Busix

I c g

D a

b

c

a

A R R A A

J L M O Q

K E C H H T I M P

1

t i p

m E p p p

(

K

0 h

Research Policy 42 (2013) 975– 988

Contents lists available at SciVerse ScienceDirect

Research Policy

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / r e s p o l

s environmental innovation embedded within high-performance organisational hanges? The role of human resource management and complementarity in reen business strategies

avide Antonioli a,∗, Susanna Mancinelli a, Massimiliano Mazzanti a,b,c

University of Ferrara, Department of Economics & Management, Via Voltapaletto 11, 44121 Ferrara, Italy CERIS CNR Milan, Italy SEEDS (Sustainability, Environmental Economics and Dynamics Studies), Italy

r t i c l e i n f o

rticle history: eceived 11 October 2011 eceived in revised form 23 October 2012 ccepted 30 December 2012 vailable online 4 February 2013

EL classification: 6 53 3 55

eywords: nvironmental innovations

a b s t r a c t

This paper investigates whether firms’ joint implementation of organisational innovation and training may foster their adoption of environmental innovation (EI), and if this correlation falls within Porter Hypothesis (PH) framework. We study the relationship of complementarity between strategies of High Performance Work Practices (HPWP) and Human Resource Management (HRM) when EI adoption is the firms’ objective, using an original dataset on 555 Italian industrial firms regarding EI, HPWP and HRM, coherent with the last CIS2006-2008 survey. Results show that sector specificity matter. The only case in which strict complementarity is observed in organisational change concerns CO2 abatement, a relatively complex type of EI, but this is true only when the sample is restricted to more polluting (and regulated) sectors. This evidence is coherent with the Porter hypothesis: complementarity-related adoption of EI is an element of organisational change in firms that are subject to more stringent environmental regula- tions. Nevertheless, the fact that strict complementarity is not a diffuse factor behind the adoption of all environmental innovation indeed does not come as a surprise. At this stage in the development of green strategies, the share of eco-firms is still limited, even in advanced countries that are seeking tools for a

omplementarity RM PWP raining nnovation survey

anufacturing firms

new competitiveness. The full integration of EIs within the internal capabilities and firm’s own assets is far from being reached, even in advanced and competitive industrial settings.

© 2013 Elsevier B.V. All rights reserved.

orter hypothesis

. Introduction

Do firms’ actions in organisation and training foster the adop- ion of environmental innovation? Are environmental strategies ntegrated with organisational changes aimed at increasing firms’ erformances?

These questions, which revolve around the issue of environ- ental innovation adoption, relate to an exhaustive definition of

nvironmental Innovation (EI).1 In the MEI (Measuring EI) research

roject (Kemp and Pearson, 2007; Kemp, 2010), EI is defined as “the roduction, assimilation or exploitation of a product, production rocess, service or management or business method that is novel

∗ Corresponding author. E-mail addresses: [email protected] (D. Antonioli), [email protected]

S. Mancinelli), [email protected] (M. Mazzanti). 1 For further discussion on EI determinants see Mazzanti and Zoboli (2009a) and emp and Pontoglio (2011).

048-7333/$ – see front matter © 2013 Elsevier B.V. All rights reserved. ttp://dx.doi.org/10.1016/j.respol.2012.12.005

to the organisation (developing or adopting it) and which results, throughout its life cycle, in a reduction of environmental risk, pollu- tion and other negative impacts of resources use (including energy use) compared to relevant alternatives”2 (Kemp, 2010, p. 2).

The definition of EI is not limited to specific technologies; it also includes new organisational methods, products, services and knowledge-oriented innovations. Organisational methods are also closely linked to education and training and then to human capital formation within firms.

It is worth spending some words on the definition of organisa-

tional changes as we intend them here. The literature often adopts the term High Performance Workplace Practices (HPWP),3 to define a set of organisational changes which can be thought of as drivers of

2 Results of the MEI project can be found at http://www.merit.unu.edu/MEI/. 3 A plethora of names has been assigned to the ‘new organisational practices’

according to the practices selected and to the perspective adopted in the differ- ent studies: e.g. High Performance Work Systems (Ramsay et al., 2000; Osterman,

9 rch Po

s w o i i R t b b i o c H n B i t c o t t t s v

f m 2 J

E a t fi e i e a m d t C f i s h

v f t H 2

p

2 M

p

p fi p s m

76 D. Antonioli et al. / Resea

uperior innovative or economic performances in the firm. Coupled ith this set of practices that are related to changes in production

rganisation (e.g. autonomous or semi-autonomous teams, qual- ty circles) and labour organisation (e.g. job rotation, multitasking, ncreased workers’ responsibility), we take into account Human esource Management (HRM) practices which are linked to the raining activity sphere. The human capital embodied in employees ecomes a fundamental resource since “innovating organisation enefits from a strong skill-base” (Leiponen, 2005, p. 304), which

s able to sustain and to direct absorptive capacity. The importance f training activities4 that help generate and accumulate skills and ompetencies complementary to HPWP becomes clear. HPWP and RM practices, as intended here, are intertwined firm’s compo- ents, which, in a process of co-evolution and adaptation (Van den ergh and Stagl, 2003), influence each other and impact the firm’s

nnovative performance. Indeed, when a firm undergoes organisa- ional changes such as the introduction of HPWP, the employees an be asked to learn how to manage and how to behave in a new rganisational environment. Reconfiguring the organisational sys- em in a way that increases workforce involvement and skill base, hrough the implementation of complementary HPWP/HRM prac- ices, may be functional to the creation of an environment that moothly absorbs and exploits even more complex types of inno- ation.

The potential relationship between HPWP/HRM and EIs is ocused on as a core issue by the scholars examining the develop-

ent of the well-known Porter Hypothesis (PH) (Ambec and Barla, 006; Ambec and Lanoie, 2008; Ambec et al., 2010; Jaffe et al., 1995; affe and Palmer, 1997).

Some recent studies have tried to shed light on this issue in I-related literature. Among others, we can quote Cole et al. (2008) nd Bloom et al. (2010). The first assesses the role of foreign derived raining on a sample of African firms’ environmental performances, nding that foreign training of a firm’s decision maker, not for- ign ownership per se, does reduce fuel use. Bloom et al. (2010), nstead, survey UK manufacturing firms to assess whether energy fficiency performance is influenced by various forms of HPWP nd find mixed evidence: more general proxies of human capital anagement do not have an impact, while some others seem to

ecrease energy use. Various other papers find a positive effect of raining on EI performances (Horbach, 2008; Horbach et al., 2011; ainelli et al., 2011). Further, Kesidou and Demirel (2012) show or a sample of UK firms that organisational factors are important n determining eco innovation investment. Horbach et al. (2012) tress how organisational capabilities, among several other factors, ave to be included among the determinants of eco innovation.

Notwithstanding the above, integration of environmental inno- ation studies and the stream of organisational change research is ar from being fully satisfactory: research windows are open. In par- icular, we are not aware of studies that investigate the role of the PWP/HRM couple in the specific theme of EI adoption5 (Rennings, 000).

The aim of the paper is to investigate these somewhat unex- lored issues.

006); High Involvement Management (Bryson et al., 2005a); High Commitment anagement (Dorenbosch et al., 2005; Bryson et al., 2005b). 4 For empirical evidence on the relations between training and firms’ economic

erformance see Conti (2005) and Zwick (2004). 5 Recently, only Pekovic (2011) has tried to merge environmental and HPWP/HRM

erspectives through a study that exploits an employee-employer dataset on French rms. Environmental innovations are assumed to enhance high commitment HRM ractices, encourage employee involvement and reshape work organisation. Results how that greener firms present more labour oriented strategies and this is ulti- ately beneficial for firm-specific performance.

licy 42 (2013) 975– 988

We scrutinise whether firms’ HPWP and HRM integrated strate- gies can foster the adoption of EIs. More precisely, our main research focus is to examine if a relationship of complementar- ity exists among these practices when the adoption of EIs is the objective. We embed this analysis within the Porter Hypothesis framework. We test complementarity between strategies for all manufacturing firms and for the sub-sample of more polluting and consequentially more heavily regulated firms.

We believe that a full integration of EI in firms innovation strategies is possible and needed to evolve EI from ‘green wash- ing’ or ‘ancillary’ strategies into a key issue in firms’ redefinition of competitive advantages. Fostering green innovation strategies for growth through adequate policy interventions and studying the determinants of eco-innovation, is a central issue in the near future of developed countries (OECD, 2011; EIO, 2011).

Thus, our purpose is to investigate the extent to which environ- mental innovation is associated to human resource management (HRM) and organisational change (HPWP) implementation, by assessing their impact through the lens of complementarity theory (Milgrom and Roberts, 1990, 1995).

In particular we analyse whether the implementation of joint HRM and HPWP strategies in fostering the adoption of firms’ EIs is more evident for manufacturing firms belonging to heavily envi- ronmental regulated sectors under many aspects such as CO2, emissions and waste.6 In fact, more stringent environmental stan- dards might foster firms’ adoption of training and organisational innovation, which in turn could lead to further environmental inno- vation. The conceptual framework is that of the Porter idea of firm competitive advantages that reside in the firm value chain, within which “Strategy is manifested in the way activities are configured and linked together” (Porter, 2010).7 These ‘links’ are the comple- mentarity we investigate.

To be more precise in terms of the ample Porter-related litera- ture available (Costantini and Mazzanti, 2012), we focus here on the weak aspect of the PH. The weak version predicts that additional innovations induced by regulations present opportunity costs on the one hand, but their gross benefits may be higher. The genera- tion of those net benefits is also coherent with the assumption of initial profit maximising behaviour. Agents will be induced by new constraints to re-engineer and reorganise technology and organi- sation, to improve activity coordination and to align incentives for the purpose of meeting these constraints at a lower cost, result- ing in more efficiency and increased productivity. This view is also compatible with a neo Schumpeterian approach, as the dynamics of innovation are linked and co-evolve with appropriability con- ditions and the generation of new economic performances (Dosi et al., 2006; Malerba, 2007).

We investigate the issue by using new and original data that covers 555 Italian firms belonging to environmentally regulated manufacturing sectors over the 2006–2008 period, the same time span covered by the last CIS. We thus assure potential compa- rability of results with CIS studies (see Horbach et al., 2012 for a recent analysis on Germany).8 CIS based studies surveyed by

Mairesse and Mohnen highlight how issues regarding environmen- tal innovation have recently made their appearance (Mairesse and Mohnen, 2010). Moreover, to better explore the complementary

6 A few examples of stringent environmental standards are: the EU emission trad- ing 2003 Directive; IPPC 2008 Directive on emissions abatement and environmental technology together with its 2010 revision; the EU waste Packaging Directives of 1994 and 2003.

