; L25 ; O12 ; O17 ; O33.
2
1 In oduc ion
In o mali y is a signi ican cause and symp om o p oduc i i y s agna ion in mode n sec-
o s in de eloping coun ies and a majo conce n o policymake s (Djidonou and Fos e -
McG ego (2022)). The o maliza ion o in o mal i ms is seen as a means o imp o ing
o e all p oduc i i y, p omp ing se e al coun ies o enac policies aimed a incen i izing in-
o mal i ms o egis e . Fo ins ance, he Wes A ican Mone a y Union (WAEMU) has
adop ed a single elec onic en e p ise o maliza ion window o educe he cos s and com-
plexi y o i ms o malizing hei ac i i ies. Howe e , he esul s o such policy ini ia i es
ha e been mixed, pa ly because in o mali y is o en a delibe a e economic choice (Meaghe
(1995, 2013); B uhn (2013)). This highligh s he impo ance o examining i m dynamics
wi hin he in o mal sec o and unde s anding he ac o s d i ing p oduc i i y imp o emen s
in his sec o .
Mic o, Small, and Medium En e p ises (MSMEs) in bo h he o mal and in o mal sec o s
a e one o he main d i e s in educing po e y and inequali y by gene a ing he majo i y
o jobs, incomes, and pa hways o be e employmen oppo uni ies. As key d i e s o long-
e m economic g ow h and de elopmen , hei ole is pa icula ly signi ican in A ica, whe e
eigh ou o en wo ke s a e engaged in in o mal employmen (Kanbu (2021)). Howe e ,
he se e e decen wo k de ici s cha ac e izing he in o mal economy ha e led o a g owing
consensus ha ansi ioning o o mali y is essen ial o ealizing decen wo k o all, making
his ansi ion a p io i y on policy agendas ac oss A ica.
In o mali y is also po en ially a subs an ial d ag on agg ega e p oduc i i y ac oss Sub-
Saha an A ica (SSA) since in o mal i ms a e ypically smalle and less p oduc i e han
hei o mal coun e pa s (Ulyssea (2020); Djidonou and Fos e -McG ego (2022)). They
o en lack he manage ial and echnological capabili ies equi ed o expand hei ac i i ies.
One way o alle ia ing capabili y gaps and enhancing p oduc i i y o in o mal i ms is by
3
c ea ing links wi h o mal i ms ha ha e such capabili ies. To access such capabili ies,
howe e , in o mal i ms may need o ha e a ce ain le el o abso p i e capaci y.
In his pape , supply chain linkages a e iden i ied as a main mechanism h ough which
p oduc i i y spillo e s may occu om o mal o in o mal i ms. The p ima y ocus is on he
backwa d linkages ha exis om o mal o in o mal i ms. These linkages a e pa icula ly
no ewo hy as hey can c ea e a posi i e eedback loop ha bene i s bo h ypes o i ms. By
p o iding in o mal i ms wi h access o new ma ke s, knowledge, and esou ces, o mal i ms
can help in o mal i ms inc ease hei p oduc i i y and compe i i eness. Collabo a ion wi h
o mal i ms may u he imp o e p oduc quali y and compliance wi h highe s anda ds.
In o mal i ms may also acqui e skills and inancial suppo om o mal pa ne s, upg ading
hei ope a ions and expanding capabili ies. Pa icipa ion in he supply chain p o ides alu-
able ma ke in elligence, aiding in o med decision-making and ma ke adap a ion. Howe e ,
o mal i ms also bene i om he lexibili y, lowe cos s, local ma ke knowledge, ex ensi e
ne wo ks, and inno a ion o in o mal i ms (Ulyssea (2020)).
P oduc i i y spillo e s and knowledge ans e om o mal o in o mal i ms a e no
s aigh o wa d o au oma ic. In o mal i ms wi h in e nal compe encies a e be e posi-
ioned o bene i om linkages wi h o mal i ms. Fo example, in o mal i ms may need
highe le els o abso p i e capaci y o lea n and adop new echniques and help access he
esou ces equi ed o upg ade hei p oduc ion p ocesses. The concep o abso p i e capac-
i y has been used by esea che s o explain a ious o ganiza ional phenomena (Cohen e al.
(1990)) and is a c i ical discussion in he Global Value Chain (GVC) li e a u e (Khan e al.
(2019); Fa ole and Winkle (2015)). Abso p i e capaci y can be unde s ood as he abili y o
ecognize, assimila e, ans o m, and apply he alue o ex e nal esou ces o local economic
de elopmen (Cohen e al. (1990); Todo o a and Du isin (2007)). P e ious li e a u e has
shown ha he le el o abso p i e capaci y is d i en by he educa ion and wo k expe ience
o wo ke s (Spi ho en and Knockae (2011)). Educa ion-based human capi al imp o emen
4
helps i ms enla ge hei echnological bounda ies and abso b new echnologies (Faems and
Sub amanian (2013); B ekke (2021)). In he case o o mal-in o mal supply chains, he ab-
so p i e capaci ies o wo ke s can help in o mal i ms bene i e ec i ely om he linkages.
The ex en o which linkages wi h o mal i ms help in he p oduc i i y enhancemen o
in o mal i ms and he mode a ing ole o complemen a y asse s like human capi al in his
ela ionship is unde -explo ed, especially in he con ex o SSA.
The cu en s udy add esses his gap in he exis ing li e a u e by examining he impac
o i m linkages in supply chains on he ans e o p ac ices, echnologies, wo ke s, and
knowledge om mo e p oduc i e o mal i ms o less p oduc i e in o mal i ms in de eloping
coun ies and how educa ion media es his. Recen e idence collec ed by he In e na ional
Labou O ganiza ion (ILO) in La in Ame ica inds ha supplying o and buying om la ge
domes ic and o eign-owned i ms ac s as a ehicle o ansmi ing manage ial p ac ices o
small i ms, including in o mal i ms embedded in hei supply chains, wi h hese e ec s
incen i izing o maliza ion. The main explana ion o such esul s is ha pu chasing inpu s
om o mal i ms can be a sou ce o p oduc i i y spillo e s. Th ough hese linkages, in o mal
i ms may access new and imp o ed p oduc a ie ies and adop mo e ecen echnologies
and p ac ices ha imp o e hei p oduc i i y.
In his pape , we e alua e he implica ions o o mal linkages o he p oduc i i y o
in o mal i ms in SSA. Mo eo e , we s udy how human capi al media es he impac o o mal-
in o mal supply chain linkages on i ms’ p oduc i i y. To achie e his goal, we use a da ase ,
consis ing o phases 1 and 2 o he ”Enquˆe es 1-2-3” su ey (hence o h Su ey 1–2–3), which
p o ides de ailed in o ma ion on he linkages be ween o mal and in o mal i ms in Sub-
Saha an A ican coun ies. The Su ey 1-2-3 adop s he de ini ions o in o mali y sugges ed
by he ILO, de ining in o mal i ms as small p oduc ion uni s ha do no ha e w i en o mal
accoun s and/o a e no egis e ed wi h he ax adminis a ion1.
1Following he In e na ional Labou O ganiza ion (ILO) de ini ion o in o mal i ms, i ms a e classi ied
5
Ou esul s indica e ha o mal backwa d linkages ma e o p oduc i i y enhancemen
in in o mal i ms. Howe e , he e ec s become insigni ican when we con ol o sec o
and coun y dummies. We u he ind ha human capi al media es he impac o o mal
backwa d linkages on in o mal i ms’ p oduc i i y in all scena ios conside ed in ou analysis.
In o he wo ds, in o mal i ms wi h skilled employees bene i o a g ea e ex en in e ms o
p oduc i i y om linkages wi h o mal i ms.
