Mazo odze, B ian Ta onga
A icle
P oduc i i y and wages in Sou h A ica
Economies
P o ided in Coope a ion wi h:
MDPI – Mul idisciplina y Digi al Publishing Ins i u e, Basel
Sugges ed Ci a ion: Mazo odze, B ian Ta onga (2024) : P oduc i i y and wages in Sou h A ica,
Economies, ISSN 2227-7099, MDPI, Basel, Vol. 12, Iss. 12, pp. 1-27,
h ps://doi.o g/10.3390/economies12120330
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/329257
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h ps://c ea i ecommons.o g/licenses/by/4.0/
Ci a ion: Mazo odze, B ian Ta onga.
2024. P oduc i i y and Wages in
Sou h A ica. Economies 12: 330.
h ps://doi.o g/10.3390/
economies12120330
Academic Edi o : Ral Fendel
Recei ed: 21 Oc obe 2024
Re ised: 23 No embe 2024
Accep ed: 29 No embe 2024
Published: 2 Decembe 2024
Copy igh : © 2024 by he au ho .
Licensee MDPI, Basel, Swi ze land.
This a icle is an open access a icle
dis ibu ed unde he e ms and
condi ions o he C ea i e Commons
A ibu ion (CC BY) license (h ps://
c ea i ecommons.o g/licenses/by/
4.0/).
A icle
P oduc i i y and Wages in Sou h A ica
B ian Ta onga Mazo odze
Depa men o Accoun ing and Economics, Sol Plaa je Uni e si y, Cen al Campus Academic Building,
P i a e Bag X5008, Kimbe ley 8300, Sou h A ica; [email p o ec ed]
Abs ac : The wo ld has expe ienced apid p oduc i i y g ow h in he las h ee decades, bu has his
g ow h e lec ed in wages? In heo y, unde ce ain condi ions, wo ke s ea n hei ma ginal p oduc
so ha p oduc i i y g ow h eeds in o wages on a one- o-one basis. Gi en he con adic o y li e a u e,
his pape e isi ed he p oduc i i y–wage ela ionship in Sou h A ica using an indus y-le el panel
da ase comp ising 74 indus ies obse ed be ween 1993 and 2023. Using se e al es ima o s, ou
main indings a e obse ed. Fi s , p oduc i i y is ound o ha e g own as e han wages. Second, he
obse ed p oduc i i y–wage di e gence pa ly e lec s he squeezing o wo ke s o p o i s. Thi d,
p oduc i i y g ow h mos ly ou paced he wages o low-skilled wo ke s, wo ke s on sho - e m
con ac s, and in o mal wo ke s. Fou h, p oduc i i y g ow h la gely unde mined ake-home pay
compa ed o inge bene i s. These esul s imply ha al hough boos ing p oduc i i y g ow h may
be impo an , i s achie emen may no lead o b oad-based wage gains in Sou h A ica unless he
go e nmen pu sues policies ha ealign p oduc i i y and he pay o low-skilled wo ke s, in o mal
wo ke s, and wo ke s on sho - e m con ac s. Such policies may include sec o -based incen i es
o businesses ha imp o e wage condi ions, inc eased union suppo in high-g ow h indus ies,
imp o ed egula ion o he in o mal sec o , and oca ional aining o low-skilled wo ke s.
Keywo ds: labou p oduc i i y; eal emune a ion; eal wages; Sou h A ica
1. In oduc ion
Economic heo y p edic s ha wo ke s in compe i i e ma ke s a e paid hei ma ginal
p oduc such ha g ow h in labou p oduc i i y eeds en i ely in o eal wages on a one- o-
one basis. In con as o his p edic ion, eal wages ha e ailed o keep pace wi h labou
p oduc i i y globally in he las h ee decades (Felds ein 2008;Mishel 2012;S ansbu y
and Summe s 2018). In some s udies, i is sugges ed ha he ela ionship be ween labou
p oduc i i y and wages may ha e b oken.
This pape e isi s his discussion by es ima ing he sho - and long- e m impac o
labou p oduc i i y on eal emune a ion in Sou h A ica. I aims o un a el he he e o-
genei ies in he p oduc i i y–wage ela ionship be ween (i) low-, semi-, and skilled wo ke s
and (ii) o mal and in o mal wo ke s. In addi ion o he commonly used emune a ion pe
wo ke p oxy o labou income, i seeks o sepa a ely es ablish he po en ial he e ogeneous
impac o p oduc i i y on wages and bene i s. In a bid o es ablish he unde lying ac o s
in luencing he p oduc i i y–wage ela ionship, he pape de e mines he ex en o which
he expansion o indus ial p o i s may ha e p e en ed p oduc i i y om aising labou
income in Sou h A ica be ween 1993 and 2023.
Wi hin he li e a u e, he empi ical indings ha e la gely been inconclusi e. In Dos ie
(2011), o ins ance, he conclusion is ha wages do no de ia e signi ican ly om p oduc-
i i y, sugges ing a nea one- o-one ela ionship. This esul is simila ly obse ed in he
s udies by Van Bieseb oeck (2003) and mo e ecen ly He man (2020). Howe e , he ela-
ionship be ween p oduc i i y and wages is ound o be ei he weak, non-exis en , o less
han one- o-one. Schwellnus e al. (2017) pa icula ly ind agg ega e labou p oduc i i y
g ow h in mos OECD coun ies o ha e decoupled om he eal median compensa ion
Economies 2024,12, 330. h ps://doi.o g/10.3390/economies12120330 h ps://www.mdpi.com/jou nal/economies
Economies 2024,12, 330 2 o 27
g ow h. Thei obse a ion sugges s ha inc easing p oduc i i y may no su icien ly aise
eal wages o he ypical wo ke in he egion.
Despi e he as ness o he li e a u e linking p oduc i i y wi h wages, only Wake o d
(2004) and Tsoku and Ma a ise (2014) ha e, o ou knowledge, explici ly examined he
sho - un and long- un impac o p oduc i i y g ow h on eal emune a ion in he con ex
o Sou h A ica. These s udies eached con lic ing conclusions. Wake o d (2004) concluded
ha labou p oduc i i y in Sou h A ica eeds in o wages only in he long un, and i
does so on a less han one- o-one basis, while Tsoku and Ma a ise (2014) concluded ha
p oduc i i y g ow h does no signi ican ly eed in o wages using a speci ica ion comp ising
eal emune a ion, p oduc i i y, and he a e o unemploymen .
This pape di e s om he abo e s udies in se e al ways. Fi s , i ca ego ises labou
in o (i) o mal and in o mal wo ke s and (ii) low, semi-skilled, and skilled wo ke s. These
decomposi ions allow us o p o ide a clea e pic u e o how he eal ea nings o speci ic
g oups esponded o p oduc i i y dynamics du ing he sampling pe iod. Second, in
addi ion o he commonly used eal emune a ion measu e, i dis inguishes be ween eal
wages and eal bene i s. Wages di e om emune a ion in ha hey exclude employe
con ibu ions o social insu ance, pension con ibu ions, and employe paymen s o heal h
insu ance and o he inge bene i s (Boswo h e al. 1994). F om a esea ch s andpoin ,
Cashell (2004) a gues ha he use o wages while igno ing bene i s can be empi ically
misleading as he wo e ol e a di e en paces. F om a policymaking iewpoin , he
he e ogeneous e olu ion o wages and bene i s may ha e di e en wel a e implica ions.
As ecommended by Wake o d (2004), his pape pu sues a disagg ega ed app oach
ha uses da a on 74 indus ies. The da ase comp ises all h ee sec o s o he econ-
omy, namely, p ima y, seconda y, and e ia y. The o al sample pa icula ly comp ises
3 indus ies
om he ag icul u al sec o , 7 om mining, 42 om manu ac u ing, and
22 om se ices.
This disagg ega ed app oach enables us o exploi he ich a ia ion in
p oduc i i y g ow h and wages ac oss indus ies belonging o di e en sec o s in a bid o
a ain a e lec i e pic u e o how and why wages may ha e decoupled om p oduc i i y
g ow h in Sou h A ica. The accommoda ion o sec o al and indus ial di e si y is impo -
an , as Faggio e al. (2010) show ha ocusing on one sec o , he manu ac u ing sec o , as
did He man (2020), may unde es ima e he wage and p oduc i i y di e gence.
In addi ion, his pape no only con ols o unemploymen bu also impo pene a ion
and expo in ensi y. The li e a u e has shown ha ade can in luence bo h p oduc i i y
g ow h and wage dynamics; hence, he omission o impo pene a ion and expo in ensi y
may lead o an omi ed a iable bias. Las ly, i conside s, in an addi ional speci ica ion,
g oss ma kup as an addi ional con ol a iable. The g oss ma kup o an indus y is he ne
ope a ing su plus as a pe cen age o o al in e media e inpu s plus labou emune a ion.
Since i p oxies indus ial p o i s, i s inclusion allows us o make in e ences abou he
unde lying causes o a less han one- o-one ela ionship be ween p oduc i i y and wages.
A less han one- o-one ela ionship when g oss ma kup is con olled o would, o example,
sugges he p esence o o he unde lying ac o s besides he expansion o indus ial p o i s
p e en ing p oduc i i y g ow h om ully e lec ing in wages. A less han one- o-one
ela ionship ha u ns one- o-one once g oss ma k is con olled o would sugges ha
p oduc i i y g ow h may ha e been la gely p e en ed om aising wages ully by he
expansion o p o i s.
