Closer and greener? Firms' global value chain participation and sustainability

Semrau, Finn, Ole; Semrau, Finn Ole
Closer and greener? Firms' global value chain participation and sustainability

Author: Semrau, Finn Ole
Year: 2024
Source: https://macau.uni-kiel.de/servlets/MCRFileNodeServlet/macau_derivate_00005606/Semrau_GVCs and sustainability_thesis.pdf
Close and g eene ? Fi ms’ global alue
chain pa icipa ion and sus ainabili y
Inaugu al-Disse a ion
zu E langung des akademischen G ades eines Dok o s
de Wi scha s- und Sozialwissenscha en
de Wi scha s- und Sozialwissenscha lichen Fakul ¨a
de Ch is ian-Alb ech s-Uni e si ¨a zu Kiel
Vo geleg on:
Finn Ole Sem au, M.Sc.
aus Hens ed -Ulzbu g
Kiel, 2024
Ged uck mi Genehmigung de Wi scha s- und Sozialwissenscha lichen Fakul ¨a de
Ch is ian-Alb ech s-Uni e si ¨a zu Kiel
E s begu ach ung: P o . Aoi e Hanley, Ph.D.
Zwei begu ach ung: P o . D . Ka in Rehdanz
D i begu ach ung: P o . D . E ic S obl
Tag de m¨undlichen P ¨u ung: 28.03.2023
Con en s
1 In oduc ion 1
2 Fi ms’ global alue chain pa icipa ion and i s en i onmen al pe o mance -
A e iew o he empi ical li e a u e 11
2.1 In oduc ion....................................... 12
2.2 Fi ms’ global alue chain pa icipa ion and en i onmen al eﬃciency in p oduc ion 13
2.3 Fi ms’ global alue chain pa icipa ion and en i onmen al inno a ion adop ion . 17
2.4 Main esul s and conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3 S epping up o he ma k? Fi ms’ expo ac i i y and en i onmen al inno a-
ion in 14 Eu opean coun ies 28
3.1 Addi ional appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4 On he D i e s o Clean P oduc ion: Fi ms’ Global Value Chain Posi ioning 42
4.1 In oduc ion....................................... 44
4.2 Fi m’s GVC Posi ioning and Clean P oduc ion . . . . . . . . . . . . . . . . . . . 47
4.3 Me hodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.3.1 Da a Desc ip ion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.3.2 Empi ical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.3.2.1 Measu ing a Fi m’s Clean P oduc ion . . . . . . . . . . . . . . . 53
4.3.2.2 Measu ing a Fi m’s GVC Posi ioning . . . . . . . . . . . . . . . 54
4.3.2.3 Measu ing Exposu e o En i onmen al Policy S ingency in Ex-
po Ma ke s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.3.3 Desc ip i e S a is ics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.4 Resul s.......................................... 57
4.4.1 A Fi m’s GVC Posi ioning and Clean P oduc ion . . . . . . . . . . . . . 57
4.4.2 Ins umen al Va iable App oach . . . . . . . . . . . . . . . . . . . . . . . 60
4.4.3 The Mode a ing Roles o Expo S a us and Exposu e o EPS . . . . . . 62
4.4.4 Robus ness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.5 Conclusion ....................................... 65
4.6 Appendix ........................................ 68
4.6.1 Calcula ion o CO2emissions . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.6.2 Addi ional s a is ical ou pu s . . . . . . . . . . . . . . . . . . . . . . . . . 70
i
5 Co po a e Social Responsibili y along he Global Value Chain 76
5.1 In oduc ion....................................... 78
5.2 A Model o CSR along he Global Value Chain . . . . . . . . . . . . . . . . . . . 82
5.2.1 Incomple e Con ac ibili y o CSR . . . . . . . . . . . . . . . . . . . . . . 82
5.2.2 Se up: P e e ences and Technology . . . . . . . . . . . . . . . . . . . . . . 84
5.2.2.1 P e e ences, CSR, and Pe cei ed E hical Quali y . . . . . . . . . 84
5.2.2.2 P oduc ion o Physical Ou pu . . . . . . . . . . . . . . . . . . . 85
5.2.3 Sou cing wi h Sequen ial P oduc ion . . . . . . . . . . . . . . . . . . . . . 86
5.2.3.1 Rela ional s. Anonymous Sou cing . . . . . . . . . . . . . . . . 86
5.2.3.2 Demand and Fi m Re enue . . . . . . . . . . . . . . . . . . . . . 87
5.2.3.3 Hold-Up and Ba gaining . . . . . . . . . . . . . . . . . . . . . . 88
5.2.3.4 Inc emen al Con ibu ion a S age m............... 88
5.2.3.5 Ba gaining along he Value Chain . . . . . . . . . . . . . . . . . 89
5.2.4 CSR In es men s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
5.2.5 Equilib ium CSR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
5.2.5.1 Indus y Equilib ium . . . . . . . . . . . . . . . . . . . . . . . . 91
5.2.5.2 Equilib ium CSR along he Global Value Chain . . . . . . . . . 92
5.2.6 Discussion and Link o he Da a . . . . . . . . . . . . . . . . . . . . . . . 93
5.2.6.1 Tes able Implica ions: Pa ial s. Indus y Equilib ium . . . . . 93
5.2.6.2 A Di ec Measu e o he Non-Con ac ible Va iable . . . . . . . 93
5.3 GVC Posi ion and CSR in India . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
5.3.1 Da a Desc ip ion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
5.3.2 Measu ing CSR Spending along GVCs . . . . . . . . . . . . . . . . . . . . 95
5.3.2.1 Measu ing a Fi m’s CSR Spending . . . . . . . . . . . . . . . . . 95
5.3.2.2 Measu ing a Fi m’s Value Chain Posi ion . . . . . . . . . . . . . 96
5.3.3 Empi ical Speciﬁca ion............................. 98
5.3.3.1 Empi ical Se -Up . . . . . . . . . . . . . . . . . . . . . . . . . . 99
5.3.3.2 Desc ip i e S a is ics . . . . . . . . . . . . . . . . . . . . . . . . 100
5.3.3.3 Empi ical Resul s . . . . . . . . . . . . . . . . . . . . . . . . . . 101
5.3.4 Robus ness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
5.3.4.1 Smoo hed CSR Spending . . . . . . . . . . . . . . . . . . . . . . 103
5.3.4.2 Con olling o Visibili y . . . . . . . . . . . . . . . . . . . . . . 104
5.4 Conclusion .......................................104
5.5 Theo y Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
5.5.1 Indus y Equilib ium wi h Incomple e Con ac s . . . . . . . . . . . . . . 107
5.5.2 Comple e Con ac s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
5.5.2.1 Le el o CSR wi h Comple e Con ac s and Compa ison o In-
comple e Con ac s . . . . . . . . . . . . . . . . . . . . . . . . . 109
5.5.2.2 Remaining Indus y Equilib ium Resul s . . . . . . . . . . . . . 109
5.5.3 The Ex ended Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
5.5.3.1 Incomple e Con ac s . . . . . . . . . . . . . . . . . . . . . . . . 111
5.5.3.2 Comple e Con ac s . . . . . . . . . . . . . . . . . . . . . . . . . 112
ii
5.6 Empi ical Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
5.6.1 Wages:......................................113
5.6.2 OECD expo exposu e: . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
E kl¨a ung zum selbs ¨andigen Ve assen de A bei und Einhal ung de DFG-
Rich linien 121
Acknowledgemen 124
iii

Index o abb e ia ions
CIS Communi y Inno a ion Su ey
CMIE Cen e o Moni o ing Indian Economy
CSR Co po a e Social Responsibili y
EI En i onmen al Inno a ion
EPS En i onmen al Policy S ingency
GHG G eenhouse gas
GVC Global alue chain
IEA In e na ional Ene gy Agency
I W Kiel Ins i u e o he Wo ld Economy
IPCC In e go e nmen al Panel on Clima e Change
KCG Kiel Cen e o Globaliza ion
NAFTA No h Ame ican F ee T ade Ag eemen
PPML Poisson pseudo maximum likelihood
SDG Sus ainable De elopmen goal
WIOD Wo ld Inpu -Ou pu Da abase
i
Chap e 1
In oduc ion
The yea 2015 was a pi o al yea o global sus ainable de elopmen . By ag eeing on he
2030 Agenda o Sus ainable De elopmen and he Pa is Ag eemen , he global communi y
commi ed o sus ainable de elopmen a all le els o go e nance. Unques ionably, bo h ag ee-
men s a e an u gen call o sus ainable de elopmen , including ac ions ela ed o social and
en i onmen al domains. I is now he esponsibili y o all coun ies o ensu e ha he s a ed
goals a e eached. A he hea o he Pa is Ag eemen is he objec i e o hold “ he inc ease
in he global a e age empe a u e o well below 2°C abo e p e-indus ial le els and pu suing
eﬀo s o limi he empe a u e inc ease o 1.5°C abo e p e-indus ial le els” (Pa is Ag eemen ,
2015, p.3). A ailu e o his goal would cause emendous clima e- ela ed haza ds o ecosys ems
and humans. Among many o he s, hese include losses in biodi e si y, physical wa e sca ci y,
an inc easing numbe o ex eme wea he e en s, mo e widesp ead diseases such as dengue, a
ise in sea le els h ea ening coas al a eas and e en en i e coun ies and an inc eased p essu e
on ood p oduc ion and access (IPCC, 2022). E en acing hese emendous consequences, i
seems as i he wo ld is a oﬀ- ack eaching he goal o limi global wa ming o well below
2°C. Ne e heless, any eﬀo o limi global wa ming is desi able, as many clima e- ela ed isks
inc ease disp opo iona ely wi h ising empe a u e (IPCC, 2022). The ac ha he global
communi y s ill has a long way o go is shown, o example, by he de elopmen o CO2emis-
sions o e ime. Figu e 1.1 illus a es he de elopmen o CO2emissions om ossil uels o e
he las wo cen u ies spli by wo ld egion.
Be ween 2019 and 2020, global CO2emission dec eased sligh ly om 36.7 billion onnes o
34.81 billion onnes. Howe e , his can be explained by he economic slowdown ela ed o he
global COVID-19 pandemic. In gene al, he ﬁgu e shows ha global CO2emissions con inue o
ise and ha e no ye peaked. Be ween 2000 and 2020, CO2emissions inc eased by 38% om
25.23 billion onnes o 34.81 billion onnes. Copeland e al. (2022) emphasise ha mos emission
g ow h o igina es om de eloping and eme ging coun ies. This pa e n is also isible in he
ﬁgu e. Undoub edly, China is he elephan in he oom. China’s CO2emissions inc eased by an
as onishing 210% om 3.44 billion onnes in 2000 o 10.67 billion onnes in 2020. In 2020, China
announced i s a ge o peak CO2emissions by 2030 and achie e ca bon neu ali y be o e 2060
(In e na ional Ene gy Agency, 2021a). These a ge s equi e China o decouple i s de elopmen
om emissions a an ea lie s age o de elopmen han o he ad anced economies in he pas ,
including he US, which has his o ically been he la ges con ibu o o global CO2emissions.
E en as India walks in China’s oo s eps, i s successes in g een ans o ma ion se he pace o
1
Figu e 1.1: G aphical illus a ion by Ri chie e al. (2020) o he global CO2emissions om
ossil uels spli by egion based on da a o he Global Ca bon Budge (2021)
he wo ld o ackle clima e change. India’s CO2emissions inc eased by 149% om 0.98 billion
onnes o 2.44 billion onnes be ween 2000 and 2020. As a esul , India is now he hi d la ges
emi e o CO2emissions, al hough i s pe capi a emissions a e signiﬁcan ly lowe compa ed o
he his o ical majo emi e s in Eu ope and No h Ame ica. All in all, he inc ease in emissions
in India and China illus a es ha global success in comba ing clima e change is u gen ly
needed no only in indus ialised coun ies bu also in eme ging and de eloping coun ies. A
he same ime, eme ging and de eloping coun ies a e especially ulne able o he isks o
clima e change (IPCC, 2022). Add essing clima e- ela ed isks poses addi ional challenges o
de eloping coun ies han hey al eady ace in add essing social- ela ed sus ainable de elopmen
goals, such as e adica ing po e y and hunge , imp o ing educa ion and achie ing heal h and
well-being.
In his doc o al hesis, I ocus on he ole o ﬁ ms o each sus ainable de elopmen . Mo e
p ecisely, I con ibu e o he unde s anding o how ﬁ ms’ global alue chain (GVC) pa icipa ion
ela es o hei sus ainabili y, co e ing en i onmen al and social pe o mance. GVCs a e a
mechanism, whe e ﬁ ms om diﬀe en coun ies a e connec ed by he ans e o goods, capi al
and wo ke s. These coun ies, in u n, a e cha ac e ised by diﬀe en s ages o de elopmen and
pa e ns o echnology, knowledge and legisla ion ela ing o social and en i onmen al aspec s.
Se e al s udies show ha GVC pa icipa ion, such as by expo ing, can be accompanied by
imp o emen s in he en i onmen al pe o mance o ﬁ ms (e.g., Ba ako a and Da ies (2012);
2
Fo slid e al. (2018); Gi ma and Hanley (2015); Rich e and Schie sch (2017)). Among o he
hings, I con ibu e o he li e a u e by analysing how he cha ac e is ics o ﬁ ms’ coun y o
o igin and he coun y o which a ﬁ m is linked h ough pa icipa ion in GVCs, mode a e he
espec i e ela ionship be ween GVC pa icipa ion and sus ainable pe o mance.
Ce ainly, unde s anding he ela ionship be ween ﬁ ms’ GVC pa icipa ion and sus ain-
abili y is only one pa o a complex ne wo k o how economic ac i i y ela es o sus ainable
de elopmen . In hei ﬂagship epo , he Wo ld Bank (2020) ou lined he ise o GVCs a e he
1990s and how i con ibu ed o g ow h, jobs, and he educ ion o po e y. On he one hand,
he epo poin s o en i onmen al deg ada ion, such as ou sou cing di y p oduc ion s eps o
coun ies wi h weak en i onmen al egula ion s ingency. On he o he hand, hey emphasise
ha GVC pa icipa ion migh lead o echnology ans e , he upg ading o s anda ds and he
ansmission o in angible asse s, such as good manage ial p ac ices. O e all, I ollow Copeland
e al. (2022) who emphasise he impo ance o decision make s o unde s anding he complex
ela ionship be ween globalisa ion and he en i onmen .
Chap e 2, 3 and 4 analyse how ﬁ ms’ GVC pa icipa ion ela e o imp o emen s in en i on-
men al pe o mance.1The diﬀusion o clean echnologies and he ela ed g een ans o ma ion
o indus ies is di ec ly linked o he Pa is Ag eemen and SDGs 9 and 12. The In e na ional
Ene gy Agency (2021b) emphasises ha he echnologies needed o achie e he necessa y deep
cu s in global emissions by 2030, al eady exis . Wha is missing is an immedia e and massi e
deploymen o clean and ene gy-eﬃcien echnologies, al eady deployed o a ailable in indus y.
The ansmission o hese echnologies is especially impo an o ﬁ ms based in de eloping and
eme ging coun ies, cha ac e ised by being - on a e age - u he away om he echnology
on ie (Gu i´e ez and Teshima, 2018). As ou lined be o e, de eloping and eme ging coun ies
ep esen he main sou ce o emission g ow h, a e mo e ulne able o clima e isks and need o
decouple economic g ow h om en i onmen al deg ada ion a an ea lie s age o de elopmen
han ad anced economies.
To begin wi h, in Chap e 2, I summa ise he ele an empi ical li e a u e on he ela ionship
be ween ﬁ ms’ GVC pa icipa ion and en i onmen al pe o mance.2By discussing he ela ed
li e a u e, he li e a u e e iew also p esen s he main con ibu ions o Chap e 3 and Chap-
e 4. A he co e o he li e a u e e iew is a discussion o wo ela i ely sepa a e li e a u e
s ands. Applying insigh s om bo h li e a u e s ands, I iden i y nine key ﬁndings and possible
esea ch gaps. The ﬁ s li e a u e s and, s a s wi h he seminal con ibu ion by G ossman
and K uege (1991), ocusing on he impac o ade on he en i onmen . I is ollowed by
many o he con ibu ions, using agg ega ed-da a (G ossman and K uege , 1995; Chichilnisky,
1994; Copeland and Taylo , 1994; An weile e al., 2001). Mo e ecen ly, s udies using ﬁ m
le el da a ha e been gaining momen um (Che niwchan, 2017). By using ﬁ m le el da a, e-
1Please no e ha he w i ing s yle o each chap e diﬀe s due o he diﬀe en equi emen s o he jou nals o
which he manusc ip s ha e been submi ed.
2The chap e is an ea lie e sion o “Sem au, F. O. (2023). “Chap e 8: Fi ms’ global alue chain pa icipa-
ion and i s en i onmen al pe o mance: a e iew o he empi ical li e a u e.” Handbook on Inno a ion, Socie y
and he En i onmen edi ed by Edi ed by Fe nando J. D´ıaz L´opez, Massimiliano Mazzan i, and Robe o Zoboli,
125-139.” The published manusc ip is accessible ia h ps://doi.o g/10.4337/9781802200065.00016. Since
he chap e includes a de ailed discussion o he li e a u e ele an o he Chap e s 3, 5 and 4, i subs i u es o
some ex end he li e a u e e iew ypically pa o an in oduc ion.
3
Timme , M. P., E. Die zenbache , B. Los, R. S eh e , and G. J. V ies (2015): “An
illus a ed use guide o he Wo ld Inpu –Ou pu Da abase: The case o global au omo i e
p oduc ion,” Re iew o In e na ional Economics, 23, 575–605.
Wo ld Bank (2020): “Wo ld De elopmen Repo 2020: T ading o de elopmen in he age
o global alue chains,” A wo ld bank g oup ﬂagship epo , Wo ld Bank G oup, Washing on.
10

Chap e 2
Fi ms’ global alue chain pa icipa ion and i s
en i onmen al pe o mance - A e iew o he
empi ical li e a u e∗†
Finn Ole Sem au
Kiel Ins i u e o he Wo ld Economy (I W)
Kiel Cen e o Globaliza ion (KCG)
ﬁnn-ole.sem [email p o ec ed]
Abs ac
Global diﬀusion o p e iously de eloped clean and ene gy eﬃcien echnologies is con-
inuing o deﬁne he pace o he in e na ional ansi ion o a g een economy o e he nex
ew yea s. Empi ical ﬁ m-le el analyses oﬀe a powe ul app oach o in es iga e how a
ﬁ m’s global alue chain pa icipa ion ela es o he adop ion o g een echnologies. In
his chap e , I e iew insigh s e e ing o wo ela i ely sepa a ed s ands o li e a u e.
Fi s , schola s om he in e na ional economics li e a u e p o ide s ong e idence o an
en i onmen al eﬃciency p emium om expo s, o eign di ec in es men s and p oximi y o
ﬁnal consump ion. Second, schola s analysing he de e minan s o en i onmen al inno a ion
adop ion go beyond eﬃciency a gumen s and co e a di e se se o en i onmen al domains.
Resul s om he la e a e ambiguous and c ucially depend on he sample unde in es iga-
ion. I conclude he e iew wi h he iden iﬁca ion o nine key ﬁndings and a discussion o
possible esea ch gaps.
JEL Classiﬁca ion Numbe s: F18, F64, O33, Q55, Q56
Keywo ds: T ade and en i onmen ; En i onmen al/ eco/ g een inno a ion; echnological
change; global alue chains; impac s o globaliza ion; li e a u e e iew
∗The chap e is an ea lie e sion o : “Sem au, F. O. (2023). “Chap e 8: Fi ms’ global alue chain pa -
icipa ion and i s en i onmen al pe o mance: a e iew o he empi ical li e a u e.” Handbook on Inno a ion,
Socie y and he En i onmen edi ed by Fe nando J. D´ıaz L´opez, Massimiliano Mazzan i, and Robe o Zoboli,
125-139.” The published a icle is accessible ia h ps://doi.o g/10.4337/9781802200065.00016. Fo ci a ion
please e e o he published a icle. Fu he , he chap e should no be eused o pos ed independen ly o he
hesis wi hou u he pe mission.
†I would like o hank Aoi e Hanley, Pe e Kannen, Jakob Leh , Massimiliano Mazzan i, F auke S eglich and
Joschka Wanne o aluable commen s du ing he w i ing o his chap e .
11
2.1 In oduc ion
Immedia e and massi e deploymen o al eady de eloped clean and ene gy eﬃcien echnolo-
gies c ucially deﬁne he pace o deep cu s in global emissions by 2030 (In e na ional Ene gy
Agency, 2021). Veugele s (2012) emphasises he u gency o un a clean inno a ion machine a
ull speed in he ligh o he sizeable clima e challenge we ace. Global diﬀusion o he a ailable
g een echnologies is necessa y o each ull s eam ahead in he g een economic ansi ion, e.g.
educing g eenhouse gas emissions and o he en i onmen al nega i e impac s.
In his chap e , I e iew empi ical esea ch on how ﬁ ms’ global alue chain pa icipa ion,
including expo s, o eign owne ship, posi ioning along a global alue chain and indi ec ex-
posu e o in e na ional ade ia impo compe i ion, is associa ed wi h ﬁ ms’ en i onmen al
pe o mance. In so doing, I iden i y nine key ﬁndings and discuss possible emaining esea ch
gaps.
O e all, he impac o globalisa ion on he en i onmen is ambiguous (Che niwchan e al.,
2017; Copeland e al., 2022). The empi ical use o ﬁ m-le el da a oﬀe s a powe ul app oach
o be e unde s and he in e ace be ween globalisa ion and he en i onmen , since i en-
ables esea che s o accoun o he he e ogenei y o ﬁ ms, e en in na owly deﬁned indus ies
(Che niwchan e al., 2017; Meli z, 2003).1
Summa ising wo main s ands o empi ical li e a u e o ms he co e o his chap e . Schol-
a s pa o he ﬁ s li e a u e s and, I cha ac e ise as belonging o he in e na ional economics
li e a u e, equen ly adop he pe spec i e o he Meli z (2003) model o analyse induced e -
ec s caused by global alue chain pa icipa ion. These eﬀec s include ene gy consump ion, CO2
emissions and emissions om o he pollu an s. Schola s om he second li e a u e s and em-
ana e om he inno a ion li e a u e. He e, he ocus is on ex e nal and in e nal ac o s d i ing a
ﬁ m’s decision o adop en i onmen al inno a ion and g een managemen ools.2Su p isingly,
bo h s ands o li e a u e a e ela i ely unconnec ed. Acco dingly, ma ching heo e ical and
empi ical amewo ks ep esen s a p omising a enue o u u e esea ch.
The in e na ional economics li e a u e widely ag ees on ﬁ m-le el en i onmen al eﬃciency
gains induced by diﬀe en o ms o global alue chain pa icipa ion and exposu e o ade such
as expo s, impo compe i ion, o eign acquisi ion and p oximi y o ﬁnal consump ion (e.g.
Ba ako a and Da ies (2012); B ucal e al. (2019); Che niwchan e al. (2017); Gu i´e ez and
Teshima (2018); Sem au (2022)). Con e sely, he esul s on how a ﬁ m’s global alue chain
pa icipa ion links o he diﬀusion o en i onmen al inno a ion is less clea -cu and esul s
depend on ac o s such as local embeddedness, he idiosync asies o a ﬁ m’s home coun y, he
ype o en i onmen al inno a ion and whe he non-inno a i e ac i e ﬁ ms a e excluded om he
empi ical analysis (e.g. Cainelli e al. (2012); Del R´ıo e al. (2016); Ghise i e al. (2015); Hanley
and Sem au (2022)). Some aspec s, like posi ioning in a alue chain o impo compe i ion a e
1Analysing he whole eﬀec o ade on he en i onmen goes beyond he scope o his li e a u e e iew. To
his end, I e e o a e y ecen con ibu ion on globalisa ion and he en i onmen by Che niwchan e al. (2017)
and Copeland e al. (2022).
2I ollow Ghise i and Pon oni (2015) and use he e m en i onmen al inno a ion a he han eco-inno a ions.
E en hough many schola s use he e ms synonymously, eco-inno a ions we e ini ially concei ed as a sub-
class o en i onmen al inno a ions, aﬀec ing a ﬁ m’s economic and en i onmen al pe o mance. En i onmen al
inno a ion, pe se, limi s he ocus o a ﬁ m’s en i onmen al pe o mance.
12
widely neglec ed so a . The nex sec ion discusses he s a ed esea ch s ands in mo e de ail.
2.2 Fi ms’ global alue chain pa icipa ion and en i onmen al
eﬃciency in p oduc ion
The discussion on how ade aﬀec s he en i onmen goes back o he seminal con ibu-
ions o G ossman and K uege (1991), G ossman and K uege (1995)), Chichilnisky (1994),
Copeland and Taylo (1994), An weile e al. (2001) and F ankel and Rose (2005). Gene ally,
he agg ega ed eﬀec s o ade on he en i onmen can be spli in o a scale-, composi ion- and
echnique eﬀec . Fi s , ade inc eases he scale o economic ac i i y ha ansla es in o a
highe en i onmen al oo p in . Second, ade libe alisa ion can shi he place o p oduc ion in
line wi h a coun y’s compa a i e ad an age ha has an ambiguous eﬀec on he en i onmen .
Thi d, he echnique eﬀec e e s o ade- ela ed changes in he en i onmen al cos s o p o-
duc ion, e.g. d i en by he adop ion o mo e eﬃcien machine y and equipmen . Theo e ical
p edic ions on he o al eﬀec o ade on he en i onmen depend on he de e minan s o a
coun y’s compa a i e ad an age (Cole and Ellio , 2003). Acco dingly, empi ical in es iga ion
needs o be guided by ou cu en unde s anding ega ding he en i onmen al impac s o glob-
alisa ion. In his chap e , I widely limi he discussion o empi ical analyses based on ﬁ m-le el
da a.
Global alue chains in ol e spli ing p oduc ion p ocesses ac oss diﬀe en coun ies and al-
lows ﬁ ms o specialise in asks. The Meli z model p edic s ha mo e p oduc i e ﬁ ms sel -selec
in o expo ing (Meli z, 2003). Bus os (2011) ex ends he heo e ical model o Meli z o echnol-
ogy choice and p edic s ha expo - ela ed inc eases in e enues enables expo e s o upg ade
echnology, e.g. inc ease spending on compu e s, so wa e, echnology ans e s, pa en s, and
inno a ion ac i i ies. The p oposed mechanism is empi ically shown o A gen inean ﬁ ms o
he pe iod 1992 o 1996, a pe iod cha ac e ised by a sha p d op o a iﬀs be ween A gen ina
and B azil due o he ade libe alisa ion o MERCOSUR.
Ba ako a and Da ies (2012) apply he heo e ical p edic ions o en i onmen al- ela ed
ou comes, mo e p ecisely o ene gy consump ion o I ish manu ac u ing ﬁ ms be ween 1991 and
2007. They ﬁnd ha expo en y is posi i ely connec ed wi h ene gy use, while i is nega i ely
associa ed wi h ene gy in ensi y. In e es ingly, hey dis inguish be ween high and low ene gy
consume s. In pa icula high ene gy consume s can oﬀse highe ene gy demand by eﬃciency
gains.
Cui e al. (2016) analyse he ela ion be ween expo and en i onmen al pe o mance o
U.S. ﬁ ms in 2002, 2005 and 2008. They ﬁnd ha mo e p oduc i e ﬁ ms epo lowe emission
in ensi ies (sulphu dioxide (SO2), ca bon monoxide (CO), ozone (O3) and o als suspended
pa icula es (TSPs) pe alue o sales). Condi ioned on ﬁ ms’ p oduc i i y, expo ing ﬁ ms
en i onmen ally pe o m be e compa ed o ﬁ ms only se ing he domes ic ma ke .
Rich e and Schie sch (2017) ﬁnd e idence suppo ing an expo e ’s en i onmen al eﬃciency
p emium - measu ed as CO2p oduc i i y. The used Ge man ﬁ m-le el da a span he pe iod
2003 o 2011 and con ain de ailed in o ma ion on a ﬁ m’s uel consump ion ha allows o
a calcula ion o di ec CO2emissions om p oduc ion. They ﬁnd ha an inc ease in expo
13
in ensi y by one pe cen age poin , leads o a e age CO2emission eﬃciency gains by 0.22%.
Fo slid e al. (2018) heo e ically and empi ically shed ligh on he mechanism as o how a
ﬁ m’s expo ac i i y ela es o CO2eﬃciency when emissions a e subjec ed o an en i onmen al
ax. A e expo en y, ﬁ ms scale-up p oduc ion in o de o mee o eign demand. This scaling-
up o p oduc ion enables ﬁ ms o sp ead ﬁxed p oduc ion cos s ac oss an inc eased numbe o
uni s. Incen i ised by exposu e o en i onmen al policy egula ions, he ﬁxed cos s o aba emen
can also be alloca ed among mo e uni s, inc easing he a ac i eness o in es men s in CO2
eﬃciency. They back up hei heo e ical p edic ions wi h empi ical e idence using da a o
Swedish manu ac u ing ﬁ ms o he ime pe iod 2004 o 2011.
Holladay (2016) employs da a conce ning oxic chemical emissions om U.S. manu ac u ing
es ablishmen s, o he pe iod 1990 o 2006. He ﬁnds ha expo e s gene a e less pollu ion
emissions compa ed o non-expo e s om he same indus y once ou pu le els a e conside ed in
he equa ions. In an addi ional ﬁnding, he e eals ha impo compe i ion educes he domes ic
p ices o a ie ies and d i es he leas p oduc i e and mo e pollu ion-in ensi e es ablishmen s
ou o business.
Shapi o and Walke (2018) show ha be ween 1990 and 2008 he ai pollu ion emissions om
he U.S. manu ac u ing sec o dec eases by 60% - p ima ily d i en by wi hin-p oduc changes in
emission in ensi y induced by en i onmen al egula ion a he han on changes in p oduc i i y
o ade. Howe e , Che niwchan (2017) ocuses on eﬀec s ela ed o ade libe alisa ion and ﬁnds
empi ical e idence o a clean-up o he U.S. manu ac u ing induced by he No h Ame ican F ee
T ade Ag eemen (NAFTA). Fo he pe iod 1994 o 1998, he ﬁnds ha nea ly wo- hi ds o he
educ ion o pa icula ma e (PM10) and sulphu dioxide (SO2) emissions can be a ibu ed
o he ade libe alisa ion. The eﬀec is d i en by wo main sou ces, namely imp o ed expo
oppo uni ies o he Mexican ma ke and imp o ed access o impo s om Mexico. The la e
eﬀec is ela i ely la ge in magni ude and indica es a high impo ance o impo compe i ion
o he clean-up o ﬁ ms’ en i onmen al pe o mance.
Kong e al. (2022) poin o a bias in he li e a u e owa ds using ﬁ m-le el da a om
de eloped coun ies and in es iga e he ﬁ m-le el en i onmen al eﬀec s ela ed o he Chinese
WTO accession in 2001 – a da e ha allows hem o label China as a de eloping coun y
and emphasising subs an ial diﬀe ences be ween de eloping and de eloped coun ies. They
ﬁnd ha he ade libe alisa ion signiﬁcan ly educed ﬁ ms’ oxic emissions - including sul u
dioxide (SO2), soo and dus , ni ogen oxide (NOx) - h ough he scale o p oduc ion, expo
oppo uni ies, echnology upg ading, in e media e inpu and wi hin-ﬁ m mechanisms.
Gu i´e ez and Teshima (2018) build on Che niwchan (2017) and mani es he posi i e en i-
onmen al eﬀec o impo compe i ion. They combine Mexican plan -le el and sa elli e image y
da a o he ime pe iod 2000 o 2003 and emphasise ha ﬁ ms in de eloping coun ies can
beneﬁ disp opo iona ely om ade- ela ed posi i e impac s on echnological choice. This is
because hese ﬁ ms a e on a e age u he emo ed om he echnological on ie o bes man-
agemen p ac ices. Thei key esul s e eal ha impo compe i ion inc eases ene gy eﬃciency,
bu is likely o induce a dec ease in aba emen expendi u es. This la e p edic ion sugges s
a echnology up ake, educing he need o expendi u e on aba emen . They also a gue ha
he li e a u e on global alue chain pa icipa ion and en i onmen al eﬃciency is biased owa ds
14
de eloped coun ies. A key eason o his bias is he a ailabili y o da a con aining conc e e
and eliable measu es o en i onmen al pe o mance a he ﬁ m-le el.
The Indian ﬁ m-le el P owess da a ep esen a popula sou ce o in es iga ions in o se e al
empi ical aspec s o in e na ional economics (e.g. De Loecke e al. (2016); Goldbe g e al.
(2010a,b); Topalo a and Khandelwal (2011)). As a a e ea u e in he con ex o de eloping
coun ies, he da a co e panel in o ma ion on a ﬁ m’s ene gy inpu s. This allows esea che s
o calcula e he di ec CO2emissions embodied in p oduc ion, simila o he app oach used by
Rich e and Schie sch (2017) o Fo slid e al. (2018) o ﬁ ms om de eloped coun ies. Ba ows
and Olli ie (2018) e en apply he me hodology a he p oduc -le el, in es iga ing how a ﬁ m’s
p oduc -mix aﬀec s he in ensi y o CO2emissions. They ﬁnd ha emission eﬃciency dec eases
wi h dis ance o he co e compe ency, whe e co e compe ency is p oxied by he p oduc wi h
he highes sales ank. Expo e s a e likely o sell hei co e compe ency on o eign ma ke s, bu
mo e p oduc i e ﬁ ms a e also able o di e si y hei o eign sales and hese p oduc s may indeed
be di ie compa ed o p oduc s cha ac e ised by he co e compe ency. To his end, ﬁ m-le el
analysis on en i onmen al eﬃciency ha ails o accoun o changes in he p oduc -mix, can
be misleading. In line wi h his, Ba ows and Olli ie (2021) con ibu e a u he s udy o shed
ligh on wi hin-p oduc CO2eﬃciency gains. They ﬁnd ha inc easing o eign demand induces
expo e s o scale-up p oduc ion and induces eﬃciency gains a he p oduc -le el. The la e
ﬁnding can be in e p e ed as a echnique eﬀec , s ongly backing p io ﬁndings ha expo
ac i i y links o en i onmen al gains ollowing he adop ion o clean echnologies.
Ma ´ınez-Za zoso e al. (2020) analyse he ela ion be ween expo s and ene gy consump ion
o Indian ﬁ ms be ween 1987 and 2016. This pe iod co e s India’s 1991 economic libe alisa ion
and accession o he WTO in 1995. In line wi h p io analyses, hey ﬁnd ha expo e s spend
mo e on ene gy, bu hei p oduc ion can be cha ac e ised as ene gy eﬃcien . In pa icula , ﬁ ms
om high ene gy consuming indus ies can inco po a e ade- ela ed eﬃciency gains. These high
ene gy use s a e mo e likely o be loca ed in an ups eam posi ion along he global alue chain,
b inging hem ela i ely close o he aw ma e ial o a alue chain (Sem au, 2022; Shapi o,
2021). A nega i e co ela ion be ween ene gy consump ion and selling ﬁnished goods indica es
such a pa e n a he Indian ﬁ m-le el Ma ´ınez-Za zoso e al. (2020). Howe e , global alue
chain posi ioning does no comp ise he co e o he main analysis. Addi ionally, he measu e o
global alue chain posi ioning does no accoun o he global alue chain posi ioning o p oduc s
sold as in e media es.
An `as e al. (2012) and An `as and Cho (2018) show how Wo ld Inpu -Ou pu ables
can be used o calcula e a coun y’s indus y global alue chain posi ioning in a espec i e
yea . This measu e conside s sales di ec ly sold o ﬁnal consume s and he p oximi y o ﬁnal
consump ion o sales sold as in e media es. By applying he espec i e global alue chain
measu e, Shapi o (2021) e eals an en i onmen al bias o ade policy: Ups eam indus ies emi
subs an ially highe le els o CO2pe uni o ou pu , while acing lowe ade a iﬀs. The pa e n
be ween CO2and ups eamness is somewha mono one and can be obse ed o e a wide ange
o diﬀe en coun ies. Gene ally, ups eam p ocesses a e mo e ene gy and capi al demanding,
e.g. he ex ac ion o aw ma e ials om he g ound and downs eam s ages disp opo iona ely
use labou as a di ec inpu , which is gene ally cleane . In e na ional ade policy se es as
15

