Research Policy 41 (2012) 1770 1778
Contents lists available at SciVerse ScienceDirect
j our nal ho me p ag e: www.elsev ier .com
a r t i c l
Article history:Received 16 MReceived in reAccepted 11 JuAvailable onlin
Keywords:Clean innovatiPrivateDevelopmentDiffusionPolicy mixDemand-indu
ge chblic can tee priventionende
for ins are
pull ducinh govion acting
1. Why we need the private innovation machine for climatechange and how to turn it on
How to lenges facin(e.g. Bosetttion and adaneeds to beclean technmercializatrequire a lonthe deploymdevelopme
For cleaciently fastbe needed. and knowleprivate cleaeffective onpetition froinitial instaincentives ttechnologie
The issue is not just whether we need government interven-tion, but also how this government intervention should be designedto lower emissions at the lowest possible cost for economic
0048-7333/$ http://dx.doi.olimit climate change is one of the grand policy chal-g the world today (Stern, 2007). Simulation exercisesi et al., 2009) conrm that to keep the costs of mitiga-ptation manageable, a wide portfolio of technologies
available and used by polluters soon. Radically newologies which are not yet available or still far from com-ion will be needed to tackle climate change, but theseger-term perspective. In the shorter term, we also needent of already available cleaner technologies and the
nt of near to market cleaner technologies.n technologies to be developed and diffused suf-
and at the appropriate scale, policy intervention willIn view of the pervasive combination of environmentaldge externalities characterizing clean innovations, then innovation machine cannot be expected to be socially
its own. In addition, new clean technologies face com-m the existing more dirty technologies, who enjoy anlled base advantage. Private actors need to be providedo switch from existing dirty technologies to new cleans (Acemoglu et al., 2009).
growth.Recently developed economic models of directed technological
change (e.g. Acemoglu et al., 2009; Bosetti et al., 2009) stronglysupport the case for a portfolio of instruments including carbonprices, R&D subsidies and regulation (see Aghion et al., 2009a).Carbon prices, obtained through a carbon tax or a cap-and-tradesystem, will not only reduce the production/consumption of dirtytechnologies, they will also be important as incentive for the pri-vate sector to develop new clean technologies and accelerate theadoption of existing cleaner technologies. Expectations of futurecarbon prices and regulations are an especially important lever forprivate sector research, development and adoption of clean tech-nologies. In tandem with a sufciently high and long-term, timeconsistent carbon price as well as performance based regulation,public support for the development and adoption by the privatesector of clean technologies is needed. Public R&D support is espe-cially crucial for clean technologies which are still in the early stagesof development, neutralizing the installed base advantage of theolder, dirtier technologies. It is important that policy instrumentsare deployed simultaneously, as there are important complemen-tarities to exploit. Acemoglu et al. (2009) show that, while a carbonprice alone could deal with both the environmental and the knowl-edge externalities at the same time, using the carbon price alonewould be a more costly policy scenario, in terms of resulting in
see front matter 2012 Elsevier B.V. All rights reserved.rg/10.1016/j.respol.2012.06.012 policy instruments to induce clean inno
e Veugelers, Naamsestraat 69, B-3000 Leuven, Belgium
e i n f o
ay 2011vised form 3 June 2012ne 2012e 22 October 2012
a b s t r a c t
In view of the sizeable climate chanfull speed. Beyond the supply of pudevelopment and adoption of new cleGreen House Gas (GHG) emissions. Thchallenge. It needs government intervexternalities and overcome path depon the motives of private sector rmsinnovation survey conrms that rmtime, the high importance of demandsector agreements as drivers for introthe private innovation machine, whicbe more powerful to induce the adoptpolicy mix and time consistently, affe/ locate / respol
allenge, we need a clean innovation machine operating atlean R&D infrastructure and clean public procurement, thechnologies by the private sector needs to be assured to reduceate clean innovation machine, left on its own, is not up to this
to address the combination of environmental and knowledgencies. The rm level evidence presented in this contributiontroducing clean innovations from the latest Flemish CIS eco-responsive to eco-policy demand interventions. At the samefrom customers and voluntary codes of conduct or voluntaryg clean innovations, is a reminder of the internal strength ofernments need to leverage. Policy interventions are shown tond development of new clean technologies when designed in
future expectations. 2012 Elsevier B.V. All rights reserved.
R. Veugelers / Research Policy 41 (2012) 1770 1778 1771
lower economic growth. Similarly, when using only the subsidyinstrument, keeping the carbon price instrument inactive, wouldimply excessively high levels of subsidies compared to their levelwhen used in combination.1
Are governments deploying the right effective policies for stim-ulating clean innovations? Aghion et al. (2009b) examined therecord of government policies for clean innovation. With low,volatile and fragmented levels of carbon pricing and subsidies, theiroverall conclusion is that we are still far off from an effective policyframework capable of leveraging the power of the private sector toresearch, develop and deploy cleaner technologies.
In this contribution we look at private clean innovation in moredetail and horiented gothat differeernment poprojects, thical objectivwhole enernew low-emthe need toas well as pof new tecties associaexample of system thatgies. In ICT,supporting ponents anddevelopmechases of cstrong procenergy is a hsion oriente(2011) repodemonstratpetroleum. result in sipoor designplemented 2007).
This conture, large innovationsthis specialchange techdevelopingtransformathe developby the privat availableof this contwhich rmtions, and hhave been ipresent newule (Section
1 A way of shprice or R&D sis to express hused as a singnation. Calibra(2009a) show the rst 5 yearst 10 years.
oriented green innovation policy on how to leverage the privatesector.