Ha ing uled ou he possibili y o simul anei y h ough panel G ange causali y es s,
he empi ical analysis was g ounded in a s anda d heo e ical amewo k in which wages
a e a unc ion o labou p oduc i i y and wo ke s a e paid hei ma ginal p oduc s when
ma ke s a e compe i i e. The pape augmen ed his unc ion wi h addi ional con ols in a
bid o isola e he e ec o p oduc i i y on wages. Conside ing he la ge ime dimension
and he la ge numbe o indus ies in he sample coupled wi h he desi e o es ima e
he p oduc i i y–wage ela ionship in he sho un and long un, he analysis applied
he Pooled Mean G oup (PMG) o Pesa an e al. (1999), which accommoda es sho - un
he e ogenei y and imposes long- un homogenei y. Po en ial endogenei y likely o a ise
Economies 2024,12, 330 3 o 27
om measu emen e o and hi d ac o s nes ed in he e o e m was add essed in
obus ness exe cises h ough he Panel Dynamic O dina y Leas Squa es (PDOLS) me hod
and he Panel Fully Modi ied O dina y Leas Squa es (PFMOLS).
The es o he pape is o ganised as ollows. Sec ion 2 e iews he p oduc i i y–
emune a ion di e gence by sec o and he heo e ical and empi ical li e a u e. Sec ion 3
desc ibes he da a and speci ies he empi ical models. The esul s a e p esen ed and
in e p e ed in Sec ion 4. Sec ion 5concludes he analysis, discusses he policy implica ions
o his s udy, and sugges s a eas o u he esea ch.
2. The Decoupling o P oduc i i y and Remune a ion in Sou h A ica: S ylised Fac s
Figu e 1displays he e olu ion o p oduc i i y and eal emune a ion o all sec o s
be ween 1993 and 2023. A cu so y look a he g aph shows ha labou p oduc i i y g ew
as e han eal emune a ion du ing he sampling pe iod.
Economies 2024, 12, x FOR PEER REVIEW 3 o 28
sho - un he e ogenei y and imposes long- un homogenei y. Po en ial endogenei y likely
o a ise om measu emen e o and hi d ac o s nes ed in he e o e m was add essed
in obus ness exe cises h ough he Panel Dynamic O dina y Leas Squa es (PDOLS)
me hod and he Panel Fully Modified O dina y Leas Squa es (PFMOLS).
The es o he pape is o ganised as ollows. Sec ion 2 e iews he p oduc i i y– e-
mune a ion di e gence by sec o and he heo e ical and empi ical li e a u e. Sec ion 3
desc ibes he da a and specifies he empi ical models. The esul s a e p esen ed and in e -
p e ed in Sec ion 4. Sec ion 5 concludes he analysis, discusses he policy implica ions o
his s udy, and sugges s a eas o u he esea ch.
2. The Decoupling o P oduc i i y and Remune a ion in Sou h A ica: S ylised Fac s
Figu e 1 displays he e olu ion o p oduc i i y and eal emune a ion o all sec o s
be ween 1993 and 2023. A cu so y look a he g aph shows ha labou p oduc i i y g ew
as e han eal emune a ion du ing he sampling pe iod.
Figu e 1. Real emune a ion and labou p oduc i i y (all sec o s). Sou ce: Own compu a ion using
da a om Quan ec.
In he p ima y sec o (Figu e 2), he decoupling o p oduc i i y and eal emune a-
ion mos ly occu ed du ing he 1993–2006 pe iod and he 2018–2023 pe iod. Be ween
2007 and 2018, p oduc i i y and wages mo ed closely. In a ew ins ances (2011, 2012, and
2016), eal emune a ion ou paced labou p oduc i i y g ow h. The gene al ends in Fig-
u e 2, howe e , mask he he e ogenei ies ac oss p ima y sec o s. Wages in ag icul u e, o
example, end o be lowe compa ed o o he sec o s. This is pa ly due o he seasonal
and o en labou -in ensi e na u e o he wo k, as well as he o e all lowe alue-added in
many ag icul u al ac i i ies. Al hough heo y p edic s a s ong connec ion be ween wages
and p oduc i i y g ow h, wage inc eases in Sou h A ica’s ag icul u e sec o ha e o en
been modes due o ma ke p essu es, low ma gins, and challenges in passing on highe
p oduc ion cos s o consume s. Wages in he mining sec o , howe e , a e gene ally highe
compa ed o he ag icul u e sec o s, eflec ing he specialised skills equi ed and he haz-
a dous na u e o he wo k. In he o es y sec o , wages a e gene ally mode a e, eflec ing
he skill le els equi ed and he na u e o he wo k. Fo es y wo k is o en less mechanised
compa ed o mining, which impac s p oduc i i y and wage le els.
1993 1998 2003 2008 2013 2018 2023
Yea
Labou p oduc i i y Real emune a ion pe wo ke
Figu e 1. Real emune a ion and labou p oduc i i y (all sec o s). Sou ce: Own compu a ion using
da a om Quan ec.
In he p ima y sec o (Figu e 2), he decoupling o p oduc i i y and eal emune a ion
mos ly occu ed du ing he 1993–2006 pe iod and he 2018–2023 pe iod. Be ween 2007
and 2018, p oduc i i y and wages mo ed closely. In a ew ins ances (2011, 2012, and
2016), eal emune a ion ou paced labou p oduc i i y g ow h. The gene al ends in
Figu e 2, howe e , mask he he e ogenei ies ac oss p ima y sec o s. Wages in ag icul u e,
o example, end o be lowe compa ed o o he sec o s. This is pa ly due o he seasonal
and o en labou -in ensi e na u e o he wo k, as well as he o e all lowe alue-added
in many ag icul u al ac i i ies. Al hough heo y p edic s a s ong connec ion be ween
wages and p oduc i i y g ow h, wage inc eases in Sou h A ica’s ag icul u e sec o ha e
o en been modes due o ma ke p essu es, low ma gins, and challenges in passing on
highe p oduc ion cos s o consume s. Wages in he mining sec o , howe e , a e gene ally
highe compa ed o he ag icul u e sec o s, e lec ing he specialised skills equi ed and
he haza dous na u e o he wo k. In he o es y sec o , wages a e gene ally mode a e,
e lec ing he skill le els equi ed and he na u e o he wo k. Fo es y wo k is o en less
mechanised compa ed o mining, which impac s p oduc i i y and wage le els.
Economies 2024,12, 330 4 o 27
Economies 2024, 12, x FOR PEER REVIEW 4 o 28
Figu e 2. Real emune a ion and labou p oduc i i y (p ima y sec o s). Sou ce: Own compu a ion
using da a om Quan ec.
Figu e 3 displays he e olu ion o p oduc i i y and eal emune a ion in he second-
a y sec o . A obus seconda y sec o con ibu es o indus ialisa ion, which is essen ial
o economic di e sifica ion. I educes dependency on p ima y sec o s like mining and
ag icul u e, which a e mo e ulne able o ex e nal shocks and p ice fluc ua ions. Apa
om 2017, seconda y sec o s saw labou p oduc i i y di e gence om eal emune a ion
du ing he sampling pe iod. Much o he di e gence was obse ed be ween 2001 and
2005. F om 2006, eal emune a ion slowly na owed he gap be o e sligh ly exceeding
p oduc i i y g ow h, albei empo a ily in 2017. These ends may a y ac oss sec o s.
Cons uc ion wo ke s, o ins ance, ea n mode a e wages, wi h highe pay o specialised
oles and manage ial posi ions. Manu ac u ing wages, on he o he hand, a e highe com-
pa ed o ag icul u e bu can be lowe han hose in high- ech indus ies.
Figu e 3. Real emune a ion and labou p oduc i i y (seconda y sec o s). Sou ce: Own compu a ion
using da a om Quan ec.
1993 1998 2003 2008 2013 2018 2023
Yea
Real emune a ion pe wo ke Labou p oduc i i y
1993 1998 2003 2008 2013 2018 2023
Yea
Real emune a ion pe wo ke Labou p oduc i i y
Figu e 2. Real emune a ion and labou p oduc i i y (p ima y sec o s). Sou ce: Own compu a ion
using da a om Quan ec.
Figu e 3displays he e olu ion o p oduc i i y and eal emune a ion in he seconda y
sec o . A obus seconda y sec o con ibu es o indus ialisa ion, which is essen ial o
economic di e si ica ion. I educes dependency on p ima y sec o s like mining and
ag icul u e, which a e mo e ulne able o ex e nal shocks and p ice luc ua ions. Apa
om 2017, seconda y sec o s saw labou p oduc i i y di e gence om eal emune a ion
du ing he sampling pe iod. Much o he di e gence was obse ed be ween 2001 and
2005. F om 2006, eal emune a ion slowly na owed he gap be o e sligh ly exceeding
p oduc i i y g ow h, albei empo a ily in 2017. These ends may a y ac oss sec o s.
Cons uc ion wo ke s, o ins ance, ea n mode a e wages, wi h highe pay o specialised
oles and manage ial posi ions. Manu ac u ing wages, on he o he hand, a e highe
compa ed o ag icul u e bu can be lowe han hose in high- ech indus ies.