an indi ec subsidy o hese di y indus ies, since ups eam indus ies success ully lobby o
educed a iﬀs in o de o educe inpu cos s. Based on Shapi o’s ﬁndings, Copeland e al.
(2022) s a e as a s ylised ac on ade and he en i onmen he pa e n ha di y indus ies
a e mo e likely o be ups eam.
Sem au (2022) analyses he ela ion be ween ups eamness and di y p oduc ion using he
Indian P owess da abase. By doing so, he applies he global alue chain measu e o He kenhoﬀ
e al. (2021), who ake he indus y-le el global alue chain measu e o he ﬁ m-le el. He ﬁnds
ha once con olling o diﬀe en ﬁ m cha ac e is ics, ups eam ﬁ ms end o equi e mo e
ene gy, emi highe le els o CO2and demons a e educed eﬃciency o bo h measu es. The e
a e al e na i e explana ions o he pa e n going beyond Shapi o (2021). Fi s , consume
awa eness o en i onmen ally iendly ene gy and CO2eﬃciency p oduc ion migh inc ease
wi h p oximi y o ﬁnal consump ion (Zhu e al., 2012). Second, once ﬁ ms obse e ha p io
s ages ha e al eady made in es men s in he p oduc ion o inpu s pe cei ed as e hical, ﬁ ms a e
incen i ised o con ibu e o a cleane ﬁnal p oduc (He kenhoﬀe al., 2021). In e es ingly, he
nega i e ela ion be ween ups eamness and clean p oduc ion educes wi h inc eased exposu e
o en i onmen ally demanding expo ma ke s. In line wi h Gu i´e ez and Teshima (2018), he
la e links o highe dis ance om he echnological on ie o bes managemen p ac ises o
ﬁ ms occupying an ups eam posi ion.
Fo eign owne ship ep esen s ano he di ec o m o global alue chain pa icipa ion, cha ac-
e ised as a pa icula o m o a ﬁ m’s expo s a egy. Ho izon al o eign di ec in es men s can
se e as a subs i u e o expo s. Addi ionally, e ical o eign di ec in es men s can be iewed
as a sub-se o a ﬁ m’s in e na ionalisa ion s a egy, agmen ing p oduc ion ac oss diﬀe en
coun ies (An `as and Yeaple, 2014).
Fo eign owne ship is seen as a powe ul o ce o con e gence in p oduc i i y and p oduc ion
echnologies (A nold and Ja o cik, 2009). The li e a u e on o eign owne ship and en i onmen-
al pe o mance da es back o Eskeland and Ha ison (2003), who analyse whe he mul ina-
ionals ﬂock o de eloping coun ies, cha ac e ised as pollu ion ha ens due o lax en i onmen al
policy s ingency. Howe e , hey ﬁnd weak e idence ha o eign in es o s a e mo e likely o a -
ge sec o s wi h high le els o ai pollu ion. Con e sely, hey ﬁnd ha o eign ﬁ ms in Mexico,
Mo occo, Cˆo e d’I oi e and Venezuela p oduce en i onmen ally mo e eﬃcien ly, compa ed o
domes ic ﬁ ms.3Gene ally, empi ical esul s on he ela ionship be ween ﬁ ms’ en i onmen al
pe o mance and owne ship a e a he mixed (Cole e al., 2017).
Cole e al. (2008) analyse Ghanaian ﬁ m-le el da a o in es iga e i aining- ela ed capaci y
building o key employees, ans e s om pa en companies o hei o eign aﬃlia es. They ﬁnd
ha aining o key decision make s in he aﬃlia es does imp o e eﬃciency in uel, ene gy and
elec ici y use. Howe e , he e is no di ec link be ween o eign owne ship and uel use no
ene gy use, excep some indica ion o a swi ch o a highe sha e o elec ici y consump ion, on
a e age being a cleane inpu compa ed o al e na i e uels used.
3The ollowing analysis limi s o he la e obse ed eﬀec on he ela ion o o eign owne ship and a ﬁ m’s
en i onmen al pe o mance. Howe e , i should be no ed ha hese neglec s possible exploi a ion o pa en ﬁ ms
based in ad anced economies acing s ingen en i onmen al egula ion and migh exploi ing lax en i onmen al
egula ions in less-ad anced economies. The la e is o ins ance shown by Hanna (2010) by employing panel
da a o US ﬁ ms.
16
In a mo e ecen s udy, B ucal e al. (2019) apply a diﬀe ence-in-diﬀe ence app oach com-
bined wi h p opensi y sco e ma ching o Indonesian ﬁ m-le el da a. They show ha acqui ed
plan s inc ease ene gy demand, bu educe ene gy in ensi y by abou 30% wo yea s ollowing
acquisi ion. Addi ionally, hey ﬁnd simila esul s o a ﬁ m’s CO2in ensi y. They in e p e he
esul s as sugges ing an associa ion be ween o eign acquisi ion and imp o emen s o p oduc ion
p ocesses a ising h ough he adop ion o be e echnologies o be e managemen .
The s udies discussed so a ﬁnd e idence sugges i e o an en i onmen al p emium om
global alue chain pa icipa ion o exposu e. Expo ac i i y, impo compe i ion, downs eam
posi ion and o some ex end o eign owne ship ela e o en i onmen al eﬃciency gains. How-
e e , he esea ch e iewed so a , p edominan ly ocuses on eﬃciency measu es o ene gy, CO2
and o he emissions. Bu he concep o en i onmen al inno a ion co e s a di e se se o en i-
onmen al domains (Kemp and Pea son, 2007). The nex sec ion connec s hese esul s wi h a
ela i ely independen li e a u e on he d i e s o en i onmen al inno a ion adop ion by a ﬁ m.
2.3 Fi ms’ global alue chain pa icipa ion and en i onmen al
inno a ion adop ion
Based on he Oslo Manual, mos schola s in es iga e he d i e s o en i onmen al inno a ions
deﬁned as inno a ions ha esul in a educ ion o he en i onmen al impac in consump ion
o p oduc ion and a e no el o he ﬁ m (OECD and Eu os a , 2005; Kemp and Pea son, 2007;
Del R´ıo e al., 2016).
Building on Rennings (2000) and Jaﬀe e al. (2005), i is well-es ablished ha he d i e s o
en i onmen al inno a ion and s anda d inno a ion adop ion diﬀe . Ho bach (2008) ca ego ises
he de e minan s o en i onmen al inno a ion adop ion in o a supply side echnology-push, a
demand-pull, en i onmen al egula ion-push and o he ﬁ m speciﬁc ac o s (Ba bie i e al., 2016;
Ghise i and Pon oni, 2015). Va iables ha cap u e he ex en o a ﬁ m’s in e na ionalisa ion,
e.g. o eign owne ship o expo ac i i y, a e con ained in he lis o ﬁ m-speciﬁc d i e s.
Howe e , he esul s o hese a iables a e ambiguous and emain con o e sial.
Empi ical s udies ha cap u e a ﬁ m’s expo ac i i y as a de e minan o en i onmen al
inno a ion adop ion, o a la ge ex en , do no epo any posi i e co ela ion (Cainelli e al.,
2012; De Ma chi and G andine i, 2012; Ho bach e al., 2013; Pe˜nasco e al., 2017). The esul s
by Chia esio e al. (2015) conﬁ m ha expo ac i i y does no induce any en i onmen al
upg ading. Addi ionally, hey also con ol o he di ec ion o ade and do no ﬁnd a posi i e
impac i expo s a ge non-eme ging coun ies. The au ho s claim ha lack o a signiﬁcan
co ela ion be ween expo ac i i y and en i onmen al inno a ion adop ion migh be an a e ac
o he da a - namely he geog aphic es ic ion o hei s udy and o he simila p io s udies
o ad anced Eu opean coun ies such as I aly, Spain, Ge many o F ance. These coun ies a e
cha ac e ised by a high le el o consume awa eness and en i onmen al policy s ingency.
Keshminde and del R´ıo (2019) pa ly add ess his esea ch imbalance, e en hough hei
analysis is es ic ed o he chemical sec o in Malaysia. Thei ﬁndings sugges ha o eign
s akeholde s inﬂuence a ﬁ m’s en i onmen al s a egy, including eﬀo s o comply wi h o eign
buye s’ demand and en i onmen al egula ion. Mo eo e , ﬁ ms can beneﬁ om knowledge
17
spillo e s. Sanni (2018) uns a comp ehensi e analysis on he d i e s o en i onmen al inno-
a ion adop ion o Nige ian ﬁ ms, including a discussion on he impo ance o ﬁ ms om
de eloping coun ies o comply wi h in e na ional s anda ds. Howe e , he empi ical analysis
does no include measu es o a ﬁ m’s global alue chain pa icipa ion.
Besides he empi ical bias owa ds single-coun y s udies in ad anced coun ies, Ghise i
e al. (2015) poin o a s ong ocus on inno a i e ac i e ﬁ ms ha also aﬀec s he gene alisabili y
o any s udy ﬁndings. Ka ou os e al. (2008) claims ha a ﬁ m’s expo ac i i y allows he
ﬁ m o ap in o knowledge c ea ed in diﬀe en ma ke s, pushing he ﬁ m’s inno a i e capaci y.
Acco dingly, and in line wi h he heo e ical li e a u e o Meli z (2003) and Bus os (2011), he
sub-sample o inno a i e ﬁ ms is mo e likely o be ac i e in o eign ma ke s as well as o epo
highe p oduc i i y ha i sel links o a be e en i onmen al pe o mance (Cainelli e al., 2015;
Hanley and Sem au, 2022; Shapi o and Walke , 2018).
Gi ma and Hanley (2015) use an ins umen al a iable app oach o assess he d i e s o
en i onmen al inno a ion adop ion o ﬁ ms based in he Uni ed Kingdom. They do no es ic
he sample o inno a i e ﬁ ms and ﬁnd obus e idence o a ﬁ m’s expo ac i i y and i s
adop ion o en i onmen al inno a ion. Ghise i e al. (2015) es ic he sample o inno a i e
ﬁ ms and exploi a mul i-coun y da ase , co e ing a b oad ange o Eas e n Eu opean coun ies.
In e es ingly, he epo ed co ela ion be ween a ﬁ m’s expo s a us and he likelihood o
en i onmen al inno a ion adop ion is posi i ely signiﬁcan , bu hey do no in es iga e his
co ela ion in mo e de ail. Howe e , Ghise i and Pon oni (2015) a gue ha expo ac i i y
does no di ec ly ela e o a ﬁ m’s en i onmen al inno a ion adop ion bu a he mode a es
R&D ac i i y as pa o a ﬁ m’s echnology-push. Con e sely, Hanley and Sem au (2022)
a gue o a di ec ela ionship, poin ing o he li e a u e on ade induced knowledge-ﬂows and
lea ning-by-expo ing (De Loecke , 2007). Gene ally, he adop ion o en i onmen al inno a ion
equi es ﬁ ms o sou ce knowledge beyond hei own bounda ies (De Ma chi, 2012; Cainelli e al.,
2015). The a gumen is ha ade induced knowledge ﬂows a e ela i ely mo e impo an o
ﬁ ms based in ca ch-up economies, acing educed access o knowledge due o deﬁciencies in
he domes ic knowledge base o in labou o ce skills. By exploi ing a c oss-coun y sample and
unning a spli ocusing on Eas e n Eu opean coun ies, Hanley and Sem au (2022) show ha
expo e s a e mo e likely o adop p ocess-based en i onmen al inno a ions. These esul s a e
in line wi h Ho bach (2016) who e eals empi ical indica ion o knowledge ﬂows om Wes o
Eas Eu ope. Addi ionally, he esul s echo Pe˜nasco e al. (2017) who claim ha in e na ional
d i e s a e mo e impo an o ﬁ ms based in coun ies wi h lowe le els o domes ic demand
o en i onmen al upg ading.
Emphasising he impo ance o egula ion o en i onmen al inno a ion adop ion (Ambec
e al., 2013; Bo ghesi e al., 2015; B unne meie and Cohen, 2003; Po e and Van de Linde,
1995), Hanley and Sem au (2022) also in es iga e he ole o he di ec ion o ade, exploi ing he
OECD En i onmen al Policy S ingency index o p oxy o eign demand o en i onmen al qual-
i y (Bo a and Ko´zluk, 2014). They ﬁnd ha highe expo exposu e o mo e en i onmen ally
s ingen ma ke s inc eases bo h, he likelihood o any en i onmen al inno a ion adop ion as
well as he numbe o en i onmen al inno a ions adop ed. The posi i e co ela ion is obse able
o p oduc - and p ocess-based en i onmen al inno a ions and indica es ha ﬁ ms inco po a e
18
o eign s akeholde s’ demand in o hei p oduc ion decision.
Del R´ıo e al. (2016) poin o lack o s udies in he inno a ion li e a u e ocusing on o eign
owne ship as d i e o en i onmen al inno a ion adop ion. This is somehow su p ising, since
o eign owne ship is shown o os e a ﬁ m’s p ocess and p oduc inno a ion ac i i ies. The la e
eﬀec goes beyond a me e selec ion eﬀec , he che y-picking o o eign in es o s who migh be
emp ed o acqui e he mos p oduc i e ﬁ ms wi hin any indus y (Guadalupe e al., 2012).
Mo eo e , and simila o expo ac i i y, o eign di ec in es men s ela e o di ec knowledge
spillo e s, an impo an d i e o ﬁ ms’ en i onmen al inno a ion adop ion (Ja o cik, 2004;
A nold and Ja o cik, 2009; De Ma chi, 2012).
Howe e , Pe˜nasco e al. (2017) do no ﬁnd ha Spanish ﬁ ms wi h o eign equi y a e mo e
likely o adop en i onmen al inno a ions. The missing ﬁnding o o eign di ec in es men is
simila o e idence p oduced by Inoue e al. (2013) who, o a sample o Japanese ﬁ ms, ﬁnd
no posi i e co ela ion be ween ha ing a head oﬃce in a o eign coun y and he in oduc ion
o quali y s anda ds. Simila ly, Cainelli e al. (2012) use I alian ﬁ m-le el da a and do no ﬁnd
e idence o a signiﬁcan ly posi i e co ela ion be ween o eign di ec in es men and en i on-
men al inno a ion adop ion. Howe e , hey ﬁnd ha o eign-owned ﬁ ms a e mo e likely o
implemen en i onmen al inno a ions i hey a e embedded in local p oduc ion sys ems, e.g.
by coope a ion o supplie s o uni e si ies. Con e sely, Cainelli e al. (2015) ﬁnd a leas some
indica ion o a posi i e co ela ion be ween ﬁ ms expe iencing o eign in es men ﬂows and he
adop ion o en i onmen al inno a ions. Howe e , he eﬀec is only signiﬁcan a he en-pe cen
le el.
As simila o he ema ks ela ing o he in e ace be ween expo ac i i y and en i onmen-
al inno a ion adop ion, he composi ion o he samples used by esea che s migh explain he
diﬀe en esul s. Albo noz e al. (2014) use A gen inean ﬁ m-le el da a o in es iga e i o eign-
owne ship inﬂuence ﬁ ms’ en i onmen al ac ions. They ﬁnd o eign-owned ﬁ ms show compa ed
o domes ic ﬁ ms’ highe ac i i ies in he adop ion o six ou o eigh en i onmen al measu es.
Also, Kannen e al. (2021) analyse da a om non-ad anced coun ies. They ex end he da a
co e age o a c oss-coun y sample con aining ﬁ ms om Eas e n Eu ope, A ica and Asia a
diﬀe en s ages o de elopmen . They ﬁnd ha o eign di ec in es men posi i ely co ela es
wi h he implemen a ion o a g een managemen , measu ed as he adop ion o a g een manage
o g een s a egy, and show ha g een managemen go hand in hand wi h be e en i onmen-
al pe o mance, e.g. highe en i onmen al inno a ion adop ion a es. By di ec ly measu ing
he link be ween o eign owne ship and g een managemen , hey pin down he mechanism o
imp o ed managemen p ac ices sugges ed by B ucal e al. (2019) on how a o eign owne ship
ela es o a ﬁ m’s en i onmen al pe o mance. Such a ela ion be ween o eign owne ship and
g een managemen is also obse able o o he en i onmen al- ela ed managemen ou ines.
Fo ins ance, Zhu e al. (2012) explo e Chinese su ey da a om 2006 and ﬁnd ha ISO 14001
en i onmen al managemen ce iﬁca ion is mo e likely adop ed in o eign owned ﬁ ms.
Summing up, he e is a dis inc lack o consensus in he li e a u e on o eign owne ship
and en i onmen al inno a ion, sugges ing mo e wo k emains o be done. The ﬁnal sec ion
summa ises he main esul s o he li e a u e e iew and connec s he ﬁndings o gaps in he
li e a u e.
19
Ja o cik, B. S. (2004): “Does o eign di ec in es men inc ease he p oduc i i y o domes ic
ﬁ ms? In sea ch o spillo e s h ough backwa d linkages,” The Ame ican Economic Re iew,
94, 605–627.
Ka ou os, M. I., P. J. Buckley, J. A. Sha p, and C. Wang (2008): “The ole o
in e na ionaliza ion in explaining inno a ion pe o mance,” Techno a ion, 28, 63–74.
Kannen, P., F. O. Sem au, and F. S eglich (2021): “G een gi s om ab oad? FDI and
ﬁ ms’ g een managemen ,” Wo king Pape 2200, Kiel Ins i u e o he Wo ld Economy, Kiel.
Kemp, R. and P. Pea son (2007): “Final epo MEI p ojec abou measu ing eco-
inno a ion,” Repo , UM Me i , Maas ich .
Keshminde , J. and P. del R´
ıo(2019): “The missing links? The indi ec impac s o
d i e s on eco-inno a ion,” Co po a e Social Responsibili y and En i onmen al Managemen ,
26, 1100–1118.
Kong, D., G. Ma, and N. Qin (2022): “T ade libe aliza ion and ﬁ m oxic emissions,” Re iew
o In e na ional Economics.
Ma ´
ınez-Za zoso, I., S. Roy-Mukhe jee, F.-O. Sem au, and A. M. Voicu (2020):
“Pollu ion educ ion by a ionaliza ion in Indian ﬁ ms,” Wo king Pape 20020.01, INFER,
In e na ional Ne wo k o Economic Resea ch.
Meli z, M. J. (2003): “The impac o ade on in a-indus y ealloca ions and agg ega e
indus y p oduc i i y,” Econome ica, 71, 1695–1725.
OECD and Eu os a (2005): “Oslo Manual: Guidelines o collec ing and in e p e ing inno-
a ion da a, 3 d Edi ion,” The measu emen o scien iﬁc and echnological ac i i ies, OECD
Publishing, Pa is.
Pe˜
nasco, C., P. del R´
ıo, and D. Rome o-Jo d´
an (2017): “Analysing he ole o in e na-
ional d i e s o eco-inno a o s,” Jou nal o In e na ional Managemen , 23, 56–71.
Po e , M. E. and C. Van de Linde (1995): “Towa d a new concep ion o he en i onmen -
compe i i eness ela ionship,” Jou nal o Economic Pe spec i es, 9, 97–118.
Rennings, K. (2000): “Redeﬁning inno a ion—eco-inno a ion esea ch and he con ibu ion
om ecological economics,” Ecological Economics, 32, 319–332.
Rich e , P. M. and A. Schie sch (2017): “CO2emission in ensi y and expo ing: E idence
om ﬁ m-le el da a,” Eu opean Economic Re iew, 98, 373–391.
Sanni, M. (2018): “D i e s o eco-inno a ion in he manu ac u ing sec o o Nige ia,” Techno-
logical Fo ecas ing and Social Change, 131, 303–314.
Sem au, F. O. (2022): “On he d i e s o clean p oduc ion: Fi ms’ global alue chain posi ion-
ing,” in Bei ¨age zu Jah es agung des Ve eins ¨u Socialpoli ik 2022: Big Da a in Economics,
Kiel, Hambu g: ZBW - Leibniz In o ma ion Cen e o Economics.
26

Shapi o, J. S. (2021): “The en i onmen al bias o ade policy,” The Qua e ly Jou nal o
Economics, 136, 831–886.
Shapi o, J. S. and R. Walke (2018): “Why is pollu ion om US manu ac u ing declining?
The oles o en i onmen al egula ion, p oduc i i y, and ade,” The Ame ican Economic
Re iew, 108, 3814–3854.
Topalo a, P. and A. Khandelwal (2011): “T ade libe aliza ion and ﬁ m p oduc i i y: The
case o India,” Re iew o Economics and S a is ics, 93, 995–1009.
Veugele s, R. (2012): “Which policy ins umen s o induce clean inno a ing?” Resea ch
Policy, 41, 1770–1778.
Zhu, Q., J. Co dei o, and J. Sa kis (2012): “In e na ional and domes ic p essu es and
esponses o Chinese ﬁ ms o g eening,” Ecological Economics, 83, 144–153.
27
Chap e 3
S epping up o he ma k? Fi ms’ expo ac i i y and
en i onmen al inno a ion in 14 Eu opean coun ies∗†
Aoi e Hanley
Kiel Ins i u e o he Wo ld Economy (I W)
Kiel Cen e o Globaliza ion (KCG)
Ch is ian-Alb ech s-Uni e si ¨a zu Kiel
[email p o ec ed]
Finn Ole Sem au‡
Kiel Ins i u e o he Wo ld Economy (I W)
Kiel Cen e o Globaliza ion (KCG)
ﬁnn-ole.sem [email p o ec ed]
Abs ac
We in es iga e he abili y o expo s o igge he adop ion o en i onmen al inno a ion
(EI) in ﬁ ms, shedding ligh on he de e minan s o con e gence in en i onmen al s anda ds
o Eu ope’s ca ch-up economies. To analyse his ques ion empi ically, we measu e he
la e as he 1) p obabili y a ﬁ m adop s EI and he 2) b ead h o EI adop ion in ﬁ ms.
Applying Logi and Poisson pseudo-maximum likelihood es ima ions o ﬁ m-le el da a o
14 Eu opean coun ies, we ﬁnd ha pa icula ly Eas e n Eu opean expo e s epo highe
EI adop ion p opensi ies, an eﬀec exclusi ely d i en by p ocess-based EI and no obse able
o p oduc -based EI. Addi ionally, we e eal ha ega dless o a ﬁ m’s o igin, exposu e o
impo ing coun ies wi h high ma ke - ela ed en i onmen al policy s ingency, is linked o
he adop ion o EI. We conclude ha lea ning-by-expo ing, egula ion-push and demand-
pull mechanisms may help o explain hese ﬁndings, wi h o eign ma ke s being cha ac e ised
by a wide di e si y o s akeholde p e e ences.
JEL Classiﬁca ion Numbe s: F18, O31, Q52, Q56
Keywo ds: G een-/Eco-/En i onmen al inno a ion; Ci cula economy; En i onmen and ade;
Expo ac i i y; G een Deal
∗The chap e has been published as: Hanley, A.; Sem au, F.O. (2022). “S epping up o he ma k? Fi ms’
expo ac i i y and en i onmen al inno a ion in 14 Eu opean coun ies.” Indus y and Inno a ion, 29(5), 672-
700. The a icle is accessible ia h ps://doi.o g/10.1080/13662716.2021.2021865. In addi ion an ea lie
e sion o he manusc ip has been published as: Sem au, F. O.; Hanley, A. (2019). “S epping up o he ma k?
Fi ms’ expo ac i i y and en i onmen al inno a ion in 14 Eu opean coun ies.” Bei ¨age zu Jah es agung des
Ve eins ¨u Socialpoli ik 2019: 30 Jah e Maue all - Demok a ie und Ma k wi scha , Kiel, Hambu g: ZBW -
Leibniz In o ma ion Cen e o Economics.
†We would like o hank Richa d F anke, Holge G¨o g, Jens Ho bach, Bea a Ja o cik, Thilo K oege , Jakob
Leh , Da id Maddison, Gio anni Ma in, Massimiliano Mazzan i, Ho s Raﬀ, Ch is ina Raasch, Ka in Rehdanz,
Claas Schneide heinze, F auke S eglich, Ruijie Tian, Julian Veh ke, o he esea che s o he KCG and he highly
help ul commen s by wo anonymous e e ees. Addi ionally, we hank semina /con e ence/wo kshop pa icipan s
a he T ini y Business School Semina Se ies 2021, he Annual Con e ence o he Ve ein ¨u Socialpoli ik 2019, he
EEARE 2019, he DRUID 2019, Annual CERCIS Wo kshop 2019 in Fe a a, he INFER Wo kshop in Coimb a
2019, he G¨o inge Wo kshop in In e na ional Economics 2019, he Kiel Cen e o Globaliza ion Wo kshop
2018 he Kiel Ins i u e Young Resea che s Semina 2018 and he Aa hus-Kiel Wo kshop 2017. Mo eo e , we
acknowledge ﬁnancial suppo om he Leibniz Associa ion h ough he Leibniz Science Campus KCG (G an
numbe is SAS-2016-I W-LWC).
‡Co esponding au ho
28
3.1 Addi ional appendix
The ollowing appendix supplemen s he published manusc ip accessible ia h ps://doi.
o g/10.1080/13662716.2021.2021865.
Desc ip ion o dependen a iable
As desc ibed in Sec ion 3.1, he ha monised Communi y Inno a ion Su ey (CIS) wa es
2008 and 2014 include an expanded lis o en i onmen al inno a ion (EI) a iables. Howe e ,
he e a e sligh diﬀe ences be ween bo h wa es. Figu es 3.1 and 3.2 p esen he exac ques ions
used o c ea e he dependen a iables.
Figu e 3.1: Lis o ques ions ela ed o EIs co e ed in he ha monised CIS 2008
29
Figu e 3.2: Lis o ques ions ela ed o EIs co e ed in he ha monised CIS 2014
We deﬁne ﬁ e p ocess-based EIs by using he ollowing a iables:
•ECOMAT
•ECOENO, we combine ECOEN and ECOCO in o a single a iable in he CIS 2008
•ECOPOL, does also include wa e in CIS 2014
•ECOSUB
•ECOREC
In addi ion, we deﬁne h ee p oduc -based EIs by using he ollowing a iables:
•ECOENU, does also include CO2 oo p in in CIS 2014
•ECOPOS
•ECOREA
Finally, we exclude wo domains in oduced in CIS 2014 om he analysis, namely ECOREP
and ECOEXT.
30
Desc ip ion o main independen a iables
Sec ion 3.2 desc ibes he econome ic s a egy, including a discussion o he main indepen-
den a iables: expo ac i i y and EPS exposu e. Nex , mo e de ails o hese a iables a e
p esen ed.
Expo ac i i y
Figu es 3.3 and 3.4 include he espec i e ques ion used o deﬁne a ﬁ m’s expo ac i i y.
Bo h ques ions a e widely simila be ween bo h su ey wa es.
Figu e 3.3: Ques ion used o deﬁne a ﬁ m’s expo ac i i y in he ha monised CIS 2008
Figu e 3.4: Ques ion used o deﬁne a ﬁ m’s expo ac i i y in he ha monised CIS 2014
We deno e h ee ca ego ies o a ﬁ m’s expo s a us. The ﬁ s ca ego y applies i a ﬁ m does
no se e o eign ma ke s, so only MARLOC is selec ed. The second ca ego y expo e deno es a
ﬁ m se ing o eign ma ke s bu whe e he bulk o he ﬁ m’s u no e a e gene a ed domes ically.
Acco dingly, MARNAT, MAREUR o MAROTH a e selec ed bu no o eign geog aphic ma ke
is selec ed as la ges ma ke in e ms o u no e (LARMAR). Finally, expo domina es deno es
ha a o eign geog aphic ma ke is selec ed as he la ges ma ke .
31