7 Taken from Michael Porter’s lecture at the Montreal 2010 event ‘Porter +20’, organised by Sustainable Prosperity (the citation is in slide 4, where the role of HRM in the value chain is stressed).

8 See, among others, Bocquet et al. (2004), Cozzarin and Percival (2006, 2008), Gomez and Vargas (2009) and Schmiedeberg (2008).

rch Po

r s t

o S s t

2 H

m s a “ n t b m a p e o f t m w p o f a a h s w o

r h a l c a s r p a c i t i w

t e

a t

a t t

In our specific case, we consider the ‘Environmental Innovation function’ of firm j (EIj) as the firm’s objective function and we focus

D. Antonioli et al. / Resea

elationship within PH framework, we deepen our analysis on a ubset of firms belonging to the most polluting sectors, which are hose most challenged by environmental regulations.

The paper is structured as follows: Section 2 presents the the- retical framework and lays out the main research hypotheses; ection 3 presents the survey and the original dataset; Section 4 hows the econometric analyses and complementarity tests; Sec- ion 5 concludes.

. Environmental innovation and complementarity among PWP/HRM practices: concepts and methods

What economists investigate through the analysis of comple- entarity is the extent to which different elements of strategy,

tructure and managerial processes in a firm fit with one another nd generate higher performances. Ballot et al. (2011, p. 2) affirm: the complementarities perspective is not itself a theory of orga- isational design, but rather an approach to help researchers o understand relational phenomena and how the relationships etween parts of system create more value than individual ele- ents of the system”. Since the seminal applied work by Mohnen

nd Roller (2005), devoted to testing empirical evidence for com- lementarities in national innovation policies, great deal of the conomic literature has revolved around the empirical analysis in rder to test complementarities in firms’ innovation practices.9 In act, firms’ innovation activity is a complex outcome deriving from he influence of many factors that are interrelated through comple-

entary relationships which might give “rise to systemic effects, ith the whole being more than the sum of its parts” (Roberts, 2006, . 37). Remaining within the innovation sphere, the importance f deepening empirical analysis of complementarity among dif- erent firms’ training and organisational innovation strategies has lready been explored. Galia and Legros (2004), for instance, in their nalysis on complementarities between obstacles to innovation, ighlight how innovation necessarily involves the combination of a killed work force and adequate organisation. As concerns EI issues, e are not aware of studies that specifically analyse the relationship

f complementarity among HPWP/HRM strategies.1110 Recently, eminent scholars who have contributed to the envi-

onmental Porter Hypothesis (PH) debate (Ambec et al., 2010) ave newly emphasised the role of competencies and training in chieving substantial adoption of environmental innovations, high- ighting how a great part of these innovations (carbon reductions, losed material loops, recycling, etc.) call for a full restructuring of

firm’s organisational strategy. The role of adopting integrated trategies of training and organisational innovation is particularly elevant in the increasing need to adopt integrated and more com- lex green strategies and not only “end of pipe” technology. CO2 batement is surely a more complex type of innovation for firms ompared to mere cuts in emissions such as SOx–NOx. Various nternal and external drivers (Horbach et al., 2012) are relevant to rigger decarbonisation. The costly process of business decarbon- sation might be mitigated by the occurrence of complementarity

hich, for example, generates increasing returns to scale. The well-known PH states that ‘well designed’ environmen-

al regulations (e.g. economic instruments such carbon taxes and mission trading, but not only) can stimulate innovations that

9 Bloom et al. (2010) intuitively give emphasis to complementarity among man- gement practices concerning human resources and organisational changes, but hey do not report specific tests on any sort of definition for complementarity. 10 More specifically, “a lattice (X, ≥) is a set X with a partial order ≥ such that for ny x′ , x′′ ∈ X the set X also contains a smallest element under the order that is larger han both x′ and x′′ (x′ ∨ x′′ ) and a largest element under the order that is smaller han both (x′ ∧ x′′ )” (Milgrom and Roberts, 1995, p. 181).

licy 42 (2013) 975– 988 977

offset the costs of pursuing that standard and which enhance firms’ productivity (Porter, 1991; Porter and van der Linde, 1995; Costantini and Mazzanti, 2012; Mazzanti and Zoboli, 2009b). This ‘offset innovation effort’ requires an often dramatic change in the way a firm approaches the management of its resources. It is of interest here that the basis upon which Porter relies is that of a systemic view of the firm. The systemic approach already adopted in the economic literature on innovation must necessarily be extended to environmental innovation. Moreover, the integra- tion of practices such as HPWP/HRM into EI is coherent with an analysis of diffusion rather than patents. Patenting activity is also limited as a way to defend rents in economic-systems where the majority of firms are of small and medium size. Intangible ways of defining property rights are possibly more diffused and effective. We claim that the complementarity of assets is one of these, given its idiosyncratic properties and hard ‘exportability’ (Teece, 1996; Mancinelli and Mazzanti, 2009).

We are particularly interested in filling the gap existing in the analysis of the relationship between different forms of techno- organisational environmental innovations (such as CO2 abatement, emission abatement, energy efficiency, EMS/ISO adoption) and HPWP/HRM strategies.

Since HPWP/HRM and innovation practices are typically inves- tigated in discrete settings (e.g. adopting or not, adopting at an intensity higher than the average, etc.), we study complemen- tarity between these forms of actions through the properties of supermodular functions. This technical approach has the benefit of focussing on a purely economic analysis, without the need to dwell on more mathematical issues, such as particular functional forms that ensure the existence of interior optima. For example, no divisibility or concavity assumptions are needed, so that increasing returns are easily encompassed.

Following Topkis (1995, 1998), Milgrom and Roberts (1990, 1995), Milgrom and Shannon (1994), we state that two variables x′ and x′′ in a lattice11 X are complements if a real-valued function F (x′, x′′) on the lattice X is supermodular in its arguments. That is, if and only if:

F (x′ ∨ x′′) + F (x′ ∧ x′′) ≥ F (x′) + F (x′′) ∀x′, x′′ ∈ X. (1)

Or, expressed differently:

F (x′ ∨ x′′) − F (x′) ≥ F (x′′) − F (x′ ∧ x′′) ∀x′, x′′ ∈ X, (2)

that is, the change in F from x′ (or x′′) to the maximum (x′ ∨ x′′) is greater than the change in F from the minimum (x′ ∧ x′′) to x′′ (or x′): raising one of the variables raises the value of increase in the second variable as well.12 Supermodularity gives an analytical structure to the idea that “increasing the value of some variables never prevents one from increasing the others as well” (Milgrom and Roberts, 1995, p. 182).

11 From Eqs. (1) and (2) it is evident that complementarity is symmetric: increas- ing x′ raises the value of increases in x′′ . Likewise, increasing x′′ raises the value of increases in x′ .

12 The EU Emission Trading System (ETS), which followed a proposal for a Direc- tive that had been discussed since 2001, was launched by the 2003 Directive. It is currently the major EU policy aimed towards achieving Kyoto and 202020 targets. It allocates tradable CO2 permits to firms in sectors such as metallurgy, ceramics, paper and cardboard, chemical, coke and refinery as far as manufacturing is concerned. The latter two are not present in the Emilia-Romagna region. The innovation effects of (the EU) ETS (Ellerman et al., 2010), though have been extensively analysed and compared to other environmental policies at theoretical level, have not found so far a consolidated empirical testing, even in relation to the first pilot phase 2005–2007. Micro based studies on this issue are very rare.

9 rch Po

o h

E

p s a ( s d o c

H i b p s

i

a n H

o H h t f

a

E

o

E

t w a i t l o

a s i s

( t m ( o s T n p c i

a

national industrial production and about 9% of the national GDP. It is also one of the two most innovative regions (together with Lom- bardy) in the Italian context and it is classified as a medium-high

13 A substitutability relationship exists if: EIj (11, �j ) − EIj (00, �j ) ≤ [EIj (10, �j ) − EIj (00, �j )] + [EIj (01, �j ) − EIj (00, �j )], that is, the changes in the firm’s environ- mental innovation process are less when both forms of HPWP/HRM practices are increased together than the changes resulting from the sum of the separate increases of the two kinds of practice.

14 See also Laursen (2002), Michie and Sheehan (2003) and Laursen and Foss (2003), for complementarities analyses entailing HRM practices defining HRM sys- tems of practice.

15 The consistency between the diffusion of EI in our sample and the data on EI from

78 D. Antonioli et al. / Resea

n two HPWP/HRM practices that can affect the firm’s EI function, ′ and h′′:

Ij = EIj (h′, h′′, �j ) ∀j. (3) The problem of firm j is to choose a combination of HPWP/HRM

ractices, (h′, h′′)∈H, which maximise its EI function. �j repre- ents the firm’s exogenous parameters. Actually, a firm operates in n environment which is characterised by exogenous parameters such as the product market, specific sector technologies, sector- pecific environmental policy) and one could be interested in how ifferent values of the parameter � may imply different instances f the firms’ decisional problems and hence different firms’ optimal hoices concerning EI.

Complementarity between the two different practices of PWP/HRM may be analysed by testing whether EIj = (h

′, h′′, �j) s supermodular in h′ and in h′′. Since each firm is characterised y specific exogenous parameters (�j), even if the maximisation roblem is the same for all the firms, the EI function may result upermodular in h′ and in h′′ for some firms, but not for others.

Our aim is to derive a set of inequalities (such as those explicated n Eqs. (1) and (2)), that are tested in the empirical analysis.

More specifically, through the supermodularity approach we nalyse whether the probability of a firm’s adoption of EI is sig- ificantly influenced by the presence of complementarities among PWP/HRM practices.

If in its EI maximising problem, a firm chooses to adopt neither f the two practices, namely h′ = 0, h′′ = 0, the element of the set is h′ ∧ h′′={00} If a firm chooses to adopt both practices, we have ′ = 1, h′′ = 1 and the element of the set H is h′ ∨ h′′={11} Including he mixed cases as well, we have four elements in the set H that orm a lattice: H = {{00}, {01}, {10}, {11}}.

From the above we can assert that h′ and h′′ are complements nd hence that the function EIj is supermodular, if and only if:

Ij (11, �j ) + EIj (00, �j ) ≥ EIj (10, �j ) + EIj (01, �j ), (4) r:

Ij (11, �j ) − EIj (00, �j ) ≥ [EIj (10, �j ) − EIj (00, �j )] + [EIj (01, �j ) − EIj (00, �j )], (5)

hat is, changes in the firm’s environmental innovation processes hen both forms of HPWP/HRM practices are increased together

re more than the changes resulting from the sum of the separate ncreases of the two kinds of practice. Actually, increases in EI due o an increase of both h′ and h′′ from {00} to {11} are greater (or at east equal) than the sum of increases in EI due to separate increases f h′ and h′′ from {00} to {10} ({01}).