Economies 2024, 12, x FOR PEER REVIEW 4 o 28
Figu e 2. Real emune a ion and labou p oduc i i y (p ima y sec o s). Sou ce: Own compu a ion
using da a om Quan ec.
Figu e 3 displays he e olu ion o p oduc i i y and eal emune a ion in he second-
a y sec o . A obus seconda y sec o con ibu es o indus ialisa ion, which is essen ial
o economic di e sifica ion. I educes dependency on p ima y sec o s like mining and
ag icul u e, which a e mo e ulne able o ex e nal shocks and p ice fluc ua ions. Apa
om 2017, seconda y sec o s saw labou p oduc i i y di e gence om eal emune a ion
du ing he sampling pe iod. Much o he di e gence was obse ed be ween 2001 and
2005. F om 2006, eal emune a ion slowly na owed he gap be o e sligh ly exceeding
p oduc i i y g ow h, albei empo a ily in 2017. These ends may a y ac oss sec o s.
Cons uc ion wo ke s, o ins ance, ea n mode a e wages, wi h highe pay o specialised
oles and manage ial posi ions. Manu ac u ing wages, on he o he hand, a e highe com-
pa ed o ag icul u e bu can be lowe han hose in high- ech indus ies.
Figu e 3. Real emune a ion and labou p oduc i i y (seconda y sec o s). Sou ce: Own compu a ion
using da a om Quan ec.
1993 1998 2003 2008 2013 2018 2023
Yea
Real emune a ion pe wo ke Labou p oduc i i y
1993 1998 2003 2008 2013 2018 2023
Yea
Real emune a ion pe wo ke Labou p oduc i i y
Figu e 3. Real emune a ion and labou p oduc i i y (seconda y sec o s). Sou ce: Own compu a ion
using da a om Quan ec.
Economies 2024,12, 330 5 o 27
Te ia y sec o s ha e he la ges con ibu ion o g oss domes ic p oduc (GDP) in Sou h
A ica. These sec o s saw he la ges decoupling o p oduc i i y and wages be ween 1993
and 2023. In 2023, hey accoun ed o 57 pe cen o o al GDP compa ed o he 31 pe cen
and 12 pe cen con ibu ions o seconda y and p ima y sec o s, espec i ely. The la ges
con ibu ion, coupled wi h he la ges p oduc i i y and eal emune a ion di e gence
(Figu e 4), he e o e implies ha e ia y sec o s may ha e had a signi ican bea ing on
o e all income inequali y in he Sou h A ican economy be ween 1993 and 2023. Wi hin
he e ia y sec o , inancial se ices expe ienced signi ican p oduc i i y g ow h, d i en by
ad ancemen s in echnology, such as digi al banking, in ech inno a ions, and au oma ion
o ou ine asks.
Wages in inancial se ices a e gene ally high compa ed o o he sec o s, e lec ing
he specialised skills equi ed and he sec o ’s p o i abili y. Roles in in es men banking,
inancial analysis, and managemen end o command pa icula ly high sala ies. In he
e ail sec o , p oduc i i y imp o ed h ough echnology adop ion, such as e-comme ce
pla o ms, supply chain managemen sys ems, and in en o y au oma ion. No wi hs anding
he p oduc i i y g ow h, he e ail sec o con inues o ace se e al challenges, anging om
luc ua ing consume demand and high compe i ion o he need o adap o changing
echnology and supply chain dis up ions due o powe sho ages.
Economies 2024, 12, x FOR PEER REVIEW 5 o 28
Te ia y sec o s ha e he la ges con ibu ion o g oss domes ic p oduc (GDP) in
Sou h A ica. These sec o s saw he la ges decoupling o p oduc i i y and wages be ween
1993 and 2023. In 2023, hey accoun ed o 57 pe cen o o al GDP compa ed o he 31
pe cen and 12 pe cen con ibu ions o seconda y and p ima y sec o s, espec i ely. The
la ges con ibu ion, coupled wi h he la ges p oduc i i y and eal emune a ion di e -
gence (Figu e 4), he e o e implies ha e ia y sec o s may ha e had a significan bea ing
on o e all income inequali y in he Sou h A ican economy be ween 1993 and 2023. Wi hin
he e ia y sec o , financial se ices expe ienced significan p oduc i i y g ow h, d i en
by ad ancemen s in echnology, such as digi al banking, fin ech inno a ions, and au o-
ma ion o ou ine asks.
Wages in financial se ices a e gene ally high compa ed o o he sec o s, eflec ing
he specialised skills equi ed and he sec o ’s p ofi abili y. Roles in in es men banking,
financial analysis, and managemen end o command pa icula ly high sala ies. In he
e ail sec o , p oduc i i y imp o ed h ough echnology adop ion, such as e-comme ce
pla o ms, supply chain managemen sys ems, and in en o y au oma ion. No wi hs and-
ing he p oduc i i y g ow h, he e ail sec o con inues o ace se e al challenges, anging
om fluc ua ing consume demand and high compe i ion o he need o adap o changing
echnology and supply chain dis up ions due o powe sho ages.
Figu e 4. Real emune a ion and labou p oduc i i y ( e ia y sec o s). Sou ce: Own compu a ion
using da a om Quan ec.
In wha ollows, he analysis desc ibes da a and specifies he empi ical models.
2.1. Li e a u e Re iew
In de ail, he ela ionship be ween p oduc i i y and wages can be explained using
he ma ginal p oduc i i y heo y o wages. The ma ginal p oduc i i y heo y o wages
posi s ha wages a e de e mined by he ma ginal p oduc i i y o labou . In a compe i i e
labou ma ke , employe s will pay wo ke s a wage equal o he alue o hei ma ginal
p oduc . This ensu es ha labou is compensa ed ully based on i s con ibu ion o he
company’s e enue.
The heo y can be used o explain wage diffe en ials ac oss indus ies based on di -
e ences in he ma ginal p oduc i i y o wo ke s. Wo ke s in indus ies wi h highe
1993 1998 2003 2008 2013 2018 2023
Yea
Real emune a ion pe wo ke Labou p oduc i i y
Figu e 4. Real emune a ion and labou p oduc i i y ( e ia y sec o s). Sou ce: Own compu a ion
using da a om Quan ec.
In wha ollows, he analysis desc ibes da a and speci ies he empi ical models.
Li e a u e Re iew
In de ail, he ela ionship be ween p oduc i i y and wages can be explained using
he ma ginal p oduc i i y heo y o wages. The ma ginal p oduc i i y heo y o wages
posi s ha wages a e de e mined by he ma ginal p oduc i i y o labou . In a compe i i e
labou ma ke , employe s will pay wo ke s a wage equal o he alue o hei ma ginal
p oduc . This ensu es ha labou is compensa ed ully based on i s con ibu ion o he
company’s e enue.
Economies 2024,12, 330 6 o 27
The heo y can be used o explain wage di e en ials ac oss indus ies based on
di e ences in he ma ginal p oduc i i y o wo ke s. Wo ke s in indus ies wi h highe
ma ginal p oduc i i y will ea n highe wages compa ed o wo ke s employed in indus ies
wi h lowe ma ginal p oduc i i y. Fo ins ance, highly skilled wo ke s in mining indus ies
a e mo e likely o ea n highe wages due o hei highe ma ginal p oduc i i y compa ed
o wo ke s employed in he clo hing indus y.
Despi e being simplis ic and in ui i e, he ma ginal p oduc i i y heo y o wages has
ha dly li ed up o expec a ions o a a ie y o easons. One o i s key weaknesses is he
assump ion o a pe ec ly compe i i e labou ma ke whe e employe s and wo ke s ha e
comple e in o ma ion and he e a e no ba ie s o en y o exi . Labou ma ke s o en ha e
impe ec ions, which may p e en a ull pass- h ough e ec o p oduc i i y o wages.
The abo e a gumen is especially ele an in he Sou h A ican con ex , whe e he
labou ma ke is mos ly dual and cha ac e ised by s uc u al ba ie s ha make i igid.
Sou h A ica’s labou ma ke is pa icula ly cha ac e ised by o mal and in o mal sec o s,
limi ed labou mobili y, egula o y challenges, and a skills gap.
The o mal labou ma ke in Sou h A ica ypically enjoys s ong labou p o ec ions,
including minimum wage laws, social secu i y bene i s, heal hca e, and e i emen plans.
Employmen con ac s a e legally binding, and wo ke s a e en i led o dispu e esolu ion
h ough es ablished legal amewo ks. The o mal sec o is also cha ac e ised by highe
wages and be e wo king condi ions compa ed o he in o mal sec o , al hough he sec o
con inues o su e he coun y’s widesp ead challenges, which include skills gaps and
limi ed job oppo uni ies.
In s a k con as , he in o mal labou ma ke is ma ked by low-wage, uns able, and
o en unp o ec ed o ms o employmen . This sec o includes ac i i ies such as domes ic
wo k, s ee ending, casual labou , and o he o ms o sel -employmen . Wo ke s in he
in o mal sec o gene ally lack access o o mal con ac s, social secu i y, heal hca e, o o he
employmen bene i s. The in o mal sec o is la gely un egula ed; hence, wo ke s in his
sec o end o ace p eca ious condi ions, such as inconsis en wo k hou s, lack o legal
p o ec ion agains un ai dismissal, and limi ed job secu i y.