En i onmen al policy exposu e
The OECD EPS Index is used o cons uc a p oxy o a ﬁ m’s exposu e o o eign EPS. In so
doing, we dis inguish be ween ma ke - ela ed and non-ma ke ela ed EPS exposu e. O e all,
he OECD EPS Index includes 14 en i onmen al policy ins umen s. Nine ins umen s a e
ma ke - ela ed and ﬁ e ins umen s a e non-ma ke ela ed (Bo a and Ko´zluk, 2014). The aw
da a al eady includes a iables cap u ing he espec i e ma ke - ela ed and non-ma ke ela ed
EPS o a coun y. These a iables a e a weigh ed mix o he diﬀe en policies. Ma ke - ela ed
EPS alloca es a weigh o 0.33 o axes, ading schemes and eed-in a iﬀs, espec i ely. The
espec i e ins umen s conside ed a e:
•Tax CO2
•Tax Diesel
•Tax NOx
•Tax SOx
•T ading schemes CO2
•T ading schemes g een ce iﬁca es
•T ading schemes whi e ce iﬁca es
•Feed-in a iﬀs sola
•Feed-in a iﬀs wind
Non-ma ke ela ed EPS includes s anda ds and R&D subsidies. To each policy a weigh ing
ac o o 0.5 is alloca ed and he espec i e ins umen s a e:
•Emission limi alues NOx
•Emission limi alues PMx
•Emission limi alues SOx
•Sulphu con en limi o diesel
•Renewable ene gy public RD&D budge
Table 3.1 includes he o e all EPS anking o he coun ies co e ed in he OECD EPS Index.
The able is so ed by he EPS sco es in 2006.
32
Table 3.1: OECD EPS sco es so ed by he sco e in 2006
Yea /Coun y 2006 2007 2008 2009 2010 2011 2012
F ance 3.28 2.86 2.90 3.69 3.15 3.70 3.57
Denma k 3.16 2.83 2.96 4.07 4.03 3.98 3.85
Finland 3.15 2.82 3.08 3.25 3.21 3.48 3.43
Sweden 3.03 2.70 2.92 3.34 3.09 3.23 3.10
Ge many 3.00 2.67 2.64 3.06 3.02 3.14 2.92
Ko ea 2.96 2.96 3.38 3.52 3.52 3.44 2.63
Spain 2.96 2.75 2.70 3.00 2.72 2.85 2.22
Czech Republic 2.88 2.55 2.72 2.89 2.89 2.37 2.38
Aus ia 2.82 2.44 2.91 3.33 3.33 3.08 2.95
Ne he lands 2.80 2.64 3.23 3.69 4.13 3.51 3.63
I aly 2.72 2.34 2.60 2.73 2.84 2.79 2.77
Po ugal 2.71 2.21 2.26 2.47 2.54 2.27 2.13
Hunga y 2.59 2.30 2.55 2.66 2.77 2.68 2.63
Belgium 2.40 2.20 2.34 2.58 2.60 2.53 2.47
Uni ed Kingdom 2.29 1.95 2.40 2.58 3.62 3.47 3.29
Poland 2.26 2.08 2.26 2.96 2.96 2.96 2.58
I eland 2.23 1.71 2.05 2.16 2.22 2.43 2.05
Canada 2.17 3.27 3.31 3.85 3.35 3.67 3.42
Swi ze land 2.13 2.13 2.67 3.19 3.33 3.29 3.29
No way 2.13 2.05 2.34 3.19 3.19 3.19 3.26
Uni ed S a es 2.13 2.34 2.47 2.93 2.68 2.47 3.17
Aus alia 2.01 2.01 2.26 2.69 2.50 3.34 3.72
G eece 1.84 1.92 1.83 2.08 2.33 2.33 2.13
Slo ak Republic 1.78 1.40 1.53 2.39 2.30 3.05 2.99
Japan 1.63 1.69 1.69 1.73 2.03 2.96 3.50
Tu key 1.50 1.50 1.50 1.54 2.06 2.21 1.83
China (People’s Republic o ) 0.77 0.77 0.81 0.98 1.10 1.35 2.04
India 0.67 0.63 0.63 1.13 1.20 1.26 1.30
Russia 0.65 0.65 0.60 0.60 0.60 0.60 0.60
Sou h A ica 0.52 0.52 0.48 1.52 1.75 1.71 0.71
Indonesia 0.50 0.50 0.50 0.50 1.17 1.17 1.17
B azil 0.42 0.42 0.42 0.42 0.42 0.38 0.38
Slo enia n.a. n.a. 1.64 1.85 2.46 2.42 2.28
Desc ip ion: The OECD En i onmen al Policy S ingency (EPS) index is a coun y-speciﬁc and
in e na ionally-compa able measu e o he s ingency o en i onmen al policy. S ingency is deﬁned as
he deg ee o which en i onmen al policies pu an explici o implici p ice on pollu ing o en i onmen-
ally ha m ul beha iou . The index anges om 0 (no s ingen ) o 6 (highes deg ee o s ingency).
The index co e s 28 OECD and 6 eme ging coun ies o he pe iod 1990-2012. The index is based on
he deg ee o s ingency o 14 en i onmen al policy ins umen s, p ima ily ela ed o clima e and ai
pollu ion (Bo a and Ko´zluk, 2014). Ve sion: 24.05.2018
33
Desc ip ion o o he independen a iables
This pa o he addi ional appendix complemen s Sec ion 3.2 by p o iding mo e in o ma ion
how he independen a iables a e cons uc ed using he CIS 2008 and CIS 2014 wa es.
MNC cap u es a o eign owne ship dummy. The dummy de e mines whe he a ﬁ m belongs
o an en e p ise g oup wi h a headqua e loca ed ou side o he ﬁ m. Fo he cons uc ion, he
ollowing ques ions o he CIS 2008 and CIS 2014 a e used:
Figu e 3.5: Ques ion used o deﬁne MNC in he ha monised CIS 2008
Figu e 3.6: Ques ion used o deﬁne MNC in he ha monised CIS 2014
We cap u e he numbe o ex e nal coope a ion pa ne s in ol ed in he inno a ion p og ess
by including ex . coop and he espec i e squa ed e m. In o al, we conside se en diﬀe en
coope a ion pa ne s based on he su ey ques ions p esen ed in Figu es 3.7 and 3.8. In he CIS
2014, in o ma ion abou coope a ion wi h clien s and cus ome s is spli in o he p i a e and
public sec o . Bo h ca ego ies a e ma ched in o a single ca ego y o imp o e he compa abili y
ac oss su ey wa es.
34
Figu e 3.7: Ques ion used o deﬁne ex . coop in he ha monised CIS 2008
Figu e 3.8: Ques ion used o deﬁne ex . coop in he ha monised CIS 2014
35
Chap e 4
On he D i e s o Clean P oduc ion: Fi ms’ Global
Value Chain Posi ioning∗†
Finn Ole Sem au
Kiel Ins i u e o he Wo ld Economy (I W)
Kiel Cen e o Globaliza ion (KCG)
ﬁnn-ole.sem [email p o ec ed]
Abs ac
Indus ies ha occupy ups eam posi ions in global alue chains (GVCs) - being posi-
ioned close o he aw p oduc - p oduce p opo iona ely mo e CO2-in ensi e. Howe e ,
ﬁ ms a e he e ogeneous, e en in na owly deﬁned indus ies. In his pape , I empi ically
in es iga e whe he he ela ionship be ween ups eamness and CO2emissions, measu ed in
absolu e and ela i e e ms, holds wi hin indus ies a he ﬁ m le el. Using g anula da a
o Indian manu ac u ing ﬁ ms and con olling o es ablished d i e s o clean p oduc ion, I
e eal ha ﬁ ms p oducing p oduc s close o ﬁnal consump ion p oduce less CO2-in ensi e.
I co obo a e he ﬁnding by using a 2-SLS ins umen al a iable app oach. In e es ingly, I
ﬁnd ha exposu e o impo ing coun ies wi h s ingen en i onmen al policies a enua es
he link be ween ups eamness and di y p oduc ion. The la e ﬁnding sugges s he impe -
a i e o echnology upg ading o di y ups eam p oduce s aiming o emain compe i i e
in in e na ional ma ke s.
JEL Classiﬁca ion Numbe s: F14, F18, O13, Q56.
Keywo ds: En i onmen and ade; en i onmen and de elopmen ; CO2in ensi y; global alue
chains; sus ainable de elopmen .
∗An ea lie e sion o he manusc ip has been published as Sem au, F. O. (2022). “On he D i e s o Clean
P oduc ion: Fi ms’ Global Value Chain Posi ioning.” Bei ¨age zu Jah es agung des Ve eins ¨u Socialpoli ik
2022: Big Da a in Economics, Kiel, Hambu g: ZBW - Leibniz In o ma ion Cen e o Economics. In addi ion, a
e sion o he manusc ip has been submi ed o a Special Issue on sus ainable alue chains in Wo ld De elopmen
in July 2022. I has been one o eigh pape s pa o he p oposal o he Special issue by he “Resea ch Ne wo k
Sus ainable Global Supply Chains”.
†I hank Eckha d Bode, Ca olyn Fische , And eas Fuchs, Robe Gold, Holge G¨o g, Aoi e Hanley, Tobias
Heidland, Thilo K oege , Jakob Leh , Chia a Lodi, S e an Pahl, Sonja Pe e son, L´eo Pica d, Niklas Po a ke,
Ma in Quaas, Ka in Rehdanz, Claas Schneide heinze, F auke S eglich, Raine Thiele and No ihiko Yamano o
hei aluable commen s and discussions on ea lie e sions. Fu he mo e, I hank pa icipan s o he Annual
Con e ence o he Ve ein ¨u Socialpoli ik in Basel 2022, he Resea ch Con e ence “Sus ainabili y in Global
Value Chains” 2021 join ly o ganised by he Uni ed Na ions Indus ial De elopmen O ganiza ion (UNIDO)
and he “Resea ch Ne wo k Sus ainable Global Supply Chains”; he 26 h Annual Con e ence o he Eu opean
Associa ion o En i onmen al and Resou ce Economis s (EAERE) 2021; he Semina o he Uni e si y Fe a a
2022, he Kiel Ins i u e o he Wo ld Economy (I W) and he Kiel Cen e o Globaliza ion Resea ch Semina
2022; he G¨o inge Wo kshop on In e na ional Economics 2020; he Aa hus-I W Wo kshop 2018; he B own
Bag Wo kshop o he Kiel Uni e si y 2018; and he HSU-I W Wo kshop 2018. Mo eo e , I acknowledge ﬁnancial
suppo om he Leibniz Associa ion h ough he Leibniz Science Campus KCG (G an numbe is SAS-2016-
I W-LWC
42

Highligh s
•In his pape , I empi ically in es iga e he ela ion be ween ﬁ ms’ CO2emissions and
GVC posi ions
•Iﬁnd ha wi hin indus ies ﬁ ms ha occupy ups eam GVC posi ions p oduce mo e
CO2-in ensi e
•Exposu e o expo ma ke s wi h s ong en i onmen al egula ion mode a es (nega i ely)
he ela ion be ween ups eamness and di y p oduc ion
•Aﬁ m’s expo s a us does no signiﬁcan ly mode a e he ela ion be ween ups eamness
and di y p oduc ion
•Accele a ing he en i onmen al aba emen o ﬁ ms in ups eam GVC posi ions is key
eaching global goals on CO2emissions
43
4.1 In oduc ion
P omo ing he g een ansi ion o indus ies is a p essing challenge o ou ime and i is
co e ed in he Uni ed Na ions Sus ainable De elopmen Goals 9 and 12. E en in he ligh o
ene gy eﬃciency gains o e he las decades, economic g ow h s ongly co ela es wi h ene gy
consump ion and ene gy consump ion i sel s ongly co ela es wi h g eenhouse gas (GHG)
emissions (Cos a-Campi e al., 2015). Global wa ming, a consequence o GHG emissions, ep-
esen s a se ious h ea o u u e gene a ions h ough inc easingly ola ile wea he condi ions,
changed ag icul u al p oduc i i y, h ea ened biodi e si y and ele a ed sea le els. Eme ging
and de eloping coun ies a e especially ulne able o he economic isks ela ed o global wa m-
ing. A he same ime, wi h hei economic ise, mos o he global inc ease in GHG emissions
o igina es om hese coun ies (Copeland e al., 2022). Acco dingly, beyond he u gency o a
signiﬁcan decoupling be ween economic g ow h and GHG emissions in indus ialised coun ies,
majo successes in GHG emission educ ion o de eloping and eme ging coun ies a e c ucial i
he e is any hope o a aining he Pa is Ag eemen ’s a ge o limi ing global wa ming o well
below 1.5°Celsius.
This pape , con ibu es o he unde s anding o he de e minan s o ﬁ ms’ clean p oduc ion,
measu ed as educ ions o ﬁ ms’ CO2emissions and CO2in ensi y, in de eloping coun ies. In so
doing, I complemen p io s udies on he ela ionship be ween a ﬁ m’s global alue chain (GVC)
pa icipa ion and i s en i onmen al pe o mance - he ocus being a ﬁ m’s GVC posi ioning,
a he han o eign owne ship, expo , o impo s a us (Ba ako a and Da ies, 2012; B ucal
e al., 2019; Gu i´e ez and Teshima, 2018). To his end, I empi ically analyse da a o ﬁ ms based
in India, he hi d la ges global emi e o CO2and a coun y wi h his o ically high a es o
emissions g ow h. A main con ibu ion o he li e a u e is whe he Shapi o’s (2021) ﬁnding o a
nega i e ela ionship be ween indus ies’ ups eamness and clean p oduc ion can be conﬁ med
a he ﬁ m le el. Speciﬁcally, his means conﬁ ming whe he ﬁ ms loca ed mo e ups eam along
he GVC - being posi ioned close o he aw p oduc han hei pee s p oducing a simila co e
p oduc - demons a e a wo se en i onmen al pe o mance. Addi ionally, I in es iga e whe he a
ﬁ m’s expo s and exposu e o expo ma ke s cha ac e ised by s ingen en i onmen al policies,
nega i ely mode a es he ela ionship o ups eamness.
The nega i e ela ionship be ween indus ies’ ups eamness and clean p oduc ion is
mono onous and somewha linea ac oss indus ies and coun ies, a all s ages o de elopmen
(Shapi o, 2021). In line wi h his ﬁnding, Copeland e al. (2022) lis as a s ylised ac ha
ups eam indus ies do indeed ca y a hea ie en i onmen al oo p in . Howe e , ﬁ ms a e he -
e ogeneous, e en wi hin na owly deﬁned indus ies (Meli z, 2003). Acco dingly, ela ionships
obse ed a he indus y le el canno commen on ﬁ m le el phenomena such as p oduc i i y,
expo s o p oduc mix. These and o he ﬁ m le el idiosync asies, which po en ially ela e o
he en i onmen al pe o mance o a ﬁ m, mo i a e his pape . The e a e s ong in ui ions why
hese ﬁ m le el phenomena ma e . Fi s ly, diﬀe ences in he p oduc mix a e expec ed o con-
nec bo h o a ﬁ m’s GVC posi ioning as well as he en i onmen al impac o i s p oduc ion.
A manu ac u ing ﬁ m ha also sells mining p oduc s is likely o epo wo se CO2in ensi y
han i s pee s because mining p oduc s equi e mo e ene gy, on a e age, du ing he p oduc ion
p ocess. A he same ime, mining p oduc s a e mo e p edominan in ups eam GVC posi ions.
44
Second, p oduc i i y a ies subs an ially be ween ﬁ ms wi hin an indus y and p oduc i e ﬁ ms
a e mo e likely o pa icipa e in global ade (Meli z, 2003). A ﬁ m’s ade pa icipa ion can
induce addi ional p oduc i i y gains, echnology upg ading, and imp o emen s in i s en i on-
men al pe o mance (Bus os, 2011; Che niwchan e al., 2017; Copeland e al., 2022; Shapi o
and Walke , 2018).
Gu i´e ez and Teshima (2018) a gue ha de eloping coun ies, in pa icula , can beneﬁ
en i onmen ally om ade-induced echnology adop ion because ﬁ ms based in hese coun ies
a e, on a e age, u he emo ed om he echnology on ie . I a gue ha a simila claim migh
hold o he ela ionship be ween ups eamness and cleane p oduc ion. I ﬁ ms in ups eam
alue chain posi ions a e cha ac e ised by a highe dis ance o he echnology on ie , hey migh
beneﬁ disp opo iona ely om ade- ela ed d i e s o cleane p oduc ion. These beneﬁ s om
ade could a ise om lea ning-by-expo ing, demand-pull, and egula ion-push (B unne meie
and Cohen, 2003; De Loecke , 2007; Ho bach, 2008). Addi ionally, hey a e mo e p onounced o
ﬁ ms exposed o coun ies wi h highe deg ees o en i onmen al policy s ingency (EPS) (Hanley
and Sem au, 2022). The la e migh link o a nega i e mode a ing ole o EPS exposu e on
he ela ionship be ween ups eamness and di y p oduc ion. Howe e , s ong en i onmen al
egula ion in impo des ina ions can cause ﬁ ms in de eloping coun ies o ully specialise in
he p oduc ion o (di y) inpu s no con o ming o hese s anda ds (Acemoglu e al., 2014).
Ou sou cing o di y p oduc ion o coun ies wi h lax en i onmen al egula ion is in line wi h
he pollu ion ha en hypo hesis, as ou lined by An weile e al. (2001) and Eskeland and Ha ison
(2003).
Simila o Ba ows and Olli ie ’s (2021) ecen con ibu ion on he nexus be ween he GVC
pa icipa ion o ﬁ ms and hei en i onmen al pe o mance, I use he P owess da abase. I deno e
clean p oduc ion by ﬁ ms’ CO2emissions and CO2emissions pe uni o sales. Addi ionally, I
ollow Da da i and Saygili (2021), measu ing ﬁ ms’ CO2in ensi y by using CO2emissions pe
uni o alue added. In his way, alue added is calcula ed by sub ac ing spending on in e me-
dia es om he sales o each ﬁ m, which co e s he alue o a good acqui ed in p io s ages o he
GVC. P oducing hese in e media es is linked o indi ec emissions along he GVC, emissions
ha a e emi ed du ing he p oduc ion o he in e media es and anspo a ion. Acco dingly,
using CO2pe alue added p o ides a measu e o a ﬁ m’s en i onmen al pe o mance a a
speciﬁc s age o he alue chain.
Exploi ing ﬁ m le el in o ma ion, he CO2emissions o a ﬁ m a e calcula ed using da a
on he quan i ies o ene gy consumed, o each ene gy sou ces used. Such de ailed ﬁ m le el
in o ma ion is a e in he con ex o a de eloping coun y and makes i possible o ega d
diﬀe ences in CO2emissions be ween diﬀe en sou ces o ene gy, in a way ha is no cap u ed
i ene gy cos s a e applied as a p oxy o clean p oduc ion (Ba ows and Olli ie , 2018, 2021).
To measu e he GVC posi ion o a ﬁ m, I apply he indus y le el GVC measu e o An `as
e al. (2012) and An `as and Cho (2018). Speciﬁcally, I ollow he me hodology o He kenhoﬀ
e al. (2021), calcula ing he Indian indus y-yea -speciﬁc GVC posi ions by employing he
WIOD a ailable o he pe iod om 2000 o 2014 (Timme e al., 2015). I hen combine he
indus y in o ma ion wi h he p oduc le el in o ma ion componen o P owess o calcula e
ﬁ m-speciﬁc GVC posi ions (He kenhoﬀe al., 2021). Empi ically, I apply an OLS eg ession,
45
con olling o es ablished d i e s o clean p oduc ion. By including indus y ﬁxed eﬀec s, he
empi ically exploi ed a ia ion in GVC posi ions o igina es om diﬀe ences in he p oduc mix
o ﬁ ms wi hin an indus y (He kenhoﬀe al., 2021). In wo al e na i e speciﬁca ions, I include
espec i e in e ac ion e ms - ﬁ s , be ween a ﬁ m’s ups eamness and expo s, and second,
be ween a ﬁ m’s ups eamness an i s exposu e o ma ke s cha ac e ised by a high EPS.
The main empi ical ﬁnding conﬁ ms he s ylised ac ha ﬁ ms occupying ups eam GVC
posi ions epo highe CO2emissions om hei p oduc ion. The eﬀec is s a is ically signi -
ican and economically meaning ul. A change in he ups eamness measu e by one uni , such
as he equi alen o swi ching om he a e age GVC posi ion o ﬁ ms in he ex iles indus y
o he basic me als indus y, inc eases he es ima ed CO2emissions by 28.15%, CO2pe uni
o sales by 26.87% and CO2pe uni o alue added by 35.39%, espec i ely. A compa ison o
ﬁ ms in e y downs eam posi ions, wi h ﬁ ms in e y ups eam posi ions u he illus a es he
economically meaning ul diﬀe ence. The es ima ed CO2emissions inc ease by 63.23%, CO2pe
uni o sales by 60.0%, and CO2pe uni o alue added by e en 82.21%. The la e co esponds
o an inc ease om 0.76 kg o 1.39 kg CO2emissions pe US dolla alue added.
Beyond he main ﬁnding, I e eal ha he ela ionship be ween ups eamness and di y
p oduc ion dec eases wi h exposu e o s ong en i onmen al egula ion. Fo ﬁ ms in ups eam
posi ions exposed o s ong en i onmen al egula ion in expo ma ke s, he ela ionship be-
ween ups eamness and di y p oduc ion e en disappea s comple ely. This ﬁnding links o
he adop ion o mo e eﬃcien echnologies by ﬁ ms aiming o emain compe i i e in en i on-
men ally demanding expo ma ke s. Lea ning-by-expo ing, egula ion-push, and demand-pull
help ﬁ ms o adop ion o hese echnologies and a e especially ele an o di ie p oduce s in
ups eam posi ions, a guably mo e emo e om he echnology on ie . Fo downs eam ﬁ ms,
en i onmen al pe o mance de e io a es wi h exposu e o s ong en i onmen al egula ion in
expo ma ke s. The ﬁnding sugges s ha , on a e age, p oduce s in downs eam posi ions spe-
cialise in he p oduc ion o di y inpu s, while p oduce s in ups eam posi ions demons a e
en i onmen al gains om exposu e o s ong en i onmen al egula ion.
In e es ingly, I do no ﬁnd any mode a ing eﬀec o expo s on he ela ionship be ween
ups eamness and clean p oduc ion. The la e esul sugges s ha ﬁ ms loca ed in mo e
ups eam posi ions do no beneﬁ disp opo iona ely om he ade-induced echnique eﬀec .
The key esul s on he ela ionship be ween ups eamness and clean p oduc ion a e obus
o a a ie y o al e na i e speciﬁca ions. These include he use o an al e na i e measu e o
en i onmen al pe o mance by ﬁ ms, ex ending he numbe o co e ed ﬁ ms in he es ima ion
sample and he applica ion o a 2-SLS ins umen al a iable app oach. In he la e , I use
ad e ising in ensi y (ad e ising expendi u e o sales) among ﬁ ms p oducing a simila co e
p oduc . The ele ance o he ins umen is mo i a ed by a highe incen i e o ad e ising
expendi u e o ﬁ ms in downs eam posi ions, such as a ge ing ﬁnal consume s. By excluding
aﬁ m’s ad e ising spending and sales om he measu e, he ins umen is exogenously gi en
oaﬁ m and canno be inﬂuenced by i s manage s o o he s akeholde s.
The unde lying analysis akes place agains a backd op in which policy-make s, keen o os e
he ansi ion owa ds a g een economy, a e seeking o unde s and he de e minan s o clean
p oduc ion. I ocus on India, a coun y ha aces mul iple challenges. Apa om s i ing
46
o deal wi h en i onmen al deg ada ion, India aces o he challenges in i s eﬀo s o achie e
sus ainable de elopmen , including po e y elimina ion and ene gy secu i y. The win a ge s
o ans o ma ion owa ds a g een economy and social de elopmen a e s ongly in e wined.
I is impe a i e o b eak he de imen al dynamic be ween economic de elopmen and GHG
emissions o achie e he ul ima e goal o sus ainable de elopmen . The p esen s udy co e s he
pe iod 2000 o 2014, du ing which India’s sha e o global CO2emissions inc eased om 4.2%
o 6.2%. Signiﬁcan ly, India’s absolu e CO2emissions mo e han doubled wi hin his pe iod,
and in absolu e e ms, India being he hi d la ges emi e in he wo ld, e en hough i s CO2
emissions pe capi a a e a ound 9.5 imes less han he equi alen alue o he Uni ed S a es
(The Wo ld Bank, 2019).1
One app oach os e ing a global g een economic ansi ion is o design ade policy in such
a way ha ewa ds clean p oduc ion along he GVC and os e s a g een economic ansi ion. A
CO2bo de adjus men ax could ep esen such an ins umen , as i penalises di ie p oduc-
e s along he GVC by a ge ing di ie p oduc ion wi h highe a iﬀs (Aichele and Felbe may ,
2015). Howe e , Shapi o (2021) unco e s an en i onmen al bias in ade policy, whe e indus-
ies in downs eam posi ions can be con on ed wi h highe a iﬀs s. indus ies in ups eam
posi ions, e en i hei p oduc ion is, on a e age, mo e en i onmen ally eﬃcien . Such a policy
wo ks agains he desi ed ou come because i subsidises di y p oduc ion in ups eam indus-
ies, he eby unde mining he o e all a ge o educing global CO2emissions. My esul s lend
s ong suppo a he ﬁ m le el o Shapi o’s conclusion o ade-induced dis o ions, poin ing o
e idence o an en i onmen al bias be ween ela i ely di ie ups eam, and cleane downs eam
p oduce s.
The emainde o his pape is s uc u ed as ollows. The nex sec ion summa ises s udies
connec ed o GVCs and he en i onmen , based on which I o mula e he es able hypo heses.
Sec ion 4.3 p esen s he da abase used in he analysis, he empi ical me hodology, a desc ip ion
o he a iables used, and he desc ip i e s a is ics. Nex , he empi ical esul s, he ins umen al
a iable app oach, and obus ness checks a e p esen ed. I conclude wi h a summa y o he esul s
and hei implica ions.
4.2 Fi m’s GVC Posi ioning and Clean P oduc ion
Signiﬁcan eﬃciency gains in GHG emissions pe uni o ou pu a e needed in o de o u he
decouple economic g ow h and GHG emissions. Howe e , ﬁ ms canno ully app op ia e he
wel a e gains o in es men s in cleane p oduc ion echniques, la gely because o he double-
ex e nali y cha ac e is ic o hese in es men s (Jaﬀe e al., 2005; Rennings, 2000). Acco dingly,
ﬁ ms in es less han he social op imum in GHG emissions educ ion. Key de e minan s in
co ec ing ma ke ailu es, which boos en i onmen al inno a ion and in es men in cleane
p oduc ion, a e egula ion-push, ma ke -pull, echnology push, and ﬁ m-speciﬁc ac o s (e.g.:
Ba bie i e al. (2016); B unne meie and Cohen (2003); Cainelli e al. (2012); del R´ıo Gonz´alez
(2009); De Ma chi (2012); Ghise i and Pon oni (2015); Ghise i e al. (2015); Ho bach (2008);
Ho bach e al. (2012); Ho bach (2016); Kesidou and Demi el (2012); Rennings (2000)).
1Downloaded on 20 Decembe 2019
47

