MANOJKUMAR M S
4JN22RI018
IMPORANCE OF MACHINE LEARNING IN DRIVING
INNOVATION
Abstract
This paper examines the effect of drivers in the development of eco-innovation from
a system dynamics perspective. While previous literature has made important
contributions in identifying factors that influence the development of eco-innovations,
there remains limited understanding of how these drivers act and interact in
promoting its development. Therefore, there is a need to develop a framework of
relationships and drivers that encourage and support eco-innovation in companies.
This paper develops an integrated framework encompassing key internal, market
and governmental factors and their complex interactions using principles of system
dynamics and machine learning to address this gap. The research questions how
these drivers interact in a dynamic and non-linear manner to influence the
development of eco-innovation in companies and how can these interactions be
effectively modelled and understood, considering the complexities of sustainable
business practices and the limitations of traditional linear approaches. We empirically
test these questions by using the Spanish Technological Innovation Panel database.
The findings demonstrate that eco-innovation is not solely driven by isolated factors;
instead, it emerges from the complex interplay between internal capabilities,
governmental policies and market dynamics. By emphasising the synergistic effects
of these drivers, the research offers a nuanced understanding of their systemic
interactions. Furthermore, our analysis highlights the varying efficiency levels of
different drivers, underscoring the pivotal role of environmental corporate policies
and the strategic allocation of financial resources. In contrast, cooperation, market
forces and regulations exhibit lower efficiency in driving eco-innovation processes.
These insights not only advance theoretical knowledge but also provide valuable
guidance for businesses and policymakers, offering a more holistic approach to
fostering sustainable innovation.
1 INTRODUCTION
In the context of pursuing a sustainable economy, eco-innovation has emerged as a common
and necessary strategy for firm innovation (see, e.g., Arranz et al., 2020; Baldassarre et
al., 2019; Dangelico et al., 2017). In fact, eco-innovation is identified as the most important
contributor towards sustainable development by the European Parliament in the Lisbon
Strategy targets for economic growth and competitiveness (Rodriguez et al., 2010). As a
result of the increased demand for environmentally friendly products and services, as well as
the growing societal relevance of sustainable development, firms have directed towards
sustainable developments respectful with the environment. Grin et al. (2010) conclude that
‘industrial transformation’ involves adjustments in the production process, as well as in the
required patterns when pursuing the route of sustainable development.
Since the recognition of the relevance of eco-innovation in achieving a sustainable
environment and growth (see, e.g., García-Granero et al., 2020), a primary concern has been
the need to develop a dynamic system for eco-innovation development, creating a framework
of relationships and factors that encourage and support eco-innovation in companies
(Kesidou & Demirel, 2012; Kiefer et al., 2019; Scarpellini et al., 2020). Thus, from an
institutional point of view, administrations have understood this need and implemented
actions, mainly regulation and financial support (Qi et al., 2021). Moreover, research has not
been immune to it. Consequently, the literature, from natural resourced-based view (NRBV)
and stakeholder theories, has considered not only the importance of factors internal to
companies in the development of eco-innovation (Hart, 1995) but also the need for firms to
establish relationships with stakeholders, as critical factors in the development of eco-
innovation (Andersén, 2021; Jové-Llopis & Segarra-Blasco, 2018). This has made it possible
to identify companies' internal and external factors as incentives and facilitators of eco-
innovation. Notably, the literature has acknowledged three different categories of drivers for
eco-innovation in companies: market forces, internal forces, and regulatory and policy forces
(Arranz et al., 2021; Kiefer et al., 2019).
However, while the literature has made important contributions in identifying factors that
influence the development of eco-innovation, there are limitations in understanding how
these factors act in promoting its development. The main limitation arises from the fact that
previous research has analysed the relationship between drivers and eco-innovation without
considering that this process is dynamic and complex,1 including the interaction between
drivers in the development of eco-innovation. This has meant that results on the eco-
innovation process have not been conclusive in determining factors and explaining how they
interact (see, e.g., Cheng & Shiu, 2012; Horbach, 2016 and Jové-Llopis & Segarra-
Blasco, 2018).
First, most papers have approached the study of eco-innovation development from an
external perspective, considering how input variables (drivers) directly affect eco-innovation,
forgetting about the effect that interactions between drivers can have in producing eco-
innovations. The importance of the interactions has been highlighted, especially in recent
research in innovative development, which has pointed out the significance of investigating
the interactions between processes, as generators of synergistic and complementary effects
(Ballot et al., 2015; Doran, 2012), being able to produce surprising effects on output
variables. Therefore, investigating how drivers interact in the eco-innovation development
process, facilitating it or generating synergistic processes, is an important issue to be studied.