Ano he compe ing heo y is he e iciency heo y o wages, which, unlike he ma ginal
heo y o wages, posi s a unidi ec ional causali y ha uns om wages o labou p oduc-
i i y. Acco ding o he e iciency wage heo y, employe s, o a ious easons, will pay
abo e he ma ke -clea ing a e o d i e p oduc i i y g ow h (Wu and Ho 2012). The con-
adic ions be ween he ma ginal p oduc i i y heo y o wages and he e iciency heo y
o wages demons a e he need o empi ically es ing he di ec ion o causali y be o e
unning single-equa ion models.
Empi ical li e a u e linking p oduc i i y and wages has signi ican ly e ol ed o e
he yea s om ea lie s udies such as Da is and Hi ch (1949) and Boswo h e al. (1994)
o he mo e ecen li e a u e. The con adic ion o indings has led many o conside
he e ogenei ies in he ela ionship anging om he o mali y o in o mali y o labou , he
skills, and he dis inc ion be ween long- un and sho - un pe iods.
In S auss and Woha (2004), he long- un ela ionship be ween eal wages and a e age
labou p oduc i i y is conside ed a he indus y le el o a panel o 459 US manu ac u ing
indus ies obse ed be ween 1956 and 1996. Resul s ejec bo h coin eg a ion and a one-
o-one ela ionship be ween p oduc i i y and wages. An inc ease in labou p oduc i i y
is ound o inc ease wages by less han uni y. In Mahlbe g e al. (2013), he age-based
wage and p oduc i i y di e en ials a e conside ed in he con ex o Canada using da a
obse ed be ween 1999 and 2005. Obse ed in hei empi ical analysis is ha wages do no
signi ican ly de ia e om p oduc i i y, indica ing a nea one- o-one ela ionship.
Economies 2024,12, 330 7 o 27
Using da a om sub-Saha an coun ies, Van Bieseb oeck (2003) ound a one- o-one
ela ionship be ween wages and p oduc i i y in speci ic coun ies (Zimbabwe) and a less
han uni y associa ion in coun ies such as Kenya. In Tanzania, he au ho could no ind
any link be ween wages and p oduc i i y. Konings and Ma colin (2014) simul aneously
es ima e he impac o p oduc i i y on wages using ci y-le el da a obse ed be ween 2005
and 2023. They ind he p esence o a wage-p oduc i i y di e gence gap o B ussels
and Wallonia. Mawejje and Okumu (2018) speci ically ocused on A ican manu ac u ing
indus ies. Con olling o endogenei y, hei main esul is ha wages e lec labou
p oduc i i y and he skill o wo ke s.
Mo e ecen ly, in he con ex o he O ganisa ion o Economic Co-ope a ion and
De elopmen (OECD), C uz (2023) es ima ed he impac o labou p oduc i i y on eal
wages and employmen . The au ho speci ically sough o es he compe ing heo ies on
wages and p oduc i i y, namely, he ma ginal p oduc i i y heo y and he e iciency heo y
o wages. Using a panel da ase comp ising 25 OECD economies, he au ho inds a posi i e
bidi ec ional causali y be ween labou p oduc i i y and eal wages. This esul alida es he
e iciency wage heo y, in which changes in wages can also in luence labou p oduc i i y.
In Ci illo and Ricci (2022), a i m-le el da ase is employed. Resul s om di e en
quan ile eg ession models d i en by he desi e o cap u e dis ibu ional e ec s o p o-
duc i i y g ow h show ha labou p oduc i i y is in ac nega i ely associa ed wi h wage
g ow h. The nega i e associa ion be ween labou p oduc i i y and wages is ound o de-
c ease in magni ude along he quan iles. These esul s demons a e he empi ical ambigui y
cha ac e ising he ela ionship be ween p oduc i i y and wages.
In he con ex o Sou h A ica, Wake o d (2004) examines he associa ion be ween
labou p oduc i i y and a e age eal wages a he mac o le el. Using an annual ime
se ies da ase , a long- un wage–p oduc i i y elas ici y o 0.58 is con i med, alida ing he
p oduc i i y–wage di e gence. Tsoku and Ma a ise (2014) simila ly es ima ed he impac
o p oduc i i y on eal emune a ion. Thei esul s ejec he e iciency heo y o wages, as
eal emune a ion is ound no o a ec G ange causali y esul s in labou p oduc i i y.
This pape adds o he abo e empi ical e idence by employing an indus y-le el anal-
ysis ha comp ises indus ies om all sec o s o he economy. I addi ionally decomposes
labou by o mali y and skills. Las ly, i decomposes labou income in o eal wages and
eal bene i s.
3. Ma e ials and Me hods
This sec ion p esen s he me hods u ilised o achie e he aim o his s udy. The
sec ion speci ically desc ibes he da a, ou lines he es ima ion s a egy, and speci ies he
empi ical model.
3.1. Da a and Va iables
The analysis uses a mac o-panel da ase comp ising 74 h ee-digi indus ies
(N = 74)
obse ed annually om 1993 o 2023 (T = 31). This implies a mac o-panel da ase wi h
2294 annual obse a ions. Da a on all a iables a e sou ced om Quan ec. Quan ec
is a consul ancy i m p o iding economic and inancial da a, coun y in elligence, and
quan i a i e analy ical so wa e, based in P e o ia, Sou h A ica. I has been a sou ce o
da a o p ominen s udies such as Rod ik (2008) and Jenkins (2008). F om his da a sou ce,
he analysis d aws da a on eal emune a ion pe wo ke de la ed using 2015 ou pu p ices,
labou p oduc i i y ( o o mal and in o mal wo ke s and low-skilled, semi-skilled, and
skilled wo ke s), impo pene a ion, and expo in ensi y. By de ini ion, eal emune a ion
pe employee (w) is equal o he o al compensa ion o employees (C) di ided by he
numbe o employees (N), ha is, w = C/N. Uni labou cos is equal o wage a e o
ea nings pe wo ke (w) imes he numbe o wo ke s (N) di ided by he ou pu p oduced
by he wo ke s (Q), ha is, (w
×
N)/Q, whe e w
×
N is a measu e o he cos o labou .
This means ha wi h emune a ion pe wo ke and uni labou cos s, we can algeb aically
spli labou income in o bene i s and wages pe wo ke .
Economies 2024,12, 330 8 o 27
The da a sou ce ca ego ises wo ke s in o low-skilled, semi-skilled, and skilled wo ke s.
These ca ego ies can be u he ca ego ised in o o mal and in o mal wo ke s. Skilled o mal
sec o employmen comp ises o mal sec o manage s, p o essionals, and echnicians,
and i pa icula ly includes p o essional, semi-p o essional, and echnical occupa ions;
manage ial, execu i e, and adminis a i e occupa ions; ce ain anspo occupa ions;
and pilo na iga o s. Semi-skilled o mal sec o employmen comp ises o mal sec o
cle ks, sales and se ices, skilled ag icul u e, c a s and ela ed ade, plan and machine
ope a o s, and i pa icula ly includes cle ks, se ice wo ke s and shop and ma ke sales
wo ke s, skilled ag icul u al and ishe y wo ke s, c a and ela ed ades wo ke s, plan
and machine ope a o s, and assemble s. Low-skilled o mal sec o employmen comp ises
o mal sec o elemen a y wo k and domes ic wo ke s. I includes elemen a y occupa ions,
domes ic wo ke s, and o he occupa ions.
Fo mali y and in o mali y o labou is measu ed sys ema ically. Fi s , Quan ec de i es
i s employmen da a om he wo main o icial sou ces o labou da a in Sou h A ican
he Qua e ly Employmen S a is ics (QES) and he Qua e ly Labou Fo ce Su ey (QLFS)
published by S a is ics Sou h A ica (S a s SA). The QES da a a e collec ed om a sample
o non-ag icul u al en e p ises and p o ide a o mal employmen igu e. This igu e also
excludes domes ic wo ke s. The QLFS is a household-based su ey and p o ides igu es
o o al o mal and in o mal employmen as well as he o icial unemploymen igu e.
To b idge he disc epancy be ween he o mal employmen igu es om hese da ase s,
Quan ec uses he QES o mal igu e, o which i adds o mal ag icul u al and domes ic
wo ke s. Using he o al employmen om he QLFS, in o mal employmen is calcula ed
as a esidual. Labou p oduc i i y is he a io be ween ou pu (Q) and he labou inpu (LI)
used o p oduce ha ou pu , ha is, labou p oduc i i y = Q/LI = ou pu pe uni o labou
inpu . I is he e o e essen ially ou pu pe wo ke ob ained om di iding o al ou pu by
he o al numbe o wo ke s employed.
The impo –domes ic demand a io, o impo pene a ion, is equal o o al impo s (Z)
di ided by o al domes ic demand (DD) imes one hund ed, ha is, he impo –domes ic
demand a io = (Z/DD)
×
100. Domes ic demand is equal o o al ou pu plus impo s
minus expo s. The impo –domes ic demand a io is an indica ion o how much o he
domes ic demand is sa is ied by impo s. The expo -ou pu a io is a measu e o how
much o a coun y’s ou pu hey expo . The expo -ou pu a io is equal o o al expo s
(X) di ided by o al ou pu (Q) o an economy mul iplied by one hund ed, ha is, he
expo -ou pu a io = (X/Q)
×
100. Las ly, we conside g oss ma kup as an addi ional
con ol a iable. The g oss ma kup o an indus y is he ne ope a ing su plus o ha
indus y as a pe cen age o o al in e media e inpu s plus labou emune a ion o ha
indus y. I is a measu e o indus ial p o i s.
