Content uploaded by Maryam Mazaheri
Author content
All content in this area was uploaded by Maryam Mazaheri on Nov 07, 2024
Content may be subject to copyright.
Maturity assessment of green patent clusters: Methodological implications
Maryam Mazaheri , Jaime Bonnin Roca
*
, Arjan Markus , Elena M. Tur , Bob Walrave
Eindhoven University of Technology, Department of Industrial Engineering & Innovation Sciences, P.O. Box 513, 5600 MB, Eindhoven, the Netherlands
ARTICLE INFO
Keywords:
Green patents
Environmental innovation
Technology life cycle
Growth models
Technological maturity
ABSTRACT
Patents are one of the most widely used tools to analyze environmental technologies. Organizations such as the
World Intellectual Property Organization and OECD have developed search strategies to retrieve green patents
based on their patent classication. These classications divide patents into clusters, which are aligned with
different sustainability goals. In this paper, we take advantage of this to analyze the distribution of patents across
1.221 patent classes within six clusters dened by OECD's ENV-TECH classication. We also assess the maturity
stage of each patent class by tting two commonly used S-curve models, namely logistic and Gompertz. We nd
that (a) most patent classes are still in a relatively early stage of the technology life cycle and (b) considerable
heterogeneity exists in the distribution of patents, both within and across clusters. We discuss the methodological
implications of our results and provide recommendations for scholars, drawing on green patent analyses, to
conduct future work on environmental technologies.
1. Introduction
Concerns about the negative impact of human activity on the envi-
ronment are growing (Azam, 2016) and sparked interest among gov-
ernments, rms, and scholars alike in developing and studying
technology developments that lower our environmental footprint (Liao,
2018; Martins et al., 2019). In this respect, scholars (Borghesi et al.,
2015; De Marchi, 2012; Horbach, 2008) and international organiza-
tions, such as Cvijanovi´
c et al. (2021) and Haˇ
sˇ
ciˇ
c and Migotto (2015),
now explicitly refer to environmental innovation and environmental-
related technologies in recognition of the critical role that innovation
and technology play in addressing global environmental challenges,
such as climate change, pollution, and biodiversity loss.
Meanwhile, research on environmental technologies is thriving. A
large portion of studies use patents to operationalize the development of
environmental technologies (Mazaheri et al., 2022). To identify green
patents, it is common practice to use the well-known ENV-TECH clas-
sication as proposed by the OECD (Haˇ
sˇ
ciˇ
c and Migotto, 2015), which is
similar to the one offered by the World Intellectual Property Organiza-
tion (WIPO) (Corrocher and Mancusi, 2021; Fusillo, 2023; Hoang et al.,
2020; Nelson et al., 2022). Both classications cluster technologies into
groups, which may encompass a wide variety of technologies. Scholars
have used these clusters in their macroeconomic analyses (e.g. Cohen
et al., 2020; Fusillo, 2023; Perruchas et al., 2020). However, such
clustering of technologies may not allow us to capture and understand
the differences in the growth of (and between) various technologies and,
as such, hinders the development of our understanding of the factors
that underlie technological progress (Lee et al., 2011).
This paper then serves to study the validity of such cluster-level
analyses. Specically, we analyze the OECD ENV-TECH classication
system at the patent class level. We analyze the number of patents in
each (sub)cluster, and estimate the maturity of each patent class by
tting two widely adopted S-curve models commonly used to predict the
evolution in patenting activities. Overall, our results raise serious con-
cerns about the validity of clustering technologies into large groups to
assess the efcacy of sustainability-oriented policies. The key reason is
that we observe signicant intra-cluster variation in patenting activity,
which substantially problematizes patent counts as a proxy for maturity.
We observe no correlation between the cumulative number of patents in
a specic patent class and the predicted level of technological maturity,
which may have important implications for the work of scholars ana-
lysing the technological life-cycle of green patents.
The rest of the paper is organized as follows. Section 2 contains a
theoretical background on the use of patents as a measurement tool of
sustainable innovation. Section 3 discusses our data collection strategy
and how we conducted the maturity analysis for each patent class.