2. Data on clean innovations
In this section, we take a look at data on clean innova-tions. A rst important observation to note is the poor qualityof standard data on clean innovations. No common denitionsare used, with terminology ranging from clean, sustainable pro-duction to eco-technologies, antipollution technologies and theirinnovations. While each source uses its own denitions, a com-
enominator of clean or eco-innovations is that they are at generating substantial improvements for the environ-These environmental effects include CO2 emission reductionate change, but also involve other types of pollution reduc-aste treatment as well as the environmental gains from ause of resources. Clean energy and energy efciency arer subsector of eco-innovation. In reporting data, our focusrivate rms activities in innovations for climate change,pending on data availability, we also report broader cate-
a on private R&D expenditures are not reported bylogy.2 There is only information available by the economicin which the R&D expending rms are active (NACE clas-on). es secto
(likmms is thOSTAor ofbora
thattion t wavcludl innm, one.rmaive a
e is ong rms, chemeasun mo09, a
the an-EU sn assevey ined thanergy echnorgy.previ
som on ital benect quatelye are tents easuow this process can be stimulated by a green missionvernment policy. Climate change has unique featuresntiate this challenge from other mission oriented gov-licies (Jaffe, 2011). Unlike in the Manhattan or Apolloe mission is not focused on a specic isolated technolog-e. What is needed is a pervasive transformation of thegy-economic system, mobilizing polluters to switch toission systems. It shares with health related missions
include support for development of new technologiesolicies on the demand side to accelerate the adoptionhnologies. The ICT area lacks the negative externali-ted with pollution, but it provides perhaps the nearesta similarly scaled transformation of the socio-economic
is needed to fully leverage the power of new technolo- many OECD governments played a major role beyond(mission-oriented) research through purchases of com-
systems, particularly in the early phases of technologynt. In energy, one could similarly envisage public pur-lean solutions in areas where the government has aurement interest (e.g. electric military vehicles). Thatighly specic and challenging case for designing a mis-d technology policy is demonstrated by its history. Jafferts on the 1970s large scale policy initiatives in the US toe the commercial feasibility of technologies to replaceThe large scale synfuels and related projects failed tognicant commercial outcomes and because of their
and implementation crowded out rather than com-private investments (see also Yang and Oppenheimer,
tribution will not discuss the public research infrastruc-scale public projects or public procurement for clean
(several of these topics are covered in the paper in issue by Davis). Our focus is on the part of a climatenology policy that to leverage the private sector into
and adopting clean technologies, to induce the neededtion of the energy-economic system. We look both atment and the deployment of new clean technologiesate sector. To this end, we rst provide a quick look
data on clean innovations (Section 2). The major partribution is dedicated to a micro-economic analysis ofs have been creating and/or adopting clean innova-ow strong which types of government interventionsn affecting these decisions (Section 3). To this end we
evidence from the Flemish CIS eco-innovation mod- 4). We close with some suggestions for a mission
owing the higher costs when using only 1 instrument (i.e. the carbonubsidies), rather than a combination of carbon pricing and subsidies,ow high the optimal carbon price or subsidies would have to be whenleton instrument relative to its optimal level when used in combi-ting this scenario in the Acemoglu et al. (2009) model, Aghion et al.that the carbon price would have to be about 15 times bigger duringrs, while subsidies would have to be on average 115% higher in the
mon daimedment. for climtion, wbetter a majois on pbut degories.
Dattechnosector sicaticlassimajor energyR&D insectors
A covationby EURbehaviin collawhereInnovathe lasas it inmentaBelgiumodul
2 Therexpendi(like carreliable (and eve
3 In 20nies from1000 no2010). Aeld surconrmout by eenergy tand ene
4 The providesquestionronmenmore dirunfortun
5 TherFirst, pamainly mThe EC-JRC-IPTS Scoreboard on large R&D spendersrms and their R&D expenditures on the basis of theirr of activity. Although in this classication, alternativee of the sectors considered, it fails to capture the clean
ments of rms whose major sector of activity is in othere GE or Siemens).3
only used data source for measuring private sector inno-e Community Innovation Survey, organized bi-annuallyT/OECD. The survey collects evidence on the innovative
companies, not only innovations developed in-house ortion with others, but also innovations developed else-
are adopted by rms. Unfortunately, the CommunitySurvey offers few insights into eco-innovation.4 Onlye, CIS-VI (20062008), partly overcomes this problem,es an optional one-page set of questions on environ-ovation. In Section 4 we will use these results frome of the countries that included the environmental
tion on patent applications can be used to measurectivities related to environmental protection.5 To be
nly information available by the economic sector in which the R&Ds are active. In most sectors important for greenhouse gas emissions
micals, petroleum), overall innovative activities cannot be used as are for clean innovations, as the innovations in these sectors are alsostly) related to other motives.mong the 1000 largest R&D spenders in the EU, 12 dedicated compa-lternative energy sector managed to get into the list, another 2 in thependers. They hold an R&D intensity of 3.5% (EC-JRC-IPTS Scoreboard,ssment of R&D for low-carbon energy innovations, using websites andterviews, was performed for an EU funded project (SRS (2008)) andt innovation in the energy sector may not predominantly be carriedcompanies. Industries with elevated research activities in low-carbonlogies include companies active in industrial machinery, chemicals,
ous CIS waves include evidence on motives for innovation, whiche links to environmental innovations, although very imperfect. Themproving energy efciency as an innovation motive relates to envi-ets, but does not necessarily reect an explicit green motive. Theestion on reducing environmental impact as an innovation motive,
is merged with health and safety motives.a number of limitations in using patents to measure eco-innovations.measure inventive activity, not innovations. Second, eco-patentsre identiable inventions that underlie clean product innovations and
1772 R. Veugelers / Research Policy 41 (2012) 1770 1778
Graph 1. Growcounted on thebased on the
Source: On thethe gap betwe
picked up aassociated wronmental While the Othe second aclassicatiomain categoand storagecombined c
Until thecertainly inued to growupwards. Osions and tstarted aroupward treenergy eldsince 2000.ing rates in activity. Biomal are growstage of dev
If we lopatenting,6
holds aboutized in Cleain a particuimportant pdespite its 1any single cacross variohomogeneo
end of pipe tecvations of thepatent analysi
6 Patents arwhich in most
Table 1Whos who in CET patenting?
Size Specialization ConcentrationShare of country inworld CET patents
RTA in CET patents Herndahl acrossCET technologies
TOP 6Japan 29.7% 0.99 0.72US 15.9% 0.87 0.33Germany 15.2% 1.05 0.28Korea 5.6% 1.21 0.82France
wn cbridgirld CEuntry
RTA >he sh
the indas going? econ
his pin inth rates of patents (applications) for selected CET. Notes: Patents are basis of claimed priorities (patent applications led in other countriesrst led patent for a particular invention).
basis of UNEP/EPO/ICTSD (2010). Patents and clean energy: bridgingen evidence and policy.
s an eco-patent, the environmental effects should beith a patent class linked to clean technologies or envi-
effects should be described in the patent application.ECD and WIPO use the rst approach, EPO also usespproach. Applying this approach to clean energy, EPOsn of Clean Energy Technologies (CET patents) includes 6ries: solar (both thermal and PV), wind, carbon capture, hydro, geothermal, biofuels and integrated gasicationycle (IGCC) (UNEP/EPO/ICTSD, 2010) (Graph 1).