To sum up, complementarity between the two decision vari- bles (h′ and h′′) exists if the EIj function is shown to be upermodular in these two variables and this happens when either nequality (4) or inequality (5) or other derived inequalities are atisfied.

It is worth highlighting what Milgrom and Roberts (1995) show in their fourth and fifth results) that a firm’s optimal choice related o a decisional factor may initially be zero. Nevertheless, if environ-

ental change leads to an increase in the level of another variable which has become more profitable), then the new optimal choice f the first variable may become positive if it shows a relation- hip of complementarity with the factor that has been increased. hus, increasing both variables may become more attractive in a ewly changed ‘environment’. Hence the adoption of both com- lementary practices by a firm may be an optimal choice in some

ircumstances but not in others even if its behaviour is maximising n both cases.

‘Environmental changes’ may be represented as both dynamic nd horizontal variations. In our analysis, which is static, we

licy 42 (2013) 975– 988

consider only the second type of variations and the parameter �j embodies the different environments that the different firms may face.

As it will become more clear in the following sections, for the scope of our analysis it is relevant to distinguish the situations in which the PH is more suitable. Indeed, our crucial question is if the joint implementation of HPWP/HRM strategies can foster the adop- tion of EIs especially in situations in which the PH can be verified, that is in situations of more stringent environmental regulations, namely for firms belonging to more polluting sectors, that, among other policies, have been subject to the EU ETS system since 2005.13

What our theoretical analysis suggests is that different HPWP/HRM strategies may result complements for some values of � but not for others.

As an example, in our specific analysis firms operating in sectors less exposed to environmental regulations and hence, following the PH, less stimulated to adopt EIs, could find it more convenient to externalise the management of workforce training. This kind of behaviour could even lead to a crowding out effect among some of the many strategies of training and organisational innovation and hence to substitutability14 among them.

We can thus set out two consequential research hypotheses: [H1]. Complementarity that refers to HPWP/HRM strategies is

relevant to fostering the adoption of various EIs (CO2 abatement, emission abatement, EMS/ISO implementation, material use reduc- tion).

[H2]. ETS firms belonging to sectors such as ceramics, metallurgy and paper cardboard might present more evident signals of com- plementarity than non-ETS firms as a way to pro-actively tackle the regulation challenge through ‘innovation offsets’.

We test [H1] by taking all manufacturing firms into account, while we coherently test [H2] by taking only the more polluting manufacturing sectors into account.

It is finally worth noting that we are more interested here in examinations of two-three way relationships among individual ele- ments of a firm’s organisational changes, rather than investigations of ‘entire’ systems of complementarity.15 Ennen and Richter assert that “complementarities are system specific phenomena. Studies of relationships among individual elements of factors can offer valuable insights, but the failure of such a study to confirm com- plementarity effects where it had been expected them may mean that the full range of factors at work and their relationships have not yet been fully understood” (Ennen and Richter, 2009, p. 3).

3. Data and empirical strategy

The empirical context of this work is the manufacturing sector of the Emilia-Romagna region in Italy (NUTS 2 level), which, with a population of around 4.5 million (ISTAT, 2010), accounts for 20% of

the Community Innovation Survey database covering 6483 Italian manufacturing firms, which shows adoption in a 13–18% range across sectors and type of EI is worth stressing. Adoptions in the northeast of Italy, where the region is located are 19% for energy efficiency and 15% for CO2 abatement (and 18% and 14% respectively for Italy as a whole).

rch Po

i 2 r c

l 5 ( t q s i u l s i w E s m s a I a r H d

m

3

i t i ‘ v i

p e t e b a a n ( (

e o u t fi a u

e

s p

D. Antonioli et al. / Resea

nnovator region at the EU27 level (Brusco, 1982; Hollander et al., 009). A leading innovative region of a developed country can rep- esent a good ‘laboratory’ to test our hypothesis about HPWP/HRM omplementary practices on EIs.

The test of research hypotheses [H1] and [H2] is based on micro evel data coming from a unique dataset concerning a sample of 55 manufacturing firms located in the Emilia-Romagna region see Appendix A for a snapshot of the questions used to construct he main variables). The information collected through a structured uestionnaire refers to the 2006–2008 period. The sample is con- tructed on the basis of a stratified random sampling technique, n order to obtain reliable results for the overall regional man- facturing context, with a stratification by province (geographic

ocation), size and sector (Table B1 in Appendix B). It is worth tressing again the proximity of our questions with those included n the CIS5 (Community Innovation Survey) carried out in 2008,

hich may allow for direct comparison with data collected at the uropean level on some specific issues. However, the information et provided by the questionnaire administered to firms’ manage- ent is even richer than that drafted by the CIS and concerns

everal sets of firm activity, spanning across issues and themes such s technological and organisational changes, training activities, CT implementation, environmental innovation and internation- lisation strategies, as well as the quality of firm level industrial elations and working conditions, for which we focus on EIs and on PWP/HRM practices in order to answer our research question as escribed below.

The parts of the questionnaire that we exploit in this paper are ainly those referring to EI adoptions and HPWP/HRM aspects.

.1. EI variables

The outcome variables stem from a set of questions concern- ng the EI activities carried out by the firms in 2006–2008. EI is hen neither sector nor technology specific and it can take place n any economic activity and not only in the still loosely defined eco-industry’ sectors. It is not limited to environmentally moti- ated innovations, but includes the “unintended’ eco-effects of all nnovations.

In formulating the questions relative to EIs we followed the MEI roject (Kemp, 2010) that informed the CIS5. For this reason we licited information (Table 1) concerning the adoption of EI for: he reduction of energy and material for unit of product (ENERGY), missions reduction in terms of CO2 (CO2), emissions reduction to etter the quality of soil, water and air (EMISSIONS) and, finally, the doption of procedures such as EMAS, ISO14001 (EMASISO). EI is

key factor in tackling the challenge of sustainable development, amely but not only the challenges posed by the reduction of CO2 set by Kyoto targets and the EU 2020 strategy) and waste reduction Mazzanti and Zoboli, 2009c; Marin and Mazzanti, in press).

In Table 1 the distribution of EI in our sample is shown.16 An xpected result emerges when the overall sample is restricted to nly those firms belonging to more polluting sectors17: the man- facture of coke, refined petroleum products and nuclear fuel;

he manufacture of chemicals, chemical products and man-made bres; the manufacture of other non-metallic mineral products nd the manufacture of basic metals and fabricated metal prod- cts, which are respectively classified as DF, DG, DJ, DI (Table B1 in

16 Marin and Mazzanti (in press) present figures and trends for these sectors’ missions. 17 Because of aggregation constraints regarding the collection of information in our urvey we are forced to include the DH sector (Manufacture of rubber and plastic roducts) in the set of the polluting sectors.

licy 42 (2013) 975– 988 979

Appendix B) according to a two-digit NACE-Rev1 classification.18

In fact, while the presence of EIs is really low in the overall sample, it gains several percentage points in distribution frequency when only the polluting sectors are considered, passing from an average of 14% to 20%.

Our analysis thus examines (i) the entire working sample of 555 firms and (ii) the sub-set composed of firms belonging to the most polluting sectors, which are those most challenged by environmen- tal regulations (ETS in primis). In line with the outlined research hypotheses, our main aim is to investigate how the joint imple- mentation of HPWP/HRM practices can foster the adoption of EIs firstly for all firms ([H1]) and secondly within the ‘Porter Hypothesis framework’ (see [H2]).

3.2. HPWP/HRM variables

Three sets of organisational aspects that can be brought back into the wider concept of HPWP/HRM practices are here taken into consideration (Table 2): changes in production organisation (ORGPROD), changes in labour organisation (ORGLAB) and train- ing activities (TRAINCOVERAGE, TRAINCOMP, TRAININVEST). They represent a comprehensive set of organisational practices aimed at increasing firms’ performances. These variables allow us to capture the within firms’ strategic decisions belonging to the organisational sphere capable of increasing the absorptive capacity of the firm towards EIs.

Starting from the organisational changes set, the questions addressed to the management provided us with the possibility to construct composite additive indexes of intensity in organisa- tional changes: the more organisational changes are implemented in both production and labour organisation, the higher the index. The items included in the indexes construction are associable to the set of items usually ascribed to HPWP practices in the liter- ature, such as, for example, the introduction of team work and quality circles as for production organisation; and improvement of competences, increase of workers autonomy and problem solving, reduction of the hierarchical layer as regards labour organisation (see Appendix A). For purposes linked to the complementarity conceptual framework analysis, the indexes were dichotomised according to the following rule: if the index was above or equal to the mean (median) then we assigned the value 1, while otherwise we assigned the value 0. We note that the necessary dichotomisa- tion of indexes and continuous variables is performed, to check the sensitivity of results, both using mean and median as statistics as clearly evidenced in section three below.

As for the training activities which refer to HRM practices, we exploit information concerning the percentage of employees cov- ered by training programmes (TRAINCOVERAGE), a variable that tells us whether the firm introduced training courses in order to develop the entire range of competences (TRAINCOMP)19 listed in the questionnaire (technical, IT, organisational and concerning eco- nomics/law) and not just some of them and finally, a variable that informs us whether the firm invested its own economic resources in training activities (TRAININVEST).

On the basis of such dichotomised HPWP/HRM variables we were able to define four states of the world, as it is shown in Table 3, where the distribution is reported. These are the ‘states’ we exploit

for complementarity assessments as described in section two.

We may argue that the occurrence of the different states of the world associated to the joint presence/absence of pairwise

18 The variable takes value 1 only when the firms aim to develop all competences expressed in the question in Appendix A.

19 We use such a taxonomy, instead of the two digit NACE REV1, in order to reduce the number of controls.

980 D. Antonioli et al. / Research Policy 42 (2013) 975– 988

Table 1 Adoption of environmental-innovations (distribution).

Whole sample By more polluting sectorsa

Freq. % Freq. %

Energy/Material reduction per unit of product (ENERGY) 82 14.77 43 22.4 CO2 reduction (CO2 ) 64 11.53 33 17.19 Emissions reduction for soil, water and air (EMISSIONS) 78 14.05 41 21.35 Adoption of procedures like EMAS and ISO14001 (EMASISO) 80 14.41 36 18.75 Obs./mean % 555 13.69 192 19.92

a Two digit classification: DF, DG, DJ, DI (and DH).

Table 2 HPWP D/HRM D variables (distribution).