Building on Shapi o (2021), I ocus on GVC posi ioning as a d i e o clean p oduc ion - a
d i e ha has no been conside ed in p io ﬁ m le el s udies. Shapi o (2021) ﬁnds ha up-
s eam indus ies a e cha ac e ised by educed en i onmen al eﬃciency. The easons sugges ed
o his pa e n a e mul iple. Fi s , ﬁ ms in ups eam indus ies ake aw ma e ials ex ac ed
om he g ound and ans o m hem. This is mo e ene gy-demanding han he p oduc ion
o downs eam ﬁ ms, which depend mo e on ela i ely cleane inpu s such as labou (Shapi o,
2021). Second, ossil ene gy inpu s such as coal a e disp opo iona ely used by he hea ies
indus ies, ypically hose occupying mo e ups eam posi ions such as elec ici y gene a ion,
cemen manu ac u ing, and s eel blas u naces (Shapi o, 2021).
The e a e also o he mechanisms, which unde pin he ela ionship be ween ﬁ ms’ ups eam-
ness and di y p oduc ion, beyond hose ou lined by Shapi o (2021). Ma ke -pull ac o s aﬀec
aﬁ m’s decision o in es in clean p oduc ion (Ho bach, 2008). Zhu e al. (2012) emphasise
ha downs eam s akeholde s - in pa icula - push ﬁ ms o adop en i onmen al measu es.
Fo hese s akeholde s, downs eam ﬁ ms a e ela i ely mo e isible han ups eam ﬁ ms; o
ins ance, downs eam p oduc s a e a ubiqui ous pa o he daily ou ine o many consume s.
I downs eam s akeholde s push o sus ainable p oduc ion, his u gency migh diminish o
less isible p oduc s, posi ioned u he ups eam he alue chain.
He kenhoﬀe al. (2021) heo e ically and empi ically analyse he ela ionship be ween he
ups eamness and co po a e social esponsibili y (CSR), which se es as ano he dimension
o ﬁ ms’ sus ainabili y in p oduc ion. They e eal ha ﬁ ms in downs eam posi ions spend
compa a i ely mo e on CSR. A key eason is ha CSR is cha ac e ised by incomple e con ac s
and CSR ac i i ies a e obse able, bu no e iﬁable. As a esul , ﬁ ms ha e a highe incen i e
o in es in CSR once hei supplie s in es in CSR. He kenhoﬀe al. (2021) empi ically conﬁ m
he p edic ion o ampli ying CSR spending along he GVC, using he Indian P owess da abase.
I can be concluded om he He kenhoﬀe al.’s s udy ha a ﬁ m occupying a mo e downs eam
posi ion migh ace a highe incen i e o in es in en i onmen ally iendly p oduc ion echniques
- an incen i e unde pinned by he p io i isa ion o en i onmen al ou comes by supplie s along
he GVC. These en i onmen al- ela ed expendi u es sha e some simila i ies wi h CSR, because
hey a e o buye s o in e media es obse able, bu no e iﬁable.
Con a y o he mechanisms desc ibed ea lie , which help o igge a nega i e ela ionship
be ween ups eamness and clean p oduc ion, a posi i e ela ionship can also be a gued. Shapi o
(2021) poin s o an en i onmen al bias o ade policy due o lowe a iﬀs o ups eam indus-
ies, bu highe CO2emissions pe US dolla o ou pu . Bus os (2011) in oduces echnology
change in o he Meli z (2003) model and ﬁnds ha A gen inian ﬁ ms ha ace ela i ely lowe
a iﬀs inc ease e enues by en e ing o eign ma ke s and a e he e o e be e able o abso b cos s
o echnology upg ading. Kong e al. (2022) show ha such echnology upg ading is also linked
o he imp o ed en i onmen al pe o mance o ﬁ ms. They ﬁnd ha China’s ade libe alisa ion,
ollowing i s Wo ld T ade O ganiza ion accession, signiﬁcan ly educed he oxic emissions o
ﬁ ms. Combining he esul s by Shapi o (2021), Bus os (2011), and Kong e al. (2022), ﬁ ms in
ups eam posi ions ace lowe a iﬀs and he e o e ha e be e access o o eign ma ke s. This,
in u n, encou ages a ﬁ m o in oduce echnology upg ading and o imp o e i s en i onmen al
pe o mance. None heless, iewing he e idence collec i ely and building on Shapi o (2021), I
48
would expec a nega i e ela ionship be ween ups eamness and clean p oduc ion o p e ail.
Howe e , ﬁ ms’ expo s migh se e o mode a e he nega i e ela ionship be ween up-
s eamness and clean p oduc ion. Recen ﬁ m le el s udies ha e ound ha expo s posi i ely
ela e o ﬁ ms’ en i onmen al pe o mance (Ba ako a and Da ies, 2012; Fo slid e al., 2018;
Gi ma and Hanley, 2015; Hanley and Sem au, 2022; Rich e and Schie sch, 2017). Gu i´e ez
and Teshima (2018) emphasise he posi i e impac o ade on echnological choice can play a
la ge ole, especially o ﬁ ms in de eloping coun ies, because such ﬁ ms a e gene ally u he
emo ed om he echnology on ie . A guably, ﬁ ms in ups eam posi ions p oduce in an en-
i onmen ally less eﬃcien ly manne and a e mo e emo e om he echnology on ie . In line
wi h his, he expec ed ela ionship be ween ups eamness and di y p oduc ion migh diminish
wi h expo s, because ﬁ ms occupying ups eam posi ions can beneﬁ - in en i onmen al e ms
- disp opo iona ely om expo s.
In gene al e ms, s udies in es iga ing he eﬀec o a ﬁ m’s ade pa icipa ion on i s en i-
onmen al ou comes in de eloping o eme ging coun ies a e sca ce. An excep ion is Ba ows
and Olli ie (2021), who simila ly use he P owess da abase and ocus on he di ec CO2emis-
sions o ﬁ ms. The me hodology used - calcula ing he CO2emissions embodied in he physical
ene gy consumed - is also applied in Ba ows and Olli ie (2018) and is no el in he con ex o
an eme ging coun y. Howe e , he concep ollows in he spi i o ela ed s udies by Rich e
and Schie sch (2017) and Fo slid e al. (2018) who use Ge man and Swedish da a, espec i ely.
Ba ows and Olli ie (2021) e eal ha o eign demand g ow h inc eases CO2emissions and
imp o es CO2eﬃciency. In e es ingly, hey also ﬁnd e idence o echnological upg ading a
he p oduc le el. In sum, o eign demand inc eases he le el o o al ou pu , bu simul aneous
imp o emen s in eﬃciency, on a e age, mi iga e he scale eﬀec , o he ex en o abou a hal
o his eﬀec .
Apa om he expec ed eﬀec o expo s, I expec he magni ude o he ela ionship be-
ween GVC posi ioning and di y p oduc ion o diminish wi h inc eased exposu e o EPS. Fi ms
in ups eam posi ions ha ace ough compe i ion in in e na ional ma ke s ha e a disp opo ion-
a e need o adop new echnologies, in o de o imp o e eﬃciency and o emain compe i i e
in en i onmen ally s ongly egula ed coun ies. In addi ion, Hanley and Sem au (2022) e-
eal ha ﬁ ms exposed o s ong en i onmen al egula ion in impo des ina ions demons a e
highe en i onmen al inno a ion adop ion a es. On a e age, consume s in coun ies wi h s in-
gen en i onmen al egula ions ha e s ong p e e ences o sus ainable p oduc ion and ﬁ ms
a e mo e likely o be equipped wi h s a e-o - he-a echnologies and skilled labou . Hanley
and Sem au (2022) in e p e his ﬁnding as a mix u e o expo -induced lea ning-by-expo ing,
demand-pull, and egula ion-push ansmi ed h ough he alue chain and augmen ed by high
sus ainabili y s anda ds in impo des ina ions (B unne meie and Cohen, 2003; De Loecke ,
2007; Ho bach, 2008; Newman e al., 2018). Howe e , Acemoglu e al. (2014) emphasise ha
s ong en i onmen al egula ion in impo des ina ions may induce specialisa ion in he p oduc-
ion o CO2-in ensi e inpu s by ﬁ ms in de eloping coun ies, whe e he la e ace ela i ely
lax domes ic EPS. In line wi h his, he pollu ion ha en hypo hesis p edic s a shi o di y
indus ies o de eloping and eme ging coun ies wi h less igo ous EPS (An weile e al., 2001;
Eskeland and Ha ison, 2003). The la e can lead o wo se en i onmen al pe o mance o hese
49
ﬁ ms. Howe e , i migh be he case ha hese ﬁ ms s ill p oduce in a mo e en i onmen ally
iendly way han hei non-expo ing domes ic pee s. Summa ising he li e a u e e iew how
GVC pa icipa ion ela e o he en i onmen al pe o mance o ﬁ ms, I empi ically in es iga e
he ollowing hypo heses:
H1: Aﬁ m’s CO2emissions inc eases wi h ups eamness.
H2: Aﬁ m’s CO2in ensi y inc eases wi h ups eamness.
H3: Expo s nega i ely mode a e he link be ween ups eamness and di y p oduc ion.
H4: Exposu e o s ingen en i onmen al policy in expo ma ke s nega i ely mode a es he
link be ween ups eamness and di y p oduc ion.
The nex sec ion p esen s he me hodology, s a ing wi h a desc ip ion o he da a used.
4.3 Me hodology
4.3.1 Da a Desc ip ion
I use he ﬁ m le el P owess da abase, me ged wi h he WIOD o es he ou hypo heses
de i ed in he p e ious sec ion. Comp ising da a on he ﬁnancial pe o mances o Indian ﬁ ms,
he P owess da a is p o ided by he Cen e o Moni o ing he Indian Economy.2Designed o
academia (and o he use s), he da a p ima y sou ce in o ma ion om income s a emen s and
balance shee s. To al p oduc ion in P owess co e s mo e han 80% o India’s GDP (Bos and
Vannoo enbe ghe, 2018).
The empi ical se -up spans he pe iod be ween 2000 and 2014. P owess da a is also a ailable
be o e and a e 2014. Howe e , I apply he me hodology o He kenhoﬀe al. (2021), c ea ing a
measu e o a ﬁ m’s GVC posi ion using he WIOD (Timme e al., 2015).3Un o una ely, he
la e da abase is only a ailable o he a o emen ioned pe iod. An `as e al. (2012) desc ibe
how Wo ld Inpu -Ou pu Tables can be used o calcula e a coun y-indus y speciﬁc con inuous
measu e o ups eamness. The app oach by He kenhoﬀe al. (2021) links he indus y-speciﬁc
GVC measu e wi h he p oduc le el da a o he P owess da abase, he la e including in o ma-
ion on he capaci y, p oduc ion, and sales, disclosed by ﬁ ms. Such p oduc le el in o ma ion
is a e in he con ex o an eme ging coun y, going back o he Indian 1956 Companies Ac ,
which manda ed o disclose such in o ma ion in he annual epo (Goldbe g e al., 2010b).
Apa om such essen ial and no o en accessible in o ma ion a he p oduc le el, he
co e age o ene gy consump ion is unusual in he con ex o an eme ging coun y. Fi ms epo
on hei physical ene gy consump ion, such as quan i ies o na u al gas, elec ici y, coal, diesel
o ag icul u al by-p oduc s consumed. The da a a ailabili y goes back o a 1988 amendmen
in he Indian Companies Ac , which o ces ﬁ ms o epo on ene gy consump ion by ene gy
ype. This le el o de ail makes i possible o calcula e he hea con en o ene gy consump ion
measu ed in MMBTU. The espec i e MMBTU o an ene gy sou ce can be e o mula ed in
2The da a we e downloaded om h ps://p owessdx.cmie.com/ on he 26 h o No embe 2019. I downloaded
he da a in US dolla s and deﬂa ed all mone a y a iables using he Wo ld KLEMS da a. The KLEMS da a
we e downloaded om h p://www.wo ldklems.ne /da a.h m, using he shee GO P , which akes 2010 p ices
as he e e ence p ice. Las download on 21 Sep embe 2018.
3I also employ he WIOD Release 2016, including Inpu -Ou pu Tables o 28 Eu opean coun ies and 15
o he majo economies (Timme e al., 2015), downloaded om h p://www.wiod.o g/home
50
e ms o i s CO2con en (Ba ows and Olli ie , 2018, 2021).
Gi en i s unique da a co e age, he P owess da abase is widely used in he economics li -
e a u e, including in e na ional economics (e.g., Ba ows and Olli ie (2018, 2021); De Loecke
e al. (2016); Goldbe g e al. (2010a,b)).
Simila o o he s udies analysing he de e minan s o a ﬁ m’s clean p oduc ion, I limi he
sample o he manu ac u ing base (Ba ows and Olli ie , 2018, 2021; Rich e and Schie sch,
2017). The manu ac u ing base is o pa icula in e es in he con ex o GVCs and clean p o-
duc ion because manu ac u ing p oduc s a e compa a i ely high in CO2emissions, compa ed o
se ice p oduc s. Addi ionally, manu ac u ing p oduc s a e mo e equen ly aded (Copeland
e al., 2022). The nex sec ion p esen s he empi ical s a egy used in he analysis.
4.3.2 Empi ical Analysis
Equa ions 4.1 and 4.2 illus a e he wo main OLS es ima ions, wi h CO2emissions and
CO2in ensi y used as dependen a iables, espec i ely.
ln(CO2emissions ) = α0+β1Ups eamness +β2X +γ +δ +� +ζ ,(4.1)
ln(CO2in ensi y ) = α0+β1Ups eamness +β2X +γ +δ +� +ζ ,(4.2)
CO2emissions is he na u al loga i hm o a ﬁ m’s CO2emissions in kilo onnes (k )
in yea de i ed om i s ene gy consump ion du ing p oduc ion. CO2in ensi y is he na -
u al loga i hm o a ﬁ m’s CO2emissions in k o e i s sales and alue added in US dolla s,
espec i ely. The cons uc ion o he dependen a iables is p esen ed mo e comp ehensi ely in
Sec ion 4.3.2.1.
Ups eamness is he main independen a iable o in e es . The alue inc eases wi h he
ups eamness o a ﬁ m in yea . The measu e a ies o e ime wi h changes in a ﬁ m’s
p oduc mix. This is desc ibed in mo e de ail in Sec ion 4.3.2.2.
X cap u es diﬀe en idiosync asies o he ﬁ m, po en ial co-de e minan s o he decision
o apply a clean p oduc ion echnology. These d i e s can be g ouped in o egula ion-push,
ma ke -pull, echnology-push, and ﬁ m-speciﬁc ac o s.4
EP S measu es he exposu e o a ﬁ m o o eign ma ke s cha ac e ised by s ong en i on-
men al egula ion. On one hand, EPS exposu e migh connec o he ansmission o egula ion-
push, demand-pull, and lea ning-by-expo ing mechanisms h ough he alue chain (Hanley and
Sem au, 2022). On he o he hand, domes ic ﬁ ms con on ing weake EPS can specialise in he
p oduc ion o di y p oduc s (Acemoglu e al., 2014). The cons uc ion o EPS is p esen ed in
mo e de ail in Sec ion 4.3.2.3.
Fi ms ha a e ac i e in di ie indus ies a e mo e likely o ade hei ou pu s (Copeland
e al., 2022). Howe e , ﬁ m le el s udies e eal ha expo e s demons a e highe ene gy and
emission in ensi ies (Ba ows and Olli ie , 2021; Ba ako a and Da ies, 2012; Fo slid e al.,
4Fo a li e a u e e iew on he d i e s o en i onmen al inno a ion check Ba bie i e al. (2016) and del
R´ıo Gonz´alez (2009).
51
Table 4.2: Indus y s a is ic abou CO2emissions and ups eamness
Indus y CO2CO2CO2/SA CO2/SA CO2/VA CO2/VA Ups . Obs.
a e age median a e age median a e age median
Chemicals 52.76 6.33 1.09 0.47 2.07 0.98 3.19 2,500
Coke & eﬁned pe oleum
p oduc s 624.97 0.54 0.57 0.12 1.68 0.36 2.75 93
Basic me als 305.77 11.97 1.89 0.60 3.73 1.31 2.67 2,545
Pape 116.43 29.08 4.62 4.00 9.34 7.71 2.62 977
O he non-me allic mine al 332.66 15.48 3.66 1.57 4.90 2.24 2.34 720
Rubbe & plas ics p oduc s 32.40 2.97 0.68 0.33 1.36 0.73 2.30 1,112
Fab ica ed me al p oduc s 15.24 4.88 0.66 0.41 1.44 0.81 2.28 646
Wood 4.46 1.15 1.92 0.91 3.79 1.61 2.17 23
P in ing & ep oduc ion o
eco ded media 0.06 0.06 0.09 0.09 0.09 0.09 2.12 1
Compu e elec onic & op ical
p oduc s 1.40 0.82 0.31 0.11 0.53 0.18 2.11 171
Elec ical equipmen 6.30 0.80 0.27 0.09 0.60 0.17 2.00 292
Machine y & equipmen 11.41 0.74 0.66 0.12 1.05 0.20 1.95 538
O he manu ac u ing 49.24 0.90 0.94 0.10 1.41 0.23 1.93 72
O he anspo equipmen 20.72 1.67 0.15 0.06 0.30 0.17 1.82 46
Mo o ehicles 8.50 4.16 0.28 0.17 0.51 0.33 1.78 379
Tex iles 22.92 7.72 1.57 1.01 2.97 1.80 1.68 3,061
Lea he 2.35 1.76 0.62 0.33 1.69 0.48 1.65 28
Wea ing appa el 2.73 1.03 0.32 0.09 0.40 0.21 1.65 82
Be e ages 41.79 5.96 0.89 0.29 1.48 0.39 1.60 236
Tobacco 94.17 1.05 0.09 0.04 0.12 0.04 1.59 18
Food 45.24 5.59 1.89 0.70 3.54 1.15 1.57 2,713
Pha maceu icals 33.07 2.91 0.88 0.18 1.31 0.29 1.57 1,289
p oduc mix o ﬁ ms p oducing in he same main indus y. Columns (1) and (4) ocus on he
de e minan s o ﬁ ms’ absolu e CO2emissions, while he emaining columns ocus on ﬁ ms’
CO2in ensi ies.
Table 4.3: OLS es ima ion: Ups eamness and clean p oduc ion
(1) (2) (3) (4) (5) (6)
VARIABLES CO2CO2/SA CO2/VA CO2CO2/SA CO2/VA
Ups eamness 0.334*** 0.289*** 0.445*** 0.248** 0.238** 0.303***
(0.055) (0.053) (0.055) (0.104) (0.108) (0.107)
EPS 0.062 0.135 0.039 -0.052 0.115 0.022
(0.273) (0.269) (0.270) (0.238) (0.233) (0.228)
Expo s a us -0.462*** -0.615*** -0.581*** -0.242*** -0.427*** -0.368***
(0.060) (0.059) (0.058) (0.055) (0.054) (0.054)
Fo eign owne ship -0.423*** -0.455*** -0.662*** -0.284*** -0.337*** -0.514***
(0.118) (0.120) (0.122) (0.109) (0.110) (0.112)
R&D dummy -0.201*** -0.320*** -0.448*** -0.009 -0.171** -0.254***
(0.075) (0.078) (0.080) (0.070) (0.070) (0.072)
SOE 1.544*** 1.332** 1.353*** 1.372*** 1.195** 1.191***
(0.552) (0.578) (0.477) (0.465) (0.471) (0.366)
(ln) p oduc i i y 0.083*** -0.179*** -0.331*** 0.037* -0.208*** -0.384***
(0.022) (0.021) (0.021) (0.021) (0.020) (0.020)
(ln) asse s 0.998*** 0.077*** 0.124*** 0.947*** 0.037** 0.074***
(0.021) (0.020) (0.020) (0.019) (0.018) (0.019)
(ln) age 0.290*** 0.112*** 0.038 0.200*** 0.036 -0.047
(0.038) (0.037) (0.037) (0.034) (0.033) (0.033)
Ag icul u al p oduc s 0.720*** 0.565*** 0.763*** 0.349*** 0.216** 0.428***
(0.115) (0.119) (0.120) (0.105) (0.109) (0.111)
Mining p oduc s 0.880*** 1.104*** 0.926*** 0.648*** 0.873*** 0.693***
(0.199) (0.197) (0.196) (0.192) (0.198) (0.190)
Se ice p oduc s -0.034 -0.203* -0.128 0.030 -0.126 -0.058
(0.114) (0.119) (0.119) (0.106) (0.109) (0.109)
Exclusi e manu ac u ing 0.172 0.154 0.160 0.093 0.096 0.091
(0.123) (0.128) (0.127) (0.113) (0.116) (0.116)
Obse a ions 17,542 17,542 17,542 17,541 17,541 17,541
R-squa ed 0.535 0.172 0.202 0.627 0.341 0.365
Yea dummy Yes Yes Yes Yes Yes Yes
Indus y dummy No No No Yes Yes Yes
S a e dummy Yes Yes Yes Yes Yes Yes
Robus s anda d e o s in pa en heses (clus e ed a he ﬁ m le el)
*** p<0.01, ** p<0.05, * p<0.1
Ups eamness is posi i ely signiﬁcan in all six speciﬁca ions. In e es ingly, he coeﬃcien s
a e highe and s anda d de ia ions a e lowe o he speciﬁca ions wi hou indus y ﬁxed eﬀec s.
Howe e , he coeﬃcien s emain signiﬁcan once indus y ﬁxed eﬀec s a e added. This s ongly
backs he esul s and policy conclusions o Shapi o (2021) in he con ex o an applied mic o-
le el app oach, con olling o se e al ﬁ m cha ac e is ics. Acco dingly, ﬁ ms p oducing in
ups eam indus ies show a di ie p oduc ion, indica ed by highe CO2emissions and CO2
emission in ensi ies. No ably, compa ing he esul s o CO2in ensi y pe uni o sales wi h CO2
58

in ensi y pe uni o alue added e eals s onge esul s o using alue added as a measu e o
CO2in ensi y. A change in he ups eamness measu e by one uni , such as swi ching om he
a e age GVC posi ion o ﬁ ms in he ex iles indus y o he basic me als indus y - inc eases
he es ima ed CO2emissions by 28.15%, CO2pe uni o sales by 26.87%, and CO2pe uni o
alue added by 35.39%.10
Figu e 4.4 illus a es he es ima ed le el o absolu e CO2emissions and CO2in ensi ies
o diﬀe en GVC posi ions, once all o he a iables a e held cons an a hei means. The
speciﬁca ions depic ed a e Columns (4), (5), and (6) o Table 4.3.
1.4
1.6
1.8
2
2.2
2.4
Linea P edic ion o (ln) CO2
p1p10 p20 p30 p40 p50 p60 p70 p80 p90 p99
Ups eamness (pe cen iles)
P edic i e Ma gins wi h 95% CIs
-1
-.8
-.6
-.4
-.2
0
Linea P edic ion o (ln) (CO2/SA)
p1p10 p20 p30 p40 p50 p60 p70 p80 p90 p99
Ups eamness (pe cen iles)
P edic i e Ma gins wi h 95% CIs
-.4
-.2
0
.2
.4
.6
Linea P edic ion o (ln) (CO2/VA)
p1p10 p20 p30 p40 p50 p60 p70 p80 p90 p99
Ups eamness (pe cen iles)
P edic i e Ma gins wi h 95% CIs
Figu e 4.2: Es ima ed ela ionship be ween GVC posi ioning and CO2emissions and CO2
in ensi ies, espec i ely (Columns (4), (5), and (6) o Table 4.3)
A he ﬁ s pe cen ile, he es ima ed le el o CO2emission is ln(1.65) a a ups eamness
o 1.43, while he eﬀec a he 99 h pe cen ile is ln(2.14) a a ups eamness o 3.39. Taking
he exponen ial unc ion o ob ain he alue in k leads o a magni ude o 5.21 k and 8.50 k ,
espec i ely. This co esponds o an inc ease o 63.23%, which is a guably a big diﬀe ence.
A simila pa e n is obse able o he es ima ed le el o CO2emissions pe uni o sales
a he espec i e pe cen iles. The es ima ed le el a he ﬁ s pe cen ile is ln(-0.81) and a he
99 h pe cen ile ln(-0.34). Taking he espec i e exponen ial unc ion o he alues leads o 0.44
kg CO2and 0.71 kg CO2emissions pe US dolla sold. This co esponds wi h 60.0% also o an
economically meaning ul inc ease.
Using alue added as a measu e o CO2in ensi y e eals a diﬀe ence o e en 82.21%. The
es ima ed le el a he ﬁ s pe cen ile is ln(-0.27) and a he 99 h pe cen ile ln(0.33). Taking he
10The de i ed change in pe cen age is calcula ed by sol ing: (exp(coeﬃcien )-1)*100.
59
espec i e exponen ial unc ion o he alues leads o 0.76 kg CO2and 1.39 kg CO2emissions
pe US dolla alue added.
All in all, he esul s s a is ically and economically back Hypo heses 1 and 2 and show ha
also wi hin indus ies, ﬁ ms in ups eam posi ions p oduce di ie han ﬁ ms in downs eam
posi ions.11 The ollowing sec ion ocuses on an ins umen al a iable app oach o deal wi h
possible sou ces o endogenei y.
4.4.2 Ins umen al Va iable App oach
The abo e-p esen ed esul s on H1 and H2 a e biased i a ﬁ m’s en i onmen al ou come
aﬀec s i s GVC posi ion o i an omi ed a iable simul aneously aﬀec s a ﬁ m’s ups eamness
and he en i onmen al ou come in p oduc ion.
An example o e e se causali y is whe e a ﬁ m in oduces a clean p oduc o he p oduc
mix o a oid epu a ional isks. Gene ally, consume s migh be mo e likely o claim sus ainable
p oduc ion on p oduc s hey a e amilia wi h, ypically in a mo e downs eam GVC posi ion.
To his end, he in oduc ion o a clean p oduc pa es he way o ge close o cus ome s, while
educing epu a ional isks.
A ela i ely simila h ea o iden iﬁca ion would a ise i an omi ed a iable simul aneously
aﬀec s GVC posi ioning and clean p oduc ion. Fo ins ance, a ﬁ m’s managemen has social
p e e ences and a ou s simul aneously B2C (business o consume s) p oduc s in downs eam
posi ions and in oduces a g een s a egy o imp o e he en i onmen al oo p in in p oduc ion.
In his case, he e ealed eﬀec occu s due o he social p e e ences o he managemen ha a e
no co e ed in he ﬁ m le el da a.
I employ an ins umen al a iable es ima ion o o e come such endogenei y conce ns. By
cons uc ion, a ﬁ m can only inﬂuence i s GVC posi ion by changes in i s p oduc mix. The
GVC posi ioning o each p oduc is al eady beyond he scope o he managemen because i is
measu ed a he indus y le el. Howe e , o exclude any inﬂuence by he managemen , I in-
s umen ups eamness wi h he na u al loga i hm o he ad e ising spending o e sales among
ﬁ ms p oducing a simila main p oduc measu ed a he ﬁ e-digi le el ((ln)ad e ising/SA).
The ins umen does no include he espec i e spending and sales o he ﬁ m.
The ins umen should mee wo c i e ia o be alid. Fi s , he ins umen needs o be el-
e an , so he co ela ion be ween ups eamness and he ins umen should be suﬃcien ly high.
I a gue ha he incen i e o a ﬁ m o ad e ising inc eases along he GVC. Acco dingly, ﬁ ms
p oducing he main p oduc ha is close o ﬁnal consump ion spend mo e on ad e ising, on
a e age. Gene ally, i a co ela ion is suﬃcien canno ﬁnally be s a ed, bu he co ela ion
should be signiﬁcan and a high F-s a is ic abo e 104.7 se es as a e y conse a i e ule o
humb o alida ion (Lee e al., 2022). Second, a alid ins umen mee s he exclusion es ic-
ion. In line wi h he exclusion es ic ion, he ins umen should no aﬀec he en i onmen al
ou come o a ﬁ m, excep h ough he GVC posi ion. The la e is objec i ely diﬃcul o e alu-
11Going beyond ups eamness and aking Column (6) as a benchma k o measu e he d i e s o ﬁ ms’ clean
p oduc ion e eals ha expo e s, o eign-owned ﬁ ms, mo e p oduc i e ﬁ ms and ﬁ ms ac i e in R&D show
lowe CO2in ensi y. Con e sely, s a e-owned ﬁ ms and ﬁ ms p oducing mining o ag icul u al p oduc s p oduce
mo e CO2-in ensi e. No eﬀec is obse able o pu ely manu ac u ing ﬁ ms o se ice ﬁ ms. Las ly, exposu e o
EPS does no ela e o any changes in a ﬁ m’s clean p oduc ion.
60
a e. Howe e , I a gue ha he exclusion o a ﬁ m’s spending on ad e ising and sales ulﬁls he
exclusion es ic ion because a ﬁ m does no ha e any inﬂuence on he alue o he ins umen .
Mo eo e , i a ﬁ m decides o ex end o educe he p oduc mix, his no mally does no aﬀec
he main p oduc . Acco dingly, sho - e m changes due o managemen p e e ences do no en e
he es ima ion and he espec i e ins umen al a iable emains exogenously gi en o a ﬁ m.
Table 4.4 includes he esul s o a 2-SLS es ima ion. The ﬁ s column epo s on he ﬁ s
s age, es ima ing he espec i e alue o a ﬁ m’s �
ups eamness. The speciﬁca ion does no
include indus y ﬁxed eﬀec s because he e is no suﬃcien co ela ion be ween he ins umen
and ups eamness a he ﬁ s s age once he es ima ion ocuses on he wi hin indus y a ia ion.
Howe e , his is no comple ely su p ising, because i is qui e es ic i e in he ligh o a high
co ela ion o a e age ad e ising spending o e sales es ima ed a he ﬁ e-digi p oduc le el
wi hin an indus y. Acco dingly, he 2-SLS es ima ion does closely link o he speciﬁca ions
co e ed in Columns (1)-(3) o Table 4.3, o e all linking o he esul s a he agg ega ed le el by
Shapi o (2021). Columns (2)-(4) o Table 4.4 e e o he main esul s o he 2-SLS es ima ion,
including he es ima ed alues o �
ups eamness as an ins umen o ﬁ ms’ GVC posi ions.
Table 4.4: 2-SLS Es ima ion
(1) (2) (3) (4)
�
Ups eam. CO2 CO2/SA CO2/VA
VARIABLES OLS 2-SLS 2-SLS 2-SLS
(ln) ad e ising/SA -0.123***
(0.007)
�
Ups eamness 1.632*** 1.215*** 1.854***
(0.169) (0.152) (0.172)
EPS -2.874*** 3.905*** 2.889*** 4.162***
(0.102) (0.573) (0.520) (0.587)
Expo s a us 0.075*** -0.540*** -0.670*** -0.663***
(0.023) (0.070) (0.066) (0.071)
Fo eign owne ship 0.166*** -0.509*** -0.512*** -0.759***
(0.046) (0.136) (0.133) (0.144)
R&D dummy -0.005 -0.060 -0.209** -0.300***
(0.031) (0.090) (0.087) (0.096)
SOE 0.178 1.387*** 1.218** 1.181**
(0.132) (0.519) (0.549) (0.523)
(ln) p oduc i i y 0.018** 0.029 -0.215*** -0.388***
(0.008) (0.026) (0.024) (0.026)
(ln) asse s -0.012* 0.990*** 0.073*** 0.116***
(0.007) (0.023) (0.022) (0.024)
(ln) age -0.029** 0.342*** 0.154*** 0.103**
(0.013) (0.044) (0.041) (0.044)
Ag icul u al p oduc s -0.294*** 1.124*** 0.864*** 1.202***
(0.037) (0.132) (0.128) (0.137)
Mining p oduc s 0.103** 0.720*** 0.988*** 0.740***
(0.043) (0.188) (0.185) (0.187)
Se ice p oduc s 0.053 -0.060 -0.225* -0.163
(0.032) (0.115) (0.116) (0.120)
Exclusi e manu ac u ing -0.032 0.274** 0.221* 0.254*
(0.036) (0.125) (0.126) (0.129)
Obse a ions 16,803 16,803 16,803 16,803
R-squa ed 0.410
F-s a 283.69
Yea dummy Yes Yes Yes Yes
Indus y dummy No No No No
S a e dummy Yes Yes Yes Yes
Robus s anda d e o s in pa en heses (clus e ed a he ﬁ m le el)
*** p<0.01, ** p<0.05, * p<0.1
The esul s o he ﬁ s s age e eal ha (ln)ad e ising/SA signiﬁcan ly nega i ely co e-
la es wi h a ﬁ m’s ups eamness. Also, he F-s a is ic is 283.69 clea ly abo e s anda d h esh-
olds. Bo h pa e ns p o ide a high le el o conﬁdence ha he ins umen is ele an .
The posi i e ela ionship be ween ﬁ ms’ ups eamness and di y p oduc ion emains un-
changed in all speciﬁca ions. In all speciﬁca ions, he posi i e eﬀec is signiﬁcan a he one-
pe cen le el, which again s ongly suppo s H1 and H2.
As s a ed in Hypo heses 3 and 4, he nega i e eﬀec o ups eamness on he en i onmen al
61
ou come migh diﬀe wi h a ﬁ m’s expo s a us and exposu e o EPS. The nex sec ion analysis
he alidi y o hese hypo heses.
4.4.3 The Mode a ing Roles o Expo S a us and Exposu e o EPS
Columns (1) o (3) o Table 4.5 show he esul s o an in e ac ion e m be ween ups eamness
and expo s a us. Columns (4)-(6) include he in e ac ion e m be ween ups eamness and
exposu e o EPS.
Table 4.5: OLS es ima ion: Mode a ing eﬀec s o expo s a us and EPS on ups eamness
(1) (2) (3) (4) (5) (6)
VARIABLES CO2 CO2/SA CO2/VA CO2 CO2/SA CO2/VA
Ups eamness 0.225** 0.206* 0.270** 0.988*** 0.768*** 0.886***
(0.112) (0.114) (0.114) (0.298) (0.289) (0.299)
Expo s a us -0.364** -0.595*** -0.543*** -0.242*** -0.427*** -0.368***
(0.181) (0.174) (0.177) (0.055) (0.054) (0.054)
EPS -0.045 0.124 0.032 1.614*** 1.307** 1.335**
(0.237) (0.233) (0.228) (0.609) (0.581) (0.592)
Ups eamness x Expo s a us 0.056 0.077 0.080
(0.081) (0.077) (0.079)
Ups eamness x EPS -0.809*** -0.579** -0.637**
(0.298) (0.277) (0.288)
Fo eign owne ship -0.284*** -0.337*** -0.514*** -0.282*** -0.336*** -0.513***
(0.109) (0.110) (0.111) (0.109) (0.110) (0.112)
R&D dummy -0.011 -0.174** -0.256*** -0.011 -0.172** -0.255***
(0.070) (0.070) (0.072) (0.070) (0.070) (0.072)
SOE 1.374*** 1.198*** 1.195*** 1.370*** 1.194** 1.190***
(0.460) (0.463) (0.359) (0.468) (0.473) (0.367)
(ln) p oduc i i y 0.037* -0.207*** -0.383*** 0.037* -0.208*** -0.384***
(0.021) (0.020) (0.020) (0.021) (0.020) (0.020)
(ln) asse s 0.947*** 0.038** 0.075*** 0.947*** 0.038** 0.074***
(0.019) (0.018) (0.019) (0.019) (0.018) (0.019)
(ln) age 0.199*** 0.035 -0.048 0.201*** 0.036 -0.047
(0.034) (0.033) (0.033) (0.034) (0.033) (0.033)
Ag icul u al p oduc s 0.351*** 0.219** 0.431*** 0.353*** 0.219** 0.430***
(0.106) (0.109) (0.111) (0.106) (0.109) (0.111)
Mining p oduc s 0.651*** 0.878*** 0.698*** 0.637*** 0.866*** 0.685***
(0.192) (0.198) (0.190) (0.192) (0.197) (0.190)
Se ice p oduc s 0.032 -0.123 -0.054 0.034 -0.123 -0.054
(0.106) (0.110) (0.110) (0.106) (0.109) (0.109)
Exclusi e manu ac u ing 0.097 0.101 0.096 0.094 0.096 0.092
(0.114) (0.116) (0.116) (0.113) (0.116) (0.116)
Obse a ions 17,541 17,541 17,541 17,541 17,541 17,541
R-squa ed 0.627 0.341 0.365 0.627 0.341 0.365
Yea dummy Yes Yes Yes Yes Yes Yes
Indus y dummy Yes Yes Yes Yes Yes Yes
S a e dummy Yes Yes Yes Yes Yes Yes
Robus s anda d e o s in pa en heses (clus e ed a he ﬁ m le el)
*** p<0.01, ** p<0.05, * p<0.1
While he o e all posi i e ela ionship be ween ups eamness and di y p oduc ion emains
signiﬁcan in all speciﬁca ions, he in e ac ion e m o ups eamness and expo s a us is no
signiﬁcan in any speciﬁca ion. Acco dingly, he e is no signiﬁcan diﬀe en slope o he ela-
ionship be ween ups eamness and di y p oduc ion depending on he expo s a us o a ﬁ m
obse able. This ﬁnding does no back H3, poin ing o possible diﬀe ences in he magni ude o
he echnique eﬀec along he alue chain.
Con e sely, he esul s o Columns (4)-(6) e eal a mode a ing ole o demanding expo
ma ke s on a ﬁ m’s en i onmen al pe o mance. The in e ac ion e m be ween ups eamness
and EPS is signiﬁcan in all speciﬁca ions, a ying be ween ﬁ e-pe cen and one-pe cen sig-
niﬁcance. This ﬁnding indica es ha he main ela ionship be ween ups eamness and di y
p oduc ion is condi ional on he in e ac ion because he posi i e slope diminishes wi h an in-
c easing le el o exposu e o s ong en i onmen al egula ion in o eign ma ke s.
Figu e 4.3 illus a es he mode a ing ole o EPS on he ela ionship be ween ups eamness
and CO2emissions pe uni o alue added (Column (6)).12 The uppe g aph illus a es he
12The espec i e g aphs o CO2emissions (Column (4)) and CO2emissions pe uni o sales (Column (5))
62
p edic ed CO2pe uni o alue added depending on ups eamness o gi en le els o EPS.
The lowe g aph illus a es he p edic ed CO2in ensi ies depending on EPS o gi en le els o
ups eamness.
-.5
-.4
-.3
-.2
-.1
0
.1
.2
.3
.4
.5
.6
P edic ed (ln) (CO2/VA)
1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2 3.4
Ups eamness
EPS = 0.7
EPS = 0.9
EPS = 1.1
EPS = 1.3
EPS = 1.5
P edic i e Ma gins
-.5
-.4
-.3
-.2
-.1
0
.1
.2
.3
.4
.5
.6
P edic ed (ln) (CO2/VA)
.7 .8 .9 1 1.1 1.2 1.3 1.4 1.5 1.6
EPS
Ups . = 1.4
Ups . = 1.8
Ups . = 2.2
Ups . = 2.6
Ups . = 3.0
Ups . = 3.4
P edic i e Ma gins
Figu e 4.3: P edic ed (ln) CO2pe uni o alue added o diﬀe en combina ions o ups eam-
ness and EPS
An EPS sco e o 1.5 links o e y s ong en i onmen al egula ion in expo des ina ion
ma ke s. Gi en ha ﬁ ms ace such a s ong egula ion, he slope o ups eamness is sligh ly
nega i e, e en i i is no signiﬁcan . Con e sely, an EPS sco e o e y low 0.7 links o a s ongly
posi i e slope. The espec i e coeﬃcien o ups eamness is 0.44 signiﬁcan a he one-pe cen
le el. Holding he EPS sco e cons an , i is also possible o de i e he economic signiﬁcance. Fo
example, o ﬁ ms exposed o weak en i onmen al egula ion in o eign ma ke s (EPS = 0.7), a
one-uni change in he ups eamness measu e is associa ed wi h an inc ease o CO2emissions o
52.46%, CO2emissions pe uni o sales o 43.72%, and CO2emissions pe uni o alue added
o 55.29%.13
In e es ingly, he ela ionship be ween EP S and di y p oduc ion becomes signiﬁcan ly pos-
i i e once he in e ac ion e m be ween ups eamness and EPS is added. Also, he ela ionship
be ween EPS and clean p oduc ion is condi ional on ﬁ ms’ GVC posi ions. As he lowe g aph
o Figu e 4.3 illus a es, ﬁ ms’ CO2in ensi y inc eases wi h EPS o ﬁ ms in e y downs eam
posi ions (ups eamness = 1.4). Howe e , he de i ed coeﬃcien o 0.44 is only signiﬁcan a
he 10-pe cen le el. Fo ﬁ ms in e y ups eam posi ions (ups eamness = 3.4), he coeﬃcien
a e pa o he Appendix (Figu es 4.5 and 4.6).
13Fo example, he espec i e coeﬃcien o CO2emissions is calcula ed by sol ing: 0.988-0.809*0.7=0.42. The
eﬀec is gi en by: (exp(0.42)-1)*100=52.46.
63