Second, previous studies have analysed the relationship between drivers and eco-innovation
without considering that this process is dynamic and complex (García-Granero et al., 2020;
Jové-Llopis & Segarra-Blasco, 2020). This means that to the classic limitations of the
diversity of surveys and the variety of measures used, which make generalisation difficult
(García-Granero et al., 2020; Horbach, 2016), we can add the limitation that econometric
models have in modelling complex relationships, as it is shown that most analyses do not
exceed 40% explained variance, generating difficulties in modelling dynamic systems for
eco-innovation.2 That is, there is abundant and important research on the identification of
factors that affect the development of eco-innovation, concluding whether they are significant
or not (see, e.g., Jové-Llopis & Segarra-Blasco, 2018, 2020); however, studies that deal with
how various drivers affect eco-innovation are limited and with inconclusive results. For
example, the question of quantifying and prioritising how drivers affect eco-innovative
development has not been resolved, which is an important issue from the perspective of
business decisions and the development of environmental policies (D'Amato et al., 2021;
Elmagrhi et al., 2019), considering the limited resources and the need to identify what the
critical factors are in the development of eco-innovation. Moreover, while there is consensus
on the positive effect of internal factors to the company (such as innovation capability or
green corporative management) and institutional factors (such as regulations and financial
support), this cannot be extrapolated to market factors. In fact, the literature shows
contradictory results in terms of how the market affects eco-innovation. While a group of
studies consider that the market has a positive impact (see, e.g., Rennings et al., 2006,
Veugelers, 2012), other researchers have not observed any statistically significant
relationships (Kesidou & Demirel, 2012; Kiefer et al., 2019). Unlike previous literature that
has argued that the lack of consensus is based on discrepancies of measures in different
geographical and sectoral areas, we argue that it derives from the low explained variance of
market factors with respect to institutional and/or internal factors of the company, which,
added to the low explanatory power of the model, justifies the variability of the results of
these factors.
Therefore, the study of eco-innovation will require the solving of previous limitations and
approaching the research from a more realistic, non-linear and complex perspective, which
will allow adequate modelling of these systems to find out how the various drivers interact. In
this context, our paper addresses this gap. First, from a methodological point of view and
using a systems approach (Bergek, 2019; Bergek et al., 2008),3 we assume, as in previous
literature, that there are three categories of eco-innovation drivers: internal, market and
governmental, considering these as input variables; and as an output variable, the eco-
innovation developed by companies. Moreover, in line with Wu and Marceau (2002), we
consider that drivers interact in non-linear and dynamic processes towards the development
of eco-innovation. To do this, following Sterman (2000), we use the theory of dynamic
systems, which combined with simulation methods will allow us to deduce the interaction
between the drivers. Thus, we will be able to solve previous limitations of the literature,
which have exclusively considered the direct impact of drivers in eco-innovation (Arranz et
al., 2021). Second, from an instrumental point of view, we will combine both regression
analysis and artificial neural networks (ANNs) in our modelling. Thus, to the explanatory
power of the regression models, we can add the capacity of ANNs in the analysis of complex
problems,4 determining all interactions through learning algorithms. This will allow us to
solve previous limitations of regression models, providing a higher level of explanatory
variance, which will result in a better understanding and quantification of how various drivers
influence eco-innovation development (Arranz et al., 2022). Last, we use the Spanish
Technological Innovation Panel (PITEC) as our database, which is the Spanish counterpart of
the EU Community Innovation Survey. The usage of this widely utilised database will enable
us to compare and generalise the results. The final sample comprises of 5,221 companies in
the manufacturing sector.
The main contribution of this paper lies in its innovative approach to understanding and
modelling the complex dynamics of eco-innovation grounded theoretically in the Natural
Resource-Based View and Stakeholder Theory. By integrating a dynamic systems
perspective and employing advanced machine learning techniques, the study goes beyond the
limitations of traditional linear analyses. It delves into the intricate interplay and feedback
loops among internal, market and governmental drivers, shedding light on how these factors
synergistically influence the development of eco-innovation. This novel methodology not
only provides a nuanced understanding of the multifaceted relationships between drivers but
also offers a robust quantitative framework for comprehending their non-linear interactions.