Variables (Dummies) Whole sample Polluting sectorsa

Freq. % Freq. %

HPWP Production organisation aspects (ORGPROD D) 350 63.06 127 66.15 Labour organisation aspects (ORGLAB D) 218 39.28 83 43.23 HRM Employees involved in training activities (TRAINCOVERAGE D) 209 37.66 87 45.31 Full set of competences covered by training activities (TRAINCOMP D) 58 10.45 18 9.38 Presence of resources invested in training (TRAININVEST D) 408 73.51 153 79.69 Obs./mean% 555 40.23 192 44.40

a Two digit classification: DF, DG, DJ, DI (and DH).

Table 3 HPWP D/HRM D states of the distribution.

Variables (Dummies) States of the world (555 obs.) whole sample % States of the world (192 obs.) polluting sectorsa %

(1,1) (1,0) (0,1) (0,0) (1,1) (1,0) (0,1) (0,0)

TRAINCOVERAGE D ORGPROD D 26.67 10.99 36.40 25.95 31.77 13.54 34.38 20.31 TRAINCOVERAGE D ORGLAB D 21.44 16.22 17.84 44.50 27.08 18.23 16.15 38.54 TRAINCOMP D ORGPROD D 8.47 1.98 54.59 34.95 7.81 1.56 58.33 32.29 TRAINCOMP D ORGLAB D 7.57 2.88 31.71 57.84 6.25 3.13 36.98 53.65 TRAININVEST D ORGPROD D 49.37 24.14 13.69 12.79 55.73 23.96 10.42 9.90

6

H p c ( g w I e w b b d t i t

3

i d t fi

TRAININVEST D ORGLAB D 32.97 40.54

a Two digit NACE-Rev1 classification: DF, DG, DJ, DI (and DH).

PWP/HRM practices provides a first insight into the likely resence of complementarity (Mohnen and Roller, 2005). Let us onsider the TRAINCOVERAGE D and the two HPWP dummies ORGPROD D and ORGLAB D). The occurrence of (1,1) plus (0,0) is reater than the occurrence of the sum of the other two states of the orld, for both the whole sample and the polluting sectors sample.

t is worth stressing that such a difference in occurrence is more vident for the polluting sectors, pointing, although in a descriptive ay, to the presence of possibly stronger complementary relations

etween the couples of our HPWP/HRM variables. The same can e said when we take into consideration training investment ummy (TRAININVEST D), while the nature of relations between he competencies addressed by training programmes and changes n labour and production organisation seems to be more oriented owards substitution rather than complementarity.

.3. Control variables

To enrich the analysis and set a comprehensive vector of

nnovation related factors (Horbach et al., 2012) we use a stan- ard set of controls, that includes size dummies, Pavitt/OECD axonomy for sectors20 and less standard aspects related to the rms’ strategic behaviour such as the “openness” to international

20 Results are available upon request.

.31 20.18 36.46 43.23 6.77 13.54

markets provided by a variable indicating if a firm is an asso- ciated company of a foreign one (INTERN OPEN) and the type of such an association (e.g. joint venture, stake below or above 50%), the presence of resources invested in R&D (R&D) and an index capturing the intensity in collaborations for technological innovations (TECH NET) (for descriptive statistics see Table B2 in Appendix B). The ratio behind the use of such variables is that they may constitute influencing structural and strategic factors for EI adoption: the openness to international markets as well as the effort devoted to R&D activities and to collaborations for technological innovations may represent positive impulses.

On the basis of the theoretical framework for complementari- ties assessment we set up a two steps procedure, described in the following section, in order to investigate the extent to which HRM and HPWP interact and eventually drive the adoption of EIs.

Our approach may by inscribed within a stream of works based on the direct utilisation of an objective function according to which we test the presence of complementary relationships among selected covariates (HPWP/HRM) over an objective variable (EI) (Mohnen and Roller, 2005). Such an approach differentiates from several others used to test the existence of complementarities, which usually do not need an objective variable, but are essentially based on ‘revealed preferences’ and are tested through correlations.

The latter may be simple bivariate correlations or more sophisti- cated ones in which controls for observable and unobservables are made (see Athey and Stern, 1998; Arora, 1996 for a full review of the different approaches).

rch Po

a

[

w H w o f H f s

4

4

b i a fi m r d s t a d t t s

y l n b d t i c b t o v c p t b t a h

d ( t e e

n a

sectors

As anticipated above we test the hypothesis [H2] for more pol- luting and regulated sectors. The heavier regulatory burden to

D. Antonioli et al. / Resea

Unlike these approaches, here we set up an objective function, n innovation function, that can be modelled as follows:

EI]i = b0i[Controls] + b1i[HPWP D(1), HRM D(1)] + b2i[HPWP D(1), HRM D(0)] + b3i[HPWP D(0), HRM D(1)] + b4i[HPWP D(0), HRM D(0)] + ui (7)

here the EI dummy variables enter a probit regression, the PWP/HRM variables are capturing the different states of the orld; it is worth noting that the constant term is suppressed in

rder to obtain coefficients for each state of the world; i stands or the i-th firm. Matching the HPWP/HRM factors generates six PWP/HRM ‘couples’ that we include among the regressors for the

our EI dependents (Tables C1 and C2 in Appendix C). In total we um up to 24 cases in the analysis (6 states and 4 types of EI).

. Empirical analysis

.1. Probit regressions

The first step of the investigation is given by a set of pro- it regressions, that show the overall good quality of the model,

n addition to both expected and unexpected signals. Both SIZE nd TECH NET matter in determining EI for the whole sample of rms (Table C1 in Appendix C), but when we look at the set of ore polluting sectors (Table C2 in Appendix C), the significant

elation with networking activities for innovation disappears. The ichotomous variables that identify the states of the world are all ignificant, although with a minus sign. We nevertheless note that hose regressions are estimated with the omission of a constant. In

standard probit with the exclusion of one of the state of the world ummies and the reintegration of a constant term instead, the signs urn out to be more consistent with what we would expect: setting he state of the world (0,0) as a benchmark and omitted case, the tate of the world (1,1) is significant and positive.21

Once we fitted the probit models,22 the second step of the anal- sis was to test hypotheses implementing a set of Wald tests. The atter allows us to test the following linear restriction, under the ull hypothesis, on the state-of-the-world-dummies coefficients: 1 + b4 = b2 + b3. The test, which is distributed as a �2 with one egree of freedom, since we are testing a single linear restriction at a ime, is not informative as we would like. Indeed, we are interested n the following inequalities, namely the sign of the scalar linear ombination of the coefficients of interest: b1 + b4 − b2 − b3 ≥ 0; 1 + b4 − b2 − b3 ≤ 0. The standard Wald test only informs us as o whether or not we can reject the null hypothesis of equality f the coefficients sum. However, coupling the information pro- ided by the Wald tests with the sign of the inequalities, also onfirmed by one-sided tests on the linear combination of the arameters, we know the direction towards which a rejection of he null leads us in terms of supermodularity or submodularity. If 1 + b4 − b2 − b3 ≥ 0 and the Wald test leads us to reject the null,

hen we can argue that we are in presence of supermodularity nd hence of complementary HPWP/HRM practices. Submodularity olds if b1 + b4 − b2 − b3 ≤ 0 and the null is rejected as well.

21 Innovation choices can be simultaneous. The empirical procedure may test ependence between environmental innovations. A set of bivariate probit models setting up seemingly unrelated probit models) were created for this purpose. We hen ran Wald tests accordingly for every couple of HPWP/HRM variables in each quation of the bivariate probit. Results, available upon request, mostly confirm the ssence of probit analysis. 22 As stated above, we also carried out one-sided tests, distributed as a standard ormal Z, that give similar outcomes and from which the signs of our inequalities re confirmed.

licy 42 (2013) 975– 988 981

We implement the set of tests on the coefficients associated to 24 cases. The complementarity hypothesis is also tested for the polluting/ETS sectors, following the same procedure and carrying out further 24 tests.

4.2. Complementarity analysis: all manufacturing sectors

In this section we scrutinise research hypothesis [H1]. Table 4 clearly shows that there are not cases of strict com-

plementarity. [H1] is thus rejected. Overall, the investigation does support strict substitutability in one case. The critical value of the Wald test23 (5% level of significance)24 is surpassed for all cases of EI adoption for the couple ‘TRAINCOMP-ORGPROD’.

The strong specificity of complementarity existence is then highlighted: training competencies – changes in re-organisation of production seem not to match well for the aim of increasing the adoption of EIs.

We note this is not in itself a ‘failure’: complementarity surely is an ‘asset’ that can improve firm performances, but trade-offs may simply illustrate that some firms are capable of managing one factor at a time. They cannot deal with complex organisational change, but they can positively correlate either training or organisational change with EI.25

This might also be coherent with recent evidence that shows how training (alone) is a determinant of EIs (Horbach, 2008; Cainelli et al., 2011). It is a signal of potential weaknesses and difficulty regarding the organisational change firms face. Further, Ennen and Richter (2009) state that (strict) complementarity can be a source of significant competitive advantage, but it is really idiosyncratic to the sector, innovation type and inputs to innovation or perfor- mance we analyse. The embeddedness in complex systems makes it hard for complementarity to be managed purposefully. Indeed, the two authors find that the evidence of (strict) substitutability among inputs, that is trade off in firm strategies, is quite diffused. Though the match of heterogeneous factors is more likely to gener- ate complementarity gains, they did not find a single factor whose co-occurrence with others invariably results in the emergence of complementarity relationships.

Complementarities are not a low hanging fruit. They might exist as a content of new organisational designs and practices for some firms which maximise their innovation performance through the exploration of the full set of possibilities related to HRM/HPWP practices. This is a message that is useful for firms and managers in rethinking their processes.

Given that firms’ heterogeneity is very relevant in the analysis of complementarities, we now analyse its presence for a specific subset of manufacturing firms.

4.3. Complementarity in a Porter framework: the more polluting

23 The two tailed test on inequality has as a null hypothesis, depending on the direction of the inequality (≥; ≤) either ‘complementarity’ or ‘substitutability’. This means that the non rejection of the null cannot allow an inference on the strong or weak content of these. The rejection of the null respectively means ‘strong substi- tutability’ and ‘strong complementarity’. In other words, strong complementarity is assessable as a rejection of the null when testing substitutability. The two tests are obviously ‘complements’ and are based on the same t statistics.

24 This is confirmed by simple probit regressions. Results available upon request. 25 We also checked whether firms in the only sectors that have reduced emissions

of CO2 in the last 20 years behave differently. Results do not change with respect to those of ‘polluting sectors’ (sectors that have reduced emissions are DB–DC; DF–DH–DG, DJ).