3.2. Es ima ion S a egy
The choice o an app op ia e baseline s a egy was na owed o es ima ion app oaches
ha a e sui able when bo h N and T a e qui e la ge, as esea ch in o he asymp o ics o
mac o-panels has aised he need o conside ing slope he e ogenei y (Phillips and Moon
2000;Im e al. 2003) and non-s a iona i y unde his ci cums ance. Bo h aspec s sugges
ha he adi ional ixed-e ec s o andom-e ec s and ins umen al a iable app oaches
may no be app op ia e in he p esen case gi en ha N and T a e qui e la ge, N = 74 and
T = 31,
and non-s a iona i y could be p oblema ic. The al e na i e app oach o a e aging
he annual obse a ions in o 5-yea a e ages o elimina e business cycle e ec s is less
desi able o a a ie y o easons. Fi s , i is associa ed wi h a loss o in o ma ion as he
ans o ma ion di ec ly dis o s he o iginal unde lying da a-gene a ing p ocess. Second, i
is also no ob ious ha a e age comple ely elimina es cyclical e ec s. Thi d, a e aging will
ake away he abili y o sepa a e sho - un and long- un e ec s o p oduc i i y on wages.
Economies 2024,12, 330 15 o 27
Table 3. Uni oo es s.
Va iable Tes Le els Fi s Di e ence O de o In eg a ion
Wages pe wo ke
( o mal and in o mal)
IPS 0.34452 −15.7344 *** I(1)
LLC −0.68803 −12.5431 *** I(1)
B ei ung 1.19109 −6.51086 *** I(1)
Bene i s pe wo ke
( o mal and in o mal)
IPS 1.73343 −15.2179 *** I(1)
LLC 0.50484 −12.0791 *** I(1)
B ei ung 2.21157 −8.89573 *** I(1)
No e: *** deno es signi icance a 1%, espec i ely.
Since he e idence appea s o indica e I(1) p ocesses, i was necessa y o pu sue coin-
eg a ion es ing. Coin eg a ion would exis i , in each case, he linea combina ion o he
non-s a iona y p ocesses abo e is o an I(0) s uc u e. Table 4p esen s he coin eg a ion
es esul s. In 5 ou o 8 speci ica ions, bo h es s ejec he null hypo hesis o no coin e-
g a ion. In he emaining 3 speci ica ions, a leas one es ejec s he null hypo hesis o no
coin eg a ion. Al hough he e idence o a coin eg a ing ela ionship is no o e whelming,
we assumed a long- un ela ion and elied on he sign and s a is ical signi icance o he
speed o adjus men e m in he e o -co ec ion model p esen ed sho ly o alida ion o
ebu al o he no so ob ious conclusion de i ed om Table 4.
Table 4. Coin eg a ion es s.
Dependen Va iable Ped oni Kao
G oup ADF-S a is ic Panel ADF-S a is ic
ADF S a is ic
Log emune a ion pe wo ke (bo h o mal and in o mal)
−2.315 ** 0.866 −6.487 ***
Log emune a ion pe wo ke ( o mal) −1.341 * 1.117 −6.063 ***
Log emune a ion pe wo ke (in o mal) −0.121 0.615 −2.425 ***
Log emune a ion pe wo ke (low skilled) −6.371 *** −4.688 *** −5.417 ***
Log emune a ion pe wo ke (semi-skilled) −5.837 *** −4.208 *** −6.041 ***
Log emune a ion pe wo ke (skilled) −5.441 *** −3.902 *** −6.539 ***
Log wage pe wo ke −6.689 *** −4.285 *** −6.077 ***
Log bene i s pe wo ke −6.981 *** −4.625 *** −6.487 ***
No e: *, **, and *** deno e signi icance a 10%, 5%, and 1%, espec i ely.
Table 5p esen s he Hausman speci ica ion es esul s. Ac oss all he eg ession
a ian s, he calcula ed chi-squa e Hausman s a is ic and i s co esponding p obabili y
alue u n ou o be s a is ically insigni ican a con en ional le els. Wi h his esul , i
is concluded ha he PMG es ima o , he e icien es ima o unde he null hypo hesis,
is p e e ed.
Table 5. Hausman speci ica ion es .
Va ian (1) (2) (3) (4) (5)
Hausman s a is ic 2.84 3.10 6.51 3.78 1.05
P ob > chi2 0.4172 0.3764 0.1640 0.1513 0.5961
4.2. P oduc i i y and Real Remune a ion Be ween Fo mal and In o mal Wo ke s
Table 6p esen s he es ima ed esul s om he PMG es ima ion. Al hough long-
un es ima es a e o p ima y impo ance, we also epo sho - un es ima es as we need
o compa e he sho - un and long- un ela ionship be ween p oduc i i y and wages.
The able epo s ou eg ession a ian s. The i s a ian uses eal emune a ion o
Economies 2024,12, 330 16 o 27
bo h o mal and in o mal wo ke s. The second and hi d a ian s use eal emune a ion
pe wo ke in he o mal and in o mal sec o s, espec i ely. The las a ian adds g oss
ope a ing ma kup as an addi ional con ol a iable. The i s a ian shows a long- un
p oduc i i y–wage es ima e o 0.76, sugges ing ha o a gi en a e o unemploymen
and impo pene a ion and expo in ensi y, eal emune a ion pe wo ke inc eases by
0.76 pe cen in esponse o a one pe cen inc ease in labou p oduc i i y. The es ima ed
coe icien is signi ican ly di e en om bo h ze o and one. The e m C (1) = 1 a he
bo om o he able es s he null hypo hesis ha he p oduc i i y e ec is uni y. Since
his null hypo hesis is s ongly ejec ed, we do no ind e idence ha labou p oduc i i y
g ow h in Sou h A ica eeds in o eal emune a ion pe wo ke on a one- o-one basis.
Al hough Wake o d (2004) made a simila inding, we ind a sligh ly highe elas ici y o
0.76 compa ed o he au ho ’s 0.58. Wi h espec o he sho - un dynamics, we ind he
coe icien s a is ically signi ican a he one pe cen le el bu conside ably lowe (0.361)
han he long- un es ima e. The la e esul is no su p ising since wo ke s a e mos ly
handed sho con ac s in he sho un, which makes i di icul o p oduc i i y g ow h o
ully aise eal emune a ion.
When decomposed in o o mal and in o mal wo ke s, a e ealing insigh is ha
labou p oduc i i y g ow h has a ela i ely la ge e ec on he eal emune a ion o o mal
wo ke s. Fo gi en le els o expo in ensi y and impo pene a ion and a gi en a e o
unemploymen , a one pe cen inc ease in labou p oduc i i y inc eases he emune a ion
o a o mal and in o mal wo ke by 0.62 and 0.47 pe cen , espec i ely, in he long un. The
ela i ely ma ginal esponse o eal emune a ion in he in o mal sec o is no su p ising
since in o mal wo ke s do no belong o labou unions. Unions ha e a well-documen ed
ole o ba gaining on behal o wo ke s, and e idence indica es ha o simila p oduc i i y
le els, union membe s ea n highe han non-union membe s.
In e es ingly, he coe icien o labou p oduc i i y inc eases o 0.964 once he model
con ols o g oss ma kup in he ou h a ian . A es o uni y does no ejec he null
hypo hesis a he 10 pe cen le el as indica ed a he bo om o he able. This obse a ion
implies ha a one- o-one ela ionship exis s once g oss ma kup is held cons an . Since
g oss ma kup is a measu e o p o i s, a ian 4 could imply ha pa o he p oduc i i y
g ow h ha ough o ha e e lec ed ully in wages culmina ed in p o i s. This esul is
consis en wi h he inding epo ed by Con i (2005). The au ho ound an insigni ican
link be ween aining, p oduc i i y, and wages and concluded ha i ms eap mo e o he
p oduc i i y e u ns.
Table 6. P oduc i i y g ow h and eal emune a ion.
Dependen Va iable: Log
Remune a ion pe Wo ke
Fo mal and In o mal Fo mal In o mal Fo mal and In o mal
ARDL
(1, 1, 1, 1, 1)
ARDL
(2, 1, 1, 1, 1)
ARDL
(2, 2, 2, 2, 2)
ARDL
(2, 2, 2, 2, 2)
LogLP 0.760 ***
(0.018)
0.619 ***
(0.019)
0.469 ***
(0.052)
0.964 ***
(0.024)
Expo −0.005 ***
(0.001)
0.005 ***
(0.001)
0.001
(0.002)
−0.005 ***
(0.001)
Impo 0.002 ***
(0.0005)
0.001
(0.001)
0.004**
(0.002)
0.012 ***
(0.001)
Unem −0.009 ***
(0.001)
−0.011 ***
(0.001)
−0.050 ***
(0.004)
−0.005 ***
(0.001)
Economies 2024,12, 330 17 o 27
Table 6. Con .