mid-1990s, CET patents stagnated and even declined, relative terms as overall patenting activities contin-. But since the late nineties, CET patents have trendedne cannot ignore the correlation between political deci-he take-off of CET technologies, as the upward trendund 1997, when the Kyoto protocol was signed. This
Source: Oenergy: 2% of woof the copatents;sum of tweightsand 1 (p
of CET CET pa
Thiciationat whewhich tion ofAsia intant hpatentmicro-cies innew anthe latanalys
As tpolicy nd holds particularly when compared to the traditionals (fossil fuels and nuclear) which have trended down
When looking at individual CET technologies, patent-solar PV, wind and carbon capture have shown the mostfuels is a more recent growth story. Solar- and geother-ing more slowly if at all, reecting their still prematureelopment.ok at which countries are active in clean energyJapan is the clearest positive outlier (Table 1). Japan
30% of all CET patents, but it is not particularly special-n Energy Technologies, and it is heavily concentratedlar CET technology, namely solar PV. Korea is anotherlayer in CET patenting, specialized in solar PV. The US,6% share of world clean patents, is not specialized inlass of Clean Energy Technologies. It is more dispersedus CET technologies. If the EU were to be counted as aus block, it would be the block with the largest share
hnologies, whose environmental impacts are specic aims and moti- inventions. For other types of innovation, such as process changes,s is less useful because many of these innovations are not patented.e assigned to countries on the basis of the location of the assignee,
cases is a corporation.
clean innovsion orientThe evidencscale publicanyway scaOppenheim
The resutention thathe privategies. Jaffe ethe empiricheavily.
Most of tof environmnologies, uis Lanjouwbetween thronmental p
7 Also some(RTA > 1), but of size): NetheHungary 18.104.22.168% 0.70 0.263.6% 0.98 0.28
32.0% 1.01 0.25
alculations on the basis of UNEP/EPO/ICTSD (2010). Patents and cleanng the gap between evidence and policy; a Top 6 country has at leastT; together the Top 6 represent 74% of world CET patents; RTA = share
in world CET patents relative to the share of the country in total world 1 measures specialization in CET patents; Herndahl is the weightedare of each CET technology in total countrys CET patents, with the
the share. The Herndahl ratio varies between 0 (maximal dispersion) concentration).
nts. In Europe, Germany is by far the largest country for.7
ck glimpse at the data is suggestive of a positive asso- active government policy, particularly when lookingan energy patents have taken off, i.e. post-Kyoto, andtries are most active, i.e. with the relatively weak posi-
US and the relatively strong position of the EU andividual classes of CET technologies. But how impor-vernment policy been in explaining patterns in CETIn the next sections we examine in more detail, usingomic evidence, the effectiveness of government poli-ivating private clean innovations. Before we presentis using the Flemish eco-innovation module results fromIS-VI survey Section 3 presents a review of existing
e on the impact of government policies to inducean innovations
aper focuses on the role of climate change technologyducing the private sector into developing and adoptingations, this section accordingly does not look at mis-ed government policies to supply clean technologies.e on the impact of public research infrastructure, large
projects or public procurement for clean innovations isnt and not very encouraging (e.g. Jaffe, 2011; Yang ander, 2007).
lts of an extensive empirical literature support the con-t environmental policies do succeed in incentivizing
sector to develop and adopt new clean technolo-t al. (2002) and Johnstone (2008) provide reviews ofal literature, on which the following paragraphs draw
he existing empirical studies concentrate on the impactental policies on the creation of new clean tech-
sing patents as an empirical proxy. An early paper and Mody (1996), who examined the relationshipe number of environmental patents granted and envi-olicy for a series of countries for the period 19711988.
other EU countries are specialized in environmental technologiesare nevertheless small players (
R. Veugelers / Research Policy 41 (2012) 1770 1778 1773
They use expenditures for pollution abatement as a very indi-rect and imperfect measure of environmental policy stringency,and nd that pollution abatement costs affect the number ofpatents successfully granted. Unfortunately, their study failed tocontrol for oJaffe and Paincorporatital innovat19771989environmention AbatemThey measupatents, witonly clean stringency sectors, but
Brunnerwork, narroAs policy inand the nututions. Covariable hasronmental Taylor et afur dioxidegas desulfuthey foundregulation lation, Poppwell as in twhether thdiffusion.
More redirty innovas instrumecles to metechnologiethat fuel prhave a greaThis is conto break rgies.
Very fewtive behaviPopp (2003able permitAmendmenfurization. of the tradprevious teevidence thronmental eintroduced lead to incr
The empmeasures, and preferaregulation, different caefuent frosions; and dthe case sturonmental policy instrnologies cothe instrum
energy, the implementation of different policy measures had ameasurable impact on innovation, with tax measures and quotaobligations being statistically signicant determinants of patentactivity. However, the effect of the different policies varied by
pe od R&ergyvatiices pos, thear envatis, puhe deuctioe an t rep
folloat ovtors,es. Qdabltionss invffect
cleant tes bylogieener. A sutivesplyinnd t
ying are: t ang exptintionstionstions
neelso ) o
rall, not tionsatinge ofn tecnd tons vectivnationnovenced pubate ther determining factors. Using US industry-level data,lmer (1997) extended Lanjouwand Modys study, byng various factors that potentially affect environmen-ion. Focusing on US manufacturing industries during, Jaffe and Palmer examined the relationship betweental stringency(as measured by higher level of Pollu-ent Costs and Expenditures (PACE)) and innovation.red innovations in terms of both R&D expenditures andhout restricting the investment or patent data to coverinnovations. They found that increased environmentaldoes increase R&D expenditures in US manufacturing
not the number of patents.meier and Cohen (2003) built on Jaffe and Palmerswing innovation to purely environmental patents.dicators, they used pollution abatement costs (PACE)mber of inspections undertaken by regulatory insti-ntrary to Jaffe and Palmer, they found that the PACE
a statistically signicant (and positive) effect on envi-patents, where as subsequent monitoring does not.l. (2003) studied the time path of patents in sul-
(SO2) control, especially activities related to uidrization. Analyzing a100 year time span (18871995),
that more patent applications were led after SO2was introduced in the 1970s. In addition to SO2 regu-
(2006) also examined NOX regulation in the US, ashe German and Japanese electricity sectors, exploringese regulations affected (inter) national innovation and
cent work by Aghion et al. (2010) studied clean versusations in the auto industry, using (taxes on) fuel pricesnt for government policies, patents for electric vehi-asure clean innovations and patents for combustions to measure dirty innovations. Their results conrmices drive clean patents more than dirty patents, butter effect on rms with a larger stock of dirty patents.sistent with the need for government interventionm-level path dependency in the old dirty technolo-
studies have compared the effects on rms innova-or of different policy instruments. Using patent data,) examined the effects of the introduction of the trad-
system for SO2 emissions as part of US Clean Air Actts on the technological efciency of uidgas desul-Comparing patent applications after the introductionable permit scheme with those submitted under thechnology-based regulatory system, Popp (2003) founde while early regulations did little to improve the envi-ffectiveness of the technologies developed, innovationsafter the introduction of the tradable permit scheme dideased environmental effectiveness.irical evidence with respect to the use of other policyparticularly subsidies for environmental R&D, solelybly in combination with market based instruments andis even more limited. Johnstone (2008) reports on threese studies: abatement technologies for wastewaterm pulp production; abatement of motor vehicle emis-evelopment of renewable energy technologies. Overall,dy evidence is supportive of the argument that envi-policy affects technological innovation, but whether aument is effective or not varies across the clean tech-nsidered and at which phase of the technology life cycleents was used. For instance, in the study on renewable
the tytargeteable enof innofuel prnot in energyfor solto innoprocesulate tintrodto havdid nograntsregula
In arm thinnovanologiand trainnovasuch amost edue toR&D su
Govof newimportnologitechnoof the changeon moof comvey foucomplThese producsecurinfor adoregulainnovainnovato thewere arelatedrm.