Ou app oach has se e al ad an ages o e pas s udies. Fi s ly, he esul s a e based on
a su ey ha is na ionally ep esen a i e and compa able ac oss coun ies. Secondly, ou
analysis ea u es a c i ical me hodological ad ance o e p e ious s udies by add essing he
endogenei y o in o mal backwa d linkages, which may a ise due o e e se causali y and
omi ed a iables. Thi dly, his is he i s s udy we a e awa e o o examine he d i e s o
possible linkages and he ole o complemen a y ac o s like human capi al in he e ec o
linkages on in o mal i ms’ p oduc i i y. This ocus is cen al o unde s anding he p ocess
o capabili y accumula ion, p oduc i i y imp o emen , and dynamics wi hin in o mal i ms,
wi h ele an policy implica ions.
The pape is o ganized as ollows: Sec ion 2 ou lines he heo e ical amewo k and
e iews ele an li e a u e. Sec ion 3 de ails he da ase used o he empi ical analysis.
Sec ion 4 o e s desc ip i e e idence on o mal and in o mal linkages. Sec ion 5 explains he
econome ic me hodology. Sec ion 6 p esen s he esul s, ocusing on he impac o o mal
backwa d linkages on in o mal i ms’ p oduc i i y and he media ing ole o human capi al.
Finally, Sec ion 7 summa izes he key indings and p o ides conclusions.
as in o mal i ms i hey do no comply wi h all egula ions ha apply o hei ac i i ies. Fo example,
in o mal i ms do no o icially egis e hei ac i i ies, a oid ax paymen s, do no hold a o mal accoun , o
do no espec employmen and ope a ing licenses.
6
p ac ices, such as be e in en o y managemen and op imized p oduc ion p ocesses. These
changes esul in ope a ional imp o emen s ha d i e p oduc i i y g ow h. Ou i s e-
sea ch ques ion seeks o empi ically assess whe he in o mal i ms wi h o mal linkages exhibi
highe p oduc i i y o p o i abili y compa ed o hose wi h only in o mal linkages.
Howe e , he ex en o which in o mal i ms bene i om o mal backwa d linkages de-
pends on hei human capi al and abso p i e capaci y. Resea ch shows ha i ms wi h mo e
skilled labo o highe abso p i e capaci y—de ined as he abili y o unde s and and u i-
lize ex e nal knowledge—a e be e equipped o capi alize on o mal linkages. Fo example,
Van Bieseb oeck (2005) inds ha in o mal i ms wi h skilled labo a e mo e likely o expe-
ience p oduc i i y gains om o mal backwa d linkages. Simila ly, G¨o g and S obl (2005)
demons a e ha i ms wi h g ea e abso p i e capaci y a e mo e e ec i e a u ilizing he
knowledge ans e ed om o mal i ms.
S udies documen he ole o human capi al in enhancing p oduc i i y h ough o mal
linkages highligh ing he c i ical unc ion o educa ion and aining (Schul z, 1961) and
(Becke , 2009). In o mal i ms wi h a mo e educa ed wo k o ce a e be e posi ioned o adop
new echnologies and p ac ices, he eby imp o ing p oduc i i y. Sonobe and O suka (2006)
shows ha in o mal i ms in de eloping coun ies wi h be e - ained employees bene i mo e
om o mal linkages. Abso p i e capaci y hen plays a mode a ing ole in he e ec i eness o
echnology ans e h ough o mal-in o mal linkages. Fi ms wi h highe abso p i e capaci y
can e ec i ely assimila e and apply knowledge gained om o mal i ms, esul ing in s onge
inno a ion and pe o mance ou comes. Con e sely, i ms wi h low abso p i e capaci y may
s uggle o ully bene i om o mal linkages, leading o incomple e echnology ans e .
To da e, empi ical esea ch has ye o ully explo e how abso p i e capaci y mode a es
he impac o o mal-in o mal linkages on he pe o mance o in o mal i ms. The e o e,
ou second esea ch ques ion aims o in es iga e whe he abso p i e capaci y, as e lec ed
by human capi al, media es he ela ionship be ween i m linkages and he p oduc i i y o
13
in o mal i ms.
3 DATA
We use da a om he Su ey 1–2–3 collec ed in 2015 o his s udy. Su ey 1–2–3 is he i s
na ionally ep esen a i e household and in o mal i m su ey conduc ed in se en WAEMU
coun ies and some o he selec ed A ican coun ies, including Senegal, Togo, Bu kina Faso,
Mali, Benin, Nige , Cˆo e d’I oi e, Came oon, he Democ a ic Republic o Congo, and Mo-
occo. Su ey 1–2–3 adop s he de ini ions sugges ed by ILO and Al e Chen (2005), de ining
in o mal i ms as small p oduc ion uni s ha do no ha e w i en o mal accoun s and/o
a e no egis e ed wi h he ax adminis a ion.
The Su ey 1–2–3 employs a s anda dized ques ionnai e and was conduc ed simul ane-
ously in he WAEMU coun ies, ensu ing compa abili y ac oss da ase s. The su eys a e
eliable due o echnical suppo om he Obse a oi e Economique e S a is ique d’A ique
Subsaha ienne (AFRISTAT)2and D´e eloppemen , Ins i u ions e Mondialisa ion (DIAL)3,
and inancial backing om he Wo ld Bank, he Eu opean Commission, he F ench Minis y
o Fo eign A ai s, he A ican De elopmen Bank (A DB), he Uni ed Na ions De elopmen
P og am (UNDP), he Depa men o In e na ional De elopmen (DFID), and he go e n-
men s o each pa icipa ing coun y. The su eys we e classi ied as o icial da a (B¨ohme and
Thiele (2014)).
The Su ey 1–2–3 ex ends he p inciple o mixed su eys on he in o mal sec o o be e
unde s and i s ole in each economy. I comp ises h ee nes ed su eys in ol ing di e -
en s a is ical popula ions: indi iduals, p oduc ion uni s, and households, and co e ed all
p o inces, communes, and ci ies, making he sample na ionally ep esen a i e. The i s
phase o his su ey consis s o employmen , unemploymen , and household ac i i y con-
2h ps://www.a is a .o g/
3h ps://dial.i d. / eche che/enque es/enque es-1-2-3/
14
di ions (phase 1: employmen su ey). The second phase consis s o su eying heads o
p oduc ion uni s on hei condi ions o ac i i y, economic pe o mance, mode o in eg a ion
in o he p oduc ion s uc u e, and p ospec s (phase 2: in o mal p oduc ion uni s su ey).
Finally, he hi d phase is a household consump ion su ey, which aims o es ima e household
wel a e, measu e he weigh o he o mal and in o mal sec o s in hei consump ion, and
analyze he de e minan s o he choice o di e en places o pu chase (phase 3: in es iga ion
o consump ion and po e y).
The employmen su ey ul ills a dual objec i e: o p o ide he main indica o s o de-
sc ibe he si ua ion o indi iduals and households in he labo ma ke and o se e as a il e
su ey o phase 2 on in o mal p oduc ion uni s. I p o ides he in o ma ion necessa y o
iden i y all in o mal uni s su eyed du ing he i s phase. In 2015, 21,454 we e su eyed in
he DRC, 38,599 in Came oon, and 32,393 in Bu kina Faso.
Du ing he second phase, 13,626 i ms we e andomly selec ed om he in o mal i ms
iden i ied in he i s phase (B illeau e al. (2005)). Heads o in o mal p oduc ion uni s we e
in e iewed and we e asked ques ions abou hei condi ions o ac i i y, hei economic pe -
o mance, he mode o inse ion in o he p oduc ion and supply chains, and hei p ospec s.
The second phase allows us o ha e speci ic in o ma ion on a iables such as u no e , p o-
duc ion cos s, p o i s, and p oduc quali y in o ma ion, he size o he en i y, he human
capi al o wo ke s, he in ensi y o inpu s, he cha ac e is ics o en ep eneu s o in o mal
i ms, and hei linkages wi h o he i ms. The 1–2–3 su eys de ine in o mal en e p ises as
small p oduc ion uni s ha do no ha e w i en o mal accoun s and/o a e no egis e ed
wi h he ax adminis a ion.