Furthermore, the use of the PITEC as a comprehensive dataset ensures the applicability and
generalisability of the findings. Overall, the paper's significant contribution lies in advancing
the field's understanding of eco-innovation processes, enabling more effective decision-
making for businesses and policymakers striving towards sustainable economic practices.
Therefore, the contribution is framed in the field of environmental management, developing
an approach to the modelling of eco-innovation from a dynamic point of view. Moreover,
unlike previous studies, our contribution focuses, first, on the interconnection and
interdependence of drivers; second, on the dynamic feedback processes between these
drivers; and, third, on the resulting behaviours, studying the systemic interaction of variables
that affect eco-innovation. In fact, our findings demonstrate that eco-innovation is not solely
driven by isolated factors; instead, it emerges from the complex interplay between internal
capabilities, governmental policies and market dynamics. By emphasising the synergistic
effects of these drivers, our research offers a nuanced understanding of their systemic
interactions. Furthermore, our analysis highlights the varying efficiency levels of different
drivers, underscoring the pivotal role of environmental corporate policies and the strategic
allocation of financial resources. In contrast, cooperation, market forces and regulations
exhibit lower efficiency in driving eco-innovation processes. These insights not only advance
theoretical knowledge but also provide valuable guidance for businesses and policymakers,
offering a more holistic approach to fostering sustainable innovation.
2 CONCEPTUAL FRAMEWORK
2.1 Eco-innovation conceptualisation
The Eco-Innovation Observatory (2018, p. 8) defines eco-innovation as the ‘introduction of
any new or significantly improved product (good or service), process, organisational change
or marketing solution that reduces the use of natural resources (including materials, energy,
water and land) and decreases the release of harmful substances across the whole life-cycle’.
In fact, this definition, in line with the Community Innovation Survey, describes eco-
innovation as a type of innovation (product, process, organisation and marketing) with the
goal of reducing pollution and keeping a sustainable economy. Horbach et al. (2012, p. 119)
corroborate this conceptualisation of eco-innovation as an innovation, defining it as ‘product,
process, marketing, and organizational innovations, leading to a noticeable reduction in
environmental burdens’. Bossle et al. (2016), for their part, specify that the objectives of eco-
innovation are to minimise the environmental effect of business operations, adhere to
environmental regulatory standards, and increase energy savings.
Therefore, considering eco-innovation with environmental innovation, it is to be expected
that there will be a parallel in the innovation and eco-innovation development process
(Arranz et al., 2020). As widely established in the literature, the development of innovation is
characterised by the uncertainty and risks of developing this process (Jalonen, 2012; Jové-
Llopis & Segarra-Blasco, 2020; López Pérez et al., 2023; Teece et al., 2016). This is, on the
one hand, the technical uncertainty, to assess both the results and the time they were obtained,
and, on the other hand, the uncertainty of the market, as a consequence of the need for
consumer acceptance of eco-innovations. Moreover, similar to the innovation literature, it is
to be expected that certain internal and external factors of companies (drivers) can facilitate
this process, reducing risk and mitigating uncertainty.
While there is a similarity between the conceptualisation of eco-innovation and other types of
innovation, there are also some differences. The first difference is that, while traditional types
of innovation target economic profits, eco-innovation targets both economic and
environmental benefits (Acebo et al., 2021; Janahi et al., 2023; Oh et al., 2020; Zhang &
Walton, 2017). The second difference between innovation and eco-innovation is shown in the
phenomenon called the ‘double externality’ (Arranz et al., 2020; Dangelico, 2016). That is,
while the development of eco-innovation supposes an internal cost for companies, the public
character of eco-innovation and its social benefit means that other companies can assume and
imitate it without incurring costs. Therefore, the company is not incentivised to invest in eco-
innovation. The last difference is the role that regulations and incentives play in eco-
innovation as compared with traditional innovations. Previous research highlights the positive
effect of environmental regulations and policies on eco-innovations (Costantini et al., 2017;
Doran & Ryan, 2016).
2.2 Drivers of eco-innovation: theoretical framework
As a theoretical framework, we employ the NRBV5 and stakeholder theory,6 which
complement each other to explain firms' decisions to eco-innovate. The theories emphasise
the role of external drivers of eco-innovation (Hart, 1995; Sarkis et al., 2010), noting that
proactive firms manage their interaction with the natural environment through the integration
of stakeholders. In fact, the NRBV highlights the engagement of stakeholders as a key driver
of pollution reduction (Andersén, 2021; Katsikeas et al., 2016; Roxas et al., 2017; Zhang &
Walton, 2017). Moreover, stakeholder theory has noted that stakeholder pressure exercised
by customers, regulators, suppliers and competitors is a driver of eco-innovation
(Horbach, 2008; Rennings & Rammer, 2011). Researchers in the area of eco-innovation have
categorised external eco-innovation drivers into two groups: regulatory and policy forces and
market forces (Horbach, 2008; Horbach et al., 2012; Kiefer et al., 2019).