Oveture isinnovastimulthe typof cleasolar) aemissifor effcombiclean iof eviding anfor climf renewable energy involved. Public expenditures onD were statistically signicant for every type of renew-. Relative prices are found to induce particular kindson. In the case of motor vehicle emissions abatement,encouraged investment in integrated innovation, butt-combustion technologies. In the case of renewable
role of electricity prices was rarely signicant, exceptergy. Other market factors can also be important spurson. In the case of bleaching technologies in the pulpingblic concerns about the environment appeared to stim-velopment of new cleaner technologies, pre-dating then of regulatory standards. Eco-labeling did not appearinuence on innovation in this case. The case studiesort on any possible complementarity between R&D
demand-side policies such as carbon pricing and/or
w-up econometric analysis, Johnstone et al. (2010) con-erall environmental policies signicantly affect private
although the strength of the effects varies over tech-uantity-based policy instruments such as obligationse certicates were found to be most effective in inducing
in wind power technology. Price-based instrumentsestment incentives, tax measures and tariffs provedive in encouraging innovation in solar. Unfortunately,igh correlation with the intercept, the effectiveness ofrt programs could not be isolated.ent policy is not only important to induce the creation
aner technologies, as measured by patents. It is alsoo drive the adoption of already developed clean tech-
rms. Far fewer studies exist on the adoption of cleans, although it is an important part of the transformationgy-economic system that is needed to tackle climatervey of rms in eight sectors in ve European countries
for adopting eco-innovations conrms the importanceg with regulations (Arundel et al., 2009). But the sur-hat there are many more important reasons besideswith regulations for introducing an eco-innovation.improving the rms image, reducing costs, and, ford service innovations, demand pressure (measured byisting markets and increasing market share as motiveg clean technologies). Compliance with environmental
was more important for adoption of pollution control than for the other types of eco-innovation. Process and recycling were often introduced in responsed to comply with regulations, but many of themintroduced to obtain cost savings (not environment-r to improve the environmental image of the
the econometric evidence from the economics litera-unfavorable for the impact of clean policies on clean. But it also highlights that policies are no panacea for
clean innovations. Although the evidence suggest that policy instrument (e.g. tariffs versus grants), the typehnology targeted (e.g. motor vehicle emissions versushe nature of the environmental effect (e.g. reducing CO2ersus enhancing energy efciency) all seem to mattereness, we still have a very incomplete view on whichn of policy instruments is most effective in stimulatingation creation and diffusion. This comes on top of a lack
on the effectiveness of public R&D infrastructure build-lic procurement as parts of the technology policy mixchange.
1774 R. Veugelers / Research Policy 41 (2012) 1770 1778
4. Micro evidence from Belgian rm level CIS-VI data on theimpact of government policies to induce clean innovations
This section discusses new evidence, drawn from the FlemishCIS eco-innovation module, on which rms have been developingand adopting clean innovations, and which types of governmentinterventions have been most effective in inuencing these pri-vate decisions. EUROSTAT/OECD introduced in the 6the CommunityInnovation innovation including ginnovationsclean goverclean innovalso allows of the rmsa cross sector panel.
Belgiuminnovation investigate vation behaits Innovatiits ambitiontion. To thiof strategic governmenheads/clusEnergy andsmart gridsincentives dies for R&Dintroduced,the policy iincentives aefcient solclean technerative R&Drenewables
We will nenergy techernment. Wof instrumeand adopt ceffects on voluntary sinduced.8 Whypotheseseco-innovasenting theinterventio
4.1. The CIS
In the CIrespondentmore envirbenets are
Six types of the inn
8 Croci (200ments (such acase studies. B
lower use of materials (ECOMAT), lower energy use (ECOEN),lower CO2 emissions (ECOCO), less use of pollutants (ECOPOL),less pollution of soil, water, air or noise control (ECOSUB),recycling (ECOREC).