Mo eo e , he 1–2–3 su eys allow us o de ine he linkages be ween o mal and in o mal
i ms. Fo example, o mal backwa d linkages pe ain o he supply o aw ma e ials, equip-
men /machine y, inance, and consume goods om o mal o in o mal i ms. In he Su ey
1–2–3, in o mal i ms a e asked he sec o o ac i i y o hei main supplie s and clien s.
15
In ou da ase , o mal backwa d linkages occu when in o mal i ms pu chase in e media e
inpu s om he public and pa a-public sec o , la ge comme cial p i a e en e p ises, o la ge
non-comme cial p i a e en e p ises. Simila ly, in o mal backwa d linkages occu when in o -
mal i ms buy om small p oduc ion uni s ha do no ha e w i en o mal accoun s and/o
a e no egis e ed wi h he ax adminis a ion. This pape mainly elies on he second phase
o he su ey, wi h he sample co e ing 4,504 i ms in he DRC, 4,598 i ms in Came oon,
and 4,524 i ms in Bu kina Faso. The andom selec ion o i ms om he i s phase o he
su ey is na ionally ep esen a i e.
Se e al s udies ha e u ilized da a om he Su ey 1–2–3 o analyze a ious aspec s o
in o mal economies in Sub-Saha an A ica. Fo example, G imm e al. (2012) used he
Su ey 1–2–3 da a o in es iga e he pe o mance o in o mal en e p ises in Wes A ica,
highligh ing he he e ogenei y wi hin he in o mal sec o . No dman and Vaillan (2014)
employed he su ey da a o s udy gende dispa i ies in ea nings wi hin he in o mal sec o .
Addi ionally, Benjamin and Mbaye (2012) u ilized he Su ey 1–2–3 o explo e he impac
o in o mali y on economic de elopmen in Sub-Saha an A ica.
4 Desc ip i e S a is ics on Fo mal and Backwa d Link-
ages
In he ealm o he i m linkages li e a u e, wo p ima y ypes a e o en discussed: consump-
ion and p oduc ion linkages, as elucida ed by Hi schman (1985). Consump ion linkages
pe ain solely o sales di ec ed owa ds inal demand, while p oduc ion linkages can be u -
he ca ego ized in o o wa d and backwa d linkages. In Su ey 1-2-3, o wa d linkages occu
when in o mal i ms sell hei goods and se ices o o he i ms, whe he o mal o in o mal.
Speci ically, a o wa d linkage is o mal when an in o mal i m sells i s goods and se ices
16
o a o mal i m. Con e sely, a o wa d linkage is in o mal when in o mal i ms sell hei
goods and se ices o o he in o mal i ms. Backwa d linkages occu when in o mal i ms
buy hei goods and se ices om o he i ms. In his con ex , a backwa d linkage is in o mal
when in o mal i ms buy hei goods and se ices om o he in o mal i ms. Con e sely, a
backwa d linkage is o mal when in o mal i ms buy hei goods and se ices om o mal
i ms.
The assump ion ha i ms canno simul aneously ha e bo h backwa d and o wa d link-
ages, o bo h o mal and in o mal backwa d linkages, s ems om he design o he su ey,
which is s uc u ed o cap u e only he p ima y supplie s o buye s o each i m. The su ey
ocuses on he mos signi ican ansac ional ela ionships, p io i izing he main supplie s o
clien s a he han de ailing he ull spec um o in e ac ions a i m may ha e. The su ey’s
coding sys em ca ego izes a i m’s p ima y buye s o supplie s, based on he en i y ha
ep esen s he highes alue o ansac ions. Each code co esponds o a di e en ype o
buye o supplie , helping o classi y he i m’s ela ionships wi hin i s supply chain4. Conse-
quen ly, i ms a e ca ego ized based on hei dominan supply chain ela ionships, ei he as
ha ing o mal backwa d linkages (sou cing om o mal i ms) o in o mal backwa d linkages
(sou cing om o he in o mal i ms), wi hou oom o o e lap be ween he wo ca ego ies.
This app oach p esen s a limi a ion, as i o e looks seconda y o less p ominen ela-
4DESTINATION / ORIGIN CODES OF PRODUCTS: Fo a i m, we seek he main des ina ion/sou ce
(in e ms o Sales), meaning he one ha ep esen s he highes alue. 1 = Public o Pa as a al Sec o :
go e nmen , public o pa as a al en e p ises; 2 = La ge P i a e Comme cial En e p ise: p i a e comme -
cial es ablishmen , egis e ed, employing mo e han 5 people; 3 = Small Comme cial En e p ise: p i a e
comme cial es ablishmen employing up o 5 people; 4 = La ge P i a e Non-Comme cial En e p ise: p i-
a e p oduc ion o se ice es ablishmen , egis e ed, employing mo e han 5 people; 5 = Small P i a e
Non-Comme cial En e p ise: p i a e p oduc ion o se ice es ablishmen employing up o 5 people; 6 =
Household/Indi idual: This ca ego y applies o indi iduals pu chasing o hei inal consump ion. Fo ex-
ample, in he case o a business making clo hes and deli e ing hem o neighbo s o wea , he des ina ion is
coded as ”Household” (code 6). Howe e , i he cus ome is an indi idual pu chasing o esale, i should be
coded as ”Small Comme cial En e p ise” (code 3). 7 = Di ec Expo : sales o a o eign pa ne ; 8 = Sel -
Consump ion: This code cap u es he amoun o p oduc ion om he in o mal p oduc ion uni (UPI) ha
is consumed by he household o i s p omo e . Fo a UPI p oducing and selling donu s, his code indica es
he amoun o donu s consumed by all membe s o he p omo e ’s household. 9 = Associa ion/NGO.
17
ionships ha could p o ide a mo e nuanced unde s anding o a i m’s posi ion wi hin he
b oade economic ne wo k. Fo example, an in o mal i m migh sou ce mos o i s inpu s
om o he in o mal i ms (classi ied as ha ing in o mal backwa d linkages) bu could also
occasionally sou ce specialized inpu s om o mal i ms. These seconda y o mal linkages,
hough no cap u ed in he su ey, may s ill in luence he i m’s pe o mance and de elop-
men . Simila ly, while he su ey assumes ha i ms canno engage in bo h backwa d and
o wa d linkages a he same ime, in p ac ice, in o mal i ms may ope a e in bo h di ec ions,
ac ing as supplie s o o he i ms while simul aneously sou cing inpu s.
By ocusing only on he p ima y supplie s o buye s, he su ey limi s he scope o he
analysis, po en ially unde ep esen ing he complexi y o he in e ac ions be ween o mal
and in o mal i ms. This es ic ed iew o i m linkages can lead o an incomple e unde -
s anding o how i ms in eg a e in o la ge supply chains, and how hese in e ac ions a ec
hei p oduc i i y and g ow h. Despi e his limi a ion, ou analysis emains ele an by
shedding ligh on he co e dynamics o o mal-in o mal backwa d linkages and hei in lu-
ence on in o mal i m pe o mance. Fu u e esea ch could bene i om mo e comp ehensi e
da a collec ion ha includes mul iple ie s o supplie s and buye s o be e cap u e he
mul i ace ed na u e o i m ela ionships.
Desc ip i e s a is ics e eal ha households a e he p edominan consume s o in o mal
goods, wi h 11,759 households epo ing demand o p oduc s om in o mal i ms. Mo e-
o e , in o mal i ms o en ely on households no only as consume s bu also as supplie s o
in e media y goods wi hin hei p oduc ion chains. Speci ically, 2,764 in o mal i ms in e-
g a e inpu s sou ced di ec ly om households. This s udy, howe e , delibe a ely excludes
any analysis o linkages be ween households and in o mal i ms. Ins ead, i aims o elucida e
he dynamics o o mal and in o mal backwa d linkages among i ms only.