Regarding the external regulatory and policy forces as drivers for eco-innovation, the
literature centres on the effect that government regulatory forces and subsidies, or financial
support, have had on the development of eco-innovations (Bimonte et al., 2023; Fischer &
Pascucci, 2017). Regulations and subsidies push firms to invest in environmental innovation
(Horbach et al., 2012; Kesidou & Demirel, 2012; Kiefer et al., 2019; Veugelers, 2012). For
example, Directive 2009/125/EC determines the framework for the eco-design requirements
for ecological products (Bovea & Pérez-Belis, 2012); XP X30-901 (AFNOR) and BS 8001
(British Standard) are certifications to promote zero waste and product recycling. Kiefer et al.
(2019) and Kesidou and Demirel (2012) point out that governments offering incentives, tax
breaks or feed-in tariffs for companies adopting renewable energy sources create a favourable
environment for eco-innovation in the energy sector. Therefore, in line with previous
research, it can be affirmed that the existence of a regulatory framework and public financial
support should enable the eco-innovation process, having a significant impact on companies'
decisions to develop them.
Market forces have traditionally been acknowledged as important external factors driving
innovation decisions (Prajogo & Ahmed, 2006). Previous studies note that eco-innovation has
recognised the environmental consciousness of consumers as a driver of eco-innovation
demand (Hojnik & Ruzzier, 2016; Jansson, 2011). The proactive attitude of consumers
towards ecological products has been considered as a driver for sustainable product
development (Demirel & Kesidou, 2019). For example, the growing consciousness of
environmental concerns and a preference for sustainable transportation have resulted in an
increased demand for electric vehicles (Larson et al., 2014). Consequently, automotive
manufacturers are addressing this demand through innovations in the electric vehicle sector,
encompassing the development of extended-range batteries, improvements in charging
infrastructure and the exploration of eco-friendly manufacturing practices. Therefore,
companies find a new market as a business opportunity, fostered by the consumer's attitude
towards the consumption of green products.
Moreover, in a similar way to innovation, NRBV has emphasised the importance of factors
internal to the company as drivers of eco-innovation. Thus, the possession of resources and
capacities has been highlighted as a key factor in the development of eco-innovation
(Horbach, 2016; Jové-Llopis & Segarra-Blasco, 2018, 2020; Kiefer et al., 2019). In addition
to this static approach to identifying resources and capabilities of a firm, we emphasise
internal processes and organisational dynamics developed with these resources, in line with
the purpose of this research, focused on the dynamics of processes. That is, grounded in the
theory of dynamic capabilities (Barney, 2001; Zahra et al., 2006), and following Eisenhardt
and Martin (2000), who have characterised dynamic capabilities as an identifiable process,
we assume that the firm must have the ability to develop eco-innovation, deploying resources
and capabilities, and using organisational processes to achieve these objectives. Moreover,
following Teece (2007), who states that firms' capabilities enable the development of
innovation processes, we consider the ability to develop certain processes as internal to the
organisation, conceptualising them as drivers of eco-innovation.
In this research, three internal drivers of eco-innovation are considered: innovation capability,
environmental corporate management (ECM) and cooperation agreements (see Table 1).
Innovation capabilities are defined as the capacity of companies to manage resources, using
organisational processes to reach some innovation goals (Teece, 2007; Teece et al., 2016). Li
et al. (2020) and Arranz et al. (2020) have found a parallel in the development of innovation
and eco-innovation, pointing out the similarity of the processes, which makes the
competencies and skills in innovation acquired by companies a driver of eco-innovation.
Indeed, possessing a strong innovation capability has acted as a catalyst for the development
of eco-innovation. For instance, Kumi (2023) emphasises the case of Tesla, where
innovations like the Powerwall and Powerpack exemplify their prowess in energy storage
solutions. These products facilitate the efficient storage of renewable energy for subsequent
use. Tesla's success in developing effective energy storage solutions directly tackles a
challenge associated with renewable energy sources, namely, intermittency. This not only
addresses a crucial issue but also adds to the wider eco-innovation landscape by facilitating
the seamless integration of renewable energy into the grid.