u ns nega i e o 0.83, again being signiﬁcan only a he 10-pe cen le el. In line wi h he
only sligh signiﬁcance and gi en ha a one-uni change o EPS ela es o a diﬀe ence in expo-
su e going sligh ly beyond he diﬀe ence o he minimum and maximum alue obse ed in he
sample, he economic signiﬁcance should no be o e -in e p e ed.14 S ill, hese ela ionships
indica e some specialisa ion in di y p oduc ion in downs eam posi ions.
All in all, he esul s e ealed in Table 4.5 s ongly back Hypo hesis 4 and i s economic
ele ance. This ﬁndings indica es ha he ela ionship be ween ups eamness and di y p o-
duc ion diminishes wi h EPS o ﬁ ms in ups eam posi ions. Acco dingly, ﬁ ms in ups eam
posi ions seem o en i onmen ally beneﬁ ou s andingly om exposu e o s ong en i onmen al
egula ions in o eign ma ke s. This insigh links o he adop ion o mo e eﬃcien p oduc ion
echniques o emain compe i i e wi h ﬁ ms in en i onmen ally demanding ma ke s, ade-
induced lea ning-by-expo ing eﬀec s, demand-pull, and egula ion-push especially impo an
o ﬁ ms being u he emo ed om he echnology on ie in ups eam posi ions. Con e sely,
ﬁ ms in downs eam posi ions show di ie p oduc ion wi h inc eased exposu e o s ong en i-
onmen al egula ion in o eign ma ke s. The la e links o a specialisa ion in di y p oduc ion
in line wi h Acemoglu e al. (2014). The nex sec ion ocuses on he obus ness o he key
esul s.
4.4.4 Robus ness
Un o una ely, no all ﬁ ms epo on hei ene gy consump ion in physical ene gy uni s.
Acco dingly, i is only possible o calcula e annual CO2emissions o a subse o ﬁ ms co e ed
in P owess. Howe e , mos ﬁ ms epo on hei ene gy spending in US dolla s.15 Ene gy cos s
a e widely used in he li e a u e o measu e ﬁ ms’ en i onmen al pe o mances, o example,
Eskeland and Ha ison (2003) and Ba ako a and Da ies (2012). Cole e al. (2008) also use
ene gy cos s o p oxy ﬁ ms’ en i onmen al pe o mance. Howe e , hey emphasise ha hei
dependen a iable should ideally cap u e a ﬁ m’s emissions. I ollow his sugges ion in he
main speciﬁca ions because he e a e wo main d awbacks o using ene gy cos s. Fi s , p ice
changes in ene gy cos s esul in changes on he en i onmen al pe o mance measu e, e en
i i does no aﬀec a ﬁ m’s en i onmen al pe o mance. Second, swi ching o cheape ene gy
sou ces esul s in an imp o emen o he measu e, e en hough he newly used ene gy sou ce can
show a wo se o be e CO2 oo p in . None heless, bo h p oxies a e highly co ela ed, which
mo i a es his obus ness check. Table 4.6 epo s on he esul s once ene gy cos s a e used o
calcula e ﬁ ms’ clean p oduc ion in absolu e and ela i e e ms. Columns (1)-(3) connec o he
main speciﬁca ions on he ela ionship be ween ups eamness and di y p oduc ion discussed
in Sec ion 4.4.1. Columns (4)-(6) link o he main esul s ou lined in Sec ion 4.4.3 on he
mode a ing ole o exposu e o EPS.
14Fo example, o ﬁ ms in e y downs eam posi ions (ups eamness = 1.4), a one-uni change in he EPS
measu e is associa ed wi h an inc ease o CO2emissions o 61.83%, CO2emissions pe uni o sales o 64.28%,
and CO2emissions pe uni o alue added o 55.77%. Tu ning o e y ups eam posi ions, he sign o he
ela ionship be ween EPS and clean p oduc ion shi s posi i ely. Fo example, o ﬁ ms in e y ups eam posi ions
(ups eamness = 3.4), a one-uni change in he EPS measu e is associa ed wi h a dec ease o CO2emissions o
67.91%, CO2emissions pe uni o sales o 48.4%, and CO2emissions pe uni o alue added o 56.43%.
15In addi ion, using ene gy cos s as dependen a iable does no equi e any assump ion abou he CO2
con e sion ac o , e.g., he emissions pe kwh can a y subs an ially be ween egions and ime.
64
Table 4.6: OLS es ima ion: Robus ness checks
(1) (2) (3) (4) (5) (6)
VARIABLES ENE ENE/SA ENE/VA ENE ENE/SA ENE/VA
Ups eamness 0.344*** 0.340*** 0.424*** 1.089*** 0.829*** 1.012***
(0.069) (0.064) (0.072) (0.210) (0.195) (0.214)
EPS 0.141 0.236** 0.207* 1.723*** 1.274*** 1.455***
(0.127) (0.119) (0.123) (0.402) (0.370) (0.395)
Ups eamness x EPS -0.798*** -0.524*** -0.630***
(0.194) (0.180) (0.193)
Expo s a us 0.145*** -0.152*** -0.076** 0.145*** -0.152*** -0.075**
(0.034) (0.032) (0.032) (0.034) (0.032) (0.032)
Fo eign owne ship 0.154** -0.025 -0.167*** 0.155** -0.024 -0.166***
(0.062) (0.060) (0.062) (0.062) (0.060) (0.062)
R&D dummy 0.069* -0.121*** -0.191*** 0.069* -0.122*** -0.191***
(0.039) (0.038) (0.039) (0.039) (0.038) (0.039)
SOE 0.746*** 0.678*** 0.779*** 0.747*** 0.678*** 0.779***
(0.264) (0.220) (0.166) (0.265) (0.221) (0.166)
(ln) p oduc i i y -0.014 -0.301*** -0.487*** -0.014 -0.301*** -0.486***
(0.014) (0.012) (0.014) (0.014) (0.012) (0.014)
(ln) asse s 0.812*** -0.121*** -0.073*** 0.812*** -0.121*** -0.073***
(0.011) (0.010) (0.010) (0.011) (0.010) (0.010)
(ln) age 0.212*** 0.066*** 0.003 0.213*** 0.067*** 0.004
(0.019) (0.018) (0.019) (0.019) (0.018) (0.019)
Ag icul u al p oduc s 0.561*** 0.432*** 0.614*** 0.565*** 0.435*** 0.617***
(0.077) (0.072) (0.077) (0.077) (0.072) (0.077)
Mining p oduc s 0.218** 0.409*** 0.280*** 0.206** 0.402*** 0.270**
(0.104) (0.102) (0.105) (0.104) (0.103) (0.105)
Se ice p oduc s 0.082 0.006 0.034 0.086 0.009 0.038
(0.080) (0.077) (0.078) (0.080) (0.077) (0.078)
Exclusi e manu ac u ing 0.027 0.135* 0.115 0.027 0.135* 0.116
(0.082) (0.080) (0.080) (0.083) (0.080) (0.080)
Obse a ions 42,762 42,762 42,762 42,762 42,762 42,762
R-squa ed 0.632 0.375 0.416 0.632 0.375 0.416
Yea dummy Yes Yes Yes Yes Yes Yes
Indus y dummy Yes Yes Yes Yes Yes Yes
S a e dummy Yes Yes Yes Yes Yes Yes
Robus s anda d e o s in pa en heses (clus e ed a he ﬁ m le el)
*** p<0.01, ** p<0.05, * p<0.1
The ela ionship be ween ups eamness and en i onmen al pe o mance is s a is ically sig-
niﬁcan a he one-pe cen le el in all six speciﬁca ions. In addi ion, he in e ac ion be ween
ups eamness and EPS is signiﬁcan in all h ee speciﬁca ions. To summa ise, he obus -
ness checks back p io esul s on H1, H2, and H4. Acco dingly, ﬁ ms in downs eam posi ions
show signiﬁcan ly cleane p oduc ion and exposu e o s ong en i onmen al egula ion in expo
ma ke s diminishes he nega i e eﬀec o ups eamness on clean p oduc ion. The las sec ion
concludes he o e all esul s o he s udy.
4.5 Conclusion
A la ge sha e o India’s popula ion s ill li es in po e y. Economic g ow h is essen ial o
elimina e po e y bu co esponds o a highe ene gy demand. Highe ene gy demand s ongly
co ela es o highe CO2emissions, a nexus ha has no been suﬃcien ly b oken. E en hough
he pe capi a CO2emissions o India a e ela i ely low on global scale, he coun y is he
hi d la ges global emi e o CO2. Al hough indus ialised coun ies a e esponsible o he
majo i y o his o ical CO2emissions, successes in de eloping coun ies, including India, a e
c ucial i he e is any hope o ulﬁlling he Pa is Ag eemen ’s a ge o limi ing global wa ming
o well below 1.5°Celsius. This is mainly because mos emissions g ow h o igina es om his
subse o wo ld’s economies (Copeland e al., 2022).
To his end, egional and global decision make s need o unde s and he de e minan s o
ﬁ ms’ en i onmen al eﬃciency in p oduc ion. Shapi o (2021) e eals ha indus ies in down-
s eam GVC posi ions emi less CO2pe uni o ou pu . Howe e , e en in na owly deﬁned
indus ies, ﬁ ms diﬀe subs an ially, o ins ance in hei p oduc i i y and p oduc mix. I empi -
65
ically con ibu e o he li e a u e on he unde s anding o he d i e s o clean p oduc ion, mo e
p ecisely on he ques ion o how ﬁ ms’ GVC posi ioning ela es o i s en i onmen al ou come.
The de i ed esul s show ha Indian manu ac u ing ﬁ ms occupying ups eam GVC posi-
ions emi signiﬁcan ly mo e CO2and epo wo se CO2in ensi ies, measu ed as CO2pe uni
o sales, and as CO2pe uni o alue added. A change in he ups eamness measu e by one
uni - such as swi ching om he a e age GVC posi ion o ﬁ ms in he ex iles indus y o he
basic me als indus y - inc eases he es ima ed CO2emissions by 28.15%, CO2pe uni o sales
by 28.87%, and CO2pe uni o alue added by 35.39%. These key ﬁndings a e obus agains
a 2-SLS es ima ion.
In addi ion, he empi ical esul s e eal ha exposu e o o eign EPS nega i ely mode -
a es he ela ionship be ween ups eamness and di y p oduc ion. Fi ms in ups eam posi ions
a e, on a e age, u he emo ed om he echnology on ie and ha e o adop mo e eﬃcien
p oduc ion echniques o emain compe i i e in en i onmen ally demanding ma ke s ab oad.
These ups eam ﬁ ms can also disp opo iona ely beneﬁ om ade- ela ed posi i e d i e s o
hei clean p oduc ion such as lea ning-by-expo ing, egula ion-push, and demand-pull. In-
e es ingly, o ﬁ ms in downs eam posi ions, en i onmen al pe o mance de e io a es wi h
inc eased exposu e o EPS in expo ma ke s - a ﬁnding linked o specialisa ion in di y p o-
duc ion (Acemoglu e al., 2014). No ably, he nega i e ela ionship be ween ups eamness and
clean p oduc ion does no diﬀe wi h he ﬁ m’s expo s a us.
My s udy con ibu es o he unde s anding o he de e minan s o ﬁ ms’ clean p oduc ion
in de eloping coun ies. Copeland e al. (2022) emphasise ha he ela ionship be ween up-
s eamness and di y p oduc ion is obse able in coun ies a all s ages o de elopmen . I would
expec simila esul s using ﬁ m le el da a o ad anced coun ies. Howe e , I expec ha
he mode a ing ole o EPS in o eign ma ke s migh dis inguish be ween s udies using da a
o ad anced coun ies. Fi ms occupying ups eam posi ions and al eady acing s ong domes-
ic en i onmen al egula ion migh no ha e o s ep up o new p oduc ion echniques, e en i
hey encoun e s ong EPS ab oad. Such an analysis could ep esen pa o a u u e esea ch
agenda. Mo eo e , in his pape , I ha e no ouched on he ole o indi ec emissions. The
ela i e impo ance o indi ec emissions ampliﬁes wi h downs eamness and i would be in e -
es ing o in es iga e whe he he e ealed ups eamness is also obse able o indi ec emissions
(Copeland e al., 2022).
The key esul s ha e diﬀe en implica ions o a ﬁ m’s managemen and poli ical decision
make s. Fi s , managemen can imp o e he o e all en i onmen al pe o mance by in oducing
ela i ely clean p oduc s in downs eam posi ions. A swi ch o he p oduc mix o include
cleane p oduc s does no equi e any echnological adop ion and migh be an a ac i e ool o
he managemen o imp o e he gene al pe cep ion o s akeholde s on sus ainable p oduc ion,
such as when publishing s a is ics ela ing o CO2emission in ensi y, o ou side in es o s.
A guably, he swi ch o downs eam p oduc s is a o m o g eenwashing because i does no
equi e he ﬁ m o make any undamen al changes in i s p ocesses - me ely, in i s p oduc mix.
Acco dingly, his s a egy is open o c i icism by mo e sophis ica ed kind o s akeholde - hose
able o moni o he ﬁ m’s pe o mance a he p oduc le el.
Second, egional decision make s ha e an incen i e o a ac downs eam indus ies o
66
imp o e hei egional CO2balance shee . Howe e , on a global scale, he nega i e eﬀec o
GHG emissions on global wa ming does no depend on he loca ion o emissions. Acco dingly,
and in line wi h he en i onmen al bias o ade policy e ealed by Shapi o (2021), a iﬀs o
downs eam indus ies should be educed, bols e ing suppo o hem ela i e o ups eam
indus ies. Cu en ly, he pa e n is comple ely e e sed, leading o an implici subsidy o he
ela i ely CO2-in ensi e ups eam indus ies.
Thi d, egula o s should posi ion he spo ligh on he CO2in ensi y o ups eam indus ies,
whe e mo e ene gy is equi ed o p oduce one uni o alue added. The issue is how o incen i ice
consume s and o he s akeholde s in e es ed in sus ainable p oduc ion, o expend highe eﬀo s
o moni o hese en i onmen al ou comes o ups eam GVC posi ions, whe e end-consume s
may be less di ec ly in es ed in he p oduc s. Some o hese ups eam indus ies a e agged as
ha d o deca bonise, e.g. chemicals, cemen o s eel p oduc ion. Howe e , his does no p eclude
egula o s om applying speciﬁc measu es. One app oach would be o educe o elimina e he
implici subsidy o ups eam indus ies by dec easing a iﬀs o downs eam posi ions. Ano he ,
would be o es ablish an app op ia e CO2p ice mechanism and o incen i ise en i onmen ally-
iendly inpu s such as enewable ene gies, enewable na u al gas o g een hyd ogen. A judicious
use o hese measu e could accele a e he g een ansi ion o hese indus ies.
Las ly, moni o ing global eﬀo s on CO2 educ ion, mul ila e al ins i u ions should con-
side diﬀe ences in he a e age GVC posi ions o indi idual coun ies. Ups eam indus ies
gene ally epo educed en i onmen al eﬃciency, bu hey a e in eg al o GVCs, whose ﬁnal
p oduc ion s ages a e o en loca ed in o he – mo e de eloped - coun ies. The la e depend
on hese inpu s. Acco dingly, hese coun ies a e indi ec ly con ibu ing o global pollu ion,
when he GVC is iewed in i s en i e y. A comp ehensi e assessmen o en i onmen al oo -
p in s - consump ion-based e alua ion s. p oduc ion-based e alua ion - could help o e come
his dis o ion. Addi ionally, he applica ion o a p ice which would co ec o he en i on-
men al ex e nali y, o example CO2bo de adjus men axes, would p o ide an incen i e o
downs eam ﬁ ms o push o sus ainable p oduc ion along he en i e GVC.
67
Fo slid, R., T. Okubo, and K. H. Ull ei -Moe (2018): “Why a e ﬁ ms ha expo
cleane ? In e na ional ade, aba emen and en i onmen al emissions,” Jou nal o En i on-
men al Economics and Managemen , 91, 166 – 183.
Ghise i, C., A. Ma zucchi, and S. Mon eso (2015): “The open eco-inno a ion mode.
An empi ical in es iga ion o ele en Eu opean coun ies,” Resea ch Policy, 44, 1080–1093.
Ghise i, C. and F. Pon oni (2015): “In es iga ing policy and R&D eﬀec s on en i onmen al
inno a ion: A me a-analysis,” Ecological Economics, 118, 57–66.
Gi ma, S. and A. Hanley (2015): “How g een a e expo e s?” Sco ish Jou nal o Poli ical
Economy, 62, 291–309.
Goldbe g, P. K., A. K. Khandelwal, N. Pa cnik, and P. Topalo a (2010a): “Impo ed
in e media e inpu s and domes ic p oduc g ow h: E idence om India,” The Qua e ly
Jou nal o Economics, 125, 1727–1767.
——— (2010b): “Mul ip oduc ﬁ ms and p oduc u no e in he de eloping wo ld: E idence
om India,” The Re iew o Economics and S a is ics, 92, 1042–1049.
Gu i´
e ez, E. and K. Teshima (2018): “Aba emen expendi u es, echnology choice, and
en i onmen al pe o mance: E idence om ﬁ m esponses o impo compe i ion in Mexico,”
Jou nal o De elopmen Economics, 133, 264–274.
Hanley, A. and F. O. Sem au (2022): “S epping up o he ma k? Fi ms’ expo ac i i y and
en i onmen al inno a ion in 14 Eu opean coun ies,” Indus y and Inno a ion, 29, 672–700.
He kenho , P., S. K au heim, F. O. Sem au, and F. S eglich (2021): “Co po a e
Social Responsibili y along he global alue chain,” Wo king Pape 9498, CESi o, Munich.
Ho bach, J. (2008): “De e minan s o en i onmen al inno a ion - New e idence om Ge man
panel da a sou ces,” Resea ch Policy, 37, 163 – 173.
——— (2016): “Empi ical de e minan s o eco-inno a ion in Eu opean coun ies using he
communi y inno a ion su ey,” En i onmen al Inno a ion and Socie al T ansi ions, 19, 1–14.
Ho bach, J., C. Ramme , and K. Rennings (2012): “De e minan s o eco-inno a ions by
ype o en i onmen al impac - The ole o egula o y push/pull, echnology push and ma ke
pull,” Ecological Economics, 78, 112–122.
Ja e, A. B., R. G. Newell, and R. N. S a ins (2005): “A ale o wo ma ke ailu es:
Technology and en i onmen al policy,” Ecological Economics, 54, 164–174.
Kannen, P., F. O. Sem au, and F. S eglich (2021): “G een gi s om ab oad? FDI and
ﬁ ms’ g een managemen ,” Wo king Pape 2200, Kiel Ins i u e o he Wo ld Economy, Kiel.
Kesidou, E. and P. Demi el (2012): “On he d i e s o eco-inno a ions: Empi ical e idence
om he UK,” Resea ch Policy, 41, 862–870.
74

Kong, D., G. Ma, and N. Qin (2022): “T ade libe aliza ion and ﬁ m oxic emissions,” Re iew
o In e na ional Economics.
Lee, D. L., J. McC a y, M. J. Mo ei a, and J. Po e (2022): “Valid - a io In e -
ence o IV,” Wo king Pape 29124, Na ional Bu eau o Economic Resea ch Wo king Pape ,
Camb idge, MA 02138.
Le insohn, J. and A. Pe in (2003): “Es ima ing p oduc ion unc ions using inpu s o
con ol o unobse ables,” The Re iew o Economic S udies, 70, 317–341.
Manj´
on, M., J. Manez, e al. (2016): “P oduc ion unc ion es ima ion in S a a using he
Acke be g-Ca es-F aze me hod,” S a a Jou nal, 10, 288–308.
Meli z, M. J. (2003): “The impac o ade on in a-indus y ealloca ions and agg ega e
indus y p oduc i i y,” Econome ica, 71, 1695–1725.
Newman, C., J. Rand, F. Ta p, and N. T i ko ic (2018): “The ansmission o socially
esponsible beha iou h ough in e na ional ade,” Eu opean Economic Re iew, 101, 250–
267.
Olley, G. S. and A. Pakes (1996): “The dynamics o p oduc i i y in he elecommunica ions
equipmen indus y,” Econome ica, 64, 1263–1297.
Rennings, K. (2000): “Redeﬁning inno a ion - eco-inno a ion esea ch and he con ibu ion
om ecological economics,” Ecological Economics, 32, 319–332.
Rich e , P. M. and A. Schie sch (2017): “CO2emission in ensi y and expo ing: E idence
om ﬁ m-le el da a,” Eu opean Economic Re iew, 98, 373–391.
Shapi o, J. S. (2021): “The en i onmen al bias o ade policy,” The Qua e ly Jou nal o
Economics, 136, 831–886.
Shapi o, J. S. and R. Walke (2018): “Why is pollu ion om US manu ac u ing declining?
The oles o en i onmen al egula ion, p oduc i i y, and ade,” The Ame ican Economic
Re iew, 108, 3814–54.
The Wo ld Bank (2019): Wo ld Bank De elopmen Indica o s.
Timme , M. P., E. Die zenbache , B. Los, R. S eh e , and G. J. V ies (2015): “An
illus a ed use guide o he wo ld inpu –ou pu da abase: he case o global au omo i e
p oduc ion,” Re iew o In e na ional Economics, 23, 575–605.
Zhu, Q., J. Co dei o, and J. Sa kis (2012): “In e na ional and domes ic p essu es and
esponses o Chinese ﬁ ms o g eening,” Ecological Economics, 83, 144–153.
75
Chap e 5
Co po a e Social Responsibili y along he Global
Value Chain∗†
Philipp He kenhoﬀ
Uni e si y o Mainz
philipp.he kenhoﬀ@uni-mainz.de
Sebas ian K au heim‡
Uni e si y o Passau
CESi o
sebas ian.k [email p o ec ed]
Finn Ole Sem au
Kiel Ins i u e o he Wo ld Economy (I W)
Kiel Cen e o Globaliza ion (KCG)
ﬁnn-ole.sem [email p o ec ed]
F auke S eglich
Kiel Ins i u e o he Wo ld Economy (I W)
auke.s e[email p o ec ed]
Abs ac
Loca ing subs an ial pa s o he p oduc ion p ocess in de eloping and eme ging
economies, many ﬁ ms ace an inc easing demand by s akeholde s o Co po a e Social
Responsibili y (CSR) along hei alue chains. Con ac ual incomple eness be ween ﬁ ms
and hei supplie s a diﬀe en s ages o p oduc ion can exace ba e he abili y o mee hese
demands. We analyze a model o sequen ial p oduc ion wi h incomple e con ac s whe e
CSR by independen supplie s diﬀe en ia es he ﬁnal p oduc in he eyes o ca ing con-
sume s. P og essing down he alue chain, ou model p edic s an inc easing CSR p oﬁle
om ups eam supplie s wi h low CSR o downs eam supplie s wi h highe CSR. We con-
ﬁ m his p edic ion using Indian ﬁ m-le el da a – compu ing a ﬁ m’s alue chain posi ion
by combining i s p oduc -le el sales in o ma ion wi h he Wo ld Inpu -Ou pu Da abase.
We ﬁnd ha mo e downs eam ﬁ ms epo highe CSR expendi u es as measu ed by a
combina ion o s aﬀwel a e spending and social communi y spending.
∗The chap e is an ea lie e sion o : “He kenhoﬀ, P., K au heim, S., Sem au, F. O., & S eglich, F. (2024).
“Co po a e Social Responsibili y along he global alue chain”. Jou nal o De elopmen Economics, 167, 103236.”
The published a icle is accessible ia h ps://doi.o g/10.4337/9781802200065.00016. An ea lie e sion o
he manusc ip has been published as: He kenhoﬀ, P.; K au heim, S.; Sem au, F.O. and S eglich, F. (2021).
“Co po a e Social Responsibili y Along he Global Value Chain.” CESi o Wo king Pape No. 9498. Please no e
ha he w i ing s yle o he chap e diﬀe s due o submission equi emen s, such as w i ing in Ame ican English.
†We would like o hank Julia Cajal-G ossi, Ron Da ies, Ben Fabe , Gab iel Felbe may , Sou a el Gi ma,
Robe Gold, Holge G¨o g, Aoi e Hanley, Roza Khoban, K isz ina Kis-Ka os, Ca ol Newman, Ma celo Ola eaga,
Gianma co O a iano, Philip Sau ´e, Jan Schymik, Raine Thiele, Fa id Toubal, Gonzague Vannoo enbe ghe, Yo o
Yo o , pa icipan s o he HSU-I W Wo kshop 2019, he PEGNe Con e ence 2019, he EGIT 2020, he launch
e en o he Resea ch Ne wo k Sus ainable Global Supply Chains 2021, he CESi o A ea Con e ence on Global
Economy 2021, he Wo kshop ‘Shaping Globaliza ion’ a JGU Mainz 2021, he ETSG 2021, he G¨o ingen
Wo kshop on In e na ional Economics 2022, he TRISTAN Wo kshop a he Uni e si y o Bay eu h 2022, as
well as semina pa icipan s a T ini y College Dublin, Uni e si y o Passau, Uni e si y o No ingham, JGU
Mainz, Kiel Ins i u e o he Wo ld Economy (I W), Kiel Cen e o Globaliza ion (KCG), and he Uni e si y
o G¨o ingen. K au heim g a e ully acknowledges ﬁnancial suppo by he Ge man Resea ch Founda ion (DFG)
g an no. 421074032; Sem au and S eglich acknowledge ﬁnancial suppo by he Leibniz Associa ion h ough he
KCG g an no. SAS-2016-I W-LWC.
‡Co esponding au ho
76
JEL Classiﬁca ion Numbe s: F12, F61, F63, L23, M14, O12
Keywo ds: Co po a e Social Responsibili y; global alue chains; incomple e con ac s; sus ainable de el-
opmen ; e hical sou cing; India
77
5.1 In oduc ion
Pa icipa ion in global alue chains (GVCs) is o en conside ed a pa h o de elopmen o low and
middle-income coun ies (OECD/WTO, 2013; Wo ld Bank, 2020). Conce ns abou nega i e social and
en i onmen al impac s o GVCs in hose coun ies a e, howe e , widesp ead and boos ed by agedies
such as he Rana Plaza collapse in 2013. Consume s ha e esponded o hese conce ns by becoming
inc easingly ocal abou hei demand o e hical s anda ds along he en i e alue chain. Wi h la ge
pa s o oday’s alue chains loca ed in de eloping and eme ging economies cha ac e ized by high leniency
wi h ega d o en i onmen al o labo egula ion and/o en o cemen , his equi es ha supplie s in es
in Co po a e Social Responsibili y (CSR). CSR in es men s – by deﬁni ion – ex end beyond local legal
equi emen s (e.g., McWilliams and Siegel, 2001) and a e he e o e key o mee ing he demands o an
e e g owing numbe o e hically conce ned consume s and, mo e b oadly, o he gene al accep ance (o
no ) o globalized p oduc ion in he public deba e.
Pe sis en and sizable compliance issues conce ning labo and en i onmen al p ac ices in he alue
chains o a la ge numbe o ﬁ ms indica e, howe e , how diﬃcul i can be o ﬁ ms o impose CSR
le els on hei in e na ional supplie s, e.g., h ough legally binding codes o conduc (e.g., Locke, 2013b).
A he same ime, elying on he indi idual sel -in e es o a la ge numbe o independen supplie s a
diﬀe en posi ions along he alue chain is unlikely o deli e op imal le els o CSR ( om he pe spec i e
o he ﬁ m).
Wi h his pape , we con ibu e o a be e unde s anding o he ac o s ha d i e CSR along he
global alue chain. Mo e speciﬁcally, we ask he ollowing ques ions: Which ole does he alue chain
posi ion o supplie s play in hei CSR in es men s? Wha a e he esul ing pa e ns o CSR along a
alue chain, om ups eam o downs eam p oduc ion s ages?
We make wo main con ibu ions. Fi s , we analyze a model o non-con ac ible CSR in es men s by
supplie s in a sequen ial p oduc ion p ocess (a alue chain). In ou model, CSR in es men s o supplie s
in he Global Sou h p o ide ho izon al and e ical p oduc diﬀe en ia ion when selling he ﬁnal p oduc
o ca ing consume s in he Global No h. We ﬁnd ha he equilib ium CSR in es men s a e low in
ups eam s ages, inc ease along he alue chain, and peak a he mos downs eam s ages. Second,
we es his p edic ion employing Indian ﬁ m-le el da a. Combining de ailed in o ma ion on a ﬁ m’s
p oduc mix wi h he Wo ld Inpu -Ou pu Da abase (WIOD), we cons uc a measu e o a ﬁ m’s GVC
posi ion o empi ically in es iga e how i aﬀec s a ﬁ m’s mone a y CSR spending. Ou empi ical analysis
suppo s he heo e ical p edic ion: CSR expendi u e inc eases along he alue chain.
In ou model, ﬁ ms and consume s a e loca ed in he Global No h, while supplie s a e loca ed in
he Global Sou h and ope a e unde lenien egula ion. We in oduce a g oup o ‘ca ing’ consume s who
app ecia e a p oduc ’s ‘e hical quali y’ embodied in en i onmen al and wel a e imp o emen s (CSR)
implemen ed by supplie s along he alue chain.1In line wi h he ﬁndings in Ba ling e al. (2015), in
ou model, CSR along he alue chain p o ides p oduc diﬀe en ia ion o he ﬁnal p oduc . Highe CSR
in es men s by supplie s he e o e allow he ﬁ m o cha ge a highe ma kup.2
A cen al p emise o ou analysis is ha supplie s make hei decisions on CSR in an en i onmen
cha ac e ized by incomple e con ac s: he ﬁ m selling he ﬁnal good o consume s canno w i e binding
con ac s p esc ibing p oduc ion s anda ds in supplie ac o ies. We a gue in de ail in Sec ion 5.2.1 ha
o he case o CSR, con ac s a e incomple e as he supplie ’s ac ion is “obse able, bu no e iﬁable”
1Su eys (e.g., O’Rou ke, 2005, and Lou ei o and Lo ade, 2005) as well as ﬁeld expe imen s wi h eal pu -
chasing decisions (e.g., Hiscox and Smy h, 2011, and Hainmuelle e al., 2015) sugges ha consume s do ca e
abou he e hical con en o hei consump ion and in ac ha e a highe willingness o pay o ‘e hical’ p oduc s.
2This implies ha we abs ac om massi e consume boyco s o well-media ized scandals, which dep ess
demand o a ge ed ﬁ ms. Such e en s may play a ole o gene a ing o expanding he g oup o ca ing consume s.
We ake he exis ence o ca ing consume s as gi en and a he model CSR in es men s along he alue chain as
oppo uni ies o inc ease he pe cei ed (e hical) quali y o he p oduc by inc emen ally con ibu ing CSR om
p oduc ion s age o p oduc ion s age (see Sec ion 5.2.2.1).
78
(Ha and Moo e, 1999, p. 118). Cen al o he a gumen is he diﬀe ence be ween ci cums an ial
e idence on a supplie ’s in ingemen s o e hical s anda ds (obse abili y) on he one hand and, on he
o he hand, e idence ha will hold in a cou o law ( e iﬁabili y). While ci cums an ial e idence can
be suﬃcien o ca ing consume s o e-assess hei alua ion o he p oduc and he e o e demand, i is
insuﬃcien in a cou o law, especially so in coun ies wi h weak legal ins i u ions.3
We in oduce he abo e elemen s – ca ing consume s and incomple e con ac abili y o CSR – in o
a model o sequen ial p oduc ion along he lines o An `as and Cho (2013). We ﬁnd an inc easing
p oﬁle o he supplie s’ equilib ium CSR in es men s along he alue chain: while CSR in es men s a
ups eam s ages a e low, hey inc ease along he alue chain owa ds mo e downs eam s ages. This
pa e n a ises because unde incomple e con ac s, he cumula i e CSR in es men in p e ious s ages
p o ides he supplie wi h an incen i e o choose a highe CSR le el. We de i e he comple e indus y
equilib ium wi h ee en y and ﬁnd ha endogenizing he numbe o ﬁ ms has impo an implica ions o
iden i ying es able implica ions. While in he pa ial equilib ium ( aking agg ega e a iables as gi en),
CSR in es men s inc ease in he o al expendi u e on e hically diﬀe en ia ed goods, his eﬀec anishes
in he indus y equilib ium. This lea es us wi h a single and clea es able implica ion o ou model: he
inc easing p oﬁle o CSR along he alue chain, which we ake o he da a in Sec ion 3.
Se e al policy implica ions di ec ly eme ge om ou heo y. Fi s , as highe CSR le els a ups eam
s ages encou age CSR in es men s a mo e downs eam s ages (sequen ial complemen a i y), policy
in e en ions a ge ing CSR a ups eam s ages may beneﬁ CSR along he en i e alue chain. Second,
o go e nmen s aiming a decen wo k and economic g ow h (as speciﬁed in he UN’s Sus ainable
De elopmen Goal 8), he ac ha he mo e ups eam s ages o p oduc ion ea u e he lowes CSR le els,
p o ides an incen i e o s a egically ocus indus ial policy owa ds specializa ion in mo e downs eam
p oduc ion s ages. Thi d, policy ini ia i es like he EU Di ec i e on Co po a e Sus ainabili y Due
Diligence, he F ench Du y o Vigilance Law o he Ge man Due Diligence Law shi he esponsibili y
o supplie ac ions o ﬁ ms. Ou analysis hin s a he limi ed abili y o a ﬁ m o impose e hical s anda ds
on i s supplie s and highligh s ha his p oblem is pa icula ly se e e o supplie s a ups eam s ages
o p oduc ion. Fu u e due diligence egula ion should ake his asymme y along he alue chain in o
accoun .
We es ou model’s p edic ion using he Indian ﬁ m-le el da ase P owess p o ided by he Cen e
o Moni o ing Indian Economy (CMIE), which has been used by Goldbe g e al. (2010a,b); De Loecke
e al. (2016); Bos and Vannoo enbe ghe (2018) as well as Ba ows and Olli ie (2021), among o he s.
I is a special ea u e o his da ase ha i con ains in o ma ion on he ﬁ ms’ s aﬀwel a e spending as
well as social communi y spending. We use hese i ems o cons uc a measu e o a ﬁ m’s CSR spending
in mone a y uni s, which we obse e o 15,512 medium-size and la ge ﬁ ms be ween 2000 and 2013 in
India.
Ou app oach o compu ing a ﬁ m’s alue chain posi ion pa allels he p ocedu e in Cho e al.
(2021). Fi s , we use he Wo ld Inpu -Ou pu Da abase (WIOD) and apply he me hodology om
An `as e al. (2012) and An `as and Cho (2018) o compu e he yea ly ups eamness o each Indian
indus y. Second, o each p oduc in a ﬁ m’s po olio, we assign he alue o annual ups eamness o i s
indus y and calcula e he ﬁ m’s alue chain posi ion, i s ups eamness, as he sales-weigh ed a e age o
i s p oduc s’ ups eamness alues.
We es ima e he ela ionship be ween a ﬁ m’s alue chain posi ion on i s CSR expendi u e. In ou
analysis, we con ol o a la ge se o ﬁ m-le el de e minan s o CSR iden iﬁed in he li e a u e (e.g.,
3C i o and Fo ge (2015) su ey a li e a u e on CSR in managemen sciences and indus ial o ganiza ion.
They lis impe ec compe i ion and con ac incomple eness as cen al d i e s o CSR (see, e.g., hei Figu e 1,
p. 114). Schille (2018) mo i a es his pu ely empi ical wo k alluding o he same no ion o incomple e con ac s:
he inabili y o a ﬁ m o ully con ol CSR in es men s o i s supplie s. Mo eo e , he e is a ecen and quickly
expanding li e a u e on ela ional con ac s in de eloping coun ies su eyed in Macchia ello (2021), which lends
suppo o ou modeling on incomple e con ac s in alue chains in ol ing de eloping and eme ging economies.
79