To p o ide insigh s in o he da ase , Table 1 p esen s de ailed in o ma ion on he numbe
o in o mal i ms in ol ed in bo h o mal and in o mal backwa d linkages. I highligh s ha
18
app oxima ely 85 pe cen o he o al linkages obse ed a e backwa d linkages, unde sco ing
hei signi icance o in o mal i ms in Sub-Saha an A ica. Addi ionally, he able p esen s
compa a i e de ails on he numbe o in o mal i ms pa icipa ing in o mal and in o mal
o wa d linkages, highligh ing he ole o in o mal i ms as supplie s o o he en i ies. The
indings demons a e ha only a mino i y o in o mal i ms a e in ol ed in o wa d linkages,
cons i u ing 15 pe cen o o al linkages. Only, 3.03 pe cen o o al linkages a e o mal
o wa d linkages, while 11.87 pe cen a e in o mal o wa d linkages.
This pape excludes he analysis o bo h o mal and in o mal o wa d linkages om i s
scope, as i s p ima y ocus is o jux apose o mal backwa d linkages agains in o mal ones.
Gi en his, we de ine he backwa d linkage a iable as a bina y ou come, wi h he o mal
backwa d linkage a iable aking he alue 1 i he supplie s o in o mal i ms a e o mal i ms
and 0 i he supplie s a e in o mal i ms. As such, he analysis ocuses only on in o mal i ms
wi h a backwa d linkage o ei he o mal o in o mal i ms. The da a indica es ha a ound
16 pe cen o o al linkages a e o mal backwa d linkages, while he majo i y, o e 69 pe cen ,
a e in o mal backwa d linkages, indica ing ha a subs an ial p opo ion o in o mal i ms
sou ce hei inpu s and inal goods om o he in o mal i ms. This end can be a ibu ed
o he p e e ence o in o mal i ms o a o dable inpu sou ces (Ranis and S ewa (1999)).
Despi e in o mal backwa d linkages being p e alen , he cu en pape ocuses on he
po en ial o in e ac ions wi h o mal i ms h ough backwa d linkages o impac pe o mance
h ough spillo e e ec s. In his sense and gi en he ocus on in o mal i ms wi h backwa d
linkages only, he in o mal i ms wi h backwa d linkages o o he in o mal i ms se e as he
con ol g oup.
To p o ide an ini ial compa ison be ween in o mal i ms wi h o mal and in o mal back-
wa d linkages, Table 2 p o ides in o ma ion on he a e age alues (and he s anda d de i-
a ion) o a se o key i m pe o mance indica o s o i ms wi h o mal backwa d and wi h
in o mal backwa d linkages (see Table 3 o a ull de ini ion o he a iables). The indings
19
indica e ha in o mal i ms wi h o mal backwa d linkages end o be, on a e age, mo e
capi al-in ensi e and o employ a highe -skilled wo k o ce ela i e o in o mal i ms wi h in-
o mal backwa d linkages. Addi ionally, hese i ms end o be la ge in e ms o alue-added,
enjoy be e access o elec ici y and c edi acili ies, and exhibi highe p o i s.
To examine whe he he e a e di e ences in p oduc i i y (de ined as sales pe wo ke ) and
p o i abili y (de ined as p o i pe wo ke ) be ween in o mal i ms wi h o mal and in o mal
backwa d linkages, Figu e 1 compa es he p oduc i i y dis ibu ion o hese wo ypes o
in o mal i ms. The igu e demons a es ha in o mal i ms wi h o mal backwa d linkages
end o exhibi highe p oduc i i y le els han hose wi h in o mal backwa d linkages. Figu e
2 epo s simila in o ma ion, bu uses da a on p o i abili y a he han p oduc i i y, wi h
a simila ou come. While such esul s a e sugges i e o he ole o o mal backwa d linkages
in enhancing he p oduc i i y o in o mal i ms, i could also be ha o mal i ms display a
p e e ence o engaging in supply chains wi h in o mal i ms ha possess s onge capabili ies.
This aises he issue o e e se causali y, which will be add essed subsequen ly.
Table 1: F equency and Sha e o In o mal Fi ms wi h Di e en Linkage Types
Backwa d Fo wa d To al
Fo mal 1795 (15.93%) 342 (3.03%) 2137 (18.97%)
In o mal 7792 (69.16%) 1337 (11.87%) 9129 (81.03%)
To al 9587 (85.09%) 1679 (14.90%) 11266 (100%)
No e: The igu es ep esen he numbe o in o mal i ms wi h backwa d and o wa d linkages. The igu es in pa en heses
ep esen he sha e o he linkages in o al linkages.
Sou ce: Au ho s’ calcula ion based on 1-2-3 su ey.
20
Table 2: Compa ison o pe o mance measu es be ween in o mal i ms wi h o mal and
in o mal backwa d linkages
Fo mal Backwa d In o mal Backwa d
Obs Mean S d. de . Min Max Obs Mean S d. de . Min Max
P oduc i i y 1759 11.8 1.40 6.5 16.6 7648 11.088 1.33 1.79 19.0
P o i abili y 1759 11.4 1.69 3.7 16.6 7645 10.766 1.50 3.26 19.0
Educa ion 1795 6.73 4.85 0 18 7792 5.243 4.44 0 20
Capi al 1560 10.5 2.17 3.3 17.7 7010 9.687 1.9 2.67 16.5
Value Added 1790 10.7 1.49 4.4 15.9 7758 10.055 1.46 0.996 16.7
Numbe o wo ke s 1795 1.49 1.10 1 12 7792 1.311 0.893 1 17
Access o C edi 1795 0.065 0.246 0 1 7792 0.037 0.19 0 1
Age o wo ke s 1794 19.1 27.5 0 94 7786 20.361 28.3 0 94
Fi m Age 1795 36.5 11.1 12 77 7792 36.734 12.5 7 89
Gende o Wo ke s 1792 0.084 0.277 0 1 7782 0.047 0.212 0 1
Belong o Associa ion 1795 0.607 0.488 0 1 7792 0.336 0.472 0 1
Access o Suppo 1236 0.453 0.498 0 1 5175 0.449 0.497 0 1
Expe ience 1794 18.2 25.9 0 89 7787 19.433 26.9 0 89
Access o Wa e 1795 0.059 0.236 0 1 7792 0.025 0.156 0 1
Sou ce: Au ho s’ calcula ion based on 1-2-3 su ey.
Table 3: De ini ion o Va iables
Va iable De ini ion
Backwa d Linkage A bina y a iable ha equals 1 i he i m has a o mal backwa d linkage and 0 i he
i m has an in o mal backwa d linkage.
P oduc i i y Log o o al sales pe wo ke
P o i abili y Log o o al p o i pe wo ke
Educa ion A e age numbe o yea s o schooling o wo ke s
Capi al Log o capi al s ock o i ms
Value Added Log o o al alue added
Numbe o wo ke s To al numbe o wo ke s
Access o C edi A bina y a iable equal o 1 i i m has access o inance and 0 o he wise
Age o wo ke s A e age age o wo ke s in yea s
Fi m Age Numbe o yea s o i m exis ence
Gende o Wo ke s The sha e o wo ke s ha a e male.
Belonging o Associa ion A bina y a iable equal o 1 i he i m belongs o an associa ion and 0 o he wise.
Fi ms we e asked whe he hey we e membe s o p o essional associa ions o ecei ed
help om p o essional associa ions.
Access o Suppo A bina y a iable equal o 1 i he i m has access o p o essional suppo om
associa ions and 0 o he wise.
Expe ience A e age numbe o yea s spen by wo ke s in a i m
Access o Wa e A bina y a iable equal o 1 i i m has access o wa e and 0 o he wise.
Figu e 1: P oduc i i y dis ibu ion o in o -
mal i ms wi h o mal and in o mal back-
wa d linkages
Sou ce: Au ho s’ calcula ion based on
1-2-3 su ey.
Figu e 2: P o i abili y dis ibu ion o in o -
mal i ms wi h o mal and in o mal back-
wa d linkages
Sou ce: Au ho s’ calcula ion based on
1-2-3 su ey.