982 D. Antonioli et al. / Research Policy 42 (2013) 975– 988

Table 4 Complementarities tests in a discrete setting. Linear restriction on states of the world coefficients from probit regressions.a

HPWP D/HRM D variables ECOINNO

(Mean value used for dicotomisation)

ENERGY CO2 EMISSIONS EMASISO

Wald testb

Sign of the linear combination (b1 + b4) + (−b2 − b3)

Wald testb

Sign of the linear combination (b1 + b4) + (−b2 −b3)

Wald testb

Sign of the linear combination (b1 + b4) + (−b2 −b3)

Wald testb

Sign of the linear combination (b1 + b4) + (−b2 − b3)

TRAINCOVERAGE D ORGPROD D 0.01 ≤0 0.31 ≥0 0.1 ≤0 0.34 ≤0 TRAINCOVERAGE D ORGLAB D 0.74 ≥0 1.2 ≥0 0.24 ≥0 0.11 ≥0 TRAINCOMP D ORGPROD D 7.64*** ≤0 8.00*** ≤0 10.65*** ≤0 7.13*** ≤0 TRAINCOMP D ORGLAB D 0.03 ≥0 0.74 ≤0 0 ≤0 0.76 ≥0 TRAININVEST D ORGPROD D 0.35 ≥0 0.28 ≤0 0 ≤0 2.12 ≥0 TRAININVEST D ORGLAB D 0.47 ≥0 2.47 ≥0 1.76 ≥0 0 ≥0 a Tests conducted on marginal effects provide the same results (not reported for space constraint but available from the authors upon request). b Since we are testing one linear restriction at a time the Chi2 distribution has 1 degree of freedom as the number of the linear restrictions. Critical values of Chi2 (1)

distribution: 6.63, 3.84 and 2.71 (***1%, **5% and *10% level of significance respectively); N = 555, (b1 + b4) + (−b2 −b3) ≥ 0 is index of supermodularity. (b1 + b4) + (−b2 − b3) ≥ 0 is index of submodularity.

Table 5 Complementarities tests in a discrete setting. Linear restriction on states of the world coefficients from probit regressions (Polluting sectors).a

HPWP D/HRM D variables ECOINNO

(Mean value used for dicotomisation)

ENERGY CO2 EMISSIONS EMASISO

Wald testb

Sign of the linear combination (b1 + b4) + (−b2 − b3)

Wald testb

Sign of the linear combination (b1 + b4) + (−b2 −b3)

Wald testb

Sign of the linear combination (b1 + b4) + (−b2 −b3)

Wald testb

Sign of the linear combination (b1 + b4) + (−b2 −b3)

TRAINCOVERAGE D ORGPROD D 0.41 ≥0 4.15** ≥0 0.2 ≥0 0.3 ≥0 TRAINCOVERAGE D ORGLAB D 1.03 ≥0 0.48 ≥0 0.34 ≥0 1.89 ≥0 TRAINCOMP D ORGPROD D 0.44 ≤0 1.83 ≤0 1.89 ≤0 0.25 ≤0 TRAINCOMP D ORGLAB D 0.06 ≤0 0.28 ≤0 0.02 ≤0 0.51 ≥0 TRAININVEST D ORGPROD D 1.09 ≥0 0.39 ≥0 0.31 ≥0 2.23 ≥0 TRAININVEST D ORGLAB D 0.4 ≤0 n.f. n.f. 2.01 ≥0 0.02 ≤0 a Tests conducted on marginal effects provide the same results (not reported for space constraint but available from the authors upon request); n.f. means the state of the

world TrainInvest = 0 and OrgLab = 1 predict failures perfectly in the probit estimation, hence the variable is dropped and the test cannot be computed. b Since we are testing one linear restriction at a time, the Chi2 distribution has 1 degree of freedom as the number of the linear restrictions. Critical values of Chi2 (1)

distribution: 6.63, 3.84 and 2.71 (***1%, **5% and *10% level of significance respectively); N = 555, (b1 + b4) + (−b2 −b3) ≥ 0 is index of supermodularity. (b1 + b4) + (−b2 −b3) ≥ 0 i

w t b t

r

o t o d p i i c i E o

u

u

industry has performed over the past decades. CO2 emissions, whose reduction requires a full redefinition of economic, energy and technological strategies (clean integrated technologies), are

40 60

80 10 0

s index of submodularity.

hich more polluting sectors are subject might increase the impor- ance of EI and the related likelihood of using complementarity ased strategies to redesign organisation in the face of the regula- ion challenge.

The evidence in Table 5 is in fact somewhat different with espect to what we found in Table 4.

For this sub sample of firms belonging to sectors that are n the ‘frontier’ of environmental (climate change) challenges, he weakness regarding the linking of training competencies and rganisation of production is not relevant.26 As an example of quite ifferent evidence, in one case (training coverage – organisation of roduction) we do find evidence in support of strict complementar-

ty, at the 5% significance level. This shows that complementarity s present as an option in the firm HPWP/HRM tool kit to tackle the omplex challenge of CO2 abatement27 It is worth noting that Italy

s among those countries (we find a rare exception in Northern U) that have not to cutting down on carbon dioxide. As evidence f the complexity of the challenge, Fig. 1 shows how the Italian

26 If we use median values as a benchmark the result is confirmed. Results available pon request. 27 If we use median values as a benchmark the result is confirmed. Results available pon request.

20

1990 1995 2000 2005 2010 Year

CO2 NOx SOx

Fig. 1. Emissions and CO2 trends in Italian economy, industry and services (1990 = 100).

Source: NAMEA, Italian Statistical Agency, environmental accounting datasets.

rch Po

c o r s t

d c

o t t d i – a h 1

[ t t c r l fi t n H b s o a

5

o r n t r p s h n r

b t t b U c a e s w s c

t a t

D. Antonioli et al. / Resea

onstant, while emissions, which are to a large extent dependent n filters, have been considerably cut. This poor picture also eflects innovation weaknesses. Heavier and more regulated ectors seem to react and adopt different strategies according to he PH.

Firms at the frontier of environmental challenges do respond ifferently than the average firm, though they still fail to exploit omplementarities in extended ways.

As we recalled at the beginning, this is highly in the spirit f the Porter idea of competitive advantages stemming from he extension of the firm’s aims and the use of multiple ways o reshape their organisation. It is then possible that properly esigned regulations bring about conditions – such as boost-

ng the demand for green products, pricing scarce resources making unexploited technologies available (Wagner, 2006) nd opening up the set of choices constrained by production abits towards a re-engineering of routines (Sinclair-Desgagnè, 999).

We believe that the evidence around research hypothesis H2], which shows that strict complementarity is present only in he CO2 case28 we commented on, is dependent upon the fact hat most firms have tended to rely on single factors (training, ooperation with clients or universities, etc.) to adopt the envi- onmental innovations they needed. This is well documented in EI iterature. Nevertheless, this does not appear to be currently suf- cient to increase the adoption of green innovations and enhance he possibility to witness EI as fully integrated strategies. Inter- al drivers, such as the reorganisation of firm production and RM, are also needed for this purpose. EI adoption can thus ecome part of the asset stock possessed by firms which is con- tituted both by mere adoption and by the integration of EI with ther competitiveness strategies (the complementarity intangible sset).

. Conclusions

In the aim of providing new understanding about the effects f firms’ organisational changes on EI adoption, we study the elationships between human resource management and inter- al processes of organisational change in labour and production hrough the lens of the complementarity theory. Though the elevance of HPWP/HRM for developing relatively new and com- lex forms of innovations such as EI has been noticed by cholars that contributed to the development of the Porter ypothesis, the lack of integration between environmental eco- omics and HRM disciplines has blocked research in this specific ealm.

We analyse diverse situations of potential complementarity etween HRM and organisational changes, covering 4 different ypes of EI (CO2 abatement, emissions reduction, EMS/ISO adop- ion, energy/material efficiency). We show that for EI adopted y firms located in a densely industrial region of the European nion which is highly exposed to international competition, strict omplementarity is rarely present. In contrast, when looking t the full sample of manufacturing firms, strict substitutability merges in one case. Training in key competencies and organi- ational changes in production seem to suffer from a mismatch

hen considering their integration which highlights how green

trategies are not fully embedded within firms’ reorganisation hanges.

28 We may argue that, for example, implementing organisational changes such as eam working or quality circles and coupling them with wide training activities cre- tes new knowledge and competencies that in turn foster the (absorptive) capacity o introduce energy saving practices.

licy 42 (2013) 975– 988 983

Though the fact that EI development in countries such as Italy is still in a non-mature phase might be part of the explanation for this, the evidence can signal ‘problems’. We cannot say that observing substitutability is a weakness, given that EIs are possibly correlated to single factors. We note a lack of systemic innovation capability, which is one of the brakes behind the poor competitiveness and environmental performance of some (southern) EU countries at the moment.

The evidence confirms the well-known fact that complementar- ity is not to be taken for granted: it is industry, innovation and factor specific. Its achievement requires a full screening of firms’ ‘existent assets’ and of those that could be ‘created’ (e.g. complementar- ity between assets as immaterial source of competitiveness). This requires proper investments in the re-engineering of firm organi- sation.

Firms that are on the frontier of environmental technological challenges (more polluting firms, more heavily regulated firms) instead present some evidence which does not reject the ‘Porter hypothesis’ and which we here enrich with complementarity con- cepts.

Complementarity emerges for CO2 abatement, through the integration of training coverage and organisation of production strategies. Sector specificity, namely heavier environmental reg- ulations, influences the way firms behave with respect to the setting up of complementarity strategies. We observe comple- mentarity related adoption of EI as an element of organisational change in firms that are subject to more stringent regula- tions.

Nevertheless, the fact that strict complementarity is not a diff- used factor behind the adoption of all environmental innovations comes in no way as a surprise. At this stage in the development of green strategies, the share of eco-firms is still limited even in advanced countries that are seeking for new competitive tools. Inte- gration of EIs with the internal capabilities and firm’s own assets is far from being achieved even in advanced and competitive indus- trial settings.

Further research should be aimed both at extending evi- dence to an EU level (through the CIS2008) and at assessing the effects of EI (among EIs and between EI and other techno- organisational changes) on economic and environmental firms’ performance by also using a complementarity based perspec- tive.

Acknowledgements

The authors acknowledge the financial support of the Emilia- Romagna Region and the support of Confindustria. The present work was also co-funded by the FAR project, of the University of Ferrara (2007–2009), on Innovation, Industrial Relations and Produc- tivity in Local Systems, responsible Prof. Paolo Pini. The comments and suggestions of two anonymous referees, which helped to improve the paper, are also gratefully acknowledge. The usual dis- claimers apply.

Appendix A.

Selected questions used to construct our HPWP/HRM and EI

variables. The answers refer to the period 2006–2008.

HPWP Q1: Which of the following organisational practices do you

adopt?