Newman e al., 2018; G¨o g e al., 2018; Schille , 2018). These include a measu e o in e na ionaliza ion
(sha e o expo s in o al sales), exposu e o mo e demanding cus ome s (measu ed by he ac ion o
expo s o OECD coun ies), local embeddedness (sha e o domes ically sou ced inpu s), ﬁ m size ( o al
sales), age and ﬁ m owne ship (dummies o s a e owned and o eign owned ﬁ ms). Impo an ly, we
con ol o he possibili y ha wage le els may a y sys ema ically along he alue chain using a ﬁ m-
speciﬁc wage measu e ha combines indus y-le el in o ma ion on labo compensa ion wi h a ﬁ m’s
p oduc mix.4
In ou es ima ions, we ely on wo diﬀe en ypes o a ia ion. Fi s , in a c oss-sec ional speciﬁca ion,
we con ol o se e al es ablished ﬁ m-le el d i e s o CSR and indus y-yea as well as s a e ﬁxed eﬀec s.
This app oach exploi s a ia ion in he alue chain posi ion o ﬁ ms ha sha e he same main indus ies.
Second, in a mo e demanding speciﬁca ion, we exploi he ime dimension o ou da a. By including
ﬁ m ﬁxed eﬀec s, we iden i y he eﬀec om changes in a ﬁ m’s GVC posi ion o e ime. In bo h cases,
we ﬁnd s ong suppo o ou model’s p edic ion: mo e downs eam ﬁ ms ha e highe CSR le els.
Compa ing ﬁ ms a he 5 h and he 95 h pe cen ile, ou p e e ed speciﬁca ion indica es ha p edic ed
CSR spending is 55.3% highe o he downs eam ﬁ m compa ed o he ups eam ﬁ m.
Ou esul s also hold in se e al obus ness checks. In one o hem, we add ess he possible conce n
ha mo e downs eam p oduce s migh simply be mo e isible o ﬁnal consume s and he e o e ace
highe p essu e o in es in CSR. We con ol o isibili y by adding ma ke ing and ad e ising expenses
as an addi ional a iable. We also smoo h CSR spending o e wo yea s o con ol o lumpy CSR
spending.
Ou wo k ela es o he ecen li e a u e s udying he “globaliza ion backlash”, he massi e skep icism
economic globaliza ion encoun e s in many coun ies (Pa cnik, 2017), which is su eyed in Colan one
e al. (2022). We a e in e es ed in an addi ional and complemen a y sou ce o skep icism owa ds
globaliza ion: he ongoing iola ion o e hical s anda ds in in e na ional p oduc ion - a ac ing c i icism
om consume s and ac o s in ci il socie y. Mo eo e , while he li e a u e abo e ocuses on public
egula ion, we a e in e es ed in pa e ns o olun a y sel - egula ion along in e na ional alue chains.
The e a e se e al s ands o li e a u e ha ocus on he (consequences o ) he demand o e hical
minimum s anda ds in in e na ional alue chains. In he absence o eﬀec i e in e na ional egula ion
(Ba aglini and Ha s ad, 2020), his demand mani es s i sel as p essu e on ﬁ ms o sel - egula e in e -
na ional p oduc ion, a p ocess Ba on (2003) e ms p i a e poli ics. In Ba on (2012), he a gues ha key
ac o s o p i a e poli ics a e ad ocacy NGOs which use bo h con on a ional s a egies (like consume
boyco s) as well as coope a i e app oaches (like labels) o di ec ly inﬂuence he ac ions o ﬁ ms (in-
s ead o aﬀec ing go e nmen policies in he case o public poli ics). Ou analysis he e o e ela es o
he li e a u e on labels, wi h he ai ade label being he bes documen ed and mos esea ched case
in poin .5The o he s and o li e a u e is a ecen and expanding li e a u e on he ole o in e na ional
ad ocacy NGO ac i i y o in e na ional p oduc ion and de elopmen .6The la e is mo e ocused on
NGO campaigns and consume boyco s. Ou objec o s udy, CSR along he alue chain, is close o he
o me , as i emphasizes he oppo uni y o ﬁ ms o inc ease demand and ma kups, a he han he isk
4Con olling o ﬁ m-speciﬁc wages add esses he ollowing conce n: I s aﬀwel a e spending is a pa o
he o e all compensa ion package, downs eam ﬁ ms migh oﬀe sys ema ically highe pay due o highe skill
in ensi y, o example. A nega i e co ela ion be ween ups eamness and CSR spending (as measu ed in pa
by s aﬀwel a e spending) could hen be in e p e ed as ac ually eﬂec ing he nega i e ela ionship be ween
ups eamness and employee compensa ion.
5The ai ade label is s udied empi ically, among o he s, by Hainmuelle e al. (2015), de Jan y e al.
(2015), and D agusanu e al. (2022), and heo e ically by S ¨ahle and Richa dson (2014) and Podho sky (2015).
See D agusanu e al. (2014) o a su ey. Fo s udies o labels mo e gene ally, see Fische and Lyon (2014), Heyes
and Ma in (2017), as well as Po e (2019).
6See, e.g., Aldashe and Ve die (2009), Aldashe and Ve die (2010), Aldashe e al. (2015), K au heim and
Ve die (2016), Koenig e al. (2021), Fon agn´e and Lima di (2020).
80
o acing a de as a ing boyco . In ac , he p o ision o labels by NGOs can be seen as a mechanism o
s eng hen he posi i e eﬀec o CSR in es men s on demand. In ou pape , we ake his posi i e eﬀec
as a gi en and s udy how i shapes CSR in es men s along he alue chain.7
The e is a ecen and quickly expanding li e a u e on ela ional con ac s in de eloping coun ies
su eyed in Macchia ello (2021), which suppo s ou no ion o he ele ance o incomple e con ac s in
sou cing ela ionships especially in ol ing de eloping and eme ging economies.8Mos in o ma i e o
ou modeling (see Sec ion 5.2.3.1), Cajal-G ossi e al. (2020) show ha supplie s in he Bangladeshi
ga men sec o ob ain highe p ices when selling o ela ional buye s han when selling o spo buye s.
While hese pape s p o ide aluable insigh s in o he con ac ing ela ionships o o e come con ac ual
incomple eness in an en i onmen wi h weak legal ins i u ions, hey do no analyze he ela ion o CSR
in es men s.
Ou heo y is oo ed in he la ge li e a u e s udying he ole o incomple e con ac s o he in e -
na ional bounda ies o he ﬁ m, based on he seminal con ibu ions o An `as (2003) and An `as and
Helpman (2004), su eyed in An `as and Yeaple (2014) and An `as (2016). We build ou modeling on
An `as and Cho (2013) whe e a ﬁ m’s alue chain is modeled as a sequen ial p oduc ion p ocess. Like
mos o his li e a u e, hey ocus on he in eg a ion s. ou sou cing decision along he alue chain.
Cos ino e al. (2013) also model in e na ional sou cing as a sequen ial p oduc ion p ocess, bu wi hou
incomple e con ac s. They ocus on he endogenous agmen a ion o he alue chain ac oss coun ies.
We abs ac om bo h he in e na ional make-o -buy decision and he alloca ion o asks ac oss coun-
ies. Ou main ocus, in con as , is on CSR in es men s along a sequen ial alue chain cha ac e ized
by incomple e con ac s. Ou modeling opens up bo h he heo e ical and he empi ical oolki o he
li e a u e on in e na ional sou cing wi h incomple e con ac s o he analysis o social and en i onmen al
issues in globalized p oduc ion ha ea u e p ominen ly in he public deba e. Closes o ou pape in
his espec is ecen wo k by He kenhoﬀand K au heim (2022), who analyze a model ea u ing ca -
ing consume s, consume boyco s and incomple e con ac s o s udy he eﬀec s on he in e na ional
make-o -buy decision in a se ing wi h a single supplie . We, in con as , s udy he de e minan s o CSR
in es men s based on a model ha ea u es a con inuum o supplie s along a sequen ial alue chain. This
allows us o s udy he eﬀec o a supplie ’s alue chain posi ion on i s CSR in es men s and he e o e
he p oﬁle o CSR along he alue chain.
The e is an ex ensi e empi ical s and o his li e a u e, wi h Al a o e al. (2019) being a ecen
example. These pape s di ec ly ela e o he heo e ical unde pinnings, whe e incomple e con ac s lead
o unde in es men in a ela ionship-speciﬁc non-con ac ible a iable (inpu s, quali y-adjus ed inpu s,
e c.). As hese non-con ac ible a iables a e no o iously diﬃcul o obse e in he da a, he empi ical
s udies u n o he obse able op imal esponse o he ﬁ m o he unde in es men : he in eg a ion s.
ou sou cing decision. An `as and Cho (2013), o example, in es iga e he ole o ups eamness o a sup-
plie o i s – empi ically unobse able – quali y-adjus ed in es men . To gene a e es able implica ions,
hey he e o e u n o he obse able op imal esponse o he ﬁ m o he unde in es men : depending on
he ups eamness o he supplie , ei he in eg a ion o ou sou cing deli e s he igh incen i es minimiz-
ing he nega i e consequences o he unde in es men . Ou empi ical analysis complemen s he exis ing
li e a u e as we, in con as , di ec ly obse e he non-con ac ible a iable o in e es : CSR spending.
This allows us o in es iga e mo e di ec ly he ole o he GVC posi ion o supplie in es men s unde
7Closes o ou model in e ms o he ype o policy conside ed is Al a o-U e˜na e al. (2021) who de elop a
quan i a i e gene al equilib ium model o s udy he eﬀec s o esponsible sou cing in o igin coun ies based on
da a on aﬃlia es o mul ina ionals in Cos a Rica.
8Macchia ello (2010) uses a model wi h incomple e con ac s o s udy he eﬀec o e ical in eg a ion on he
pe asi e use o subcon ac ing a angemen s among small ﬁ ms in de eloping coun ies. Casabu i and Macchi-
a ello (2019) show how impe ec con ac en o cemen gene a es ba ie s o en y in he Kenyan dai y sec o .
Machia ello and Mo ja ia (2021) highligh he impo ance o long- e m ela ional con ac s in he Rwandan coﬀee
sec o . B ugu´es (2020) analyzes he ole o impe ec con ac s in he ex ile, pha maceu ical and cemen sec o s
in Equado depending on ma ke powe .
81
incomple e con ac s.
The e is a la ge li e a u e on CSR a he in e sec ion o he ﬁelds o managemen science and
indus ial o ganiza ion. Ki zmulle and Shimshack (2012), C i o and Fo ge (2015), and Pisani e al.
(2017) p o ide su eys o his li e a u e. This includes a la ge body o empi ical s udies on CSR which
mainly ocuses on he eﬀec , a he han he de e minan s, o CSR. We ake a diﬀe en angle, con ibu ing
o a be e unde s anding o he de e minan s o CSR by analyzing he impac o a ﬁ m’s alue chain
posi ion. Mo eo e , mos o his li e a u e uses indices combining quali a i e measu es ob ained by
manda o y CSR epo ing o lis ed ﬁ ms in de eloped a he han eme ging economies (e.g., Ma ano
and Kos o a, 2016; Pisani e al., 2017). We, in con as , use da a om India, a key eme ging economy,
which no only allows us o cons uc a con inuous mone a y measu e o CSR, bu also o s udy he
beha io o non-lis ed medium-sized and la ge ﬁ ms.
Ou empi ical analysis complemen s a small numbe o empi ical s udies assessing he ole o global
alue chain ela ionships o CSR. No ewo hy con ibu ions a e Schille (2018) and Newman e al.
(2018). Schille (2018) uses da a es ic ed o la ge publicly lis ed companies o show ha co po a e
en i onmen al and social policies o ﬁ ms a e ansmi ed o supplie s in hei alue chain and s udies he
eﬀec o hese policies on ﬁ m pe o mance. He p o ides e idence ha is consis en wi h he in e p e a ion
ha despi e incomple e con ac s ﬁ ms can s ill aﬀec hei supplie ’s CSR choices o some ex en . We,
in con as , a e in e es ed in how he limi ed con ol o a ﬁ m o e i s supplie s shapes hei CSR choices.
Mos ela ed o ou empi ical analysis is Newman e al. (2018). They exploi CSR indica o s ha
cap u e compliance wi h labo s anda ds, communi y and managemen - ela ed CSR p ac ices o analyze
he ansmission o socially esponsible beha io h ough ade in Vie namese ﬁ ms. They ﬁnd ha a
ﬁ m’s CSR inc eases wi h expo ing and impo ing, a ying ac oss expo des ina ions. The in e na ion-
aliza ion o supply chain ela ionships he e o e seems o ma e o CSR ac i i ies o ﬁ ms. Mo eo e ,
Newman e al. (2018) conjec u e ha in addi ion o he in e na ionaliza ion o buye -selle ela ionships,
he supplie ’s posi ion along he alue chain may ma e and con ol o i using ﬁ m ﬁxed eﬀec s. We,
in con as , use a ime- a ying, ﬁ m-speciﬁc measu e o a ﬁ m’s alue chain posi ion ha allows us o
di ec ly measu e he GVC posi ion o a ﬁ m and o analyze i s ela ionship o he ﬁ m’s CSR spending.
The emainde o he pape is s uc u ed as ollows. Sec ion 2 p esen s ou heo e ical analysis o
CSR along he global alue chain. Sec ion 3 ou lines ou empi ical analysis. Sec ion 4 concludes.
5.2 A Model o CSR along he Global Value Chain
In his sec ion, we inco po a e consume alua ion o e hical p oduc ion p ac ices and CSR in es -
men in o a model o sequen ial p oduc ion wi h incomple e con ac s based on An `as and Cho (2013).
We assume ha each ﬁ m in he Global No h has a alue chain wi h uni measu e o equi ed inpu s,
which a e all p o ided by independen supplie s in he Global Sou h, whe e en i onmen al and labo
egula ion is insuﬃcien om he pe spec i e o ca ing consume s in he Global No h. We de i e he
indus y equilib ium o he model and show ha ou heo y implies an inc easing CSR p oﬁle along he
alue chain – a p edic ion which we ake o he da a in Sec ion 3.
5.2.1 Incomple e Con ac ibili y o CSR
Be o e u ning o he ou line o ou model, we ﬁ s plead he case o he ele ance (and he ype)
o incomple e con ac s o CSR along global alue chains. Assu ing suﬃcien CSR le els by supplie s is
no o iously diﬃcul o achie e. Nike’s esponse o he “swea shop campaign” in he 1990s is p obably
he bes documen ed and mos esea ched case in poin . P o es s agains wo k condi ions in Indonesian
supplie ac o ies led Nike o implemen compulso y, legally binding codes o conduc o all i s sup-
82
plie s.9Ha ison and Sco se (2010) show ha he campaigns caused an inc ease in he eal wages in
Indonesian ac o ies manu ac u ing o Nike. Locke (2013a), howe e , shows ha mo e han a decade
a e he in oduc ion o legally binding codes o conduc combined wi h an audi ing sys em, subs an ial
compliance p oblems pe sis ed in Nike’s supply chain. He analyzes da a om ac o y audi s o wo king
condi ions in mo e han 900 o Nike’s supplie s loca ed ac oss ﬁ y coun ies and ﬁnds ha , despi e
he ac ha all supplie s a e obliged o sign he codes o conduc and despi e la ge in es men s in o
moni o ing, he e s ill is a subs an ial numbe o supplie s ha a e anked as “noncomplian ”. In Asia,
he noncomplian supplie s e en cons i u e he majo i y. While Nike is a pa icula ly well-documen ed
case in poin , hese p oblems a e by no means speciﬁc o Nike o he oo wea and appa el sec o . Based
on da a om one o he wo ld’s la ges supply chain audi ing ﬁ ms con aining 16,795 audi s o 5,819
ac o ies in 13 indus ies ac oss 66 coun ies o e a pe iod o six yea s, Sho , Toﬀel and Hugill (2016,
2020) documen widesp ead iola ions o codes o conduc in a eas like child labo , o ced labo , wo king
hou s, occupa ional heal h and sa e y, he minimum wage and disciplina y p ac ices.
In ou iew, his calls o a model o CSR along GVCs ha explici ly accoun s o incomple e con ac s
be ween a ﬁ m and i s supplie s. One s anda d way o mo i a ing con ac ual incomple eness is ha i
is p ohibi i ely cos ly - i possible a all - o co e all con ingencies in a con ac . This does no seem o
be a majo conce n in ou con ex : codes o conduc and simila con ac s can speci y minimum social,
labo and en i onmen al s anda ds. In he case o CSR in es men s, he con ac ual incomple eness
s ems om ano he s anda d a gumen : he ac ions o he supplie a e “obse able bu no e iﬁable”
(e.g., Ha and Moo e, 1999). This means ha i is cos less (o a leas ela i ely cheap) o de e mine
wha is going on in a supplie ’s ac o y bu is p ohibi i ely cos ly o hold he supplie accoun able o
he induced losses in a cou o law.
The e a e qui e gene al and well-known p oblems wi h e i ying labo and en i onmen al s anda ds
implemen ed in p oduc ion plan s, especially so in low- egula ion coun ies. Sa e y equipmen may be
dis ibu ed one day, bu no he o he , ﬁ e exi s may be accessible one day, bu blocked he o he , wo ke s
may ea he loss o hei jobs, i hey epo o ced and excessi e o e ime, oxic was e may be disposed
in o a i e when no-one is wa ching.10 I is ne e heless ela i ely cheap o , say, an ad ocacy NGO o
alk o wo ke s, measu e oxins in was e wa e o send an unde co e agen in o a ac o y o ﬁnd ou
abou he ac ual p oduc ion condi ions. Such ci cums an ial e idence, possibly ob ained by b eaking
domes ic laws, is insuﬃcien in a cou o law. Howe e , in many cases i is suﬃcien o con ince ca ing
consume s o e-assess hei alua ion o he p oduc and o educe demand.11 P oducing he ype o
e idence ha would lead o a con ic ion o he supplie - hough in p inciple possible - will in many cases
be p ohibi i ely cos ly o p oduce. I would equi e a massi e and con inuous p esence o audi o s on si e
and e en hen, he e is no gua an ee ha hey will win he ca -and-mouse game wi h local managemen .
This highligh s ha he no ion in he incomple e con ac s li e a u e ha he ac ion is “obse able bu
no e iﬁable” (e.g. Ha and Moo e, 1999) applies o CSR in es men s along GVCs.12
9In May 1998, Nike CEO Phil Knigh amously made he ollowing s a emen : “The Nike p oduc has become
synonymous wi h sla e wages, o ced o e ime, and a bi a y abuse. I uly belie e he Ame ican consume
doesn’ wan o buy p oduc s made unde abusi e condi ions.” (New Yo k Times, 1998).
10Fo a i id desc ip ion o he ypical p oblems o audi o s in he ﬁeld, see Locke (2013b), p. 35-37.
11This is all he mo e so as NGOs a e epea edly epo ed o be he mos us ed ins i u ions
among consume s be o e go e nmen s and he media. See, o example, he Edelman T us Ba om-
e e o 2020, a h ps://cdn2.hubspo .ne /hub s/440941/T us %20Ba ome e %202020/2020%20Edelman%
20T us %20Ba ome e %20Global%20Repo -1.pd .
12In ou modeling, we assume o simplici y ha he ﬁ m and consume s can obse e he ac ual le el o CSR
by he supplie - bu ha i is no e iﬁable by a cou . In case ha some eade s ake issue wi h he no ion
ha he ac ual CSR le el is obse able bu non- e iﬁable, no e ha his assump ion can be elaxed - u he
b oadening he gap be ween he cos o obse ing and he cos o e i ying. Fo he mechanics o ou model o
unc ion, we do no need he ﬁ m and consume s o obse e he ac ual CSR in es men s. The ﬁ m only needs
o know he CSR le el which consume s obse e o pe cei e (howe e misguided hei pe cep ion may be) and
which hey use o o m hei opinion on he e hical quali y, ε(ω). This obse able/pe cei ed CSR le el is wha
83
p e ious CSR le els. Unde he main ained assump ion discussed in Sec ion 5.2.2.1 ha α∈(0,ρ), CSR
in es men a s age mis inc easing in p e ious CSR le els. This allows us o s a e he ollowing lemma:
Lemma 1 Wi h α∈(0,ρ), CSR in es men s along he alue chain a e sequen ial complemen s, i.e.,
highe le els in p e ious s ages lead o a highe op imal le el o CSR in s age m.
P oo : In he ex .
When he incen i es o CSR in es men o supplie s depend on p e ious in es men s by o he
supplie s, he posi ion in he alue chain ela i e o o he supplie s can become a c ucial de e minan o
CSR in es men . We he e o e now u n o he sequence o CSR in es men s along he comple e alue
chain. Combining equa ion (5.11) wi h equa ion (5.9), we ob ain
�(m) = 1
α�1−β
cs�α
1−α�ρAα(1−ρ)
ρq(ω)α�1
1−α (m)ρ−α
ρ(1−α).(5.12)
Using he ini ial condi ion (0) = 0, sol ing his diﬀe en ial equa ion deli e s
(m) = �1−ρ
1−α�ρ(1−α)
α(1−ρ)�1−β
cs
ρ�ρ
1−ρ
A q(ω)ρ
1−ρmρ(1−α)
α(1−ρ).(5.13)
We can now plug his in o equa ion (5.11), which deli e s CSR in es men s a s age mas
s(m) = �1−ρ
1−α�ρ−α
α(1−ρ)�1−β
cs
ρ�1
1−ρ
A q(ω)ρ
1−ρmρ−α
α(1−ρ).(5.14)
This exp ession o CSR in es men s a s age maccoun s o CSR in es men s in all p e ious s ages,
which aﬀec he op imal choice a s age macco ding o Lemma 1. Equa ion (5.14) shows ha he
s anda d implemen ed a s age mdepends on he size o he o de q(ω) placed by he ﬁ m, which we
de e mine nex .27
To al physical ou pu q(ω) is chosen by he ﬁ m in he ﬁ s s age o he game. The ﬁ m knows ha
i will ei he ha e o pu chase inpu s on he inpu ma ke o has o compensa e i s supplie s o hei
p oduc ion cos s. In ei he case, cos s a e gi en by �1
0cxq(ω)dj =cxq(ω). The p oblem o he ﬁ m is
he e o e gi en by:
maxq(ω)πF(ω) = β (ω)−cxq(ω)− . (5.15)
Based on equa ion (5.13) e alua ed a m= 1, we can w i e o al e enue as
(ω) = �1−ρ
1−α�ρ(1−α)
α(1−ρ)�1−β
cs
ρ�ρ
1−ρ
Aq(ω)ρ
1−ρ.(5.16)
Diﬀe en ia ing πF(ω) wi h espec o q(ω) and se ing he de i a i e equal o ze o gi es he op imal
quan i y q(ω) as
q(ω) = �1−ρ
1−α�ρ(1−α)
α(1−2ρ)�1−β
cs
ρ�ρ
1−2ρ�ρ
1−ρ
β
cx
A�1−ρ
1−2ρ
.(5.17)
27No e ha he ac ha s(m) inc eases in he size o he o de q(ω) placed by he ﬁ m implies ha he ﬁ m
has an incen i e o o de he en i e quan i y o inpu m om one single supplie , as his maximizes he le el
o he s anda d unde which p oduc ion akes place and he e o e also maximizes he le el o pe cei ed e hical
quali y o he ﬁnal p oduc .
90