Section 4.1 presents the basic characteristics of the ENV-TECH dataset,
showing the number of patents per cluster and subcluster and the growth
* Corresponding author.
E-mail addresses: m.mazaheri@tue.nl (M. Mazaheri), j.bonnin.roca@tue.nl (J. Bonnin Roca), a.markus@tue.nl (A. Markus), e.m.mas.tur@tue.nl (E.M. Tur), b.
walrave@tue.nl (B. Walrave).
Contents lists available at ScienceDirect
Technological Forecasting & Social Change
journal homepage: www.elsevier.com/locate/techfore
https://doi.org/10.1016/j.techfore.2024.123813
Received 18 October 2023; Received in revised form 17 September 2024; Accepted 5 October 2024
Technological Forecasting & Social Change 209 (2024) 123813
Available online 19 October 2024
0040-1625/© 2024 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY license (
http://creativecommons.org/licenses/by/4.0/ ).
in patenting activity in the last ve years. Section 4.2 reports the results
of the maturity analysis. Section 5 discusses the methodological impli-
cations of our ndings and provides researchers with practical recom-
mendations for using green patent data.
2. Theoretical background
Knowing the maturity of a technology has critical strategic impli-
cations for governments and rms alike (Huenteler et al., 2016; Taylor
and Taylor, 2012). Most environmental technologies are relatively
young, as modern societies are still transitioning towards a more sus-
tainable future. Sustainable technological alternatives not only compete
against the current fossil-fuel-based system but also need to compete
against each other in a game usually won by the technologies with the
highest expectations (Alkemade and Suurs, 2012). However, these ex-
pectations may never become a reality, as younger technologies are also
characterized by higher levels of uncertainty regarding development
time, costs, and quality (Haessler et al., 2023; Rotolo et al., 2015).
To manage technological risks and diversify R&D portfolios, re-
searchers have developed a plethora of methods to assess technological
maturity and forecast the evolution of technology in the short- and
medium-term (Porter et al., 2011). Some methods rely on the opinions of
experts surveyed individually (Morgan, 2014) or in groups (Flostrand
et al., 2020). Others use maturity scales such as the technology readiness
level (Vik et al., 2021) or technology-specic maturity models (Senna
et al., 2023). Another large group of methods rely on using R&D in-
dicators, such as R&D expenditures, academic publications, patents, and
news sources (Jeffrey et al., 2014; Lezama-Nicol´
as et al., 2018).
Most maturity assessment methods are based on product or tech-
nology life cycle theory (Kim, 2003; Taylor and Taylor, 2012). Ac-
cording to life cycle theory, the performance of a technology, as a
function of time, follows an S-curve. During the early stages of the life
cycle, progress is slow, and there is typically intense competition be-
tween technological alternatives (Tushman and Anderson, 1986). Once
a dominant design appears, there is a period of rapid technological
development and performance growth, and development efforts start to
progressively shift from product innovation to process innovation
(Utterback and Abernathy, 1975). At the end of the life cycle, progress
slows down again as R&D efforts present decreasing returns to scale, and
only a few rms remain (Klepper, 1996).
For decades, technological forecasting scholars have developed
mathematical models to represent the shape of the S-curve (Martino,
2003; Meade and Islam, 1998). In these models, it is assumed that the
R&D output for a specic technology will follow a sigmoidal shape as
society accumulates knowledge in a technological domain until it rea-
ches a saturation point (Altuntas et al., 2015). To complement these
analyses, scholars have also used citation data to study how the net-
works of researchers or inventors evolve over time (Huang et al., 2022)
and patent classes to study patterns of technological convergence and
hierarchical interdependence (Luan et al., 2024). The robustness of
these methods is highly reliant on the ability of researchers to accurately
collect and pre-process data sources that relate to the technological eld
they are studying (Porter and Cunningham, 2004).