21
5 Econome ic Me hodology
In o de o in es iga e ou i s esea ch ques ion o whe he in o mal i ms wi h o mal
linkages ha e highe p oduc i i y (o p o i abili y) han hose wi h in o mal linkages, we
begin by es ima ing a linea eg ession model. The econome ic model akes he ollowing
o m:
Yic =β1+β2BLic +Xλ +αj+αc+ϵic (1)
whe e Yic is he indica o measu ing in o mal i m p oduc i i y o p o i abili y, BLic is
he bina y a iable cap u ing he p esence o o mal backwa d linkages o i m iin coun y
c, and Xis a ec o o addi ional con ol a iables ha a e discussed below. The eg ession
speci ica ion u he includes bo h coun y (αc) and sec o (αj) dummies o con ol o
unobse ed he e ogenei y ac oss sec o s and coun ies. The measu e o p oduc i i y ha
we adop is sales pe wo ke , which is an es ima e o labou p oduc i i y. Sales pe wo ke
is a commonly used measu e in economic and business esea ch o assess he e iciency and
ou pu gene a ed by each wo ke wi hin a i m (Sy e son (2011)). Highe sales pe wo ke
indica e ha a i m is gene a ing mo e e enue ela i e o he numbe o wo ke s employed,
sugges ing a highe le el o ou pu e iciency and p oduc i i y (Hsieh and Klenow (2009)).
Howe e , i is impo an o acknowledge ha sales alone may no cap u e he ull pic u e.
One limi a ion o sales pe wo ke is i s inabili y o accoun o changes in in en o ies, which
can a ec he accu acy o p oduc i i y es ima es. Changes in in en o ies, such as luc ua ions
in s ock le els, can in luence he o e all e iciency and p oduc i i y o a i m (Olley and Pakes
(1992)).
As a obus ness check, we use p o i pe wo ke o measu e i m p o i abili y. The inclu-
sion o p o i pe wo ke as a measu e o i m pe o mance p o ides aluable insigh s in o
he po en ial cos -sa ing e ec s and ou pu enhancemen s esul ing om spillo e s ecei ed
22
Table 4: E ec s o Fo mal Backwa d Linkages on In o mal Fi ms’ P oduc i i y: OLS and
Sec o and Coun y Dummies
(1) (2) (3) (4) (5) (6)
P oduc i i y P o i abili y P oduc i i y P o i abili y P oduc i i y P o i abili y
Fo mal Backwa d 0.343*** 0.133*** 0.357*** 0.146*** 0.228*** 0.145***
(0.036) (0.033) (0.036) (0.031) (0.035) (0.028)
log (Value Added) 0.768*** 0.899*** 0.769*** 0.894*** 0.745*** 0.839***
(0.011) (0.010) (0.011) (0.010) (0.015) (0.011)
log (Capi al S ock) 0.066*** 0.155*** 0.067*** 0.135*** 0.075*** 0.067***
(0.009) (0.008) (0.009) (0.008) (0.009) (0.008)
Numbe o wo ke s -0.478*** -0.463*** -0.476*** -0.471*** -0.438*** -0.459***
(0.021) (0.019) (0.021) (0.018) (0.030) (0.028)
Access o C edi 0.193*** -0.018 0.191*** 0.003 0.218*** 0.106***
(0.054) (0.049) (0.054) (0.047) (0.050) (0.037)
Fi m Age -0.002*** -0.001** -0.002 -0.002 -0.000 -0.000
(0.001) (0.000) (0.002) (0.002) (0.001) (0.001)
Access o Suppo -0.061** 0.048* -0.067** 0.028 -0.049* -0.031
(0.031) (0.028) (0.031) (0.026) (0.028) (0.022)
Associa ion membe ship 0.042 -0.020 0.061 0.012 0.041 0.017
(0.055) (0.050) (0.055) (0.048) (0.046) (0.039)
Access o Elec ici y 0.083* 0.231*** 0.081* 0.182*** -0.017 -0.074*
(0.043) (0.039) (0.043) (0.037) (0.042) (0.038)
Access o Wa e -0.081 -0.176*** -0.086 -0.092* 0.081 0.082*
(0.065) (0.058) (0.065) (0.056) (0.056) (0.047)
Educa ion o wo ke s 0.015*** 0.050*** 0.001 0.003
(0.003) (0.003) (0.004) (0.002)
Age o wo ke s 0.002 -0.000 0.002 0.002**
(0.001) (0.001) (0.001) (0.001)
Sex o Wo ke s -0.100*** 0.104*** -0.083*** 0.063***
(0.033) (0.028) (0.030) (0.024)
Expe ience 0.001 0.000 0.001 0.001
(0.002) (0.002) (0.001) (0.001)
Expo s 0.216 0.347** 0.264 0.387*
(0.181) (0.156) (0.252) (0.219)
Coun y and Sec o Dummies YES YES
Cons an 3.472*** 0.608*** 3.509*** 1.112*** 3.727*** 2.741***
(0.111) (0.100) (0.123) (0.107) (0.360) (0.179)
Obse a ions 3,028 3,027 3,024 3,023 3,023 3,022
R-squa ed 0.689 0.808 0.692 0.826 0.748 0.878
No es: The i s 2 columns con ol only o i m cha ac e is ics, columns 3 and 4 con ol o bo h i m and wo ke cha ac e is ics,
and he las 2 columns p esen he esul s, including i m and coun y dummies alongside i m and wo ke cha ac e is ics. Figu es
in pa en hesis a e clus e ed obus s anda d e o s; ∗∗∗ p < 0.01, ∗∗ p < 0.05, ∗p < 0.1
es ablishing causali y equi es u he in es iga ion, conside ing he po en ial o e e se
causali y in hese ela ionships.
6.2 Iden i ica ion h ough He e oskedas ici y
In his sec ion, we epo he esul s o he impac o o mal backwa d linkages on he pe -
o mance o in o mal i ms using he iden i ica ion app oach known as iden i ica ion h ough
29
he e oskedas ici y, ollowing he me hodology p oposed by (Lewbel (2012)). As discussed
abo e, we migh ace he issue o e e se causali y when es ima ing he impac o o mal
backwa d linkages on i m pe o mance. Fo example, i could be ha mo e p oduc i e and
p o i able in o mal i ms a e mo e likely o es ablish o mal backwa d linkages, a he han
hese linkages d i ing he pe o mance imp o emen s. To add ess his endogenei y issue, we
use he iden i ica ion h ough he e oskedas ici y me hod de eloped by Lewbel (2012), which
u ilizes he he e oskedas ici y p esen in he da a o gene a e a se o ins umen s (Baum
e al. (2012)).
Table 5 p esen s he esul s, wi h columns 1 and 2 showing he ou comes when con olling
o i m cha ac e is ics, columns 3 and 4 when con olling o bo h i m and wo ke cha -
ac e is ics, and columns 5 and 6 when u he including coun y and sec o dummies. The
ou comes o ou a iable o in e es emain consis en wi h hose ob ained using OLS wi hou
coun y and sec o dummies, hus con i ming ou p e ious indings ha o mal backwa d
linkages a e posi i ely and signi ican ly associa ed wi h highe p oduc i i y and p o i abili y
o in o mal i ms. Howe e , when con olling o coun y and sec o dummies, he impac
o o mal backwa d linkages on he pe o mance o in o mal i ms is no longe signi ican ,
al hough he coe icien s emain posi i e.
The esul s indica e ha in o mal i ms wi h o mal backwa d linkages exhibi p oduc-
i i y and p o i abili y le els ha a e 31% and 49% highe , espec i ely, compa ed o i ms
wi h in o mal backwa d linkages when only i m cha ac e is ics a e con olled. Simila e-
sul s a e ound when con olling o bo h i m cha ac e is ics and wo ke cha ac e is ics, wi h
signi icance a he 1% le el. Howe e , when addi ionally con olling o coun y and sec o
dummies, he impac s become lowe and insigni ican , sugges ing ha o mal backwa d link-
ages alone a e no su icien o i ms o achie e highe p oduc i i y and p o i abili y.