Dependen Va iable: Log
Remune a ion pe Wo ke
Fo mal and In o mal Fo mal In o mal Fo mal and In o mal
ARDL
(1, 1, 1, 1, 1)
ARDL
(2, 1, 1, 1, 1)
ARDL
(2, 2, 2, 2, 2)
ARDL
(2, 2, 2, 2, 2)
GM −0.010 ***
(0.001)
COINTEQ01 −0.295 ***
(0.025)
−0.338 ***
(0.029)
−0.226 ***
(0.015)
−0.238 ***
(0.025)
∆log emune a ion pe wo ke (−1) ---- 0.103 ***
(0.020)
−0.112 ***
(0.024)
0.067 ***
(0.031)
∆logLP 0.361 ***
(0.018)
0.314 ***
(0.033)
0.591 ***
(0.053)
1.187 ***
(0.047)
∆logLP (−1) ---- ---- −0.012
(0.052)
0.044
(0.045)
∆Expo 0.002
(0.002)
0.004
(0.004)
−0.020 **
(0.009)
−0.006 **
(0.003)
∆Expo (−1) ---- ---- −0.031 **
(0.014)
0.017
(0.001)
∆Impo −0.005
(0.004)
−0.006
(0.004)
−0.030 **
(0.013)
0.007
(0.004)
∆Impo (−1) ---- ---- 0.046 **
(0.022)
−0.002
(0.004)
∆Unem −0.005 ***
(0.001)
−0.003 ***
(0.001)
0.041 ***
(0.002)
0.0001
(0.001)
∆Unem (−1) ---- ---- 0.031 ***
(0.002)
−0.001
(0.001)
∆GM −0.024 ***
(0.001)
∆GM (−1) −0.001
(0.001)
C2.601
(0.219)
−1.445
(0.178)
−1.145
(0.040)
−0.738
(0.104)
@T end 0.003 ***
(0.001)
0.004 ***
(0.001)
0.015 ***
(0.001)
0.001 ***
(0.0001)
Obse a ions 2144 2070 2069 2069
C (1) = 1 169.38 *** 409.68 *** 105.45 *** 2.16
No e: ** and *** deno e signi icance a 5%, and 1%, espec i ely. Figu es in pa en heses a e s anda d e o s.
As expec ed, unemploymen en e s nega i ely, signi ican ly in he long un. A ise in
unemploymen signals a dec ease in demand o labou , which pu s downwa d p essu e on
eal emune a ion. Excep o one speci ica ion, expo s and impo s a e mos ly associa ed
wi h inc eases in eal emune a ion o gi en p oduc i i y le el and unemploymen a e in
he long un. The e o -co ec ion e m is nega i e and s a is ically signi ican , indica ing
a coin eg a ing ela ionship be ween eal emune a ion and i s speci ied de e minan s.
The speed o adjus men is, howe e , low, as only abou 23–34 pe cen o a sho - un
disequilib ium is elimina ed annually. This means a sho - un disequilib ium clea s a e
abou 3–4 yea s, which compa es wi h he 3 yea s epo ed by Wake o d (2004). The slow
speed o adjus men is no su p ising gi en he well-documen ed p esence o ic ions ha
make Sou h A ica’s labou ma ke igid. The end componen is posi i e and s a is ically
signi ican ac oss all he eg ession a ian s, indica ing ha eal emune a ion inc eased
du ing he sampling pe iod.
Economies 2024,12, 330 18 o 27
4.3. P oduc i i y and Real Remune a ion Be ween Low-Skilled, Semi-Skilled, and Skilled Wo ke s
Table 7ca ego ises labou in o low, semi-skilled, and skilled wo ke s. Fo skilled
wo ke s, he end componen en e ed insigni ican ly and was he e o e d opped o a oid
unnecessa y model o e i ing. The p e e ed speci ica ion o skilled wo ke s is he e o e
he las a ian , which excludes he end componen . We made wo obse a ions. One
is ha he ela ionship be ween p oduc i i y g ow h and eal emune a ion emains less
han one- o-one e en when labou is decomposed by he le el o skill. Two is ha in he
long un, p oduc i i y g ow h la gely inc eases he eal emune a ion o semi-skilled and
skilled wo ke s compa ed o low-skilled wo ke s. A one pe cen inc ease in p oduc i i y
pa icula ly aises eal emune a ion o semi-skilled and skilled wo ke s by 0.76 pe cen
and 0.73 pe cen o semi-skilled wo ke s, espec i ely, o a gi en a e o unemploymen ,
expo in ensi y, and impo an pene a ion. This esul indica es ha al hough p oduc i i y
g ow h ou paced eal emune a ion o all h ee ca ego ies o wo ke s du ing he sampling
pe iod, low-skilled wo ke s we e le u he behind.
Table 7. P oduc i i y g ow h and eal emune a ion.
Dependen Va iable: Log
Remune a ion pe Wo ke
Low-Skilled Semi-Skilled Skilled
ARDL
(2, 1, 1, 1, 1)
ARDL
(2, 1, 1, 1, 1)
ARDL
(1, 1, 1, 1, 1)
ARDL
(1, 1, 1, 1, 1)
LogLP 0.621 ***
(0.017)
0.761 ***
(0.016)
0.619 ***
(0.022)
0.734 ***
(0.034)
Expo −0.004 ***
(0.001)
0.004 ***
(0.001) −0.006 *** (0.001) −0.0027 **
(0.001)
Impo 0.004
(0.0005)
0.0018 ***
(0.0005)
0.0011 *
(0.0006)
−0.004 ***
(0.001)
Unem −0.015 ***
(0.001)
−0.011 ***
(0.001)
−0.001
(0.001)
−0.025 ***
(0.002)
COINTEQ01 −0.362 ***
(0.030)
−0.325 ***
(0.028)
−0.273 ***
(0.027)
−0.136 ***
(0.014)
∆log emune a ion pe wo ke (−1) 0.118 ***
(0.021)
0.091 ***
(0.022) ---- ----
∆logLP 0.299 ***
(0.030)
0.294 ***
(0.034)
0.354 ***
(0.030)
0.417 ***
(0.028)
∆Expo 0.004
(0.004)
0.003
(0.004)
0.0054 *
(0.003)
0.001
(0.002)
∆Impo −0.005
(0.004)
−0.005
(0.004)
−0.0087 *
(0.0049)
−0.002
(0.002)
∆Unem −0.004
(0.003)
−0.004 ***
(0.001)
−0.0063 ***
(0.001)
−0.004 ***
(0.001)
C−1.709
(0.135)
−1.706
(0.152)
−0.980
(0.102)
−0.466
(0.047)
@T end 0.006 ***
(0.001)
0.005 ***
(0.001)
−0.0003
(0.001) ----
C (1) = 1 490 *** 219 *** 290 *** 60 ***
Obse a ions 2070 2070 2144 2144
No e: *, **, and *** deno e signi icance a 10%, 5%, and 1%, espec i ely. Figu es in pa en heses a e s anda d e o s.
Economies 2024,12, 330 19 o 27
4.4. P oduc i i y, Wages, and Bene i s
Table 8uses wages and bene i s pe wo ke as he dependen a iables. The esul s
show ha p oduc i i y g ow h, despi e no ully e lec ing in labou income, had a la ge
impac on bene i s han wages. In he long un, a one pe cen inc ease in labou p oduc i i y
was associa ed wi h 0.69 and 0.76 pe cen inc eases in wages and bene i s pe wo ke ,
espec i ely, o a gi en a e o unemploymen , impo pene a ion, and expo in ensi y.
The e o -co ec ion e m is co ec ly signed and s a is ically signi ican , alida ing a
coin eg a ing ela ionship.
Table 8. P oduc i i y g ow h, eal wages, and eal bene i s.
Dependen Va iable: Log Real
Wages o Bene i s pe Wo ke
Wages Bene i s
ARDL
(1, 1, 1, 1, 1)
ARDL
(1, 1, 1, 1, 1)
LogLP 0.691 ***
(0.018)
0.760 ***
(0.018)
Expo −0.006 ***
(0.001)
−0.005 ***
(0.001)
Impo 0.002 ***
(0.001)
0.002 ***
(0.001)
Unem −0.004 ***
(0.001)
−0.008 ***
(0.001)
COINTEQ01 −0.306 ***
(0.026)
−0.295 ***
(0.026)
∆logLP 0.402 ***
(0.021)
0.361 ***
(0.021)
∆Expo 0.0004
(0.003)
0.0004
(0.003)
∆Impo −0.008 *
(0.005)
−0.006
(0.004)
∆Unem 0.0003
(0.001)
−0.005 ***
(0.001)
C0.411
(0.049)
2.600
(0.218)
@T end 0.003 ***
(0.001)
0.003 ***
(0.001)
Obse a ions 2144 2144
C (1) = 1 309 *** 169 ***
No e: * and *** deno e signi icance a 10%, and 1%, espec i ely. Figu es in pa en heses a e s anda d e o s.
In he sho un, he es ima ed coe icien s associa ed wi h labou p oduc i i y a e
posi i e bu conside ably smalle , sugges ing ha wages and bene i s a e ela i ely inelas ic
in he sho un compa ed o he long un. E idence u he shows ha expo dynamics
had an insigni ican impac on wages and bene i s pe wo ke in he sho un.