2.1. Use of patents to study the development of environmental
technologies
Societies in developed countries have determined that developing
technologies that lower our environmental footprint should be a priority
(Liao, 2018; Martins et al., 2019). As such, there is a growing interest in
examining the development of methods to measure environmental
technologies, which would allow for assessing the efcacy of govern-
ment and corporate sustainability measures. However, measuring sus-
tainable innovation is a challenging task, given that it is a
multidimensional construct encompassing aspects of product, process,
and organizational innovation (García-Granero et al., 2020).
Environmental technologies are more complex vis-`
a-vis non-green
technologies, typically requiring more collaborative efforts to be
developed and adopted successfully (Barbieri et al., 2020). In response,
researchers have developed a myriad of instruments to measure both
innovation inputs, such as R&D expenditures (Leiponen and Helfat,
2010); outputs, such as patents (Griliches, 1990), scientic publications
(Cheng and Shiu, 2012); and measures of non-technological innovation
(Albitar et al., 2023).
A common way to measure sustainable innovation is using patent
analyses (Afeltra et al., 2021). In this case, patent classes proxy for
technological domains related to sustainable activities, products, and
processes. Scholars have used patents, for instance, to analyze interna-
tional networks of collaborators in sustainable innovation(Corrocher
and Mancusi, 2021); the effects of regulatory policies on sustainable
innovation (Fabrizi et al., 2018); the impact of design capabilities on
sustainable innovation (Ghisetti et al., 2021); the relationship between
geographic proximity and dissemination of environmental technologies
(Losacker, 2022); or the effect of foreign direct investment in a regional
technological specialization (Castellani et al., 2022).
When using patents, a key question is how to distinguish between
patents that refer to environmental technologies and those that do not.
To solve this problem, international organizations such as the World
Intellectual Property Organization (WIPO) and the OECD have proposed
methodologies to sample environmental technologies based on their
patent class (International Patent Classication, IPC, or Cooperative
Patent Classication, CPC), which are nowadays widely used in studies
related to environmental technology (e.g. Corrocher and Mancusi, 2021;
Fusillo, 2023; Hoang et al., 2020; Nelson et al., 2022). The success of
these classications is partially due to their ease of use and replicability.
However, it also presents limitations. Notably, there is no standard
method for assigning patent classes, and the nal decision depends on
each individual patent examiner and patent ofce. Due to individual
biases, language and cultural differences, and the inuence of private
lobbying, some patent ofces may assign, on average, a larger number of
patent classes to each patent than others (Lerner and Seru, 2022; Reif-
fenstein, 2009). In addition, the patent classication is constantly
growing. New patent classes are created as technology evolves; there-
fore, patent classes need to be assigned retroactively to patents pub-
lished in the past (Wang et al., 2016). In recent years, scholars have
proposed as an alternative the use of text mining and topic modeling
techniques, based on natural language processing and large language
models, to overcome some of these limitations (Bekamiri et al., 2024;
Kamateri et al., 2024).
The classication proposed by the OECD, for instance, by Haˇ
sˇ
ciˇ
c and
Migotto (2015), consists of 1375 patent classes divided into six large
technology clusters based on six pre-determined policy objectives:
environmental management; water; energy; greenhouse gases' capture,
storage, and sequestration; transportation; and buildings. Because of the
mere existence of these clusters, previous work has started to analyze
green patents at the cluster level (e.g. Cohen et al., 2020; Fusillo, 2023;
Perruchas et al., 2020). This is potentially problematic because of the
variability at the patent class level (Adamuthe and Thampi, 2019; Yuan
and Cai, 2021).
In the case of maturity assessments, such clustering of technologies
conicts with common practice in the forecasting literature, which since
the early days places a strong emphasis on focusing on specic tech-
nological areas or products to increase the explanatory power of the
mathematical models (Martino, 2020; Porter et al., 2011). Clustering
does not allow us to capture and understand the differences in the
growth and maturity levels of (and between) various technologies and,
as such, hinders the development of our understanding of the factors
that underlie technological progress (Lee et al., 2011). However, to date,
there is a lack of understanding of the extent to which the heterogeneity
within clusters of environmental technologies may impact the results of
large-scale studies. This paper aims to ll this gap by examining two
types of heterogeneity across patent classes and clusters: rst, in terms of
M. Mazaheri et al.
Technological Forecasting & Social Change 209 (2024) 123813
2
patenting activity, or quantity of patents published per patent class, and
second, in terms of technological maturity.