Three typuse of the
nt EG),ctedEGF)ts, in
for eing ovationtaryts to
s secos, inhe l
e subleanlic reivateces oh setrmaforts
formtestimediel eth a re
in tEU-Ely 6 cecte are(e.g. . It is
wite poations in
h allllow Survey (CIS), covering the period 20062008, an eco-module. This module surveyed rms on what factors,overnment policy, affected the introduction of eco-. These data thus allow us to examine the impact ofnment policy on the likelihood of rms introducingations. As the module is part of the larger CIS-survey, itus to control for other innovation related characteristics
surveyed. The disadvantage of the CIS data is that it ision for a single period of time rather than a time series
was one of the countries that included the eco-module in its CIS, and we use the Flemish data tothe impact of government policies on rms clean inno-vior. Flanders is one of the three Belgian regions. Inon Communication for 20092013 Flanders expressed
to be part of the top 5 regions in European innova-s end it has developed a New Industrial Policy typeframework (Flanders in Action) coordinating existingt budgets, instruments and stakeholders around spear-ters and large projects. One of these spearheads is
Environment, with large scale innovation projects in and electric vehicles. In addition, Flanders providesfor adoption of clean technologies as well as subsi-
generally. Environmental regulations have also been many of which originate at EU level. In addition tonstruments at the Flanders regional level, there are taxnd nancial support for the adoption of clean or energyutions at the national (Belgian) level. The EU also has aology policy that includes a mixture of support for coop-, energy efciency regulations, carbon pricing (ETS) and
targets (see Veugelers, 2011).ot be evaluating the individual instruments of the cleannology policy framework of these three levels of gov-e instead look only at the extent to which the packagents has affected the private sectors incentives to createlean innovations. The data only allow us to separate therms behavior of subsidies, regulations and taxes, andectoral agreements, many of which are government-e rst briey discuss the data (Section 4.1), the research
(Section 4.2), descriptive statistics on who introducestions (Section 4.3) and why (Section 4.4), before pre-
econometric results on the importance of governmentn for clean innovation (Section 4.5).
-VI eco-innovation data
S-VI eco-innovation module, a rst set of questions askss if they have introduced an innovation with one oronmental benets (ECO). Nine types of environmental
of environmental benets that can occur during the useovation by the enterprise (ECOOWN):
5) present an analysis of examples of environmental voluntary agree-s benchmarking convenants on energy efciency) through a series ofenchmarking convenants are commonly used in Flanders.
Whof ecoare as(ECOU
A sintroduintroduset of f
Thipolicietives. Tincludits for cof publate prinuen
Botno infotion efsimplenitive small, (Arund
Witperiodclaim R&D acsample
Webershipof the ior. Onmay reschemactive surveysystemon largobservon rm
Witdata aes of environmental benets that can occur during the innovation by the end user (ECOUSER):rgy use (ECOENU), less pollution (ECOPOLU), recycling).
e rst six types of benets ow from the adoptionvations (ECOOWN), the other three types of benetted with the rms development of eco-innovations.d set of question asks about different motives for
environmental innovations. It asks whether the rm environmental innovations as reaction to the followings:
environmental regulations or environmental taxes
environmental regulations or environmental taxes,cluding R&D subsidies, or other public nancial incen-nvironmental innovations (ENGRA),r expected demand from customers for environmentalns (ENDEM), and
codes of practice used in the sector or sectoral agree-stimulate eco-friendly practices (ENAGREE).
nd set of questions covers a wide range of governmentcluding regulations, taxes and public nancial incen-atter can include R&D subsidies, but these could alsosidies for the adoption of clean technology and tax cred-
innovations. The list of policies does not include the usesearch infrastructure or public procurement to stimu-
eco-innovations, although these also may be importantn rm-level innovation.s of questions are asked on a simple yes or no basis, withtion on the relative size of the environmental innova-, nor on the relative importance of specic policies. Theat of the questions resulted from two rounds of cog-
ng with the managers of 20 enterprises, representingum and large rms in eight EU countries and six sectors
al., 2009).sponse rate of 44%, the Flemish CIS-VI data covering the62008, holds 2963 observations. Of these rms, 42%
active in innovation, 33% are engaged in intra-muralies and 10% have applied for at least 1 patent. 88% of thes are SMEs, and 45% are in the service sector.ed the Flemish rms in the CIS-data to data on mem-he EU Emission Trading System to test for the impactTS scheme on rms environmental innovative behav-ompanies in the CIS also are in the ETS scheme, which
the fact that a large number of Flemish rms in the ETS not in the CIS-sample because they are not innovationrms in building material) or did not respond to the CIS
also a reection of the highly targeted nature of the ETSh respect to sectors and rms, concentrating as it doeslluting sectors and rms. Due to the small number ofs, we are unable to evaluate the ETS schemes inuencenovative behavior.
the caveats in mind, the Flemish CIS-VI eco-moduleus to test the impact of clean government policy on
R. Veugelers / Research Policy 41 (2012) 1770 1778 1775
Table 2Which sectors introduce clean innovations?
ALL FOOD CHEM ELEC AUTO NUTS OTHER MANUF TRANS PORT OTHER SERV
ECOCO 20 23 35 19 17 48 17 34 13ECOOWN 8 46 46 30ECOUSER 6 25 27 24ECO 2 49 48 33
the likelihoor elsewheinvestigate ments.
We are pernment i(ENGRA)
We will aeffect from
Comparinversus ENof manag
As the CIto test for a
When tefor the intrdetermininclean innovimportanceagreementsistics that min general, lWe also conthe rm hamarketing information
4.3. Who in
Overall 4they introd43% have in(ECOOWN)oped cleanclean develicantly moroperations:innovation
If we losimilar pensaving (ECOpollution (E
Companbenets frothat reduceenergy consversa 67% oalso develo
For all threductions more generand clean in
Not surphighest on services sec
F NT AND
nspoTableEs arnter2 redl disthat ocesso intpers
montrodary agreements between regulators and polluters Govern-regulations and taxes are mentioned by almost one thirdco-innovators as motive. Financial incentives (grants) arened least frequently, in only 15% of the cases, although theyewhat more inuential for CO2 and energy saving innova-urrent and future regulations and taxes are more inuential
2 reductions than for energy saving. This and other ndingsect the fact that CO2 reduction innovations are relatively-intensive, increasing the inuence on adoption decisions ofal factors (Table 4).
large enterprises, regulations are signicantly more impor-rivers in development or adoption of environmentallogies and appear to be almost as important as voluntaryents, perhaps reecting the tendency of government policys on larger rms. Financial grants also are more inuential inolvement by large companies in eco-innovations. One thirdrge enterprises listed grants as a factor having induced theirnovation activities.43 47 65 52 59 628 26 41 48 41 446 48 67 59 64 7
od of rms introducing clean innovations, either ownre developed. More specically, these data allow us toand compare the effects of different clean policy instru-
articularly interested in analyzing which types of gov-nterventions, if any, are most inuential: R&D subsidiesor environmental regulations and taxes (ENREG).lso investigate whether there is any complementary
both types of instruments.g the effect of current and expected regulations (ENREGREGF) allows us to analyze the impact on rm behaviorers perceptions of the long-term consistency of policy.