9 rch Policy 42 (2013) 975– 988

Yes/No

arked with Yes; 0 otherwise

Yes/No

anged)

ents

anisation and on quality of process/product

usiness section d competency exchanges

ked with Yes; 0 otherwise

i

p

a

g

w

p b

Yes/No

ling)

b

o

i Yes/No

o

84 D. Antonioli et al. / Resea

n= Production organisation practices (x)

1 Quality circles and/or improvement teams 2 Team working 3 Just-in-time 4 Total quality management

ORGPROD = ∑4

n=1 xn

4 where x assumes value 1 if the organisational practice is m

n= Labour organisation practices (z)

1 Task rotation and/or job rotation (with tasks unch 2 Widening of the tasks and/or assignments 3 Higher autonomy in performing tasks and assignm 4 Broadening of competencies 5 Training associated to organisational needs 6 Higher autonomy in problem solving 7 Structured discussion/confrontation on labour org 8 Definition of goals for employees 9 Employee performance evaluation systems

10 Ex-post rewards based on the performance 11 Ex-ante rewards in order to develop competencies 12 Reduction of hierarchical layers within the same b 13 Techniques to manage information, knowledge an

ORGLAB = ∑13

n=1 zn

13 where z assumes value 1 if the organisational practice is mar

HRM Q2: Please provide the percentage of permanent employees

nvolved in training programmes: Permanent employees . . .. . ..%

TRAINCOVERAGE = . . .%/100 Q3: Which kinds of competencies were addressed by training

rogrammes? Typologies of competencies Yes/No

1. Computer science competencies 2. Technical/specialised competencies 3. Organisational/relational competencies 4. Law/economic competencies

TRAINCOMP = 1 if all the four types of competences are ddressed; 0 otherwise

Q4: Did the firm invest its own resources in training pro- rammes related to innovative activities? Yes/No

TRAININVEST = 1 if firms invested its own resources; 0 other- ise

ENVIRONMENTAL INNOVATION (EI) Q5: Did the firms adopt “environmental” products and/or

rocess technological innovations that induced the following enefits? Benefits

1. Reduction in the use of materials and/or energy by output unit (including recyc 2. CO2 emissions reduction 3. Emission reductions that improve the quality of soil, water and air

ENERGY = 1 if Reduction in the use of materials and/or energy y output unit (included recycling) marked as Yes; 0 otherwise

CO2 = 1 if CO2 emissions reduction marked as Yes; 0 otherwise EMISSIONS = 1 if Emission reductions that improve the quality

f soil, water and air; 0 otherwise Q6: Does the firm have procedures that structurally identify

ts environmental performance? Procedure

1. EMAS 2. ISO 14001 3. Others such as LCA, ISO14040, . . .. . .. . .. . .. . .. . .. . .. . .(specify)

EMASISO = 1 if EMAS or ISO14001 or Others is marked as Yes; 0 therwise

.

D. Antonioli et al. / Research Policy 42 (2013) 975– 988 985

Appendix B.

Table B1 Population and sample distribution (%) by sector and size.

Population distribution (%) Size

Sector (NACERev1) 20–49 50–99 100–249 250+ Total Total (a.v.)

Food (DA) 5.65 1.94 1.16 0.64 9.39 382 Textile (DB-DC) 6.17 1.47 0.71 0.37 8.73 355 Wood, paper and other industries (DD–DD–DN) 7.79 1.67 0.79 0.42 10.67 434 Chemical and rubber (DF–DG–DH) 5.01 1.87 1.11 0.42 8.41 342 Non metallic mineral products (DI) 3.81 1.23 1.18 0.79 7.01 285 Metallurgy (DJ) 16.99 3.29 1.18 0.25 21.71 883 Machinery (DK–DL–DM) 21.44 6.37 4.06 2.24 34.10 1387 Total 66.86 17.85 10.18 5.11 100.00 Total (a.v.) 2720 726 414 208 4068

Sample distribution (%) Size

Sector 20–49 50–99 100–249 250+ Total Total (a.v.)

Food (DA) 2.88 3.78 1.62 0.54 8.83 49 Textile (DB–DC) 2.70 1.44 1.62 0.54 6.31 35 Wood, paper and other industries (DD–DD–DN) 3.60 2.88 1.08 0.90 8.47 47 Chemical and rubber (DF–DG–DH) 3.78 3.42 1.80 1.08 10.09 56 Non metallic mineral products (DI) 1.62 2.16 1.62 2.16 7.57 42 Metallurgy (DJ) 8.83 5.77 2.16 0.18 16.94 94 Machinery (DK–DL–DM) 14.05 15.32 7.39 5.05 41.80 232 Total 37.48 34.77 17.30 10.45 100.00 Total (a.v.) 208 193 96 58 555

T D

able B2 escriptive statistics.

Whole

Mean (

Outcome variables Energy/material reduction per unit of product (ENERGY) 0.147 CO2 reduction (CO2 ) 0.115 Emissions reduction for soil, water and air (EMISSIONS) 0.140 Adoption of procedures like EMAS and ISO14001 (EMASISO) 0.144 HPWP/HRMa

Production organisation aspects (ORGPROD/HPWP) 0.484 Labour organisation aspects (ORGLAB/HPWP) 0.247 Employees involved in training activities (TRAINCOVERAGE/HRM) 0.378 Full set of competences covered by training activities (TRAINCOMP/HRM) 0.104 Presence of resources invested in training (TRAININVEST/HRM) 0.735 Controls Size dummies (5 Pavitt/OECD sector dummies: labour intensive (LI),

resource intensive (RI), science based (SB), scale intensive (SI), specialised suppliers (SS))

–

Sector dummies (4 size dummies: 20–49 employees; 50–99 emp.; 100–249 emp.; more than 249 emp.)

–

INTERN OPEN 0.021 R&D 0.800 TECH NET 0.101

a Where appropriate we report the statistics of the indexes, since the distributions of t

sample Polluting sectors

555 obs.) StDev Min/Max Mean (192 obs.) StDev Min/Max

0.355 0/1 0.223 0.417 0/1 0.319 0/1 0.171 0.378 0/1 0.347 0/1 0.213 0.410 0/1 0.351 0/1 0.187 0.391 0/1

0.329 0/1 0.474 0.342 0/1 0.173 0/1 0.234 0.169 0/1 0.369 0/1 0.428 0.393 0/1 0.306 0/1 0.093 0.292 0/1 0.441 0/1 0.796 0.403 0/1

– 0/1 – – 0/1

– 0/1 – – 0/1

0.066 0/0.83 0.016 0.053 0/0.33 0.400 0/1 0.776 0.417 0/1 0.114 0/0.74 0.089 0.108 0/0.74

he dichotomised variables are in the text.

9 8

6 D

. A

n to

n io

li et

a l.

/ R

esea rch

P o licy

4 2

(2 0

1 3

) 9 7

5 –

9 8

8 Appendix C.

Table C1 Probit results for all dependents (555 firms).

Sectors ENERGY CO2 EMISSIONS EMASISO ENERGY CO2 EMISSIONS EMASISO ENERGY CO2 EMISSIONS EMASISO Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

50–99 emp. 0.062 −0.005 −0.213 0.160 0.053 0.001 −0.192 0.172 0.141 0.047 −0.155 0.199 (0.193) (0.209) (0.197) (0.205) (0.192) (0.207) (0.197) (0.207) (0.193) (0.210) (0.197) (0.207)

100–249 emp. 0.359* 0.334 0.438** 0.758*** 0.317 0.301 0.416** 0.742*** 0.317 0.267 0.394* 0.702*** (0.211) (0.223) (0.204) (0.214) (0.208) (0.218) (0.202) (0.212) (0.208) (0.219) (0.206) (0.215)

>249 emp. 0.278 0.085 0.147 0.605** 0.284 0.129 0.174 0.631** 0.245 0.057 0.168 0.578** (0.249) (0.266) (0.246) (0.249) (0.250) (0.274) (0.249) (0.253) (0.253) (0.270) (0.250) (0.254)

TECH NET 1.797*** 2.228*** 2.087*** 1.998*** 1.879*** 2.375*** 2.242*** 2.115*** 1.986*** 2.456*** 2.357*** 2.145*** (0.646) (0.658) (0.630) (0.634) (0.635) (0.655) (0.624) (0.633) (0.628) (0.653) (0.618) (0.607)

R&D 0.324 0.131 0.047 −0.268 0.430* 0.253 0.135 −0.189 0.376* 0.192 0.082 −0.230 (0.229) (0.235) (0.203) (0.196) (0.221) (0.229) (0.204) (0.196) (0.227) (0.234) (0.202) (0.194)

INTERN OPEN 0.154 0.758 0.169 1.310 0.414 1.066 0.366 1.504 0.403 0.981 0.477 1.573 (0.915) (0.937) (0.946) (1.097) (0.916) (0.943) (0.946) (1.101) (0.904) (0.928) (0.976) (1.104)

STATES OF THE WORLD TRAINCOV D/ORGPROD D TRAINCOV D/ORGLAB D TRAINCOMP D/ORGPROD D

11 −1.237*** −1.361*** −1.035*** −1.249*** −1.383*** −1.582*** −1.204*** −1.373*** −1.736*** −2.026*** −1.881*** −1.737*** (0.363) (0.401) (0.358) (0.359) (0.373) (0.407) (0.361) (0.364) (0.414) (0.466) (0.440) (0.417)

10 −1.833*** −2.125*** −1.451*** −1.556*** −1.601*** −1.716*** −1.279*** −1.441*** −1.180** −1.403*** −0.927** −1.012** (0.412) (0.444) (0.401) (0.398) (0.364) (0.395) (0.369) (0.378) (0.500) (0.538) (0.473) (0.462)

01 −1.794*** −1.930*** −1.529*** −1.674*** −2.066*** −2.332*** −1.815*** −1.906*** −1.691*** −1.845*** −1.434*** −1.563*** (0.350) (0.390) (0.331) (0.334) (0.372) (0.411) (0.347) (0.368) (0.340) (0.380) (0.329) (0.327)

00 −2.415*** −2.486*** −2.049*** −2.173*** −2.036*** −2.122*** −1.745*** −1.872*** −2.485*** −2.735*** −2.092*** −2.111*** (0.374) (0.439) (0.369) (0.372) (0.348) (0.392) (0.335) (0.334) (0.380) (0.427) (0.378) (0.373)

N 555 555 555 555 555 555 555 555 555 555 555 555 Chi2 198.39 289.94 293.36 305.04 298.86 299.5 299.48 308.10 276.02 266.42 274.46 279.82

50–99 emp. 0.091 0.027 −0.175 0.182 0.035 −0.018 −0.204 0.112 0.034 −0.001 −0.191 0.127 (0.188) (0.202) (0.193) (0.205) (0.191) (0.206) (0.192) (0.205) (0.190) (0.202) (0.191) (0.203)