3. Data and methods
3.1. Growth curves and maturity
To assess the maturity of each patent class, we used non-linear least
squares to t the logistic yt=L
1+ae−btand Gompertz yt=Le−ae−bt
models to the cumulative number of patents per class. Both curves are
dened by three coefcients: a, b, and L. Parameters a and b determine
the position and shape of the curve, while L represents the upper limit to
growth (Martino, 1992). In our case, L represents the cumulative num-
ber of patents expected at the point of technological decline. The two
curves do have a different geometry. While the logistic curve is sym-
metric, the Gompertz curve is asymmetric, exhibiting a maximum
growth of about 37 % of the upper limit (Marinakis, 2012). Therefore,
tting both curves allows us to cover various technological growth
patterns. To avoid overtting, we establish an upper boundary for L of
500.000 patents, which is well above our sample's largest patent class
size (63.396).
Both logistic and Gompertz curves are considered best-in-class
(Meade and Islam, 1998; Tattershall et al., 2021) and have been
widely used by scholars to analyze technology life cycles with patent
data (Adamuthe and Thampi, 2019; Huang et al., 2022; Jiang et al.,
2022). To select the most appropriate model (logistic vs. Gompertz) per
patent class, we employed four commonly used different statistical
measures: the R-squared (R
2
), the Root Mean Squared Error (RMSE),
Akaike Information Criterion (AIC), and the Bayesian Information Cri-
terion (BIC) (Jiang et al., 2022; Kodama, 2004; Lee et al., 2016; Savin
and Winker, 2012). Our calculations of all the statistical measures for
every patent class are available in the supplementary data included with
this submission. R
2
and RMSE are two commonly used statistical mea-
sures in regression analysis and forecasting. R
2
is a measure that eval-
uates how well a regression model ts the observed data (Sohrabpour
et al., 2021). R
2
ranges from 0 to 1, with 1 indicating that the model
perfectly ts the observed data (Van Sark, 2008). RMSE measures the
differences between the predicted values of the dependent variable and
the actual values and is used to indicate the accuracy of the predictions
made by the model (Adamuthe and Thampi, 2019). Lower RMSE values
indicate a better t of the model to the observed data (Gupta and Jain,
2012). Compared to R
2
and RMSE, which are measures of goodness-of-
t, AIC and BIC are used to compare the quality of different models
(Ganguly et al., 2010) by balancing goodness-of-t with model
complexity. The model with the lowest AIC and BIC value is generally
considered the best-tting model (Chen et al., 2022; Hinge et al., 2021).
To choose between logistic and Gompertz, we picked the model which
had a larger R2, and lower RSME, AIC, and BIC.
Subsequently, we assessed the technology maturity of every patent
class. To do so, we calculated the ratio between the cumulative number
of patents in the last year of our dataset and the upper limit L of the
corresponding S-curve. Following previous studies (Adamuthe and
Thampi, 2019; Gao et al., 2013; Huang et al., 2022), we then continued
to categorize each patent class into one of four maturity levels: those
with a maturity lower than 25 % of the upper limit, representing the
early stages of the introduction of a technology; those with a maturity
between 25 % and 50 %, representing early growth; those with a
maturity between 50 % and 75 %, representing late growth; and those
with a maturity higher than 75 %, representing technological maturity
and market decline (P´
erez and Soete, 1988). In addition, we computed
the growth in their cumulative number of patents over the last ve years.