S data are not structured as panel data, it is not possibleny path dependency or other dynamic effects.sting the importance of clean government instrumentsoduction of clean innovations, we control for otherg factors. First we include other potential inuences onations that are not related to government policies: the
of demand from customers (ENDEM) and voluntary (ENAGREE). Secondly, we control for rm character-ay inuence a rms decision to introduce innovationsike rm size, rm age and technology/sector of the rm.trol for the innovative prole of the rm, i.e. whethers introduced other product, process, organizational orinnovations. The main modules of CIS-VI provide the
for these controls.
troduces clean innovations?
6% of the total 2894 rms in the sample respond thatuced a clean innovation in the period 20062008 (ECO).troduced a clean innovation in their own operations
. These are the clean adopters. 8% report having devel- innovations for their users (ECOUSER). These are theopers. Note that the clean developers are also signif-e likely to introduce clean innovations in their own
of all ECOUSER rms, 85% have also introduced a cleanin their own operations.ok at the different types of clean innovations, we seeetration rates. Reducing CO2 (ECOCO) holds 20%, energyEN) 22%, reducing pollutants (ECOPOL) 21%, reducingCOSUB) 22%.ies also report a mix of different types of motives for orm environmental innovations. For instance, those rms
CO2 are also signicantly more likely to reduce theirumption: 72% of ECOCO rms also are ECOEN and vicef ECOEN rms also report ECOCO; 63% of all ECOCO rmsp clean innovations for their users (ECOUSER).ese reasons, although our prime area of interest is CO2(ECOCO), we will also look at clean process innovators
Table 3Which t
Table 4Which m
The trations (
SMlarge efor COsectoraFirms uct, prlikely tdeveloshows
Thetheir ivoluntment of all ementioare somtions. Cfor COmay recapitalnanci
Fortant dtechnoagreemto focuthe invof all laeco-inally (ECOOWN) as well as clean developers (ECOUSER)novations overall (ECO) (Table 2).risingly, the chemicals and the utilities sectors scoreclean innovations, particularly for CO2 reductions. Thetors are less likely to be involved in clean innovation.
9 The paperbecause of theufacturing. rms introduce clean innovations?
ALL SMEs Large YOUNG INNOV PROCESS INNOV
20 17 43 20 30 3643 40 71 38 60 6928 26 46 28 42 4546 44 76 38 66 72
s for clean innovations?
ECO ECOCO ECOOWN ECOUSER ECOEN
32 42 33 34 3825 37 37 29 3215 22 15 18 2121 29 22 29 2839 51 41 45 50
rt sector is over represented in terms of clean innova- 3).9
e less likely to introduce clean innovations compared toprises. This difference in rm size is most pronounceductions. This is perhaps related to the supra reportedtribution, with CO2 sensitive sectors being larger scale.are active in introducing other innovations, be it prod-, organizational or marketing innovation are also moreroduce clean innovations. With the exception of clean
(ECOUSER), the introduction of process innovationstrongest complementarity with clean innovations.
s for clean innovations
tive most frequently identied by rms as leading touction of eco-innovations is participation in sectoral industry also is over represented in eco-innovations (54%), but low number of observations, this sector is included in other man-
1776 R. Veugelers / Research Policy 41 (2012) 1770 1778
Table 5Which motives for whom?
ALL ECO SMEs LE FOOD CHEM ELEC AUTO NUTS OTHER MANU TRANS PORT OTHER SERV
ENREG 32 28 61 52 45 43 25 28 28 32 22ENREGF 25 22 49 30 41 28 21 33 21 28 20ENGRANT 15 14 33 20 21 12 19 17 13 23 10ENDEMAND 21 19 38 15 28 31 23 28 17 13 27ENAGREE 39 35 67 44 51 47 50 39 35 34 35
CURR & FUTURE 19 16 43 25 34 24 17 19 16 21 13REG & GRANT 10 9 27 16 17 10 4 11 8 14 7
Governmanother in lations andto rate granrms that rlations as imwith compl
The timanalyzed byventions. Wand taxes atant inuenboth currenmotive corregulationsregulationsover time, r
The secttaxes is thereporting tsector has anticipatedtor, voluntaGrants are ithey are somfood and ca
4.5. Economclean innova
In the laernment poare particulicy instrumcontrolling tary agreemcharacteristtive inputs also separattions (ECOO(ECOUSER).particularly(ECOCO) an
The resu(ENDEM) antant motiveregulators types of ec
10 All cross-m11 The Chi-2 dence at the 112 The Pearsoicant at the 1%
l less important drivers for eco-innovations compared tod-pull and voluntary agreements. Regulations and taxes) are more important to incite the adoption of clean tech-
es (ECOOWN) than for the development of clean technologiesSER). Government subsidies (ENGRA) are less importantces.en we examine the adoption of eco-innovations, we com-ose aimed at reducing CO2 emissions (ECOCO) and thosece energy consumption (ECOEN). Voluntary agreements aree most important factor for inducing ECOCO and ECOEN. Fortions that reduce energy consumption, nancial incentiveso be the most potent government lever. For innovationsat reducing CO2, anticipated future regulations and taxes
strongest policy inuence, corroborating the importance ofrm policies in the implementation of CO2 innovations.able 7 we test the importance of consistency in government
We do this by identifying 3 cases:
current and future regulations and taxes are drivers, current regulations and taxes are drivers, and future regulations and taxes are drivers.
sults on the drivers for introducing eco-innovations.
ECO ECOUSER ECOOWN ECOCO ECOEN(1) (2) (3) (4) (5)
e (Ems incluport ote pro
includ It shoent policies are signicantly correlated10 with oneaffecting rms behavior. Companies which rate regu-
taxes as decisive motive are signicantly more likelyts as decisive (29% compared to 9%) and vice versa,ate grants as important are more likely to rate regu-portant (60% compared to 27%).11 This is consistent
ementarity between government instruments.e-consistency of government interventions can be
comparing the impact of current and/or future inter-hile 12% of eco-innovators list only current regulationss inuential and no more than 6% cite as an impor-ce only future regulations, 19% of eco-innovators listt and future regulations as decisive. Among the cross-relations, the correlation between current and future
is the strongest.12 This is consistent with the idea that and taxes will be more inuential if they are consistentather than subject to instability.or that appears to be most sensitive to regulation and
food sector, with more than half of the companieshis factor (ENREG) as inuential. Also the chemicalsa high sensitivity to regulation and taxes, including
regulations and taxes. In the car manufacturing sec-ry agreements are the most frequently reported drivers.n no sector among the most important drivers, althoughewhat more decisive in transport services, chemicals,
r manufacturing (Table 5).