4.5. Robus ness Exe cises
As indica ed ea lie in he me hodology sec ion, he analysis in oked addi ional
es ima ions o obus ness pu poses. Table 9pa icula ly p esen s he esul s om he
PDOLS, while Table 10 epo s esul s om he PFMOLS. Bo h es ima ions a e based on
he g ouped me hod wi h a ime end in he coin eg a ing equa ion. Bo h es ima ions
co obo a e he esul s epo ed in he baseline me hod. A one- o-one ela ionship be ween
p oduc i i y and eal emune a ion is s ongly ejec ed, and in o mal wo ke s a e le
u he behind compa ed o o mal wo ke s. The inclusion o g oss ma kup in he ou h
Economies 2024,12, 330 20 o 27
a ian inc eases he p oduc i i y- eal emune a ion coe icien om 0.576 o 0.956. In
bo h he PDOLS and he FMOLS, he null hypo hesis o C (1) = 1 canno be ejec ed a he
10 pe cen le el once g oss ma kup is con olled o . This sugges s ha he expansion o
p o i s may be esponsible o he decoupling o p oduc i i y and eal emune a ion, as
simila ly obse ed in he baseline me hod. These obus ness exe cises con i m he e o e
ha he decoupling o p oduc i i y and eal emune a ion is no d i en by endogenei y.
Table 9. P oduc i i y g ow h and eal emune a ion—panel DOLS.
Dependen Va iable: Log
Remune a ion pe Wo ke Fo mal and In o mal Fo mal In o mal Fo mal and In o mal
LogLP 0.576 ***
(0.042)
0.606 ***
(0.040)
0.453 ***
(0.075)
0.956 ***
(0.029)
Expo 0.049 ***
(0.009)
0.049 ***
(0.009)
−0.007
(0.017)
0.029 ***
(0.008)
Impo −0.006
(0.007)
−0.006
(0.007)
0.074 ***
(0.025)
−0.015 ***
(0.004)
Unem −0.015 ***
(0.002)
−0.013 ***
(0.002)
0.019 ***
(0.004)
−0.015 ***
(0.001)
GM −0.022 ***
(0.001)
Obse a ions 2018 2024 2006 2003
C (1) = 1 101.99 *** 94.35 *** 51.91 *** 2.365725
No e: *** deno es signi icance a 1%, espec i ely. Figu es in pa en heses a e s anda d e o s.
Table 10. P oduc i i y and eal emune a ion— ully MOLS.
Dependen Va iable: Log
Remune a ion pe Wo ke Fo mal and In o mal Fo mal In o mal Fo mal and In o mal
LogLP 0.627 ***
(0.017)
0.622 ***
(0.017)
0.385 ***
(0.031)
0.953 ***
(0.009)
Expo 0.015 ***
(0.002)
0.015 ***
(0.002)
−0.016 **
(0.006)
0.002
(0.002)
Impo 0.001
(0.002)
−0.0003
(0.002)
0.021 *
(0.012)
0.011 ***
(0.001)
Unem −0.009 ***
(0.001)
−0.010 ***
(0.001)
0.028 ***
(0.001)
−0.011 ***
(0.0004)
GM −0.021 ***
(0.0003)
Obse a ions 2144 2144 2144 2144
C (1) = 1 472.75 *** 468.75 *** 377.10 *** 0.03
No e: *, **, and *** deno e signi icance a 10%, 5%, and 1%, espec i ely. Figu es in pa en heses a e s anda d e o s.
Table 11 epo s he esul s om he DCCE, which con ols o po en ial c oss-sec ional
dependence. As simila ly obse ed in he baseline es ima ion, only abou 0.6 pe cen o
a one pe cen ise in labou p oduc i i y eeds in o eal emune a ion. A compa ison
be ween o mal and in o mal wo ke s also econ i ms he disp opo iona e e ec o labou
p oduc i i y on he wo ca ego ies. Fo mal wo ke s end o sligh ly bene i mo e om
p oduc i i y g ow h compa ed o in o mal wo ke s. When he model con ols o g oss
ma kup, he coe icien on labou p oduc i i y ises om 0.626 in a ian (1) o 0.938 in
a ian (2), as obse ed in ea lie eg essions. G oss ma kup en e s nega i ely, signi ican ly
alida ing he hypo hesis ha he expansion o indus ial p o i s in Sou h A ica may ha e
come a he expense o he eal emune a ion o wo ke s. The DCCE also eplica es he
Economies 2024,12, 330 21 o 27
esul ha labou p oduc i i y has a sligh ly la ge e ec on eal emune a ion in he long
un compa ed o he sho un.
Table 11. P oduc i i y g ow h and eal emune a ion—DCCE.
Dependen Va iable: Log (1) (2) (3) (4)
Remune a ion pe Wo ke Bo h Bo h In o mal Fo mal
L. ∆. log emune a ion 0.053 * 0.022 0.0595 * −0.008
(0.029) (0.016) (0.0309) (0.031)
∆. LogLP 0.227 *** 0.553 *** 0.203 *** 0.300 ***
(0.031) (0.025) (0.033) (0.048)
∆. Impo −0.010 ** −0.009 −0.0102 ** −0.040
(0.005) (0.007) (0.005) (0.028)
∆. Expo 0.009 0.003 0.009 0.001
(0.007) (0.004) (0.007) (0.008)
∆. GM −0.015 ***
(0.001)
COINTEQ01 −0.520 *** −0.415 *** −0.529 *** −0.408 ***
(0.191) (0.133) (0.166) (0.101)
LogLP 0.626 *** 0.938 *** 0.610 *** 0.779 **
(0.195) (0.232) (0.187) (0.381)
Impo 0.0003 0.003 0.001 −0.001
(0.009) (0.016) (0.006) (0.034)
Expo 0.0004 −0.001 0.0004 0.001
(0.005) (0.008) (0.009) (0.016)
GM −0.014 *
(0.008)
Cons an −0.023 −0.897 −0.400 −0.122
(0.754) (0.709) (0.767) (0.264)
Obse a ions 2070 2070 2070 2070
R-squa ed 0.661 0.909 0.656 0.633
Numbe o g oups 74 74 74 74
CD-s a is ic −0.67 −0.83 −1.08 −0.73
CD-s a is ic (p- alue) 0.5052 0.4089 0.2809 0.4625
S anda d e o s a e in pa en heses. *** p< 0.01, ** p< 0.05, and * p< 0.1.
The CD s a is ic and i s co esponding p obabili y alue en e insigni ican ly ac oss
all he eg ession a ian s. This indica es ha he es ima ed DCCE adequa ely con ols o
c oss-sec ional dependence. The es ima ed baseline esul s a e he e o e ue e en a e
con olling o bo h endogenei y and po en ial c oss-sec ional dependence.
While he baseline app oach, he PMG app oach, accommoda es sho - un he e ogene-
i y, he assump ion o long- un homogenei y could be deemed oo es ic i e gi en he
a ying economic s uc u es o p ima y, seconda y, and e ia y sec o s. I could be a gued
ha ag icul u al, mining, manu ac u ing, and se ice indus ies may no ha e he same
long- un slope. To explo e he po en ial long- un he e ogenei y in he p oduc i i y–wage
ela ionship, we conside he mixed-e ec s model clus e ed a indus y and sec o le els.
We pa icula ly allow he slope coe icien o p oduc i i y o a y ac oss indus ies and
sec o s based on he ollowing hie a chy. Table 12 p esen s he esul s. Va ian s (1) and
(2) a e based on an indus ial clus e , while a ian s (3) and (4) a e based on he sec o al
clus e model. Va ian s (1) and (2) he e o e con ain 4 g oups o indus ies, each comp ising
ag icul u e, mining, manu ac u ing, and se ices. Va ian s (3) and (4) con ain 3 g oups,
each comp ising p ima y, seconda y, and e ia y sec o s. In all cases, we only allow he
coe icien o labou p oduc i i y o a y; o he wise, he con ol a iables in a ian s (2)
and (4) a e only included in he g an equa ion, also known as he ixed e ec s model.
Economies 2024,12, 330 22 o 27
Se e al esul s a e no ewo hy. Fi s , we no ice ha p oduc i i y elas ici y is di e en
om one a he indus y le el bu no signi ican ly di e en om one a he sec o al le el.
This is suppo ed by he elas ici ies o 0.618–0.659 a he indus y le el and 1.157–1.190
a he sec o le el and he Chi2 es o b = 1, as epo ed on he lowe pa o Table 12.
In e es ingly, he in a-clus e co ela ion (ICC) epo ed a he bo om o he able ises om
0.71 in a ian 2 (indus y clus e wi h con ols) o 0.90 in a ian 4 (sec o al clus e wi h
con ols). This is s ong e idence o sec o al clus e ing, and i sugges s s ong co ela ions
wi hin sec o s. The likelihood a io es s a e in a ou o a mixed-e ec s model as opposed
o a s anda d linea model wi h homogenous slopes. When we look a he a iance
pa ame e o in e es (Va (loglp)), we in e es ingly no e ha he a iance o he p oduc i i y
elas ici y is much highe in a ian s (3) and (4) compa ed o a ian s (1) and (2). This
sugges s ha sec o al he e ogenei y (p ima y, seconda y, and e ia y sec o s) plays a much
mo e impo an ole in explaining he di e ences be ween he p oduc i i y– emune a ion
elas ici ies compa ed o he e ogenei y wi hin indus ies (i.e., wi hin he ag icul u e indus y,
mining, manu ac u ing, and se ice indus y). O e all, his esul co obo a es he iew
ha p oduc i i y in luences he emune a ion o wo ke s di e en ly ac oss indus ies
and sec o s.