3.2. Patent data
To identify green technologies, we used the OECD ENV-TECH
classication system developed by Haˇ
sˇ
ciˇ
c and Migotto (2015), widely
recognized as the standard framework for research on green technolo-
gies (Costantini et al., 2017a; Fabrizi et al., 2018; Losacker, 2022). Given
that Haˇ
sˇ
ciˇ
c and Migotto (2015) used PATSTAT to conduct their analysis,
and we aim to replicate their approach, we also extracted patent data
from PATSTAT (5.15, Spring 2020 version), the ofcial patent database
of the European Patent Ofce (EPO). PATSTAT is a publicly available
and widely used patent database covering bibliographic information and
legal events on over 100 million patents from over 40 patent authorities.
Our dataset spans the years from 1877 to 2020. For the purpose of this
study, we focus on data on the technological classications on patents,
both the international patent classication (IPC) and its extension, the
cooperative patent classication (CPC).
The ENV-TECH classication connects IPC and CPC classes to seven
environmentally related domains and 23 subgroups, allowing for tar-
geted analysis of green technology patents within specic environ-
mental policy objectives. We used the search query proposed by Haˇ
sˇ
ciˇ
c
and Migotto (2015, pp. 46-58) to retrieve 3,739,354 patents belonging
to 2,281,362 DOCDB
1
families in 1375 different patent classes (767 CPC
and 608 IPC). We provide the search query in Appendix B, and the list of
the patent classes in the Supplementary Data. Because of the limited
explanatory power of growth curves for small data samples (Martino,
2003), we excluded patent classes for which we had fewer than 15 years
of data. After ltering, our nal dataset contained 3,724,268 patents in
2,273,399 DOCDB families and 1221 patent classes. We computed the
cumulative number of patents per year for every patent class. In cases
where a patent was assigned to multiple green patent classes, we
counted it in each of those classes.
As explained in Section 2.1, we choose between the logistic and the
Gompertz model depending on four goodness-of-t measures. Based on
this, 491 patent classes are best described by a Gompertz model, and the
other 728 best described by a logistic model. We checked and found that
for every patent class, the model with the highest R
2
was also the model
with the lowest RMSE, AIC, and BIC, in line with Huang et al. (2022).
The average R
2
was 0.98, with an average standard deviation of 0.03.
4. Findings
4.1. Patenting activity across ENV-TECH clusters
When we analyzed the patents in our dataset, we found that they are
unevenly distributed among the six clusters as dened by OECD. More
specically, while cluster 1 (‘Environmental management,’ see Table 1)
accounts for more than half of the green patents, cluster 4 (‘Capture,
storage, sequestration, or disposal of greenhouse gases’) only accounts
for about 0.5 % of the patents in our database. See Table 1 for more
details on the six clusters. Interestingly, such a skewed distribution is
also visible at the subcluster level. For instance, the OECD divided
cluster 1 into 14 subclusters. We observe that subclusters 1.1.2.1 Post
combustion technologies (Emissions abatement from mobile sources)
and 1.1.3.1 Post combustion technologies (Not elsewhere classied),
both related to post-combustion technologies, each represent more than
30 % of the total patents in the cluster. Conversely, subcluster 1.3.1.
(‘Solid waste collection’) represents only about 0.1 % of the patents in
this particular cluster.
The weight distribution of subclusters in the database (see Fig. 1)
reveals that the most prevalent subclusters are those related to emissions
reduction. Specically, the subclusters of post-combustion technologies
for emissions abatement from mobile sources, and not elsewhere clas-
sied, account for 17.65 % and 16.26 % of the weight, respectively.
Other well-represented technologies in the database include waste
management, material recovery, recycling, reuse, and renewable energy
1
All patents in a DOCDB family cover a same invention, usually in different
patent ofces.
M. Mazaheri et al.
Technological Forecasting & Social Change 209 (2024) 123813
3
sources such as solar photovoltaic and wind energy. On the other hand,
subclusters such as nuclear fusion reactors, greywater treatment, and
elevators and escalators have a very low weight in the database. Ap-
pendix A contains a table with the number of patents, weight, and 5-year
growth of all the 70 subclusters.
We also analyzed the weight of each patent class in the entire dataset.