100–249 emp. 0.309 0.286 0.393* 0.711*** 0.232 0.248 0.371* 0.628*** 0.215 0.232 0.356* 0.625*** (0.204) (0.216) (0.201) (0.211) (0.211) (0.224) (0.207) (0.220) (0.206) (0.219) (0.204) (0.216)

>249 emp. 0.216 0.095 0.140 0.552** 0.156 0.016 0.100 0.471* 0.183 0.060 0.123 0.513** (0.245) (0.266) (0.243) (0.251) (0.246) (0.263) (0.244) (0.250) (0.243) (0.265) (0.245) (0.250)

TECH NET 1.947*** 2.439*** 2.356*** 2.163*** 1.835*** 2.318*** 2.197*** 2.041*** 1.834*** 2.386*** 2.250*** 2.088*** (0.626) (0.628) (0.613) (0.628) (0.615) (0.632) (0.618) (0.618) (0.610) (0.627) (0.615) (0.620)

R&D 0.453** 0.268 0.168 −0.141 0.314 0.157 0.062 −0.269 0.394* 0.264 0.138 −0.208 (0.223) (0.232) (0.204) (0.198) (0.229) (0.236) (0.204) (0.197) (0.224) (0.235) (0.208) (0.200)

INTERN OPEN 0.596 1.120 0.567 1.726 0.395 0.958 0.454 1.593 0.544 1.215 0.591 1.656 (0.892) (0.908) (0.939) (1.081) (0.907) (0.919) (0.953) (1.116) (0.894) (0.921) (0.941) (1.064)

STATES OF THE WORLD TRAINCOMP D/ORGPLAB D TRAININVEST D/ORGPROD D TRAININVEST D/ORGLAB D

11 −1.539*** −1.956*** −1.620*** −1.491*** −1.402*** −1.686*** −1.293*** −1.333*** −1.480*** −1.778*** −1.342*** −1.455*** (0.411) (0.464) (0.418) (0.407) (0.345) (0.384) (0.338) (0.348) (0.354) (0.401) (0.351) (0.355)

10 −1.837*** −1.705*** −1.765*** −2.022*** −2.046*** −2.301*** −1.766*** −1.844*** −1.727*** −1.974*** −1.548*** −1.584*** (0.490) (0.515) (0.538) (0.510) (0.356) (0.400) (0.358) (0.363) (0.332) (0.376) (0.339) (0.340)

01 −1.765*** −1.945*** −1.529*** −1.680*** −1.951*** −1.882*** −1.505*** −1.955*** −2.175*** −2.667*** −1.978*** −1.880*** (0.347) (0.379) (0.331) (0.340) (0.380) (0.396) (0.339) (0.351) (0.496) (0.524) (0.447) (0.434)

00 −1.977*** −2.113*** −1.687*** −1.780*** −2.348*** −2.760*** −1.981*** −1.902*** −2.121*** −2.062*** −1.623*** −1.996*** (0.329) (0.369) (0.328) (0.327) (0.484) (0.574) (0.444) (0.425) (0.381) (0.415) (0.361) (0.366)

N 555 555 555 555 555 555 555 555 555 555 555 555 Chi2 285.51 297.88 296.33 295.70 275.4 274.81 293.13 297.70 269.84 292.11 296.31 294.28

Notes: ***1%, **5% and *10% level of significance respectively; standard errors in parenthesis.

D .

A n

to n

io li

et a

l. /

R esea

rch P

o licy

4 2

(2 0

1 3

) 9 7

5 –

9 8

8 9

8 7

Table C2 Probit results for all the dependents (polluting sectors: 192 firms).

Sectors ENERGY CO2 EMISSIONS EMASISO ENERGY CO2 EMISSIONS EMASISO ENERGY CO2 EMISSIONS EMASISO No No No No No No No No No No No No

50–99 emp. 0.358 0.199 −0.081 0.495* 0.360 0.209 −0.047 0.508 0.286 0.085 −0.166 0.460 (0.269) (0.286) (0.277) (0.296) (0.280) (0.294) (0.275) (0.313) (0.264) (0.285) (0.273) (0.296)

100–249 emp. 0.769** 0.413 0.456 0.974*** 0.778** 0.485 0.480 0.980*** 0.601* 0.228 0.286 0.888*** (0.321) (0.342) (0.315) (0.345) (0.322) (0.341) (0.312) (0.346) (0.310) (0.331) (0.311) (0.332)

>249 emp. 0.948** 0.710* 0.549 1.224*** 0.917** 0.684* 0.525 1.203*** 0.736** 0.407 0.370 1.102*** (0.383) (0.391) (0.380) (0.395) (0.388) (0.398) (0.381) (0.403) (0.368) (0.384) (0.374) (0.387)

TECH NET 0.066 0.652 0.131 0.306 0.349 1.016 0.404 0.467 0.536 1.107 0.754 0.563 (0.977) (0.996) (0.951) (0.979) (0.980) (0.991) (0.956) (0.991) (1.037) (1.066) (1.039) (1.012)

R&D 0.389 0.185 0.366 −0.002 0.463 0.198 0.462 0.056 0.369 0.114 0.358 −0.014 (0.322) (0.313) (0.315) (0.307) (0.309) (0.320) (0.308) (0.311) (0.318) (0.327) (0.319) (0.311)

INTERN OPEN 0.412 1.116 −1.126 1.175 0.850 1.427 −0.790 1.634 0.793 1.328 −0.489 1.380 (1.879) (1.914) (2.073) (1.849) (1.989) (2.031) (2.117) (1.910) (1.824) (1.895) (2.084) (1.821)

STATES OF THE WORLD TRAINCOV D/ORGPROD D TRAINCOV D/ORGLAB D TRAINCOMP D/ORGPROD D

11 −1.006*** −0.910*** −0.815** −1.191*** −1.309*** −1.349*** −1.118*** −1.338*** −1.329*** −1.300*** −1.599*** −1.388*** (0.303) (0.311) (0.326) (0.316) (0.326) (0.347) (0.337) (0.351) (0.473) (0.485) (0.537) (0.504)

10 −1.719*** −2.256*** −1.575*** −1.584*** −1.266*** −1.093*** −1.153*** −1.462*** −1.336 −0.867 −1.089 −1.214 (0.407) (0.508) (0.471) (0.392) (0.320) (0.335) (0.359) (0.341) (0.824) (0.855) (0.834) (0.857)

01 −1.679*** −1.643*** −1.279*** −1.600*** −2.262*** −2.196*** −1.716*** −2.118*** −1.273*** −1.096*** −0.964*** −1.364*** (0.313) (0.319) (0.325) (0.314) (0.398) (0.449) (0.385) (0.421) (0.275) (0.295) (0.300) (0.292)

00 −2.064*** −1.749*** −1.806*** −1.721*** −1.743*** −1.594*** −1.485*** −1.559*** −1.881*** −1.919*** −1.726*** −1.661*** (0.403) (0.427) (0.412) (0.397) (0.285) (0.318) (0.299) (0.309) (0.319) (0.404) (0.376) (0.329)

N 192 192 192 192 192 192 192 192 192 192 192 192 Chi2 75.66 78.78 66 78.52 77.89 84.27 66.94 79.62 72.94 76.03 62.94 79.46

50–99 emp. 0.272 0.131 −0.137 0.513* 0.227 0.038 −0.114 0.428 0.199 0.119 −0.072 0.441 (0.273) (0.292) (0.272) (0.306) (0.268) (0.273) (0.274) (0.298) (0.272) (0.293) (0.277) (0.304)

100–249 emp. 0.633** 0.336 0.372 0.956*** 0.547* 0.220 0.374 0.852** 0.538* 0.295 0.427 0.885** (0.315) (0.344) (0.313) (0.338) (0.323) (0.344) (0.321) (0.347) (0.321) (0.346) (0.315) (0.345)

>249 emp. 0.758** 0.538 0.423 1.130*** 0.708* 0.423 0.441 1.093*** 0.662* 0.556 0.518 1.105*** (0.373) (0.389) (0.371) (0.398) (0.366) (0.376) (0.371) (0.388) (0.372) (0.412) (0.384) (0.396)

TECH NET 0.645 1.197 0.765 0.702 0.379 0.813 0.377 0.444 0.686 1.057 0.489 0.667 (1.035) (1.052) (1.023) (1.065) (1.017) (1.042) (1.009) (1.022) (1.023) (1.027) (0.996) (1.027)

R&D 0.454 0.239 0.483 0.029 0.355 0.112 0.380 0.001 0.438 0.162 0.436 0.018 (0.308) (0.317) (0.310) (0.312) (0.315) (0.322) (0.322) (0.311) (0.309) (0.329) (0.318) (0.311)

INTERN OPEN 0.955 1.502 −0.485 1.600 0.787 1.218 −0.795 1.468 1.050 1.747 −0.661 1.516 (1.849) (1.915) (2.075) (1.829) (1.839) (1.887) (2.049) (1.847) (1.844) (2.084) (2.094) (1.857)

STATES OF THE WORLD TRAINCOMP D/ORGPLAB D TRAININVEST D/ORGPROD D TRAININVEST D/ORGLAB D

11 −1.510*** −1.622*** −1.652*** −1.327** −1.161*** −1.016*** −1.035*** −1.291*** −1.451*** −1.375*** −1.219*** −1.490*** (0.519) (0.562) (0.549) (0.543) (0.299) (0.306) (0.337) (0.308) (0.336) (0.348) (0.353) (0.354)

10 −1.318** −1.067 −1.626** −1.788*** −1.825*** −1.776*** −1.769*** −1.751*** −1.378*** −1.292*** −1.383*** −1.432*** (0.658) (0.652) (0.746) (0.651) (0.343) (0.367) (0.416) (0.331) (0.274) (0.279) (0.317) (0.273)

01 −1.533*** −1.547*** −1.256*** −1.610*** −1.631*** −1.438*** −1.113*** −1.871*** −1.520*** 0.000 −1.853*** −1.552*** (0.318) (0.336) (0.327) (0.338) (0.407) (0.400) (0.353) (0.522) (0.462) (.) (0.478) (0.459)

00 −1.530*** −1.417*** −1.348*** −1.496*** −1.657*** −1.759*** −1.524*** −1.297*** −1.827*** −1.380*** −1.185*** −1.576*** (0.250) (0.272) (0.282) (0.274) (0.426) (0.574) (0.461) (0.410) (0.397) (0.394) (0.351) (0.442)

N 192 192 192 192 192 192 192 192 192 192 192 192 Chi2 70.98 83.24 64.84 82.4 73.65 77.19 65.26 78.35 77.21 72.04 71.87 79.84

Notes: ***1%, **5% and *10% level of significance respectively; standard errors in parenthesis.