We can combine equa ions (5.14) and (5.17) and use he deﬁni ion o A=EP ρ
1−ρ o ge he op imal
CSR in es men a s age m,
s(m) = �1−ρ
1−α�ρ−α(1−ρ)
α(1−2ρ)�1−β
cs
ρE�1−ρ
1−2ρ�ρ
1−ρ
β
cx
P�ρ
1−2ρ
mρ−α
α(1−ρ).(5.18)
We can now s a e he ollowing lemma:
Lemma 2 Equa ion (5.18) cha ac e izes how supplie s along a ﬁ m’s alue chain op imally choose hei
CSR le els aking agg ega e a iables as gi en (pa ial equilib ium). The p oduc ion s anda d is highe in
mo e downs eam s ages, i.e., supplie s ha a e close o he ﬁnal consume in es mo e in CSR. Wi h
he p ice index P aken as gi en, CSR in es men s also inc ease in he o al expendi u e on e hically
diﬀe en ia ed goods E.
P oo : This ollows di ec ly om equa ion (5.18) and ρ<1/2.
Lemma 2 shows ha he supplie ’s CSR in es men inc eases in m, i.e., CSR is highe in mo e
downs eam s ages. The esul eﬂec s he ac ha CSR in es men s a e sequen ial complemen s along
he alue chain, as poin ed ou in Lemma 1. In addi ion, he o e all le el o CSR in es men ac oss
all s ages is inc easing in agg ega e expendi u e on e hically diﬀe en ia ed goods. We will see below
ha Edisappea s om he equa ion when we sol e o he indus y equilib ium o ou model. The e, a
supplie ’s posi ion in he alue chain will eme ge as he key de e minan o i s CSR expendi u es in ou
model.
5.2.5 Equilib ium CSR
We now de i e he indus y equilib ium o he model wi h ee en y o ﬁnal goods p oduce s.
This allows us o analyze he equilib ium de e minan s o CSR in es men s a he diﬀe en s ages o
p oduc ion.
5.2.5.1 Indus y Equilib ium
To sol e o he indus y equilib ium, we assume ee en y and ﬁnal good p oduce s wi h homoge-
neous le els o p oduc i i y no malized o 1 so ha i mus hold ha πF(ω)=0∀ω.
Se ing equa ion (5.15) equal o ze o, plugging in he op imal q(ω) om equa ion (5.17), and using
he ac ha A=EP ρ
1−ρgi es an exp ession o he p ice index Pas
P=�1−α
1−ρ�ρ−α
αρ �1−α
βE�1−ρ
ρcscx
ρ2(1 −β)�
1−2ρ�1−2ρ
ρ
.(5.19)
Nex , we sol e o he equilib ium numbe o ﬁ ms using he op imal p icing ule and he deﬁni ion o
he ideal p ice index om demand.28 F om his, we ge a second exp ession o he p ice index as a
unc ion o he numbe o ﬁ ms nas
P=n−1−ρ
ρ
cxcs
ρ2(1 −β)
1−2ρ
�1−α
1−ρ�1−α
α
.(5.20)
Combining he wo esul s o he p ice index, one om ee en y, equa ion (5.19), and one om demand,
equa ion (5.20), gi es he equilib ium numbe o ﬁ ms as
n=βE
1−ρ
1−2ρ
.(5.21)
28De ails o he de i a ions a e p o ided in Appendix 5.5.1.
91
Combining equa ion (5.19) wi h equa ion (5.17) on he one hand, and he deﬁni ion o A=EP ρ
1−ρ
on he o he hand, yields exp essions o o de size q(ω) and he ma ke size pa ame e Ain indus y
equilib ium as
q(ω) = ρ
cx
1−2ρ(5.22)
and
A=�1−α
1−ρ�ρ−α
α(1−ρ)1−α
β�cscx
ρ2(1 −β)�ρ
1−ρ�
1−2ρ�1−2ρ
1−ρ
.(5.23)
This allows us o s a e he ollowing lemma:
Lemma 3 In indus y equilib ium wi h ee en y in he ﬁnal goods sec o , ﬁnal goods p oduce size q(ω)
and he equilib ium ma ke size pa ame e Aa e bo h independen o exogenous agg ega e expendi u e E.
P oo : Follows di ec ly om equa ions (5.22) and (5.23).
Ma ke and ﬁ m size a e bo h independen o he agg ega e le el o expendi u e Ein he economy.
The eason is ha in ou se up o ee en y wi h homogeneous ﬁnal good p oduce s, any inc ease in
agg ega e spending will be ully compensa ed by addi ional en y in o he ﬁnal goods sec o .
5.2.5.2 Equilib ium CSR along he Global Value Chain
Combining equa ions (5.18) and (5.19), he CSR le el implemen ed a s age mcan be w i en as
s(m) =
1−2ρ
(1 −α)ρ
cs
1−β
βmρ−α
α(1−ρ).(5.24)
I only depends on pa ame e s and he p oduc ion s age m. We can now s a e he ollowing p oposi ion:
P oposi ion 1 – Equilib ium CSR along he Global Value Chain. Wi h CSR in es men s ac oss
p oduc ion s ages being sequen ial complemen s, CSR expendi u es s(m)in indus y equilib ium a e in-
c easing along he alue chain om ups eam o downs eam s ages.
P oo : This ollows di ec ly om equa ion (5.24).
This cons i u es he main heo e ical esul o ou pape : equilib ium CSR le els inc ease along he
alue chain as supplie s ge close o he ﬁnal consume . The esul is d i en by he ac ha in es men s
in CSR a e sequen ial complemen s, as es ablished in Lemma 1: he la ge CSR in es men s a a gi en
s age m, he highe he incen i e o mo e downs eam supplie s o in es mo e. Compa ison o he
pa ial equilib ium exp ession (5.18) and he equilib ium exp ession in (5.24) shows ha he inc easing
p oﬁle, ou cen al ﬁnding, pe sis s in indus y equilib ium. The eﬀec o o al expendi u e Ees ablished
in Lemma 2, howe e , does no su i e endogenizing he p ice index P. We a gue below ha we use his
diﬀe ence o d aw he line be ween he co e es able implica ion o ou model, which holds in indus y
equilib ium, and a iables we me ely con ol o in he empi ical analysis, as hey play a ole in he
pa ial equilib ium analysis only.
Compa ison o P oposi ion 1 o he ou come unde comple e con ac s (see Appendix 5.5.2) shows
ha he inc easing p oﬁle o CSR esul s om he se ing o incomple e con ac s. We a gued abo e ha
CSR in es men s a e in many cases obse able bu non- e iﬁable (Ha and Moo e, 1999). Ou model
he e o e hin s a a gene al p oblem ha may well be a ypical ea u e o CSR along global alue chains:
when CSR in es men s a e sequen ial complemen s ( he case o which we a gued in Sec ion 5.2.2.1),
he ea ly s ages o he p oduc ion p ocess a e he ones ha a e c ucial o he choices along he en i e
92
alue chain; a he same ime, hese a e he s ages wi h he lowes CSR in es men s. This implies ha
o bo h headqua e s and policy make s who seek o inc ease o e all CSR le els along he alue chain,
he ups eam s ages o p oduc ion a e he mos impo an ones o a ge .
5.2.6 Discussion and Link o he Da a
Be o e we u n o he empi ical analysis, we pause o ﬁ s discuss how we link ou heo e ical esul s
o he da a. We hen place ou analysis in he con ex o he li e a u e on incomple e con ac s in
in e na ional economics and explain why ou heo y-in o med empi ical analysis closes a gap in his
li e a u e.
5.2.6.1 Tes able Implica ions: Pa ial s. Indus y Equilib ium
As poin ed ou abo e, compa ison o he pa ial equilib ium exp ession o CSR (5.18) and i s
indus y equilib ium coun e pa (5.24) shows ha he cen al p edic ion o ou heo y – he inc easing
p oﬁle o CSR along he alue chain – also holds in indus y equilib ium. We b ing his p edic ion o he
da a in Sec ion 5.3, whe e we also p o ide a de ailed desc ip ion o he da a. A his s age, simply no e
ha we use Indian ﬁ m-le el da a oge he wi h inpu -ou pu ables o compu e a ﬁ m-le el measu e o
ups eamness. Mo eo e , we obse e CSR in es men s a he ﬁ m le el. This allows us o in es iga e
he nega i e ela ionship be ween ups eamness and CSR in es men s es ablished in P oposi ion 1. I
we had ended he heo e ical analysis wi h he pa ial equilib ium exp ession o CSR in es men s in
equa ion (5.18), we could ha e deduced addi ional es able implica ions om he heo y. Speciﬁcally,
aking he p ice index Pas gi en, Lemma 2 s a es ha he CSR in es men s by he supplie a s age m
o he p oduc ion p ocess inc ease in o al expendi u es on e hically diﬀe en ia ed goods, E.
One could es his implica ion om pa ial equilib ium empi ically by aking, e.g., he sha e o
sales going o OECD coun ies as a p oxy o being exposed o s onge demand o e hical quali y.
P oposi ion 1, howe e , shows ha when he equilib ium p ice index is accoun ed o , he eﬀec o E
anishes. This is an implica ion o he simple K ugman (1980) s uc u e o ou modeling o he e hically
diﬀe en ia ed homogeneous ﬁnal good p oduce s, which leads hem o ha ing a cons an equilib ium size.
Adjus men s o changes in agg ega e a iables a e exclusi ely d i en by adjus men s in he numbe o
ﬁ ms (see equa ions (5.21) and (5.22)). As he impac o demand o e hical p oduc s does no su i e
he compu a ion o he indus y equilib ium, we do no conside i a co e es able implica ion o ou
model. The pa ial equilib ium analysis ne e heless poin s o i s ele ance, and we he e o e include in
ou eg essions a p oxy o OECD expo exposu e as a con ol a iable.
5.2.6.2 A Di ec Measu e o he Non-Con ac ible Va iable
As indica ed abo e, ou da a allow us o ake a e y di ec app oach o linking ou model o he da a.
In ou opinion, his app oach ﬁlls a gap in he empi ical li e a u e on he ole o incomple e con ac s in
in e na ional economics. The li e a u e is su eyed in An `as and Yeaple (2014) and An `as (2016), wi h
Al a o e al. (2019) being a ecen example. The empi ical s udies a e g ounded in heo e ical models,
whe e incomple e con ac s lead o unde in es men in a ela ionship-speciﬁc non-con ac ible a iable
( ypically quali y-adjus ed inpu s). In ou model, his non-con ac ible a iable is CSR in es men ,
which has some simila i ies wi h he quali y dimension in he men ioned li e a u e. I goes wi hou
saying ha such non-con ac ible a iables a e diﬃcul o measu e. The li e a u e he e o e ypically
eso s o he obse able o ganiza ion o p oduc ion as a dependen a iable, which – acco ding o he
heo y – is d i en by he unobse able non-con ac ible a iable. Nunn (2007) sheds empi ical ligh
on his heo e ical mechanism showing ha ela ionship-speciﬁci y shapes he pa e ns o in e na ional
ade. Based on Nunn (2007), Nunn and T eﬂe (2008) cons uc a coun y and indus y speciﬁc measu e
93
o con ac ual comple eness. They ﬁnd ha imp o ed con ac ing o he inpu s p o ided by he supplie
a o s in eg a ion o e ou sou cing. Ou analysis adds an addi ional and complemen a y dimension:
To ou knowledge, ou s is he ﬁ s pape o use a di ec measu e o he non-con ac ible a iable.
This becomes possible as we shi he a en ion om some echnical, bu by cons uc ion ex e nally
unobse able, p oduc o se ice cha ac e is ic o CSR in es men s, o which we do ha e a ﬁ m-le el
measu e in ou da a. We can he e o e di ec ly s udy he eﬀec – in ou case o a supplie ’s alue chain
posi ion – on he non-con ac ible a iable wi hou ecu ing o he o ganiza ion o p oduc ion as a
subs i u e obse able ou come o he unobse able (unde )in es men decision.29
5.3 GVC Posi ion and CSR in India
In his sec ion, we ake he model o he da a by es ing whe he CSR spending is la ge when ﬁ ms
a e posi ioned mo e downs eam in global alue chains. We begin wi h he desc ip ion o he da a.
5.3.1 Da a Desc ip ion
Fo ou analysis, we combine he Wo ld Inpu -Ou pu Da abase (WIOD, Timme e al., 2015) wi h
he P owess da abase.30 Ou sample co e s he yea s 2000 o 2013. The P owess da abase con ains a
wide ange o Indian ﬁ m-le el in o ma ion. I s co e age o he Indian economy is comp ehensi e, wi h
o al p oduc ion o all companies in P owess accoun ing o mo e han 80% o India’s GDP (Bos and
Vannoo enbe ghe, 2018). The da a a e collec ed by he Cen e o Moni o ing Indian Economy (CMIE),
which mainly sou ces in o ma ion om annual epo s o ﬁ ms, s ock exchanges and egula o s. India is
pa icula ly well-sui ed o b ing he heo e ical p edic ions o ou model o he da a.31
In ou heo y, CSR is any ac ion by a supplie ha leads o an inc ease in he pe cei ed e hical
quali y o he ﬁnal p oduc . This can be any ac ion ha makes he supplie appea mo e e hical in he
eyes o consume s. We he e o e use de ails on ﬁ ms’ spending on s aﬀwel a e and expenses o he
beneﬁ o socie y o he communi y in gene al as ou measu e o a ﬁ m’s CSR ac i i ies, ou dependen
a iable. I is a unique ea u e o he da a, and especially a e in an eme ging economy con ex , ha we
can obse e his in o ma ion a he ﬁ m le el. We desc ibe he a iable in mo e de ail in Sec ion 5.3.2.1.
Mo eo e , he da a include de ailed in o ma ion on gene al ﬁ m cha ac e is ics such as annual sales,
expo ac i i y, owne ship, inpu sou cing, age and he p oduc mix. The la e is key o ou empi ical
s a egy.32 We exploi diﬀe ences in he p oduc mix ac oss ﬁ ms as well as wi hin ﬁ ms o e ime o
cons uc a ﬁ m-speciﬁc measu e o he global alue chain posi ion. To do so, we combine he P owess
da a wi h indus y-le el in o ma ion om WIOD. In a ﬁ s s ep, we ollow An `as e al. (2012) and An `as
and Cho (2018) and use WIOD o cons uc a ime- a ying measu e o Indian indus ies’ ups eamness.
29We do acknowledge ha he analysis o he de e minan s o he in eg a ion s. ou sou cing decision o he
ﬁ m is an impo an endea o in i s own igh . He e, we simply wan o highligh ha ou app oach allows us o
in es iga e he non-con ac ible a iable di ec ly and he eby closes a gap in he p e ious li e a u e.
30We use he P owess in age o Decembe 2017. Da a we e downloaded om h ps://p owessdx.cmie.
com on Feb ua y 1, 2018. The WIOD elease 2016 was downloaded om h p://wiod.o g/home. Da a we e
downloaded on July 22, 2019.
31Fi s , he educ ion in ade ba ie s ollowing India’s accession o he WTO in 1995 subs an ially inc eased
India’s in eg a ion in o GVCs. Second, compa ed o mos de eloped coun ies, ﬁ ms in India a e much less
cons ained by labo and en i onmen al s anda ds lea ing ample oom o olun a y CSR in es men s. Finally,
ela ed o ou assump ion o incomple e con ac s, he conges ed cou sys em (Boehm and Obe ﬁeld, 2020)
ampliﬁes he p oblem ha con ac s be ween he supplie and he headqua e a e obse able bu no e iﬁable
(Ha and Moo e, 1999).
32Goldbe g e al. (2010b) emphasize ha i is a a e ea u e o he P owess da abase ha i cap u es annual
changes in he p oduc mix a he ﬁ m le el. This ea u e s ems om he Indian 1956 Companies Ac equi ing
ﬁ ms o disclose p oduc -le el in o ma ion on capaci ies, p oduc ion and sales in hei annual epo s. Se e al
p e ious s udies ha e used his in o ma ion, e.g., Goldbe g e al. (2010a), Goldbe g e al. (2010b), De Loecke
e al. (2016), Ba ows and Olli ie (2018), Bos and Vannoo enbe ghe (2018) and Ba ows and Olli ie (2021).
94
Linking WIOD indus ies o he ﬁ ms’ p oduc s om P owess, we use he p oduc -le el sales sha es as
weigh s in he calcula ion o an a e age, ﬁ m-speciﬁc ups eamness, ou measu e o a ﬁ m’s alue chain
posi ion. We desc ibe he calcula ion o ou main independen a iable in mo e de ail in Sec ion 5.3.2.2.
No ably, mul i-p oduc ﬁ ms accoun o 47% o Indian manu ac u ing ﬁ ms and 80% o manu ac-
u ing ou pu (Goldbe g e al., 2010b, p.1043). The e o e, a key ea u e o ou app oach is ha e en o
ﬁ ms wi hin he same main indus y, measu ed GVC posi ions may diﬀe ac oss ﬁ ms, ei he because o
diﬀe ences in he p oduc mix o because o diﬀe ences in he sales weigh s o iden ical p oduc mixes.
Ou measu e also a ies o e ime and wi hin ﬁ ms i a ﬁ m’s p oduc sales weigh s change o i i adds
o d ops p oduc s. In addi ion, a ﬁ m’s alue chain posi ion may change o e a longe ho izon i he
posi ion o i s p oduc s changes.
Sec ion 135 o he 2013 Indian Companies Ac obliges ﬁ ms abo e a ce ain u no e and p oﬁ abili y
h eshold o spend 2% o hei a e age ne p oﬁ s on CSR. Compliance wi h he law changes he CSR
decision o a ﬁ m and migh also change i s epo ing beha io .33 To a oid such a discon inui y in hei
CSR spending, we limi he analysis o he pe iod be ween 2000 and 2013, al hough ou da a a e a ailable
up o 2014.34 In he nex sec ion, we desc ibe ou key a iables o in e es , CSR spending and a ﬁ m’s
GVC posi ion, in mo e de ail.
5.3.2 Measu ing CSR Spending along GVCs
A complex se o mo i es can d i e a ﬁ m’s s a egy on CSR spending (B´enabou and Ti ole, 2010;
C i o and Fo ge , 2015). C i o and Fo ge (2015) desc ibe a ﬁ m’s CSR decision as a esponse o ma ke
impe ec ions wi h he aim o sa is ying social p e e ences o a leas one s akeholde . In he con ex o
GVCs, sa is ying hese social p e e ences becomes e en mo e complex as ﬁ ms ace no jus domes ic bu
also o eign s akeholde demands (Newman e al., 2018). Consume s in de eloped coun ies inc easingly
demand ha minimum s anda ds o wo king condi ions and en i onmen al aspec s o p oduc ion a e
obse ed also a p oduc ion si es ab oad. Be o e we ocus on he empi ical speciﬁca ion, we explain ou
measu e o CSR spending.
5.3.2.1 Measu ing a Fi m’s CSR Spending
The e is no gene al ag eemen abou he deﬁni ion o CSR no a common way o quan i ying CSR
spending (C i o and Fo ge , 2015; Newman e al., 2020). Shi odka e al. (2018) claim ha in he con ex
o de eloping and eme ging coun ies, CSR p ima ily e e s o he commi men o beha e e hically and
o con ibu e o economic de elopmen . McWilliams and Siegel (2001, p.117) deﬁne CSR mo e s ic ly
as “ac ions ha appea o u he some social good, beyond he in e es s o he ﬁ m and ha which is
equi ed by law”. In his pape , we ollow his concep and a la ge li e a u e by deﬁning ou dependen
a iable as social spending beyond wha is equi ed by law (B´enabou and Ti ole, 2010; C i o and Fo ge ,
2015; Newman e al., 2018, 2020).
We a gue ha his app oach is especially well-sui ed in he con ex o GVC in eg a ion o ﬁ ms
o igina ing om de eloping o eme ging coun ies because hese ﬁ ms a e ope a ing in a ela i ely lax
social and en i onmen al egula o y se ing. Facing lenien domes ic egula ion, ﬁ ms espond o ma ke
ailu e wi h highe CSR spending o sa is y s akeholde demands (B´enabou and Ti ole, 2010; C i o and
Fo ge , 2015). In he con ex o GVC in eg a ion, he CSR decision o ﬁ ms loca ed in de eloping
coun ies is de e mined by ela i ely highe demand o CSR by o eign s akeholde s (Newman e al.,
2018).
33See Dha mapala and Khanna (2018) o mo e de ails on he legisla ion and an e alua ion o he eﬀec s on
ﬁ m’s CSR ac i i ies.
34To con ol o po en ial an icipa ion eﬀec s, we ca y ou a obus ness es (a ailable upon eques ) whe e
we exclude he yea s 2011 o 2013. We ﬁnd ha ou esul s a e unaﬀec ed.
95

Ou da a allow us o di ec ly obse e mone a y CSR spending on own s aﬀwel a e and expenses
o he beneﬁ o he socie y o he communi y. We use his in o ma ion o measu e wha we call
p oduc ion s anda d o CSR in es men in he heo e ical model. S aﬀwel a e spending e e s o a ious
ameni ies ha a e made a ailable o he employees o hei gene al wel a e and go beyond egula
compensa ion in he o m o sala ies. S aﬀwel a e spending includes ee o subsidized medical ea men ,
anspo a ion acili ies, ec ea ion acili ies, s aﬀ ood, and can een expenses. These aspec s all co e
labo wel a e, which is he p edominan componen o CSR spending in eme ging economies (Newman
e al., 2018). Socie y o communi y spending e e s o expenses on building o main aining public pa ks,
ga den main enance, building emples, cons uc ing oads o con ibu ing o social occasions, e c. These
communi y- ela ed expenses closely e e o he ca alog o communi y- ela ed CSR ac i i ies applied by
Newman e al. (2018) and Newman e al. (2020).35
Ou abili y o obse e he money amoun o CSR spending allows us o in es iga e he con inuous
ela ionship be ween CSR spending and a ﬁ m’s posi ion in GVCs. Acco dingly, we can di ec ly exploi
a ia ion in he le el o CSR spending ac oss ﬁ ms and wi hin a ﬁ m o e ime. We a gue ha ou ocus
on obse able da a o he mone a y expenses ela ed o CSR, bu no labeled CSR, subs an ially educes
a possible bias due o sel - epo ing o ﬁ ms as discussed in Newman e al. (2020).
5.3.2.2 Measu ing a Fi m’s Value Chain Posi ion
We now u n o he key explana o y a iable o in e es : a ﬁ m’s GVC posi ion. Ou model p edic s
ha a ﬁ m’s CSR spending depends c ucially on i s posi ion in he alue chain. Wi hou ecu ing o
a heo e ical model, Newman e al. (2018) con ol o he ole o he alue chain posi ion by using a
ime-in a ian ﬁ m ﬁxed eﬀec . Ou da a allow us o no only con ol o a possible impac o he alue
chain posi ion o CSR bu o es ima e he eﬀec . To do so, we calcula e a ime- a ying and ﬁ m-speciﬁc
measu e o a ﬁ m’s posi ion in GVCs based on a combina ion o alue chain in o ma ion o indus ies
om he Wo ld Inpu -Ou pu Da abase and he P owess in o ma ion on p oduc -le el sales o ﬁ ms.
In a ﬁ s s ep, we calcula e Indian indus ies’ posi ion in GVCs based on con ibu ions by An `as e al.
(2012) and An `as and Cho (2018). They show how Wo ld Inpu -Ou pu Tables can be used o calcula e
an indus y’s ups eamness, measu ing he dis ance o an indus y’s sales om ﬁnal consump ion. In
line wi h his app oach, we s a by compu ing ups eamness U
ko indus y in coun y k(India) by
compu ing
U
k=1F
k
Y
k
+ 2�W
w=1 �L
l=1 a w
kl Fw
l
Y
k
+ 3�W
w=1 �L
l=1 �W
x=1 �L
m=1 a w
kl awx
lm Fx
m
Y
k
+...,
(5.25)
whe e F
k/Y
kis he sha e o global ﬁnal consump ion in indus y ’s g oss ou pu in coun y k. The
ollowing e ms cap u e he p oximi y o ﬁnal consump ion o indus y in coun y k h ough i s sales
in in e media es. Fo ins ance, he second summand includes he sha e o ou pu om indus y in
coun y k, which is consumed by indus y win coun y land sold o ﬁnal consume s, mul iplied by
a w
kl , which measu es he dolla amoun o indus y ’s ou pu in coun y kneeded o p oduce one dolla
wo h o indus y w’s ou pu in coun y l.
The ups eamness o an indus y is he weigh ed a e age o he numbe o s ages ha sepa a e he
ypical uni o ou pu in indus y om ﬁnal consume s. The weigh ing scheme gi es highe weigh o
35Ano he impo an CSR componen is spending ela ed o en i onmen al issues. Al hough we obse e
expenses on en i onmen al causes in ou da a, we do no include his dimension in ou main speciﬁca ion. The
eason is ha we canno be su e ha hese expenses a e uly olun a y. This a iable could also cap u e, e.g.,
compulso y in es men in en i onmen ally iendly echnologies equi ed by law. The esul s o he ex ended
CSR measu e including en i onmen al expenses s ay obus and a e a ailable upon eques .
96
he mo e indi ec sales. I indus y ’s ou pu in coun y kis comple ely sold o ﬁnal consume s, U
k
equals uni y. I a pa o he ou pu is sold as in e media es, U
kis abo e one and a highe alue o
U
kassocia es wi h highe ups eamness. Acco dingly, la ge alues o U
ka e associa ed wi h inc eased
dis ance o ﬁnal consump ion.
We use he WIOD o apply his me hodology o he Indian economy. We exclude non- adable
indus ies in he cons uc ion o ou key dependen a iable.36 Table 5.1 shows summa y s a is ics
o he de i ed indus y ups eamness measu e o 34 adable 2-digi indus ies based on he ISIC-4
classiﬁca ion. In ou pe iod, he a e age ups eamness o Indian indus ies a ies be ween 1.22 (wa e
Table 5.1: Indus y ups eamness measu e o India
ISIC-4 Indus y mean sd min max
Mining and qua ying 3.80 0.18 3.52 4.08
Manu ac u e o chemicals and chemical p oduc s 3.33 0.06 3.21 3.40
Manu ac u e o coke and eﬁned pe oleum p oduc s 2.79 0.20 2.52 3.05
Manu ac u e o basic me als 2.71 0.07 2.60 2.80
Manu ac u e o pape and pape p oduc s 2.63 0.08 2.48 2.74
Manu ac u e o o he non-me allic mine al p oduc s 2.34 0.03 2.29 2.42
Manu ac u e o ubbe and plas ic p oduc s 2.33 0.03 2.28 2.40
Manu ac u e o ab ica ed me al p oduc s, excep machine y and equipmen 2.27 0.04 2.20 2.32
Telecommunica ions 2.24 0.10 2.08 2.42
Manu ac u e o wood and o p oduc s o wood and co k, excep u ni u e;
manu ac u e o a icles o s aw and plai ing ma e ials 2.17 0.06 2.05 2.24
Manu ac u e o compu e , elec onic and op ical p oduc s 2.14 0.20 1.93 2.58
Wholesale and e ail ade and epai o mo o ehicles and mo o cycles 2.10 0.06 2.02 2.20
Wa ehousing and suppo ac i i ies o anspo a ion 2.07 0.07 1.96 2.18
Re ail ade, excep o mo o ehicles and mo o cycles 2.07 0.05 1.99 2.15
Wholesale ade, excep o mo o ehicles and mo o cycles 2.07 0.05 1.99 2.15
Manu ac u e o elec ical equipmen 2.00 0.10 1.88 2.17
P in ing and ep oduc ion o eco ded media 2.00 0.07 1.88 2.12
Legal and accoun ing ac i i ies; ac i i ies o head oﬃces; managemen consul ancy ac i i ies 2.00 0.07 1.86 2.08
Land anspo and anspo ia pipelines 1.95 0.05 1.87 2.03
Manu ac u e o machine y and equipmen n.e.c. 1.94 0.07 1.83 2.05
Manu ac u e o u ni u e; o he manu ac u ing 1.90 0.10 1.77 2.08
Fo es y and logging 1.88 0.05 1.81 1.97
Manu ac u e o o he anspo equipmen 1.84 0.21 1.58 2.19
C op and animal p oduc ion, hun ing and ela ed se ice ac i i ies 1.78 0.06 1.65 1.83
A chi ec u al and enginee ing ac i i ies; echnical es ing and analysis 1.71 0.15 1.51 1.97
O he se ice ac i i ies 1.66 0.07 1.52 1.80
Manu ac u e o ex iles, wea ing appa el and lea he p oduc s 1.63 0.05 1.55 1.71
Manu ac u e o mo o ehicles, aile s and semi- aile s 1.61 0.09 1.49 1.75
Manu ac u e o ood p oduc s, be e ages and obacco p oduc s 1.54 0.06 1.42 1.61
Manu ac u e o basic pha maceu ical p oduc s and pha maceu ical p epa a ions 1.51 0.03 1.47 1.56
Ai anspo 1.51 0.14 1.26 1.71
Compu e p og amming, consul ancy and ela ed ac i i ies; in o ma ion se ice ac i i ies 1.46 0.09 1.29 1.59
Cons uc ion 1.34 0.04 1.28 1.44
Wa e anspo 1.22 0.05 1.17 1.37
Ac oss all indus ies 2.05 0.54 1.17 4.08
Sou ce: Own calcula ion based An `as and Cho (2018) and WIOD da a.
No e: Only adable indus ies a e conside ed.
anspo ) and 3.80 (mining and qua ying). Ac oss all indus ies, he a e age ups eamness is 2.05.
The maximum alue o ups eamness (4.08) is measu ed in he mining and qua ying indus y in 2008
and he minimum (1.17) is measu ed in he wa e anspo indus y in 2012. Looking a manu ac u ing
indus ies only, he mos downs eam manu ac u ing indus y is manu ac u e o basic pha maceu ical
p oduc s and pha maceu ical p epa a ions (1.51), while manu ac u e o chemicals and chemical p oduc s
(3.33) is he mos ups eam one.
Nex , we combine he indus y-le el ups eamness measu e wi h p oduc -le el sales in o ma ion o
ob ain a measu e o ups eamness a he ﬁ m-le el. Fo each ﬁ m-yea obse a ion, we use he p oduc -
le el sales sha es as weigh s o calcula e he weigh ed a e age o ups eamness o he ﬁ ms’ p oduc
po olio as illus a ed in equa ion (5.26), so ha
U =
V
�
=1
sales
sales
U ,(5.26)
36We base he classiﬁca ion o indus ies as non- adable on he expo pa e n e ealed in WIOD. To a oid
ha an indus y wi h negligibly small expo s is deﬁned as adable in one yea and as non- adable in he nex
yea , we deﬁne a h eshold, abo e which an indus y is classiﬁed as adable. The h eshold is deﬁned as he
lowes amoun ha a oids any swi ching be ween being classiﬁed as adable o non- adable du ing ou co e ed
pe iod. The exclusion o non- adable indus ies aﬀec s 20 indus ies in WIOD, e.g., public adminis a ion and
de ence; compulso y social secu i y,educa ion and accommoda ion and ood se ice ac i i ies.
97
whe e U is he le el o ups eamness o ﬁ m a ime selling p oduc s om Vdiﬀe en indus ies.37
The measu e accoun s o he p oduc mix by weigh ing a p oduc ’s GVC posi ion U wi h i s sha e o
o al sales in yea .38
Ou measu e o a ﬁ m’s GVC posi ion U has h ee sou ces o a ia ion. Fi s , he measu e is
aﬀec ed by changes in a ﬁ m’s p oduc po olio, i.e., i a ﬁ m d ops o adds new p oduc s. Second, o
a gi en p oduc mix, he ela i e impo ance o he diﬀe en p oduc s may change. Thi d, he GVC
posi ion o he diﬀe en p oduc s in a ﬁ m’s po olio changes o e ime (e.g., because some o he ﬁ m’s
p oduc s a e inc easingly used in mo e downs eam indus ies). Figu e 5.1 illus a es hese sou ces o
a ia ion o a ﬁ m in he elec ical equipmen indus y. This ﬁ m se es as an example o an a e age
ﬁ m in ou sample in e ms o CSR spending, ups eamness and numbe o p oduc s. I has se e al
Figu e 5.1: A ﬁ m’s GVC posi ion change o e ime
1
1.5
2
2.5
3
3.5
4
Indus y Ups emaness
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Yea
p oduc sales i m-speci ic ups eamness
No es: A e age i m in e ms o ups eamness, numbe o p oduc s and CSR spending
was chosen o illus a i e pu poses. Fi m om elec ical equipmen indus y.
Example: A i m's change in GVC posi ion o e ime
p oduc s ha belong o indus ies ha diﬀe in e ms o he ups eamness posi ion. O e ime, he
ﬁ m’s p oduc po olio changes as a new p oduc is added and one is d opped and he impo ance o
p oduc s measu ed by hei sales sha e changes as well. The ﬁ m’s ups eamness posi ion (black dashed
line) a ies o e ime in line wi h he changes in he p oduc po olio.
5.3.3 Empi ical Speciﬁca ion
We s a he discussion o he empi ical me hodology wi h a desc ip ion o he main empi ical se up,
including he p esen a ion o o he independen a iables, which may d i e CSR decisions o ﬁ ms. A e
discussing desc ip i e s a is ics, we p esen he empi ical esul s.
37Ou app oach is compa able o he app oach by Cho e al. (2021), who also combine an indus y GVC
measu e wi h a ﬁ m’s p oduc in o ma ion. They use ﬁ ms’ expo s (impo s) in an indus y in o al expo s
(impo s) as weigh s o de i e a ﬁ m-speciﬁc GVC posi ion o Chinese ﬁ ms. Simila o ou de i ed GVC
measu e, he espec i e GVC measu e a ies o e ime wi hin a ﬁ m.
38Fo some ﬁ ms, he e is no p oduc -le el in o ma ion a ailable. In hese cases, we ake he ups eamness o
he main indus y o he ﬁ m as he measu e o he GVC posi ion. This aﬀec s 8.49% o he obse a ions, which
en e ou empi ical es ima ion.
98
5.3.3.1 Empi ical Se -Up
P oposi ion 1 p edic s ha CSR expendi u es inc ease along he alue chain om ups eam o
downs eam s ages. To empi ically es his p edic ion, we es ima e he ollowing wo equa ions.
Y =α0+β1U −1+β2X −1+γ +δs+� ,(5.27)
Y =α0+β1U −1+β2X −1+ζ +τ +� ,(5.28)
whe e Y measu es CSR spending o ﬁ m ac i e in main indus y in yea . In ou main spec-
iﬁca ions, we apply wo diﬀe en es ima o s, OLS and Poisson Pseudo-Maximum Likelihood (PPML).
To apply OLS, we ake he na u al loga i hm o a ﬁ m’s CSR spending (log(CSR spending )). This
emo es ﬁ ms wi h ze o CSR spending om ou sample. Ou al e na i e speciﬁca ion applying PPML
wi h CSR spending as he dependen a iable allows us o keep ze os in he da a.39 Ou main ex-
plana o y a iable o in e es is a ﬁ m’s GVC posi ion a ime −1, U −1. In all eg essions, we expec
β1<0, which indica es lowe CSR spending o mo e ups eam ﬁ ms.
In ou choice o ﬁxed eﬀec s, we use wo diﬀe en app oaches o analyze he ela ion be ween CSR
spending and ups eamness. Fi s , equa ion (5.27) is a epea ed c oss-sec ion in which we include s a e
δsand indus y-yea γ ﬁxed eﬀec s. We include s a e ﬁxed eﬀec s δs o cap u e diﬀe ences be ween
Indian s a es which may aﬀec he le el o CSR spending in all ﬁ ms alike, o example, diﬀe ences in
labo egula ion. The indus y-yea ﬁxed eﬀec s γ cap u e shocks common o all ﬁ ms in a pa icula
indus y and yea . Some indus ies may be unde igh e public sc u iny in gene al (e.g., obacco), o
he e migh be indus ial disas e s ha aﬀec ce ain indus ies in speciﬁc yea s. Acco dingly, equa ion
(5.27) exploi s a ia ion be ween ﬁ ms wi hin he same s a e, main indus y, and yea .
Second, equa ion (5.28) exploi s ime-se ies a ia ion wi hin ﬁ ms. We only include ﬁ m ﬁxed eﬀec s
ζ and yea ﬁxed eﬀec s τ .ζ cap u es unobse able ﬁ m cha ac e is ics, which a e ime-in a ian . τ
cap u es yea -speciﬁc shocks o CSR spending common o all ﬁ ms. In his speciﬁca ion, we hus exploi
wi hin-ﬁ m a ia ion o e ime. In ou da a, we obse e a ﬁ m 6.8 yea s on a e age.
X −1includes ﬁ m-le el con ol a iables o cap u e al e na i e d i e s o CSR spending. Simila
o U −1, he con ol a iables en e wi h a one-yea ime lag. We nex p esen hese con ol a iables
in mo e de ail.
The a iable log(Wages) is a p oxy o ﬁ m-le el wage paymen s o con ol o he possibili y ha
wages a y sys ema ically wi h he alue chain posi ion. One could conside s aﬀwel a e spending a
pa o he o e all compensa ion package o employees. I downs eam ﬁ ms oﬀe be e compensa ion
packages (e.g., due o highe skill in ensi y), a nega i e ela ionship be ween ups eamness and CSR
spending (pa ly measu ed by s aﬀwel a e spending) could hen be in e p e ed as ac ually eﬂec ing he
nega i e ela ionship be ween ups eamness and employee compensa ion. We con ol o his by aking
he hou ly compensa ion paid in Indian indus ies om WIOD and using he p oduc -le el in o ma ion
o ﬁ ms o gene a e a ﬁ m-le el con ol a iable ha is simila in spi i o ou ﬁ m-le el ups eamness
measu e.40 In pa icula , we calcula e wages as an indus y’s a e age hou ly compensa ion o employees
39PPML is widely used in he li e a u e o in e na ional economics and does no equi e aking he na u al
loga i hm o he dependen a iable. Thus, using PPML also allows us o explo e wi hin ﬁ m a ia ion o e ime
o hose ﬁ ms ha s a spending on CSR du ing ou analyzed pe iod. Mo eo e , i ou pe o ms i s al e na i es
in he p esence o many ze os and he e oskedas ici y (Sil a and Ten ey o, 2006).
40The wage bill is a ailable o a subse o ﬁ ms in P owess. Howe e , since he e a e many ze os and because
he p o ision o he numbe o employees is no compulso y o ﬁ ms, we decide no o include he wage bill and
p e e o use ou combined measu e.
99
o ganiza ion o p oduc ion o he analysis o CSR and e hical p oduc ion in global alue chains in ol ing
de eloping and eme ging economies. In his con ex , u u e esea ch should u he in es iga e he
ele ance o incomple e con ac s o CSR in global p oduc ion. To his end, new heo ies could be
de eloped o nea ly link exis ing o no el empi ical measu es o con ac ual incomple eness o ﬁ m’s
CSR in es men s.
106