Here, we also found that some patent classes have a disproportionate
weight, several orders of magnitude higher than others. Among the 1221
patent classes, we found that there are 22 which represent between 0.5
% and 1 % of the total number of patents. In addition, there are eight
patent classes that exhibit over 1 % of the total number of patents (see
Table 3). We also computed the cumulative distribution of patent classes
in the dataset and found that 25 % of the patents correspond to the top
31 patent classes with the highest patent count. Half of the patents in the
dataset are concentrated in only 108 patent classes, or 8 % of the total
number of patent classes that the OECD considers green (see Fig. 2,
Table 2).
In addition to a skewed distribution of patents across clusters and
subclusters, the growth rates of these (sub)clusters vary widely. Cluster
4 (‘Capture, storage, sequestration, or disposal of greenhouse gases’) is
the cluster with the lowest patent count and the cluster with the lowest
growth, with a growth rate of 26 % over the last ve years. This could
suggest that the technology is still in its infancy or that patenting activity
in this area is inherently low. Conversely, the second cluster with the
least amount of patents, cluster 2 (‘Water-related technologies’), has
experienced the highest growth rate, 60 %, over the last ve years.
In Fig. 3, we show the patent count for each subcluster, and we
compare their growth rate against the average growth of their cluster.
This gure allows us to identify subclusters with both high patent count
and higher-than-average growth rate, an indication of high investment
levels: 1.1.3.1 post-combustion technologies; 3.1.3 solar photovoltaic;
4.1 CO2 capture or storage; 5.1.3 electric vehicles; 6.2.1 lighting.
Likewise, we identied four subclusters of green patents that have
experienced substantial growth in the last ve years but whose total
number of patents remains low compared with other subclusters within
the same cluster, suggesting that they may become more relevant in the
future. These subclusters are: 1.5 environmental monitoring; 2.2.2 water
storage; 3.6.4 smart grids in the energy sector; and 5.2 rail transport.
Conversely, there are also subclusters with both a low patent count, and
a low growth rate, indicating low interest. These are: 1.3.1 Solid waste
collection, 2.1.1.1. Faucets and showers, 3.4.1. Nuclear fusion reactors,
4.2 Capture or disposal of greenhouse gases other than CO2, and 6.2.4
Elevators, escalators, and moving walkways.
Table 1
Summary of the clusters.
Cluster Name Total no. of
patents
Weight in
database
(%)
Growth
1 Environmental management 1,981,269 53.08 % 37 %
3 Climate change mitigation
technologies related to energy
generation, transmission of
distribution
786,677 21.08 % 32 %
5 Climate change mitigation
technologies related to
transportation
611,326 16.38 % 30 %
6 Climate change mitigation
technologies related to
buildings
271,429 7.27 % 48 %
2 Water-related adaptation
technologies
63,820 1.71 % 60 %
4 Capture, storage,
sequestration, or disposal of
greenhouse gases
17,750 0.48 % 26 %
Total 3,732,271 100.00 %
Fig. 1. Distribution of the weight of each cluster and subcluster in the dataset. The size of each box is directly proportional to their total.
M. Mazaheri et al.
Technological Forecasting & Social Change 209 (2024) 123813
4
4.2. Maturity analysis
In this section, we analyze the stage in the technology life-cycle of
patent classes across each of the clusters, and their subclusters. Fig. 4
shows the percentage of patent classes in each cluster in each of the four
stages of maturity. The level of maturity is calculated as the cumulative
number of patents in the last year divided by L, the upper limit coef-
cient from either the Gompertz or the logistic model, depending on
which had the better t. The clusters are organized by their level of
maturity. Cluster 2, consisting of water-related adaptation technologies,
has the highest proportion (76 %) of technologies in the 0–25 % matu-
rity range, followed by Cluster 3 (47 %). Conversely, clusters 1 and 4
have the largest proportion (34 % and 30 %, respectively) of technolo-
gies in the maturation stage. Overall, the distribution of technologies in
the early and late growth stages of maturity is highly uneven across
clusters. Even in the clusters with the highest portion of patents in the
last stage of the technology life cycle, most patent classes are still in their
early stages.