Table 12. P oduc i i y g ow h and eal emune a ion—mixed e ec s model.
Dependen Va iable: Log
Remune a ion pe Wo ke
Indus ial Clus e Sec o al Clus e
(1)
Wi hou Con ols
(2)
Wi h Con ols
(3)
Wi hou Con ols
(4)
Wi h Con ols
LogLP 0.659 ***
(0.124)
0.618 ***
(0.143)
1.157 ***
(0.280)
1.1900 ***
(0.264)
Expo 0.003
(0.009)
6.890 ***
(1.368)
0.0082 ***
(0.001)
Impo −0.00086 *
(0.0004)
−0.0032 ***
(0.0005)
Unem −0.011 ***
(0.004)
−0.0157 ***
(0.004)
Yea 0.012 ***
(0.003)
0.0099 ***
(0.009)
Cons an 8.869
(0.651)
−15.848
(6.249)
−12.931
(7.644)
Random e ec s pa ame e s
Va (loglp) 0.047 *
(0.039)
0.067 *
(0.050)
0.227 *
(0.1993)
0.200 ***
(0.0178)
Va (cons) 1.402 ***
(1.197)
0.952
(0.776)
5.435 *
(4.736)
5.093 *
(4.469)
Co (loglP, cons) 0.138 ***
(0.002)
0.131 ***
(0.003)
1.110
(1.972)
1.010 *
(0.893)
P ob > chi2 0.0000 0.0000 0.0000 0.0000
LR es s. linea model 0.0000 0.0000 0.0000 0.0000
ICC 0.786 ***
(0.143)
0.716 ***
(0.166)
0.903
(0,076)
0.901 ***
(0.078)
Chi2(1) b = 1 7.50 *** 7.04 *** 0.32 0.52
Numbe o g oups 4 4 3 3
Obse a ions 2219 2219 2219 2219
No e: * and *** deno e p< 0.1 and p< 0.01, espec i ely. S anda d e o s a e in pa en heses. ICC deno es
in a-class co ela ion.
Economies 2024,12, 330 23 o 27
Las ly, we acknowledge he need o conside addi ional socio-economic con ols
whose exclusion may po en ially mask he complex ea u es o Sou h A ica’s labou
ma ke . We conside ac o s ela ed o gende , le el o educa ion, acial composi ion, and
measu es o income inequali y. These con ols accoun o he ac ha Sou h A ica’s
labou ma ke is dual in se e al dimensions. The duali y la gely mani es s in inequali ies
be ween male and emale wo ke s, educa ion, and less-educa ed wo ke s; p i ileged and
p e iously disad an aged aces; and he ich and he poo . We cap u e he gende and
educa ion dimensions simul aneously h ough li e acy a es o males and emales. Racial
composi ion is cap u ed by he li e acy a es o h ee p e iously disad an aged aces,
namely, Black A icans, he Colou ed ace, and Indians. To a oid collinea i y, we g oup he
con ols in o wo ca ego ies, in which he i s ca ego y comp ises socio-economic con ols,
while he second ca ego y cap u es acial con ols. This is because, by measu emen , emale
and male li e acy a es a e embedded in li e acy a es by ace. Wi h espec o income
inequali y, we use economy-wide measu es o income acc uing o he bo om 10 pe cen
and he op 2.5 pe cen . The o me ep esen s he poo , while he la e ep esen s hei
ich coun e pa s. Al hough he Gini coe icien is commonly used in he li e a u e, i s da a
a Quan ec only s a in 1995.
We epo he esul s o his exe cise in Table 13. We obse e simila indings. The
addi ional socio-economic and acial con ols mos ly en e insigni ican ly. Thei inclusion
ha dly changes he main esul ha p oduc i i y eeds di ec ly in o emune a ion once
sec o al clus e ing is accoun ed o . The insigni icance o he addi ional con ol a iables
ules ou socio-economic and acial ac o s as key d i e s o he p oduc i i y–pay gap
obse ed in Sou h A ica. The wo compelling sou ces om he p esen a ion appea o be
he squeezing o wo ke s o p o i s and he he e ogenei y o sec o s.
Table 13. P oduc i i y g ow h and eal emune a ion—mixed e ec s model.
Dependen Va iable: Log
Remune a ion pe Wo ke
Indus ial Clus e Sec o al Clus e
Wi h Socio-Economic
Con ols Wi h Racial Con ols Wi h Socio-Economic
Con ols Wi h Racial Con ols
LogLP 0.6171 ***
(0.143)
0.6175 ***
(0.137)
1.197 ***
(0.406)
1.143 ***
(0.291)
Expo 0.00031
(0.0009)
0.008
(0.012)
Impo −0.000852 *
(0.005)
−0.0031
(0.003)
Unem −0.0133
0.009)
−0.011
(0.006)
Female li e acy a e 0.042
(0.130)
0.017
(0.053)
Male li e acy a e 0.043
(0.141)
−0.010
(0.046)
Log bo om 10% income 0.0155
(0.998)
0.197
(0.296)
Log op 2.5% income 0.0522
(1.125)
0.111
(0409)
Black A ican li e acy a e 0.019
(0.066)
0.045
(0.081)
Colou ed li e acy a e 0.0125
(0.198)
0.074
(0.244)
Indian li e acy a e 0.0044
(0.019)
0.022
(0.135)
Economies 2024,12, 330 24 o 27
Table 13. Con .
Dependen Va iable: Log
Remune a ion pe Wo ke
Indus ial Clus e Sec o al Clus e
Wi h Socio-Economic
Con ols Wi h Racial Con ols Wi h Socio-Economic
Con ols Wi h Racial Con ols
Yea 0.0182
(0.042)
0.032
(0.013)
0.009
(0.021)
0.003
(0.016)
Cons an −26.423
(78.392)
15.436
(25.823)
−11.709
(39.353)
14.063
(31.894)
Random e ec s pa ame e s
Va (loglp) 0.0672 *
(0.050)
0.0602 *
(0.052)
0.199 ***
(0.0108)
0.245 **
(0.114)
Va (cons) 1.402 *
(1.197)
1.101 *
(0.909)
5.059 ***
(2.667)
5.093 *
(4.469)
Co (loglP, cons) 0.9541 *
(0.779)
0.133
(0.160)
1.002 **
(0.538)
1.194 *
(1.041)
P ob > chi2 0.0000 0.0000 0.0000 0.0000
LR es s. linea model 0.0000 0.0000 0.0000 0.0000
ICC 0.71603 ***
(0.085)
0.743 ***
(0.157)
0.899 ***
(0.079)
0.910 ***
(0.071)
Chi2(1) b = 1 6.17 ** 7.77 *** 0.56 0.24
Numbe o g oups 4 4 3 3
Obse a ions 2219 2219 2219 2219
No e: *, **, and *** deno e p< 0.1, p< 0.05, and p< 0.01, espec i ely. S anda d e o s a e in pa en heses. ICC
deno es in a-class co ela ion.
The nex sec ion discusses he key indings o his s udy.
4.6. Discussion
The e idence shows ha Sou h A ica’s labou ma ke is igid, judging by he slow
speed o adjus men s obse ed in he baseline es ima o . This esul is consis en wi h
ea lie indings obse ed by Wake o d (2004) and mo e ecen ly by Habanabakize e al.
(2019). In hese s udies, Sou h A ica’s labou ma ke is ound o be simila ly igid in
adjus ing o sho - un disequilib ium a a a e anging om 4 o 37 pe cen . This ange ac-
commoda es ou a e o 13–36 pe cen . This obse a ion alida es conce ns abou complex
and s uc u al impedimen s ha cha ac e ise Sou h A ica’s labou ma ke anging om
limi ed labou mobili y and high ma ke powe o indus ies o in o mali y, dualism, and
egula o y challenges.
The esul ha labou p oduc i i y has ou paced wages and bene i s is in line wi h
Wake o d (2004), who ound a 1 pe cen ise in p oduc i i y associa ed wi h a ise o
app oxima ely 0.58 pe cen in eal wages. This is ema kably close o he 0.576, which
we ind in Table 9. The p oduc i i y–pay gap obse ed in his pape is also consis en
wi h schola s such as Webs e (1991) and mo e ecen ly Wolpe (2023), who desc ibed
Sou h A ica as a capi alis ic socie y. The la e speci ically aces he p oduc i i y–pay
gap o he coun y’s his o y o seg ega ion unde Apa heid, in which he majo i y o
wo ke s we e sou ced om ese es as cheap labou . The desc ip ion o Sou h A ica as a
capi alis ic socie y is u he alida ed by he one- o-one p oduc i i y–wage associa ion,
which eme ges once we con ol o indus ial ma kups.
When b oken down in o o mal and in o mal wo ke s, we ind a esul ha is con-
sis en wi h Sou h A ica’s expe iences, pa icula ly since he abolishmen o Apa heid
in 1994. We pa icula ly ind p oduc i i y o ha e disp opo iona ely le behind in o -
mal wo ke s. Sou h A ica’s labou ma ke has s a k dualism in which he majo i y o