The Hansen J es ’s p- alue is s a is ically insigni ican a all con en ional signi icance
le els, indica ing ha he in e nally gene a ed ins umen s used in he models a e unco -
30
ela ed wi h he e o e m. This esul suppo s he alidi y o ou ins umen s. Fo mal
backwa d linkages can p o ide in o mal i ms wi h access o be e manage ial p ac ices,
highe quali y inpu s, and mo e s able demand, which in u n can boos hei p oduc i -
i y and p o i abili y (McKenzie and Wood u (2014)). These linkages acili a e knowledge
ans e and can lead o he adop ion o mo e e icien p oduc ion echniques and imp o ed
business s a egies. Howe e , he educ ion in signi icance when con olling o coun y
and sec o dummies sugges s ha con ex ual ac o s play a c i ical ole. This inding high-
ligh s he necessi y o conside ing he b oade economic and ins i u ional en i onmen in
which i ms ope a e. Fo ins ance, coun ies wi h s onge egula o y amewo ks and be -
e in as uc u e migh enhance he e ec i eness o o mal backwa d linkages (Benne and
Es in (2007)). Con e sely, in en i onmen s wi h weak ins i u ions and poo in as uc u e,
he bene i s o hese linkages may be less p onounced.
Addi ionally, he posi i e bu non-signi ican esul s a e con olling o coun y and sec-
o dummies may imply ha in o mal i ms need mo e han jus o mal linkages o h i e.
Access o, skilled labou o example, as well as suppo i e go e nmen policies, a e c ucial
o sus aining g ow h and imp o ing pe o mance (Beck e al. (2005)). The e o e, policy
in e en ions aimed a s eng hening o mal-in o mal i m linkages should be complemen ed
by b oade e o s o imp o e he business en i onmen and p o ide di ec suppo o in o -
mal i ms. The ollowing sec ion examines he ole o human capi al in compa ing o mal
backwa d linkages wi h in o mal backwa d linkages.
6.3 Backwa d Linkages, Fi m Pe o mance, and Human Capi al
Imp o emen s in human capi al a e essen ial o gene a ing a wo k o ce capable o c ea i i y
and inno a ion, which signi ican ly impac s business de elopmen in Sub-Saha an A ica.
A low le el o human capi al hinde s i ms om pe o ming e ec i ely and e icien ly. O -
31
Table 5: E ec s o Fo mal Backwa d Linkages on In o mal Fi ms’ P oduc i i y: Iden i ica-
ion using he Lewbel (2012) me hod
(1) (2) (3) (4) (5) (6)
P oduc i i y P o i abili y P oduc i i y P o i abili y P oduc i i y P o i abili y
Fo mal Backwa d 0.308** 0.489*** 0.292** 0.375*** 0.007 0.029
(0.136) (0.125) (0.123) (0.107) (0.085) (0.068)
log (Value Added) 0.769*** 0.889*** 0.771*** 0.889*** 0.750*** 0.843***
(0.012) (0.011) (0.012) (0.010) (0.011) (0.009)
log (Le el Capi al) 0.066*** 0.149*** 0.068*** 0.132*** 0.081*** 0.071***
(0.009) (0.008) (0.009) (0.008) (0.009) (0.007)
Numbe o wo ke s -0.477*** -0.468*** -0.475*** -0.474*** -0.433*** -0.456***
(0.021) (0.020) (0.021) (0.018) (0.020) (0.016)
Access o C edi 0.195*** -0.039 0.195*** -0.010 0.229*** 0.114***
(0.054) (0.050) (0.054) (0.047) (0.049) (0.040)
Fi m Age -0.002*** -0.001* -0.001 -0.002 -0.000 -0.001
(0.001) (0.000) (0.002) (0.001) (0.002) (0.001)
Access o Suppo -0.061** 0.055* -0.068** 0.032 -0.050* -0.031
(0.031) (0.028) (0.031) (0.027) (0.028) (0.022)
Associa ion membe ship 0.044 -0.045 0.064 -0.002 0.052 0.026
(0.056) (0.051) (0.055) (0.048) (0.050) (0.040)
Access o Elec ici y 0.085* 0.204*** 0.086* 0.166*** 0.000 -0.060*
(0.044) (0.041) (0.044) (0.038) (0.041) (0.033)
Access o Wa e -0.079 -0.197*** -0.082 -0.108* 0.099* 0.096**
(0.065) (0.060) (0.065) (0.057) (0.060) (0.048)
Le el o Educa ion -0.015*** -0.050*** 0.001 0.003
(0.003) (0.003) (0.004) (0.003)
Age o wo ke s 0.002 -0.000 0.002 0.002*
(0.001) (0.001) (0.001) (0.001)
Sex o Wo ke s -0.095*** 0.088*** -0.065** 0.077***
(0.034) (0.029) (0.031) (0.025)
Expe ience -0.001 0.000 -0.001 -0.000
(0.002) (0.001) (0.001) (0.001)
Expo s 0.228 0.306* 0.304* 0.418***
(0.182) (0.158) (0.165) (0.132)
Coun y and Sec o Dummies No No No No Yes Yes
Cons an 3.464*** 0.688*** 3.495*** 1.161*** 2.956*** 1.385***
(0.114) (0.106) (0.126) (0.110) (0.125) (0.101)
Obse a ions 3,028 3,027 3,024 3,023 3,023 3,022
R-squa ed 0.688 0.800 0.692 0.823 0.744 0.876
Hansen J S a is ic 0.292 0.538 0.171 0.131 0.106 4.140
Hansen J S a is ic p- alue 0.5888 0.4633 0.6794 0.7173 0.7444 0.0419
No es: The i s 2 columns con ol only o i m cha ac e is ics, columns 3 and 4 con ol o bo h i m and wo ke cha ac e is ics,
and he las 2 columns p esen he esul s ob ained by adding i m and coun y dummies. Figu es in pa en hesis a e clus e ed
obus s anda d e o s; ∗∗∗ p < 0.01, ∗∗ p < 0.05, ∗p < 0.1.
ganiza ional alue c ea ion, os e ed by knowledge, skills, compe ency, and inno a ion, is
c i ical o bo h en ep eneu s and wo ke s. The li e a u e a gues ha a minimum le el o
educa ion, echnology, in as uc u e, and heal h is necessa y o in o mal i ms o bene i
om po en ial spillo e s (Thoum ung oje and Racela (2022)). Many in o mal i ms wi h low
le els o educa ion su e om his knowledge gap, which may explain hei ela i ely poo
32
pe o mance compa ed o p oduc i e o mal i ms.
The esul s in he p e ious sec ion show ha backwa d linkages alone a e insu icien
o i ms o be p oduc i e and gene a e mo e p o i while con olling o sec o and coun y
dummies. This sugges s ha while o mal backwa d linkages p o ide po en ial channels o
knowledge and echnology ans e , hese channels need o be coupled wi h adequa e human
capi al o ealize hei bene i s. Wi hou a skilled and educa ed wo k o ce, in o mal i ms
may no ully capi alize on he oppo uni ies p esen ed by hei linkages wi h o mal i ms.
Fo ins ance, an in o mal i m migh ha e access o high-quali y inpu s om o mal i ms,
bu wi hou he necessa y skills and knowledge, i canno u ilize hese inpu s e icien ly o
enhance p oduc i i y and p o i abili y.
In his sec ion, we examine he ole played by educa ion in he lea ning p ocess gene a ed
by linkages be ween o mal and in o mal i ms and i s impac on he pe o mance o in o mal
i ms in Sub-Saha an A ica. Table 6 summa izes he esul s om OLS eg ession wi h an
in e ac ion be ween o mal backwa d linkages and he a e age le el o educa ion o wo ke s.
Columns 1 and 2 o Table 6 show he esul s o his in e ac ion by conside ing bo h i m
and wo ke cha ac e is ics, while columns 3 and 4 add sec o and coun y dummies o he
con ol a iables.
The eg ession esul s demons a e ha educa ion media es he lea ning p ocess om
o mal backwa d linkages and signi ican ly a ec s he pe o mance o in o mal i ms in Sub-
Saha an A ica. Fo i ms wi h o mal backwa d linkages, each addi ional yea o educa ion
is associa ed wi h a 10.7% inc ease in p oduc i i y and a 7% inc ease in p o i abili y while
con olling o bo h i m and wo ke cha ac e is ics. These indings indica e a simila posi i e
e ec o an addi ional yea o wo ke educa ion on i m pe o mance when hey main ain
o mal backwa d linkages.