An analysis of the maturity levels across subclusters sheds additional
light on which technological areas are nascent and which are reaching
the end of their technology life cycle. Fig. 5 presents a comprehensive
view of the portion of patent classes within each subcluster belonging to
each maturity stage. The position of the subclusters has been sorted
according to the proportion of patent classes in the 0–25 % maturity
stage. Fig. 5 suggests a large variability in maturity levels across sub-
clusters of the same cluster. In cluster 1, for instance, which is the cluster
with highest number of mature patent classes, there are several sub-
clusters of technologies (e.g., 1.1.1.1Postcombustion technologies
(Emissions abatement from stationary sources), 1.3.3. Fertilizers from
waste and 1.5 environmental monitoring) whose patent classes are still
in the nascent stage. Cluster 2, the one which exhibits the highest per-
centage of patent classes falling in the 0–25 % maturity stage, presents
two subclusters with most of their patent classes in a later maturity
stage: 2.1.1.3 Sanitation (dual-ush toilets, dry toilets, closed-circuit
toilets) and 2.1.1.1 Faucets and showers. When looking at the other
clusters, we nd other subclusters technologies in the 75–100 % matu-
rity stage such as 3.4.2. (‘Nuclear ssion reactors’), 3.6.3. (‘Fuel cells’),
4.2 (‘Capture or disposal of greenhouse gases other than CO2’), 5.1.2
(‘Hybrid vehicles’) and 6.4 (‘Enabling technologies in buildings’). These
clusters may encompass technologies that have been in the market for a
long time, or technologies that were unsuccessful and faced decline
before their full development.
To gain a more in-depth understanding of the potential relationship
between patent counts and maturity levels, we conducted a further
analysis at the patent class level. In Fig. 6, we plotted the maturity level
of each patent class, on the x-axis, versus the total number of patents in
that same class, on the y-axis. To increase readability, we grouped patent
classes into clusters. We found no correlation between patent counts and
maturity levels. In fact, with the exception of cluster 1, the patent class
with the highest total number of patents in each cluster is associated
with 0–25 % level of maturity. Table 3 presents data on patent classes
within each cluster with the highest total number of patents but at the
lowest maturity level. This may help identify areas with the potential for
future investment or examine the policies that may be driving innova-
tion in these elds. For instance, patent classes Y02T 10/7005 cpc and
F16L 55/07 ipc are also among the classes with the highest growth of 5
years (79 % and 61 %), suggesting that these areas are receiving much
more investment than the rest and therefore experiencing accelerated
innovation.
5. Discussion
This study drew on patent data to comprehensively analyze the
growth and maturity of environmental technologies, to study the val-
idity of such type of analysis. More specically, we have used PATSTAT
data to analyze the characteristics of the OECD ENV-TECH classication
of green patents. We identied several important limitations researchers
may (and have) encounter(ed) when using this patent classication for
their studies. We discuss causes and provide recommendations to
decrease the problems that existing methods present, especially when
conducting econometric analyses.
First, we nd that the distribution of patents across clusters is highly
uneven and spans three orders of magnitude, from almost 2 million
patents in cluster 1, to less than twenty thousand patents in cluster 4.
Even within the same cluster, some patents have disproportionally more
weight than others (Fig. 1). One of the causes is that clusters 2–6 have a
much more concrete denition of their content than cluster 1. Cluster 1
includes a long list of technologies, from post-combustion catalysts to
soil remediation technologies, which do not necessarily have much in
common except their purpose of reducing the environmental footprint of
human activities. Additionally, there might be differences in patenting
behavior between technological elds, industrial sectors, and types of
organizations (de Rassenfosse and van Pottelsberghe de la Potterie,
Fig. 2. Cumulative distribution of patent classes in the dataset.
Table 2
Summary of patent classes with more than 1 % weight in database.