etric analysis of the impact of government policies ontions
st part of the analysis, we examine the impact of gov-licy on clean innovations in a multivariate analysis. Wearly interested in which of the direct government pol-ents are more important drivers for clean innovations,for other motives: demand push (ENDEM) and volun-ents (ENAGR) and controlling for rm and industryics (rm size and age, sector dummies and innova-of the rm). We look at eco-innovations in general, butely consider eco-innovations introduced in own opera-WN) and the development of eco-innovations for users
For innovations introduced in own operations, we are interested in those aimed at reducing CO2 emissionsd reducing energy consumption (ECOEN).lts reported in Table 6 indicate that demand-pulld voluntary agreements (ENAGR) are the most impor-s for ECO innovations. Voluntary agreements betweenand polluters are the most inuential driver for all
Whpare thto redustill thinnovaseem taimed are thelong-te
both only only
Table 6Probit re
Firm sizdummieA full reA bivariaaccountcant corWe alsovations.o-innovations. Government policy instruments are in
otive correlations are statistically signicant at the 1% level.test, testing independence between both variables, rejects indepen-% level.ns correlation coefcient between REG and REGF is 0.555 and signif-
Inclusion of thN = 2893.Our interest isNote: Margina(Dprobit (robusignicance le*** Represent** Represent+ Represent.302 .079 .306 .082 .058
.036 .034 .038 .029 .028*** ** *** *** **
.266 .130 .217 .146 .075
.050 .039 .051 .035 .032*** *** *** *** ***
.218 .095 .098 .087 .115
.061 .043 .060 .034 .036*** ** + *** ***
.315 .279 .186 .119 .126
.043 .038 .049 .032 .032*** *** *** *** ***
.380 .237 .421 .211 .225
.026 .029 .027 .027 .027*** *** *** *** ***
p06), rm age (Young), innovative inputs (InnovIn) and sectorded:f all coefcients is reported in Appendix A to this paper.bit for estimating (ecouser, ecoown) and (ecoco, ecoen) to take intorrelation between the dependent variables, cf supra, yields a signi-n among error terms, but gives very similar results.e the ETS dummy although it is highly skewed, covering only 7 obser-ws a high responsiveness for innovations to reduce CO2 emissions.
e ETS dummy does not affect the other coefcients.
in comparing the coefcient size across government policies.l effects reported (discrete change of dummy variables from 0 to 1)st) command in STATA). Standard errors reported in the second line,vels in the third line.s 1% signicance level.s 5% signicance level.s 15% signicance level.
R. Veugelers / Research Policy 41 (2012) 1770 1778 1777
Table 7Bivariate probit results on the importance of time consistency of clean policies.
ECO ECOUSER ECOOWN ECOCO ECOEN(1) (2) (3) (4) (5)
ENREG and E
Firm size (EmpENDEM, ENAGMarginal effec(robust) commare signicantgovernment p
Table 8Bivariate prob
Firm size (EmpENDEM, ENAGMarginal effec(robust) commare signicantgovernment p
For innositivity to ocombinatioulations anat reducingdriver thaninnovationsthe argumemakes themtions.
In Tableregulationssubsidies anonly subsidare present
For the afor reducingcombinatioential, in codevelopme
5. Conclusprivate clea
The climclean innovto publiclyment, the dthe private energy-eco
emissions. The private clean innovation machine, left on its own,is not up to this challenge. Government intervention is needed toaddress the combination of environmental and knowledge exter-nalities and to overcome path dependencies. The evidence on
rms for-innlicy ant t
evidning on ofbetw-redut foridenducinse gom thre wtrum.
inte-innoNREGF .357 .160 .360 .226 .112.041 .040 .043 .038 .035
.386 .181 .411 .139 .121
.032 .044 .038 .038 .037
.439 .343 .426 .264 .202
.030 .061 .048 .060 .058
06), rm age (Young), innovative inputs (InnovIn), sector dummies,R included. N = 2893.ts reported (discrete change of dummy variables from 0 to 1) (Dprobitand in STATA). Standard errors in the second row. All coefcients
at 1% level. Our interest is in comparing the coefcient size acrossolicies.
it results on policy mixing regulations and taxes with subsidies.
ECO ECOUSER ECOOWN ECOCO ECOEN(1) (2) (3) (4) (5)
d .371 .196 .340 .283 .253.046 .053 .056 .051 .051
DIES .343 .312 .265 .183 .158.045 .065 .058 .054 .055
AX .445 .249 .448 .214 .138.023 .031 .028 .023 .029
06), rm age (Young), innovative inputs (InnovIn), sector dummies,R included. N = 2893.ts reported (discrete change of dummy variables from 0 to 1) (Dprobitand in STATA). Standard errors in the second row. All coefcients
at 1% level. Our interest is in comparing the coefcient size acrossolicies.
ult being no impact from neither current nor future nor taxes.
currengoverntion minnovaa well-sistentfundin
Thedata angovernalbeit techno
Thecombiadoptitarity of CO2accounrent evCO2 refor thoprograin futuicy inschange
Theof ecovations aimed at reducing CO2, we nd the highest sen-nly future regulations and taxes, closely followed by an of current and future regulations. Only current reg-d taxes has the lowest impact. For innovations aimed
energy consumption, future regulations are a stronger current ones. The same holds for development of eco-
for users. We interpret this as evidence in support ofnt that the long-term nature of regulations and taxes
especially inuential in the adoption of clean innova-
8 we test the combined inuence of subsidies and and taxes, again by constructing 3 cases: one whered regulations and taxes are jointly present, one where
ies are present and one where only regulations and taxes.doption of eco-innovations aimed at reducing CO2 and
energy consumption, the results clearly suggest that an of regulations and taxes with subsidies is most inu-ntrast to the limited effects of this combination on thent of eco-innovations.
ions: linking clean government policies ton innovations: does it work?