Table 7 p esen s he esul s o he media ed ole o educa ion on i m pe o mance h ough
o mal backwa d linkages using an iden i ica ion s a egy ia he e oskedas ici y, ollowing
33
he app oach p oposed by (Lewbel (2012)). We apply Lewbel (2012) me hod o bo h he
backwa d linkage a iable and i s in e ac ion wi h educa ion. This me hod aims o enhance
he eliabili y o coe icien es ima es by add essing e e se causali y p oblems. Columns
1 and 2 show esul s using iden i ica ion h ough he e oskedas ici y, conside ing bo h i m
and wo ke cha ac e is ics, while columns 3 and 4 add sec o and coun y dummies. The
impac o o mal backwa d linkages on in o mal i m pe o mance, media ed by educa ion,
is consis en wi h p e ious eg essions. The esul s show ha i wo ke s in in o mal i ms
ha e an addi ional yea o educa ion, ha ing backwa d linkages wi h o mal i ms inc eases
hei p oduc i i y by 3.2% and p o i abili y by 7%. These esul s emain consis en when
con olling o sec o and coun y dummies. The Hansen J p- alue, epo ed in he las ow o
bo h columns, is s a is ically insigni ican a all con en ional signi icance le els, indica ing
ha he in e nally gene a ed ins umen s used in hese models a e unco ela ed wi h he
e o e m. This con i ms he alidi y o ou in e nally gene a ed ins umen s.
Simila ly, Lund Vinding (2006) ound ha he sha e o highly educa ed employees is
posi i ely co ela ed wi h a i m’s abili y o p oduce e icien ly. This inding suppo s he idea
ha in o mal i ms wi h highly educa ed wo ke s can expe ience p oduc i i y spillo e s by
pu chasing inpu s om o mal i ms. Th ough hese linkages, in o mal i ms migh access new
and imp o ed p oduc a ie ies, adop mo e ecen echnologies and p ac ices, and enhance
hei p oduc i i y (Newman e al. (2015)). Fu he mo e, Ra en e al. (2017) emphasize
ha inno a ion and en ep eneu ial capabili ies, which a e o en highe in mo e educa ed
wo k o ces, a e c ucial o i m pe o mance and compe i i eness.
Ou indings sugges ha educa ion plays a c ucial ole in media ing he lea ning p ocess
om o mal backwa d linkages and signi ican ly impac s he pe o mance o in o mal i ms in
Sub-Saha an A ica. The posi i e ela ionship be ween educa ion and bo h p oduc i i y and
p o i abili y unde sco es he impo ance o in es ing in human capi al and os e ing linkages
be ween o mal and in o mal sec o s o economic de elopmen . Enhancing he educa ional
34
Table 6: Fo mal Backwa d Linkages and In o mal Fi ms’ P oduc i i y Using OLS: Media ing
Role o Human Capi al
(1) (2) (3) (4)
P oduc i i y P o i abili y P oduc i i y P o i abili y
Fo mal Backwa d 0.528*** 0.326*** 0.324*** 0.325***
(0.063) (0.066) (0.056) (0.052)
Fo mal Backwa d ×Educa ion 0.027*** 0.028*** 0.016** 0.028***
(0.007) (0.007) (0.006) (0.005)
Educa ion (in yea s) 0.008** 0.042*** 0.005 0.010***
(0.004) (0.003) (0.004) (0.003)
log (Value Added) 0.769*** 0.891*** 0.745*** 0.837***
(0.015) (0.012) (0.014) (0.011)
log (Le el Capi al) 0.063*** 0.133*** 0.072*** 0.065***
(0.009) (0.008) (0.009) (0.008)
Numbe o wo ke s -0.396*** -0.393*** -0.366*** -0.372***
(0.025) (0.023) (0.023) (0.022)
Access o C edi (1=yes) 0.189*** 0.001 0.217*** 0.105***
(0.053) (0.043) (0.050) (0.037)
Fi m Age -0.002 -0.002** -0.001 -0.000
(0.001) (0.001) (0.001) (0.001)
Access o Suppo -0.076** 0.020 -0.056** -0.037
(0.031) (0.027) (0.028) (0.023)
Associa ion membe ship 0.055 0.001 0.036 0.004
(0.052) (0.043) (0.046) (0.039)
Access o Elec ici y 0.063 0.162*** -0.022 -0.088**
(0.045) (0.044) (0.042) (0.038)
Access o Wa e -0.066 -0.065 0.094* 0.109**
(0.059) (0.056) (0.056) (0.048)
Age o wo ke s 0.002 -0.000 0.002* 0.002**
(0.001) (0.001) (0.001) (0.001)
Sex o Wo ke s -0.101*** 0.102*** -0.080*** 0.064***
(0.033) (0.029) (0.030) (0.024)
Expe ience -0.000 0.000 -0.001 -0.001
(0.002) (0.001) (0.001) (0.001)
Expo s 0.221 0.337 0.275 0.385*
(0.269) (0.242) (0.246) (0.202)
Coun y and Sec o Dummies Yes Yes
Cons an 3.409*** 1.021*** 3.617*** 2.585***
(0.135) (0.109) (0.362) (0.176)
Obse a ions 3,063 3,062 3,062 3,061
R-squa ed 0.693 0.826 0.747 0.876
No es: Included coun y and sec o dummies a e no epo ed. Figu es in pa en hesis a e clus e obus s anda d e o ;
∗∗∗ p < 0.01, ∗∗ p < 0.05, ∗p < 0.1.
a ainmen o he wo k o ce can enable in o mal i ms o be e abso b and implemen he
knowledge and echnologies accessed h ough o mal backwa d linkages, he eby d i ing
imp o emen s in i m pe o mance and con ibu ing o b oade economic g ow h.
35
Table 7: Fo mal Backwa d Linkages and In o mal Fi ms’ P oduc i i y Using Lewbel (2012):
Media ing Role o Human Capi al
(1) (2) (3) (4)
P oduc i i y P o i abili y P oduc i i y P o i abili y
Fo mal Backwa d 0.535*** 0.372*** 0.267*** 0.278***
(0.068) (0.059) (0.062) (0.050)
Fo mal Backwa d ×Educa ion 0.023*** 0.028*** 0.009 0.021***
(0.007) (0.006) (0.007) (0.005)
Educa ion (in yea s) 0.009** 0.042*** 0.003 0.008*
(0.004) (0.003) (0.004) (0.003)
log (Value Added) 0.768*** 0.890*** 0.746*** 0.837***
(0.011) (0.010) (0.011) (0.009)
log (Le el Capi al) 0.063*** 0.132*** 0.072*** 0.065***
(0.009) (0.008) (0.009) (0.007)
Numbe o wo ke s -0.396*** -0.393*** -0.365*** -0.371***
(0.015) (0.013) (0.013) (0.011)
Access o C edi (1=yes) 0.187*** -0.001 0.216*** 0.104***
(0.053) (0.046) (0.049) (0.039)
Fi m Age -0.002 -0.002 -0.000 -0.000
(0.002) (0.001) (0.002) (0.001)
Access o Suppo -0.075** 0.021 -0.055** -0.036
(0.030) (0.026) (0.028) (0.022)
Associa ion membe ship 0.053 -0.002 0.037 0.005
(0.055) (0.047) (0.050) (0.040)
Access o Elec ici y 0.063 0.158*** -0.019 -0.085***
(0.043) (0.037) (0.041) (0.033)
Access o Wa e -0.071 -0.067 0.090 0.105**
(0.064) (0.056) (0.059) (0.047)
Age o wo ke s 0.002 -0.000 0.002 0.002**
(0.001) (0.001) (0.001) (0.001)
Sex o Wo ke s -0.103*** 0.098*** -0.079*** 0.065***
(0.033) (0.028) (0.030) (0.024)
Expe ience -0.001 0.000 -0.001 -0.000
(0.002) (0.001) (0.001) (0.001)
Expo s 0.216 0.330** 0.278* 0.388***
(0.181) (0.156) (0.164) (0.132)
Coun y and Sec o Dummies No No Yes Yes
Cons an 3.424*** 1.030*** 2.963*** 1.359***
(0.123) (0.106) (0.121) (0.098)
Obse a ions 3,063 3,062 3,062 3,061
R-squa ed 0.693 0.826 0.747 0.876
Hansen J S a is ic 7.976 11.969 40.549 44.004
Hansen J S a is ic p- alue 0.018 0.002 0.359 0.387
No es: Included coun y and sec o dummies a e no epo ed. Figu es in pa en hesis a e clus e obus s anda d e o ;
∗∗∗ p < 0.01, ∗∗ p < 0.05, ∗p < 0.1.