9 rch Po

R

A

A

A

A

A

B

B

B

B

B

B

C

C

C

C

C

C

D

D

E

E

E

G

G

H

H

H

H

I J

J

L

L

Wagner, M., 2006. A Comparative analysis of theoretical reasoning and empiri-

88 D. Antonioli et al. / Resea

eferences

they, S., Stern S., 1998. An empirical framework for testing theories about comple- mentarity in organizational design. NBER Working Paper 6600.

rora, H., 1996. Testing for complementarities in reduced-form regressions: a note. Economic Letters 50, 51–55.

mbec, S., Cohen, M., Elgieep, S., Lanoie, P., 2010. The Porter Hypothesis at 20: can Environmental Regulation Enhance Innovation and Competitive- ness?, Cahier de recherche no IEA-10-02, Institut d’économie appliqué. <http://www.hec.ca/iea>.

mbec, S., Barla, P., 2006. Can environmental regulations be good for business?, An assessment of the porter hypothesis. Energy Studies Review 14 (2), 42–62.

mbec, S., Lanoie, P., 2008. Does it pay to be green? A systematic overview. Academy of Management Perspectives 22, 45–62.

allot, G., Fakhfakh, F., Galia, F., Salter, A.,2011. Complementarities between Product, Process and Organizational Innovation in the UK and France. In: Paper presented at the DIME Final Conference, 6–8 April 2011. Maastricht UNU-MERIT & School of Economics and Business, Maastricht University.

loom, N., Genakos, C., Martin, R., Sadun, R., 2010. Modern management: good for the environment or just hot air? Economic Journal 120, 551–572.

ocquet, R., Brossard, O., Sabatier, M., 2004. Complementarities in organiztaional design and the diffusion of information technologies. Research Policy 36, 367–386.

rusco, S., 1982. The Emilian Model: productive decentralization and social integra- tion. Cambridge Journal of Economics 6, 167–184.

ryson, A., Forth, J., Kirby, S., 2005a. High-involvement management practices, trade union representation and workplace performance in Britain. Scottish Journal of Political Economy 52, 451–491.

ryson, A., Gomez, R., Kretschmer, T., 2005. Catching a wave: the adoption of voice and high-commitment workplace practices in Britain, 1984–1998. CEP Discus- sion Paper No. 676.

ainelli, G., Mazzanti, M., Zoboli, R., 2011. Environmental Innovations, complemen- tarity and local/global cooperation. International Journal of Technology, Policy and Management, 328–368.

ole, M., Elliott, R., Strobl, E., 2008. The environmental performance of firms. The role of foreign investments, training and experience. Ecological Economics 65, 538–546.

onti, G., 2005. Training, productivity and wages in Italy. Labour Economics 12, 557–576.

ostantini, V., Mazzanti, M., 2012. On the green and Innovative side of trade com- petitiveness? Research Policy 41, 132–153.

ozzarin, B.P., Percival, J.C., 2006. Complementarities between organisational strate- gies and innovation. Economics of Innovation and New Technology 15, 195–217.

ozzarin, B.P., Percival, J.C., 2008. Complementarities affecting the returns to inno- vation. Industry & Innovation 15, 371–392.

orenbosch, L., Van Engen, M., Verhagen, M., 2005. On-the-Job innovation: the impact of job design and human resource management through production ownership. Creativity and Innovation Management 14, 129–141.

osi, G., Malerba, F., Ramello, G., Silva, F., 2006. Information, appropriability and the generation of innovative knowledge four decades after Arrow and Nelson: an introduction. Industrial and Corporate Change 15, 891–901.

llerman, A.D., Convery, F.J., de Perthuis, C., 2010. Pricing Carbon: The European Union Emissions Trading Scheme. Cambridge University Press, Cambridge, UK.

IO, 2011. The Eco-Innovation Challenge. Pathways to a resource efficient Europe. Eco-Innovation Observatory, Annual Report, May 2011.

nnen, E., Richter, A., 2009. The whole is more than the sum of its parts – or is it? European Business school Research, paper series n.7, EBS.

alia, F., Legros, D., 2004. Complementarities between Obstacles to Innovation: Evi- dence from France. Research Policy 33, 1185–1199.

omez, J., Vargas, P., 2009. The effect of financial constraints, absorptive capac- ity and complementarity on multiple process technologies. Research Policy 38, 106–119.

ollander, U., Tarantola, S., Loschky, A., 2009. Regional Innovation Scoreboard (RIS) 2009. Technical report. PROINNOEUROPE, Bruxelles.

orbach, J., 2008. Determinants of environmental innovations, new evidence from German panel data sources. Research Policy 37, 163–173.

orbach, J., Belin, J., Oltra, V., 2011. Determinants and Specificities of EIs – An Econometric Analysis for the French and German Industry based on the Com- munity Innovation Survey. Cahiers du GREThA, 17, University Montesquieu Bordeaux.

orbach, J., Rammer, C., Rennings, K., 2012. Determinants of eco-innovation by type of environmental impact-The role of regulatory push/pull, technology push and market pull. Ecological Economics, Online First.

STAT, 2010. Istat – GeoDemo Statistics. <http://demo.istat.it/>. affe, A.B., Palmer, K., 1997. Environmental regulation and innovation: a panel data

study. The Review of Economics and Statistics 79, 610–619. affe, A.B., Peterson, S., Portney, P., Stavins, R., 1995. Environmental regulation and

the competitiveness of US manufacturing. What does the evidence tell us? Jour- nal of Economic Literature 33, 132–163.

aursen, K., 2002. The importance of sectoral differences in the application of com- plementary HRM practices for innovation performance. International Journal of

the Economics of Business 9, 139–156.

aursen, K., Foss, N., 2003. New human resource management practices, comple- mentarities and the impact on innovation performance. Cambridge Journal of Economics 27, 243–263.

licy 42 (2013) 975– 988

Leiponen, A., 2005. Skills and innovation. International Journal of Industrial Organi- zation 23, 303–323.

Kemp, R., 2010. Eco-innovation: definition, measurement and open research issues. Economia Politica 3, 397–420.

Kemp, R., Pontoglio, S., 2011. The innovation effects of environmental policy instruments—A typical case of the blind men and the elephant? Ecological Eco- nomics 72, 28–36.

Kemp, R., Pearson, P., 2007. Final Report of the MEI Project Measuring Eco Innovation. UM MERIT, Maastricht.

Kesidou, E., Demirel, P., 2012. On the drivers of eco-innovation: empirical evidence from the UK. Research Policy 41, 862–870.

Mairesse, J., Mohnen, P., 2010. Using Innovation surveys for econometric analysis. UNU MERIT working paper series n. 23.

Malerba, F., 2007. Innovation and the dynamics and the evolution of industries: progress and challenges. International Journal of Industrial Organizations 25, 675–699.

Mancinelli, S., Mazzanti, M., 2009. Innovation, networking activities and comple- mentarity empirical evidence on SME performances for a local economic system in Northern Italy. Annals of Regional Sciences 43, 567–597.

Marin, G., Mazzanti, M. The evolution of environmental and labour pro- ductivities dynamics. Journal of Evolutionary Economics, in press, http://dx.doi.org/10.1007/s00191-010-0199-8

Mazzanti, M., Zoboli, R., 2009c. Municipal Waste Kuznets curves: evidence on socio economic drivers and policy effectiveness from the EU. Environmental and Resource Economics 44, 203–230.

Mazzanti, M., Zoboli, R., 2009a. Embedding environmental innovation in local production systems: SME strategies, networking and industrial relations. Inter- national Review of Applied Economics 23, 169–195.

Mazzanti, M., Zoboli, R., 2009b. Environmental efficiency and labour pro- ductivity: trade-off or joint dynamics? Ecological Economics 68, 1182–1194.

Michie, J., Sheehan, M., 2003. Labour market deregulation, ‘Flexibility’ and innova- tion. Cambridge Journal of Economics 27, 123–143.

Milgrom, P., Roberts, J., 1990. The economics of modern manufacturing: technology, strategy and organization. American Economic Review 80, 511–528.

Milgrom, P., Roberts, J., 1995. Complementarities and fit strategy, structure and organizational change in manufacturing. Journal of Accounting Economics 19, 179–208.

Milgrom, P., Shannon, C., 1994. Monotone comparative statics. Econometrica 62, 157–180.

Mohnen, P., Roller, L.H., 2005. Complementarities in innovation policy. European Economic Review 49, 1431–1450.

OECD, 2011. Fostering innovation for green growth. In: OECD Green Growth Studies. OECD, Paris.

Osterman, P., 2006. The wage effects of high performance work orga- nization in manufacturing. Industrial & Labor Relations Review 59, 187–204.

Pekovic, S., 2011. Environmental standards and labor productivity. In: Paper Pre- sented at the EAERE Conference, Rome, 29 June–2 July.

Porter, M.E., 1991. America’s green strategy. Scientific American 264, 168. Porter, M.E., 2010. Reflections on a hypothesis: lessons for policy, research and

corporate practice. In: Presentation at the Porter Hypothesis at 20 Conference, Montreal, Canada, 28 June, available at www.sustainableprosperity.ca

Porter, M.E., van der Linde, C., 1995. Toward a new conception of the environment–competitiveness relationship. Journal of Economic Perspectives 9, 97–118.

Ramsay, H., Scholarios, D., Harley, B., 2000. Employees and high-performance work systems: testing inside the black box. British Journal of Industrial Relations 38, 501–531.

Rennings, K., 2000. Redefining innovation–eco innovation research and the contribution from ecological economics. Ecological Economics 32, 5–17.

Roberts, J., 2006. The Modern Firm. Organizational Design for Performance and Growth. Oxford University Press, Oxford.

Schmiedeberg, C., 2008. Complementarities of innovation activities: an empirical analysis. Research Policy 37, 1492–1503.

Sinclair-Desgagnè, B., 1999. Remarks on environmental regulation, firm behaviour and innovation. In: Paper presented at the workshop ‘Cost, innovation and envi- ronmental regulation: research and policy update’, Carnegie Mellon University Center, Washington, DC, April.

Teece, D., 1996. Firm Organization, Industrial Structure and Technological Innova- tion. Journal of Economic Behavior & Organization 31, 193–224.

Topkis, D.M., 1998. Super Modularity and Complementarity. Princeton University Press, Princeton, NJ.

Topkis, D.M., 1995. Comparative statics of the firm. Journal of Economic Theory 67, 370–401.

Van den Bergh, J., Stagl, S., 2003. Coevolution of economic behaviour and institutions: towards a theory of institutional change. Journal of Evolutionary Economics 13, 289–317.

cal studies on the porter hypothesis and the role of innovation. Zeitschrift für Umweltrecht und Umweltpolitik 3, 349–368.

Zwick, T., 2004. Employee participation and productivity. Labour Economics 11, 715–740.