5.5 Theo y Appendix
5.5.1 Indus y Equilib ium wi h Incomple e Con ac s
We use equa ion (5.19) o calcula e he alues o q(ω) and s(m) in indus y equilib ium om equa ions
(5.17) and (5.14).
q(ω) = ρ
cx
1−2ρ(5.29)
s(m) = 1−α
β
(1 −β)ρ
cs
1−2ρmρ−α
α(1−ρ)(5.30)
Using he deﬁni ion o ε(ω) we hen ge
ε(ω) = ��1
0
s(m)αdm�
1
α
=1−α
β
(1 −β)ρ
cs
1−2ρ�1−ρ
1−α�1
α
.(5.31)
Plugging his, he solu ion o q(ω), as well as he exp ession o Pin o in e se demand gi es he p icing
ule as
p(ω)=A1−ρq(ω)−(1−ρ)ε(ω)ρ=cx
ρ
1−ρ
β.(5.32)
Using ε(ω) and p(ω) in he deﬁni ion o he p ice index om demand, P−ρ
1−ρ=�n
0�ε(ω)
p(ω)�ρ
1−ρdω,
gi es equa ion (5.20), a second exp ession o he p ice index as a unc ion o he numbe o ﬁ ms, n.
Combining he wo exp essions gi es he equilib ium numbe o ﬁ ms om equa ion (5.21),
n=βE
1−ρ
1−2ρ
.(5.33)
5.5.2 Comple e Con ac s
In his sec ion, we sol e he model o he case o comple e con ac s. Unde comple e con ac s,
he headqua e can oﬀe ully speciﬁed con ac s o each supplie jand will app oach hem in he
igh o de . The headqua e oﬀe s a con ac consis ing o a physical quan i y x(j) and an acco ding
paymen w(j), as well as an implemen ed p oduc ion s anda d s(j) wi h he acco ding paymen (j). To
make compa ison wi h he baseline model easible, we deno e all endogenous a iables in he comple e
con ac s case wi h a ilde. The headqua e hen maximizes i s p oﬁ s wi h espec o he ou choice
a iables abo e, subjec o wo condi ions:
max
{˜x(j),w(j),˜s(j), (j)}j∈[0,1]
˜πF(ω) = ˜q(ω)ρ˜
A1−ρ��1
0
˜s(j)αdj�
ρ
α
−�1
0
w(j)dj −�1
0
(j)dj −
s. . w(j)−cx˜x(j)≥0
(j)−cs˜s(j)≥0
Due o he la ge numbe o a ailable supplie s, he headqua e can lowe he paymen s w(j) and (j)
so ha he cons ain s a e binding and he supplie walks away wi h ze o p oﬁ s. Because ˜x(j) = ˜q(ω),
we ha e
�1
0
w(j)dj =cx�1
0
˜x(j)dj =cx˜q(ω).(5.34)
and because o he binding cons ain s, �1
0 (j)dj =cs�1
0˜s(j)dj.
107
This implies ha we can es a e he headqua e ’s p oblem in e ms o ˜q(ω) and ˜s(j) as
max
˜q(ω),{˜s(j)}j∈[0,1]
˜πF(ω) = ˜q(ω)ρ˜
A1−ρ��1
0
˜s(j)αdj�
ρ
α
−cx˜q(ω)−cs�1
0
˜s(j)dj − (5.35)
Diﬀe en ia ing wi h espec o he choice a iables gi es he ﬁ s o de condi ions
˜q(ω) = ρ˜ (ω)
cx
˜s(j)1−α=˜ (ω)ρ
cs��1
0
˜s(j)αdj�−1
Exp essing ��1
0. . . dj�in e ms o e enue ˜ (ω) gi es ˜s(j) condi ional on a le el o ou pu ˜q(ω) as
˜s(j) = ˜s= ˜ (ω)ρ−α
ρ(1−α)˜
Aα(1−ρ)
ρ(1−α)˜q(ω)α
1−α�ρ
cs�1
1−α
∀j
The diﬀe ence o equa ion (5.11) is ha he e s age js anda d depends on o al e enue ˜ (ω) ins ead o
e enue up o ha s age (j).
Plugging his back in o he e enue exp ession om he maximiza ion p oblem gi es equilib ium
e enue ˜ (ω) as a unc ion o ˜q(ω) as
˜ (ω) = ˜
A�˜q(ω)ρ
cs�ρ
1−ρ
(5.36)
Plugging he esul o ˜ (ω) back in o ˜s om abo e gi es
˜s=˜
A˜q(ω)ρ
1−ρ�ρ
cs�1
1−ρ
∀j(5.37)
This gi es a symme ic equilib ium in es men in he s anda d ˜sa each s age jas a unc ion o exogenous
pa ame e s and condi ional on he scale o p oduc ion ˜q(ω).
We know om he ﬁ s FOC ha
˜q(ω) = ρ˜ (ω)
cx
and using he esul o ˜ (ω) om equa ion (5.36) gi es ha he op imal scale o p oduc ion is gi en
by
˜q(ω) = 
ρ�˜
A
cx�1−ρ
c−ρ
s

1
1−2ρ
(5.38)
Combining he esul s o equa ions (5.37) and (5.38) gi es equilib ium CSR in es men a each s age
108
jas
˜s=
ρ�˜
A
cs�1−ρ
c−ρ
x

1
1−2ρ
∀j(5.39)
Using he ac ha wi h ee en y, p oﬁ s a e ze o o all (homogeneous) ﬁ ms, and applying his o
equa ion (5.35) gi es he ﬁ s exp ession o he equilib ium p ice index as
˜
P=ρ−2cscx�
1−2ρ�1−2ρ
ρ
E−1−ρ
ρ(5.40)
Plugging his in o equa ion (5.39) o ˜
A=E˜
Pρ
1−ρgi es
˜s=ρ
cs
1−2ρ.(5.41)
5.5.2.1 Le el o CSR wi h Comple e Con ac s and Compa ison o Incomple e
Con ac s
Unde comple e con ac s, he op imal CSR in es men is uni o m ac oss all p oduc ion s ages jand
is gi en by ˜s=ρ
cs
1−2ρ. The uni o mi y esul a ises because all s ages in he p oduc ion p ocess en e
symme ically in o he p oduc ion o e hical quali y. The e o e, a headqua e able o w i e comple e
con ac s has no incen i e o assign diﬀe en le els o in es men s o supplie s a diﬀe en s ages along
he alue chain.
I is ins uc i e o compa e his esul wi h ou main esul om equa ion (5.24). In ac , i is
possible o exp ess s(m) as a unc ion o he comple e con ac s esul ˜sas
s(m) = (1 −α)1−β
βmρ−α
α(1−ρ)˜s. (5.42)
Unde incomple e con ac s, CSR le els a e smalle han he CSR in es men s unde comple e con ac s
o all - e en he mos downs eam - s ages o p oduc ion. To see his, conside he ollowing: Unde
ou assump ion ha α<ρand because s(m) is inc easing in m, showing ha s(1) <˜swill be suﬃcien .
This holds when β>1−α
2−α. Because 1/2>1−α
2−α, his condi ion is always sa isﬁed unde ou main ained
assump ion o β≥1/2.44 I is appa en ha he e a e wo easons o unde in es men in he incomple e
con ac s case ela i e o comple e con ac s. Fi s , he e is a e m ha applies equally o all s ages,
which esul s om indi idual p oﬁ maximiza ion o supplie s along he alue chain. Second, mdeno es
he posi ion in he alue chain. So while he e is unde in es men a each s age, he unde in es men is
mo e se e e o smalle m, i.e., u he ups eam.
5.5.2.2 Remaining Indus y Equilib ium Resul s
Fo comple eness, we also epo he emaining a iables o he indus y equilib ium wi h comple e
con ac s. Using ˜
A=E˜
Pρ
1−ρand equa ion (5.40) in equa ion (5.38), gi es he p oduc ion quan i y in
indus y equilib ium as
˜q(ω) = ρ
cx
1−2ρ.
44Recall ha β≥1/2 di ec ly s ems om An `as (2003) and 0 <α<ρis discussed in Sec ion 5.2.2.1.
109
Using his esul in in e se demand
˜p(ω) = ˜
A1−ρ˜ε(ω)ρ˜q(ω)ρ−1
as well as he deﬁni ion o ˜ε
˜ε(ω) = ��1
0
˜s(j)αdj�
1
α
= ˜s
gi es he op imal p icing ule as
˜p(ω) = cx
ρ.(5.43)
Wi h he deﬁni ion o he op imal p ice index
˜
P=��˜n
0�˜p(ω)
˜ε(ω)�−ρ
1−ρ
dω�−1−ρ
ρ
we ge a second exp ession o he p ice index,
˜
P=cscx
ρ2
1−2ρ
˜n−1−ρ
ρ,(5.44)
which, combined wi h (5.40) gi es he equilib ium numbe o ﬁ ms as
˜n= (1 −2ρ)E
(5.45)
5.5.3 The Ex ended Model
In his sec ion, we show ha ou main esul con inues o hold in an ex ended e sion o he model,
in which we in oduce an addi ional pa ame e ν∈(0,1) ha cap u es he scope o e hical quali y
diﬀe en ia ion. The ex ension gene alizes he model ou lined in Sec ion 5.2 and collapses o he baseline
e sion o ν= 1. This ex ension allows us o elax he assump ion o ρ<1/2. In pa icula , o
a bi a ily small ν, he maximum possible alue ρapp oaches one. See Foo no e 15 o a discussion.
Fo exposi ional simplici y, we main ain he same no a ion as in he main ex . I is well unde s ood
ha he exp essions in his sec ion and in he main ex a e iden ical o ν= 1 only.
In he model se up, he only change is in he u ili y unc ion, which now eads as
U=��ω∈Ω
[ε(ω)νq(ω)]ρdω�1
ρ
.(5.46)
U ili y maximiza ion subjec o he budge cons ain gi es in e se demand o a ie y ωas
p(ω)=ε(ω)νρq(ω)−(1−ρ)A1−ρ,(5.47)
whe e A=EP ρ
1−ρand P−ρ
1−ρ=�ω∈Ω�ε(ω)ν
p(ω)�ρ
1−ρdω. Using he deﬁni ion o ε(ω) gi en in equa ion
(5.3), ﬁ m e enue can be w i en as
(ω) = ��1
0
s(j)αdj�
νρ
α
q(ω)ρA1−ρ(5.48)
110
5.5.3.1 Incomple e Con ac s
Wi h incomple e con ac s, in e se demand and ﬁ m e enues a e s ill gi en by equa ions (5.47) and
(5.48).
Re enue secu ed up o s age mis gi en by
(m) = q(ω)ρA1−ρ��m
0
s(j)αdj�νρ
α
(5.49)
and he inc emen al con ibu ion o s age m o o e all e enue hen ollows as
∂ (m)
∂m= �(m) = νρ
αs(m)α (m)νρ−α
νρ q(ω)α
νA(1−ρ)α
νρ .(5.50)
As in he baseline model, each supplie maximizes i s p oﬁ s, composed o he sha e 1 −βo he
inc emen al con ibu ion o ﬁnal e enue ne o expendi u e o he p oduc ion s anda d,
max
s(m)πs(m) = (1 −β) �(m)−css(m).(5.51)
Diﬀe en ia ing hese p oﬁ s wi h espec o s(m) gi es he analog o equa ion (5.11) om he baseline
model as
s(m) = �(1 −β)νρ
cs
q(ω)α
νA(1−ρ)α
νρ �1
1−α
(m)νρ−α
νρ(1−α).(5.52)
Plugging his back in o equa ion (5.50) yields a diﬀe en ial equa ion analogous o equa ion (5.12) in he
baseline model. The solu ion gi es an exp ession o e enue secu ed up o s age mas a unc ion o
pa ame e s, A, and q(ω) analogous o equa ion (5.13) as
(m) = �1−νρ
1−α�νρ(1−α)
α(1−νρ)�1−β
cs
νρ�νρ
1−νρ
A1−ρ
1−νρ q(ω)ρ
1−νρ mνρ(1−α)
α(1−νρ).(5.53)
Plugging his back in o equa ion (5.52) gi es he p oduc ion s anda d implemen ed a s age mas a
unc ion o pa ame e s, q(ω), and Aas
s(m) = �1−β
cs
νρ�1
1−νρ �1−νρ
1−α�νρ−α
α(1−νρ)
q(ω)ρ
1−νρ A1−ρ
1−νρ mνρ−α
α(1−νρ),(5.54)
analogous o equa ion (5.14) in he baseline model.
To choose he op imal scale o p oduc ion he ﬁ m has o sol e
max
q(ω)πFω=β (ω)−cxq(ω)− . (5.55)
Using equa ion (5.53) e alua ed a m= 1, he solu ion o he maximiza ion p oblem yields
q(ω) = �1−νρ
1−α�νρ(1−α)
α[1−ρ(1+ν)] �1−β
cs
νρ�νρ
1−ρ(1+ν)�A1−ρ�β
cx
ρ
1−νρ�1−νρ�1
1−ρ(1+ν)
,(5.56)
analogous o equa ion (5.17).
Tu ning o he indus y equilib ium, using equa ion (5.53) e alua ed a m= 1 as well as equa ion
(5.56) in equa ion (5.55) and se ing i o ze o gi es a ﬁ s exp ession o he p ice index, analogous o
111

equa ion (5.19) as
P=�
1−ρ(1+ν)�1−ρ(1+ν)
ρcx
ρ�cs
(1 −β)νρ�ν
E−1−ρ
ρβ−1−νρ
ρ(1 −α)1−νρ
ρ�1−α
1−νρ�νρ−α
αρ
.(5.57)
We can use his esul o calcula e he alues o q(ω) and s(m) in indus y equilib ium using equa ions
(5.56) and (5.54).
q(ω) = ρ
cx
1−ρ(1+ν)(5.58)
s(m) = 1−α
β
(1 −β)νρ
cs
1−ρ(1+ν)mνρ−α
α(1−νρ)(5.59)
The abo e equilib ium exp ession o q(ω) shows ha q(ω) emains posi i e o a la ge ange o
alues o ρ. Mo e p ecisely, while in he case o ν= 1, we had o impose ρ<1/2, we can now impose a
weake condi ion, ρ<1/(1 + ν). As νg ows a bi a ily close o ze o, he uppe bound o he admissible
ange o ρapp oaches uni y. The equilib ium exp ession o s(m) shows ha ou main esul con inues
o hold also in his ex ended e sion. Fo ou model o p edic an upwa d sloping p oﬁle o CSR along
he alue chain, we need o assume ha α<νρ.
The emaining equilib ium a iables can be calcula ed as ollows. Using he deﬁni ion o ε(ω) we
hen ge
ε(ω) = ��1
0
s(m)αdj�
1
α
=1−α
β
(1 −β)νρ
cs
1−ρ(1+ν)�1−νρ
1−α�1
α
.(5.60)
Plugging his, he solu ion o q(ω) as well as he exp ession o Pin o in e se demand gi es he p icing
ule as
p(ω)=A1−ρq(ω)−(1−ρ)ενρ =cx
ρ
1−νρ
β(5.61)
Using ε(ω) and p(ω) in he deﬁni ion o he p ice index om demand, P−ρ
1−ρ=�n
0�ε(ω)ν
p(ω)�ρ
1−ρdω, gi es
a second exp ession o he p ice index as a unc ion o he numbe o ﬁ ms, n. Combining he wo
exp essions gi es he equilib ium numbe o ﬁ ms as
n=βE
1−νρ
1−ρ(1+ν)
.(5.62)
5.5.3.2 Comple e Con ac s
In he case o comple e con ac s, we can s a e he ﬁ m’s p oblem analogous o equa ion (5.35) as
max
˜q(ω),{˜s(j)}j∈[0,1]
˜πF(ω) = ˜q(ω)ρ˜
A1−ρ��1
0
˜s(j)αdj�
νρ
α
−cx˜q(ω)−cs�1
0
˜s(j)dj − , (5.63)
wi h he only diﬀe ence being he pa ame e νa ached o e hical quali y. Following s eps analogous
o hose ou lined in Appendix 5.5.2 gi es he ollowing esul s in indus y equilib ium. The p oduc ion
s anda d ˜s(j) = ˜s∀jis gi en by
˜s=νρ
cs
1−ρ(1+ν).(5.64)
112
Because ˜sis iden ical ac oss s ages, i also holds ha ˜ε(ω) = ˜s. The op imal quan i y o ﬁnal good
p oduc ion is gi en by
˜q(ω) = ρ
cx
1−ρ(1+ν),(5.65)
while he op imal p ice is gi en by
˜p(ω) = cx
ρ(5.66)
and he equilib ium numbe o ﬁ ms is gi en by
˜n= [1 −ρ(1+ν)] E
.(5.67)
5.6 Empi ical Appendix
This sec ion p o ides de ails on wo con ol a iables, namely, he measu e o wages and he measu e
o expo exposu e o OECD coun ies.
5.6.1 Wages:
We con ol o he possibili y ha wages a y sys ema ically wi h he alue chain posi ion. In
P owess, he wage bill is a ailable o a subse o ﬁ ms. Howe e , he e a e many ze os and he p o ision
o he numbe o employees is no compulso y o ﬁ ms so ha we canno compu e wages pe employee.
We he e o e decide no o include he wage bill om he ﬁ m-le el da a and p e e o use a p oxy o
hou ly wages aken om WIOD. We ake he hou ly compensa ion paid in Indian indus ies om WIOD
and use he p oduc -le el in o ma ion o ﬁ ms o gene a e a ﬁ m-le el con ol a iable ha is simila in
spi i o ou ﬁ m-le el ups eamness measu e. To his end, we calcula e Wages as an indus y’s a e age
hou ly compensa ion o employees weigh ed by a ﬁ m’s p oduc sales in he espec i e indus y and yea .
This deli e s ou con ol a iable o in e es , which is compu ed in he ollowing way:
Wages =
V
�
=1
sales
sales
compensa ion .(5.68)
We hus compu e Wages aking indus y ’s a e age hou ly compensa ion o employees in yea
and weigh ing i wi h a ﬁ m’s p oduc sales in he espec i e indus y. Spli ing he analyzed sample
by he median alue chain posi ion shows a median wage le el o 24.77 Rupees o he mo e ups eam
ﬁ ms and o 21.65 Rupees o he mo e downs eam ﬁ ms. Howe e , his does no con ol o o he ﬁ m
cha ac e is ics.
5.6.2 OECD expo exposu e:
The P owess da a do no co e in o ma ion on he expo des ina ions o ﬁ ms. In o de o con ol
o he e ogenei y in des ina ion ma ke s, we cons uc a p oxy o exposu e o OECD ma ke s based
on in o ma ion aken om WIOD. This measu e is cons uc ed simila o ou ups eamness and wage
113
measu es based on WIOD indus y da a as
OECD expo exposu e =
V
�
=1
sales
sales
expo s OECD
expo s o al
,(5.69)
whe e we ake an indus y ’s expo s o OECD coun ies ela i e o indus y ’s o al expo s in yea
and weigh his acco ding o a ﬁ m’s p oduc po olio.
114
Bibliog aphy
Aghion, P., R. B´
enabou, R. Ma in, and A. Roule (2020): “En i onmen al P e e ences and
Technological Choices: Is Ma ke Compe i ion Clean o Di y?” NBER Wo king Pape 26921.
Aldashe , G., M. Lima di, and T. Ve die (2015): “Wa chdogs o he In isible Hand: NGO
Moni o ing and Indus y Equilib ium,” Jou nal o De elopmen Economics, 116, 28–42.
Aldashe , G. and T. Ve die (2009): “When NGOs Go Global: Compe i ion on In e na ional
Ma ke s o De elopmen Dona ions,” Jou nal o In e na ional Economics, 79, 198–210.
——— (2010): “Goodwill Bazaa : NGO Compe i ion and Gi ing o De elopmen ,” Jou nal o De el-
opmen Economics, 91, 48–63.
Al a o, L., P. An `
as, D. Cho , and P. Conconi (2019): “In e nalizing Global Value Chains: A
Fi m-Le el Analysis,” Jou nal o Poli ical Economy, 127, 509–559.
Al a o-U e˜
na, A., B. Fabe , C. Gaube , I. Manelici, and J. P. Vasquez (2021): “Responsible
Sou cing? Theo y and E idence om Cos a Rica,” mimeo, UC Be keley.
An `
as, P. (2003): “Fi ms, Con ac s, and T ade S uc u e,” Qua e ly Jou nal o Economics, 118,
1375–1418.
An `
as, P. and D. Cho (2013): “O ganizing he Global Value Chain,” Econome ica, 81, 2127–2204.
——— (2018): “On he Measu emen o Ups eamness and Downs eamness in Global Value Chains,”
in Wo ld T ade E olu ion: G ow h, P oduc i i y and Employmen , Taylo & F ancis, 126–194.
An `
as, P., D. Cho , T. Fally, and R. Hillbe y (2012): “Measu ing he Ups eamness o
P oduc ion and T ade Flows,” Ame ican Economic Re iew, 102, 412–16.
An `
as, P. and E. Helpman (2004): “Global Sou cing,” Jou nal o Poli ical Economy, 112, 552–580.
An `
as, P. and S. R. Yeaple (2014): “Mul ina ional Fi ms and he S uc u e o In e na ional T ade,”
in Handbook o In e na ional Economics, ed. by G. Gopina h, E. Helpman, and K. Rogoﬀ, Else ie ,
ol. 4, 55–130.
An `
as, P. (2016): Global P oduc ion: Fi ms, Con ac s, and T ade S uc u e, P ince on Uni e si y
P ess.
Baldwin, R. and J. Ha igan (2011): “Ze os, Quali y, and Space: T ade Theo y and T ade E i-
dence,” Ame ican Economic Jou nal: Mic oeconomics, 3, 60–88.
Ba on, D. P. (2012): “The Indus ial O ganiza ion o P i a e Poli ics,” Qua e ly Jou nal o Poli ical
Science, 7, 135–174.
115
Einhal ung de DFG-Rich linien
Ti el des Au sa zes: S epping up o he ma k? Fi ms’ expo ac i i y and en i onmen al
inno a ion in 14 Eu opean coun ies
E scheinungso : Hanley, A.; Sem au, F.O. (2022). “S epping up o he ma k? Fi ms'
expo ac i i y and en i onmen al inno a ion in 14 Eu opean
coun ies." Indus y and Inno a ion, 29(5), 672-700.
Au o en und Au o innen: Aoi e Hanley
Finn Ole Sem au
Hie mi bes ä igen die Au o en und Au o innen, die Einhal ung de DFG-Rich linien.
Alle Au o en und Au o innen haben einen wesen lichen Bei ag zu de wissenscha lichen
Ve ö en lichung geleis e . Alle Au o en und Au o innen haben zu allen Kapi eln einen wesen lichen
An eil geleis e .
O , Da um, Un e sch i :
P o . Aoi e Hanley
O , Da um, Un e sch i :
Finn Ole Sem au
Kiel, 23.8.2022
Kiel, 30.8.2022

Einhal ungde DFG-Rich linien

Ti eldesAu sa zes:  Co po a eSocialResponsibili yalong heGlobalValueChain
E scheinungso : He kenho ,P.;K au heim,S.;Sem au,F.O.andS eglich,F.(2021).
“Co po a eSocialResponsibili yAlong heGlobalValueChain.”
CESi oWo kingPape No.9498.
Au o enundAu o innen: PhilippHe kenho 
Sebas ianK au heim
FinnOleSem au
F aukeS eglich
Hie mi bes ä igendieAu o enundAu o innen,dieEinhal ungde DFGRich linien.
AlleAu o enundAu o innenhabeneinenwesen lichenBei agzude wissenscha lichen
Ve ö en lichunggeleis e .ImWesen lichenhabenPhilippHe kenho undSebas ianK au heimam
Kapi el „A Model o CSR along he Global Value Chain“ und Finn Ole Sem au und F auke S eglich am
Kapi el„GVC Posi ion and CSR in India“ gea bei e . Alle Au o en und Au o innen haben zuden
Kapi eln„In oduc ion“ und „Conclusion“ beige agen.

O ,Da um,Un e sch i :
D .PhilippHe kenho 
O ,Da um,Un e sch i :
P o .Sebas ianK au heim
O ,Da um,Un e sch i :
FinnOleSem au
O ,Da um,Un e sch i :
F aukeS eglich
Hambu g, 8.8.2022
Kiel, 8.8.2022
F ank u , den 12.08.2022
Mainz, 22.08.22
Acknowledgemen
And I would walk 500 miles. And I would oll 500 mo e. The ly ics o he P oclaime s song cap u e
many o my eelings du ing he ime w i ing his disse a ion. The key eeling migh be he happiness o
he song which eels somehow nai e gi en ha his is a ound walking dis ance be ween he Kiel Ins i u e
and F eibu g. Diﬃcul o imagine ha you can keep all you posi i i y i you know ha you a e walking
such a dis ance. Especially, once you a i e in F eibu g, you ealise ha you ha e o oll a ound 805
km mo e. Looking a he map, he way back is e en wo se, you ha e o oll up. Howe e , I hink I can
say ha I managed somehow o keep he mo i a ion and a posi i e a i ude o e he comple e dis ance.
Thinking back on he las yea s, wi hou a doub , has no been possible wi hou he suppo o g ea
iends, my colleagues (some e en go bes iends) and amily. Wi h his acknowledgemen , I y o
name all o hem, howe e , I am su e he lis will be incomple e and I am e y so y o ha .
Le me ﬁ s go back in ime be o e I s a ed wo king as a doc o al s uden a he Kiel Ins i u e in
Oc obe 2016. The yea is 2009 and I am awa e ha deciding wha o s udy is a c oss oads. In he
end, I ha e wo opposing op ions: Oceanog aphy in Hambu g o Economics in Kiel. E en hough I lo e
he sea and i sounds coun e -in ui i e, I made he igh decision. Du ing his ime, I was obsessed wi h
unde s anding he mechanisms behind economics, which aﬀec all o ou li es and led o he ﬁnancial
c isis jus las yea . Compa ed o he g ea ﬁeld ips you can go on while s udying oceanog aphy, his
sounds lame e en o me. Bu he basic conﬁdence ha I could do some good by unde s anding how he
economy wo ks mo i a ed me. I had no idea wha ha mean .
The eye-opening momen was oge he wi h Claas Schneide heinze a ound 2011. We bo h ead he
book ”The Bo om Billion” by Jeﬀ ey Sachs, which was pa o a g ea discussion wi h William Eas e ly
on how o alle ia e global po e y. F om ha poin on, we became imme sed in he de elopmen
economics li e a u e. Now Claas is a colleague a he Kiel Ins i u e, I was his bes man a his wedding o
Ma ion and he was my bes man a my wedding o Nicola. When Raine Thiele ad e ised a posi ion a
he Kiel Ins i u e in 2012 in his lec u e on de elopmen economics, I applied di ec ly. The job in e iew
was wa m and in i ing. This posi i e a mosphe e con inued o e he nex ew yea s, du ing which
I wo ked wi h Ma cus B¨ohme, Tobias Heidland, Toman Ba sbai, Linda Kleemann, Kacana Khadja i,
Es he Ademme , F ed Go e zko and Ch is iane Geb¨uh . They all helped me o ake my ﬁ s s eps as
an economis . Raine supe ised my Mas e ’s hesis and an upda ed e sion o one chap e o i was
e en my ﬁ s publica ion in a pee - e iewed jou nal. This suppo is al eady a om no mal, bu in
addi ion, he enabled me o spend h ee mon hs a he UFRJ in Rio de Janei o in 2015.
A he end o my Mas e ’s deg ee, I wo ked as an in e n a he GFA Consul ing G oup and lea ned how
o apply my (quan i a i e) knowledge as a p ac i ione in de elopmen coope a ion. Linda and Tobias
asked me o pa icipa e in he scien iﬁc moni o ing o household su eys in en de eloping coun ies. A
challenging ask and no usual in Ge man de elopmen coope a ion a ha ime and on ha scale. I
lea ned a lo om he eam and e en go he oppo uni y o apply wha I had lea ned o an upcoming
da a collec ion in Wes e n Kenya. I will ne e o ge he ﬁ s momen on a boda boda and he iew o
Lake Vic o ia. I is deﬁni i e p oo ha wo king as an economis is no necessa ily lame and he job I
wan o do.
When a posi ion was ad e ised a he Kiel Cen e o Globaliza ion dealing wi h he in e ace o
124
global alue chains and ecological and social consequences, i was immedia ely clea o me ha I wan ed
o e u n o he Kiel Ins i u e. I am g a e ul o Sonja Pe e son, who w o e me a le e o e e ence and
suppo ed my applica ion. I is ou s anding ha he e we e only women in my job in e iew, including
my wo supe iso s Aoi e Hanley and Ka in Rehdanz. Again, I el com o able om he ﬁ s momen ,
e en hough he eam was la gely new o me. A his s age, i paid oﬀ ha I besides de elopmen
economics also specialised in en i onmen al economics du ing my mas e ’s deg ee. The link be ween
en i onmen and de elopmen con inued o be a big mo i a ion o my esea ch. Social de elopmen
is no possible wi hou success in comba ing clima e change. A key o sus ainable de elopmen is he
diﬀusion o knowledge and echnology h oughou he wo ld. Wi h my disse a ion, I am ying my bes
o con ibu e o his unde s anding.
I will summa ise my ime a li le as e a his poin because i ﬁ s wi h he eeling ha ime has
ﬂown by. Wi hou a doub , my wo supe iso s Aoi e Hanley and Ka in Rehdanz dese e an ex a scoop
o hanks and app ecia ion. Thanks o all he commen s and eedback you ha e gi en me o e he las
yea s.
I would like o especially hank my amazing eam a he esea ch cen e “Inno a ion and In e na ional
Compe i ion” ha has gi en me g ea suppo in se e al a eas, such as me hodology, p oo eading,
discussion o he opics and echnical ques ions: Thanks o Ca men Ande son, Di k Dohse, Sophia
Feh enbache , Wan-Hsin Liu, Robe Gold, Ul ich S olzenbu g, Eckha d Bode, Ch is ina Raasch, F ank
Bickenbach, Michaela Rank and Julian Veh ke.
I would like o hank he g ea eam o young esea che s a he I W, KCG, Kiel Uni e si y and
he depa men s: Mal e Becke , Sonali Chowdh y, Richa d F anke, Anke Jacksohn, Anna Jakobs, Le ke
Jessen-Thiesen, Thilo K ¨oge , Finja K ¨uge , Jakob Leh , Hend ik Mahlkow, Saskia Meuchelb¨ock, Leonie
Meißne , Alina Mulyuko a, Galina Po jagailo, Ch is oph Schweickha d , Vincen S ame , Sneha Thube,
and my iend Tim Vie gu z. The discussions wi h you a e always enligh ening. Bu mo e impo an ly,
wi hou you (coﬀee, Mobys and a e -wo k ac i i ies) i would no be possible o keep he a mosphe e so
g ea . A special hanks go o F auke S eglich, we s a ed oge he , and a e co-au ho s and bes iends.
I wan o hank my senio colleagues: Ma in Ademme , Lena De le sen, Oli ie Goda , Julian
Hinz, Cecilia Ho nok, Menusch Khadja i, and Nils Jannsen. They all played diﬀe en oles. Be i as
supe iso s o simply as an as ic colleagues.
The KCG also opened he doo o eedback om many g ea p o esso s who suppo ed me a all
s ages o he disse a ion. Holge G¨o g, Ho s Raﬀand Till Requa e dese e special men ion.
Likewise, I would like o hank my wo co-au ho s Philipp He kenhoﬀand Sebas ian K au heim.
Bo h a e excellen economis s and we ound a ETSG 2019 in Be n ha we sha e a simila passion o
unde s anding wha makes a company in es in co po a e social esponsibili y along he global alue
chain.
In addi ion o my academic wo k, I would like o hank S e an Fe and Bj¨o n Hech om he
Gesellscha ¨u In e na ionale Zusammena bei o wo king oge he on moni o ing and e alua ion
p ojec s in de elopmen coope a ion, including g ea ﬁeld ips o he Bei a Co ido in Mozambique. A
special hanks go o Linda Kleemann and he g ea eam a he GFA Consul ing G oup. I is always a
pleasu e o wo k oge he .
A balanced p i a e li e wi h g ea hobbies, iends and amily migh be he mos impo an ing edien
o he s amina you need in science. I would like o highligh my wo sis e s Lisa and Ind a, my a he
Jossie, my aun Guni and my g and a he Ge d. They all kep me g ounded, which is pe haps due o
li ing on he quie and pleasan wes coas o Schleswig-Hols ein. A e y special hank goes o my wi e
Nicola Posniak. We me igh a he beginning o ou s udies, which alone is enough o de e mine ha
I made he igh decision o s udy economics. Since hen, we ha e suppo ed each o he in all a eas,
including discussions abou ou p o essions. You ga e me s eng h o e all he yea s and you iew
om a business pe spec i e is always enligh ening o my esea ch. Fu he mo e, we sha e an in e es
125
in spo s, especially ki esu ﬁng and unning. Toge he , his gi es us he igh balance o keep ou heads
up wha e e li e h ows a us. I ’s g ea o ha e you by my side. Indeed, li e is no always on he b igh
side and I would like o dedica e pa o his wo k o Mo i z Bo he, Ma in G imm, my g andma Ch is a
and my mo he Ca men. My memo ies a e s ill e y much ali e, e en hough i is ha d o accep ha
you canno celeb a e he comple ion o his miles one wi h me.
Whe e do I s and now? Hones ly, I s ill don’ hink I unde s and all ace s o economics. So I will
con inue on my pa h as an economis and jus walk ano he 500 miles.
126
Closer and greener? Firms' global value chain participation and sustainability

Abstract

Closer and greener? Firms' global value chain participation and sustainability