ate change challenge can be met effectively only by aation machine that is operating at full speed. In addition
support for R&D infrastructure and public procure-evelopment and adoption of new clean technologies bysector is essential to the needed transformation in thenomic system for reductions in Green House Gas (GHG)
innovationsicy intervendevelopmeible and cothan curren
Finally, tvoluntary sas drivers fthe internaments shoupolicy desig
This papfor activatiinnovationsof a missionNonethelesoriented clebe compatiif the resuladopted byenergy-eco
The auththe CIS-VI dfully acknowD. Foray anhelpful.n innovation performance hints at the failure so far oft intervention to fully activate the private clean innova-e. If governments want to leverage the needed private
for clean energy technologies, they will have to providegned, time consistent policy, by a combination of con-on pricing, performance based regulations and public
ence from earlier micro-econometric studies, on patentom selected environmental technologies conrms thatt intervention can affect private sector innovations,
substantial variation among policy instruments ands.-level evidence presented in this contribution on the
introducing clean innovations using the latest Flemishovation survey conrms that rms are responsive tointerventions. The evidence is also suggestive of howhe details of the policy design are.ence supports the increased leverage of policies whenregulations and taxes with subsidies, particularly for the
innovations to reduce CO2 emissions. This complemen-een policy instruments for accelerating the adoptioncing innovations is particularly important to take into
the design of public clean subsidy policies, as the cur-ce provides little support for the efcacy of subsidies forg innovations, when used in isolation. It is a remindervernments contemplating a public clean R&D supportat the lack of a strong carbon price expected to prevailill seriously reduce the effectiveness of subsidies as pol-ent to leverage private innovative incentives for climate
rtemporal consistency of policy is relevant to all typesvations, but especially important for climate change
and more so for developers than for adopters. Pol-tions will have greater inuence on the adoption andnt of new clean technologies when designed to be cred-nsistent over time, affecting future expectations moret incentives.he high importance of demand pull from customers andectoral codes of conduct or voluntary sector agreementsor rms introducing clean innovations, is a reminder ofl strength of the private innovation machine. Govern-ld leverage this power, by a time consistent clean-techn affecting the expectations of the market.er has focused on the design of a clean tech policyng the private sector in creating and adopting eco-, and accordingly has fewer implications for the design-oriented public R&D infrastructure policy in this eld.s, this discussion underscores the need for any mission-an public R&D infrastructure policy to be designed toble with and supportive of a clean tech demand policy,ts of this public R&D are to be further developed and
the market to ensure the needed transformation of thenomic system.
or would like to thank ECOOM for providing access toata. Financial support from FWO/KUL and SEEK is grate-ledged. Comments from A. Jaffe, R. Nelson, D. Mowery,
d the participants in the Lausanne workshop were very
1778 R. Veugelers / Research Policy 41 (2012) 1770 1778
See Table A.1.
Table A.1Probit results on the drivers for introducing eco-innovations.
ECO ECOUSER ECOOWN ECOCO ECOEN(1) (2) (3) (4) (5)
Firm size .00025** .00004 .00023 .00014 .00011(Empl06) .00012 .00003 .00010** .00004*** .00004***
Firm age .033 .010 .042 .010 .023(DYoung) .029 .023 .029 .020 .019Innov inputs .236*** .164 .198 .057 .135(InnovIN) .022 .020*** .024 .017*** .018ENREG .302 .079 .306 .082 .058
.036*** .034** .038*** .029*** .028**
ENREGF .266 .130 .217 .146 .075.050*** .039*** .051*** .035*** .032***
ENGRA .218 .095 .098 .087 .115.061*** .043** .060+ .034*** .036***
ENDEM .315 .279 .186 .119 .126.043*** .038*** .049*** .032*** .032***
ENAGR .380 .237 .421 .211 .225.026*** .029*** .027*** .027*** .027***
FOOD .047 .032 .116 .049 .193.042 .036 .041*** .032* .038***
CHEM .179 .024 .243 .085 .174
Note: Margina(Dprobit (robu*** Represent** Represent* Represent+ Represent
Arundel, A., Kmethods fEnabling G
Acemoglu, D., Aghion, Ph., Bursztyn, L., Hemous D., 2009. The environment anddirected technological change, MIT, Harvard, NBER working paper (downloadedon 17 September 2012 from ).
Aghion, Ph., Hemous, D., Veugelers, R., 2009a. No Green Growth Without Innovation.Bruegel Policy Brief, 2009/07, Bruegel, Brussels.
Aghion, Ph., Veugelers, R., Serre, C., 2009b. Cold Start for the Green InnovationMachine. Bruegel Policy Contribution, 2009/12, Bruegel, Brussels.
Aghion, Ph., Dechezlepretre, A., Hemous, D., Martin, R., Van Reenen, J., 2010. Car-bon Taxes, Path Dependencies and Directed Technical Change, Evidence fromthe Auto Industry, Harvard University and LSE working paper (downloadedon 17 September 2012 from http://www.aeaweb.org/aea/2011conference/program/preliminary.php)
Bosetti, V., Carraro, C., Duval, R., Sgobbi, A., Tavoni, M., 2009. The role of R&D andtechnology diffusion in climate change mitigation: new perspectives using theWITCH model. OECD, Economics Department working paper 664.
Brunnermeier, S., Cohen, M., 2003. Determinants of environmental innovation in USmanufacturing industries. Journal of Environmental Economics and Manage-ment 45, 278293.
Croci, E. (Ed.), 2005. Handbook of Environmental Voluntary Agreements. Springer,Dordrecht, The Netherlands.
Jaffe, A., Newell, R., Stavins, R., 2002. Technological change andthe environment. Environmental and Resources Economics 22,4169.
Jaffe, A., Palmer, K., 1997. Environmental regulation and innovation:a panel data study. The Review of Economics and Statistics 79,610619.
Jaffe, A., 2011. Technology policy and Climate Change, Prepared for the NextRound of Climate Economics & Policy Research, Washington DC, October 2011(downloaded on 17 September 2012 from http://people.brandeis.edu/ajaffe/Technology%20Policy%20and%20Climate%20Change.pdf)
Johnstone, N., 2008. Environmental Policy, Technological Innovation and Patents.D Studies on Environmental Innovation, Paris.e, N., Hascic, I., Popp, D., 2010. Renewable energy policies and technologi-nnovaomics, J., Mtally r, 2003licy A., 200technGerm1.., 200ersity., Rual inn453O/ICTence
Which policy instruments to induce clean innovating?1 Why we need the private innovation machine for climate change and how to turn it on2 Data on clean innovations3 Evidence on the impact of government policies to induce private clean innovations4 Micro evidence from Belgian firm level CIS-VI data on the impact of government policies to induce clean innovations4.1 The CIS-VI eco-innovation data4.2 Research questions4.3 Who introduces clean innovations?4.4 Motives for clean innovations4.5 Econometric analysis of the impact of government policies on clean innovations
5 Conclusions: linking clean government policies to private clean innovations: does it work?AcknowledgementsReferences