7 Conclusion
This pape examines he e ec s o o mal-in o mal supply chain linkages on he pe o mance
o in o mal i ms and in es iga es how human capi al media es he impac o hese linkages
on he pe o mance o in o mal i ms. Ou s udy i s explo es he p esence o bo h o mal
36
and in o mal o wa d and backwa d linkages. We ind ha o mal backwa d linkages, - whe e
in o mal i ms use inpu s p oduced by o mal i ms - a e signi ican ly mo e p e alen han
o mal o wa d linkages, whe e o mal i ms sou ce inpu s om in o mal i ms.
We hen assess he impac o o mal backwa d linkages on in o mal i ms’ pe o mance,
measu ed bo h by p oduc i i y and p o i abili y. Ou indings indica e ha in o mal i ms
pa icipa ing in supply chains wi h o mal i ms exhibi signi ican ly highe p oduc i i y.
This pape is among he i s in he in o mal economy li e a u e o add ess he endogenei y
o o mal backwa d linkages, which can a ise due o omi ed a iables and e e se causali y.
Using iden i ica ion h ough he e oskedas ici y, we demons a e ha in o mal i ms wi h
backwa d linkages o o mal i ms expe ience enhanced p oduc i i y. Howe e , he impac s
u n insigni ican when we accoun o sec o and coun y dummies.
Fu he mo e, we examine how human capi al media es he impac o hese backwa d
linkages on he p oduc i i y o in o mal i ms in A ica. Ou indings unde sco e ha in o -
mal i ms wi h wo ke s possessing highe abso p i e capaci ies bene i mo e om backwa d
linkages in e ms o p oduc i i y gains. This unde sco es he c ucial ole o human capi al
in maximizing he bene i s de i ed om ex e nal linkages wi h o mal i ms.
Ou indings ha e signi ican implica ions bo h a he o ganiza ional le el and o policy
o mula ion. We highligh he impo ance o complemen a y asse s, such as an educa ed
wo k o ce, in le e aging ex e nal linkages wi h o mal i ms o enhance p oduc i i y. How-
e e , me ely ha ing an educa ed wo k o ce is no su icien ; i ms mus ac i ely engage in
knowledge ans e and expansion s a egies. Building linkages wi h i ms close o he e i-
ciency and knowledge on ie is essen ial o achie ing sus ained p oduc i i y imp o emen s.
A he i m le el, he esul s highligh he impo ance o complemen a y asse s, pa icu-
la ly human capi al, in maximizing he bene i s o o mal-in o mal linkages. Howe e , simply
ha ing an educa ed wo k o ce is no su icien ; i ms mus also ac i ely engage in knowledge
acquisi ion, and p ocess upg ading o sus ain p oduc i i y imp o emen s. Building s onge
37
linkages wi h i ms a he e iciency and knowledge on ie is essen ial o long- e m capa-
bili y de elopmen .
A he policy le el, ou s udy sugges s ha in e en ions aimed a enhancing skills de-
elopmen and abso p i e capaci ies among in o mal wo ke s can signi ican ly ampli y he
p oduc i i y gains om o mal-in o mal linkages. This calls o a ge ed oca ional aining
p og ams, business de elopmen se ices, and incen i es o o mal-in o mal collabo a ion
o os e inclusi e economic g ow h. Mo e b oadly, policy e o s should no only ocus on
o maliza ion bu also on c ea ing enabling condi ions o p oduc i i y enhancemen wi hin
he in o mal sec o .
By shedding ligh on he in e ac ion be ween o mal and in o mal i ms and he pi o al
ole o human capi al, his s udy con ibu es o a deepe unde s anding o how supply chain
linkages can be le e aged o d i e p oduc i i y g ow h in he in o mal economy. S eng h-
ening hese linkages h ough policy suppo , capaci y-building ini ia i es, and s a egic i m
in e ac ions can enhance he po en ial o in o mal i ms as engines o sus ainable de elopmen
and economic ans o ma ion.
38
Schmi z, H. and K. Nad i (1999): “Clus e ing and indus ializa ion: in oduc ion,”
Wo ld de elopmen , 27, 1503–1514.
Schul z, T. W. (1961): “In es men in human capi al,” The Ame ican economic e iew,
51, 1–17.
Sonobe, T. and K. O suka (2006): Clus e -based indus ial de elopmen : An Eas Asian
model, Sp inge .
Spi ho en, A. and M. Knockae (2011): “The ole o business cen es in i ms’ ne -
wo king capabili ies and pe o mance,” Science and Public Policy, 38, 569–580.
Sy e son, C. (2011): “Wha de e mines p oduc i i y?” Jou nal o Economic li e a u e,
49, 326–365.
Thoum ung oje, A. and O. C. Racela (2022): “Linking SME in e na ional ma ke ing
agili y o new echnology adop ion,” In e na ional Small Business Jou nal, 40, 801–822.
Todo o a, G. and B. Du isin (2007): “Abso p i e capaci y: Valuing a econcep ualiza-
ion,” Academy o managemen e iew, 32, 774–786.
Ulyssea, G. (2020): “In o mali y: Causes and consequences o de elopmen ,” Annual
Re iew o Economics, 12, 525–546.
Van Bieseb oeck, J. (2005): “Fi m size ma e s: G ow h and p oduc i i y g ow h in
A ican manu ac u ing,” Economic De elopmen and Cul u al Change, 53, 545–583.
Xaba, J., P. Ho n, S. Mo ala, S. Mene ee, e al. (2002): “In o mal Sec o in
Sub-Saha an A ica,” Tech. ep., In e na ional Labou O ganiza ion.
45
Yu, H., J. Zhang, M. Zhang, and F. Fan (2022): “C oss-na ional knowledge ans e ,
abso p i e capaci y, and o al ac o p oduc i i y: The in e media y e ec es o in e na-
ional echnology spillo e ,” Technology Analysis & S a egic Managemen , 34, 625–640.
46
The UNU-MERIT WORKING Pape Se ies
2025-01 De elopmen s a egies o he g een hyd ogen economy in eme ging economies
by Fabianna Bacil, An hony Black, Ma ina Domingues, Jun Jin, Rasmus Lema, Glen
Robbins and Sö en Schol in
2025-02 Do global alue chains and local capabili ies ma e o economic complexi y in EU
egions? by R. Boschma, E. He nández-Rod íguez, A. Mo ison and C. Pie obelli
2025-03 Inno a ion and p oduc i i y: he ecen empi ical li e a u e and he s a e o he a
by Jacques Mai esse, Pie e Mohnen and Ad No en
2025-04 Digi al Technologies, Hi ing, T aining, and Fi m Ou comes: E idence om AI and ICT
Adop ion in Indian i mss by Sneha Ma ydas, Nandi ha Ma hew, Gio dano De
Ma zo and Ca lo Pie obelli
2025-05 New echnologies and employmen : he s a e o he a by Ma co Vi a elli and
Guille mo A enas Díaz
2025-06 Fo mal-In o mal Supply Chain Linkages and Fi m P oduc i i y in Sub-Saha an A ica:
The Role o Human Capi al by Gbenoukpo Robe Djidonou, Neil Fos e -McG ego
and Nandi ha Ma hew