Patent class Total no.
patents
Growth Wt. subcluster
(%)
Wt. cluster
(%)
Wt. database
(%)
Subcluster
B09B 3/00, ipc 69,997 34 % 67 % 3.53 % 1.88 % 1.3.6. Waste management – Not elsewhere classied
Y02E 10/50, cpc 63,396 27 % 45 % 8.06 % 1.70 % 3.1.3. Solar photovoltaic (PV) energy
F02D 45/00, ipc 52,145 11 % 8 % 2.63 % 1.40 % 1.1.2.1 Post-combustion technologies (Emissions abatement from
mobile sources)
Y02T 10/7005,
cpc
51,134 79 % 32 % 8.36 % 1.37 % 5.1.3 Electric vehicles
B01D 46/00, ipc 45,613 114 % 8 % 2.30 % 1.22 % 1.1.3.1 Post-combustion technologies (Not elsewhere classied)
Y02E 30/40, cpc 41,766 8 % 66 % 5.31 % 1.12 % 3.4.2. Nuclear ssion reactors
B01D 50/00, ipc 40,883 194 % 7 % 2.06 % 1.10 % 1.1.3.1 Post-combustion technologies (Not elsewhere classied)
Y02E 10/40, cpc 37,954 24 % 35 % 4.82 % 1.02 % 3.1.2. Solar thermal energy
M. Mazaheri et al.
Technological Forecasting & Social Change 209 (2024) 123813
5
2009; Ghisetti and Quatraro, 2017).
As a result, regression-based studies which use control variables to
distinguish between technologies only at the cluster level (Fabrizi et al.,
2018; Girod et al., 2017; Wang and Wei, 2020) may primarily be
observing the behavior of a handful of overrepresented patent classes
while neglecting any potential variability at the patent class level
(Fig. 3). Our primary goal is to raise awareness regarding the patent
cluster choice and provide potential remedies to researchers in the green
patenting eld. Researchers may take several actions to decrease the
impact of the uneven distribution of patents. They may consider rening
the sample of patents used in their analyses, for instance by using natural
language processing techniques (Savin et al., 2022), to ensure that all
clusters are well-dened and encompass technologies with similar
characteristics (Cecere et al., 2014), or limiting their analyses to a spe-
cic sector. Doing so will improve understanding the nuances and het-
erogeneity within each cluster and avoid making overly general
conclusions that may not accurately represent every technology within a
cluster. Even though regression-based analysis examines the on-average
relationship between variables, we want to raise awareness regarding
research design choices, particularly patent class choice. In addition,
researchers should consider using control variables to account for the
characteristics and dynamics of each industrial sector, for instance, with
ISIC or NACE codes (Ghisetti and Quatraro, 2017; Wurlod and Noailly,
2018). They may also conduct additional robustness checks concerning
changes in the composition of their sample (Lerner and Seru, 2022).
Second, our paper also offers insights for researchers who use patent
data to study the technology life cycle of environmental innovations. We
have shown (Fig. 6) how the cumulative number of patents of a patent
class alone does not adequately represent its maturity level. While we
have used the OECD ENV-TECH classication for our study, we strongly
expect our ndings to hold for alternative classications of green pat-
ents, such as the World Intellectual Property Organization's (WIPO) IPC
Green Inventory (Ghisetti and Quatraro, 2017; Ning and Guo, 2022).
This is due to differences across elds in terms of patenting. To solve this
problem, we advise researchers to use comparative (relative) indicators
instead of absolute ones (Porter and Cunningham, 2004). Otherwise,
scholars risk paying disproportionate attention to those industrial ac-
tivities with high patenting activity, lowering the validity of their work.
In addition, our ndings offer relevant policy implications. Often,
policymakers rely on macroeconomic studies to assess the effectiveness
of large-scale policy instruments, such as subsidies (Bai et al., 2019;
Yang et al., 2022), taxes (Costantini et al., 2017b, 2015), or emission-
trading schemes (Hwang and Kim, 2017; Wang et al., 2020). Our re-
sults suggest that the results of such studies are highly sensitive to how
environmental technologies are classied. Using an inadequate tech-
nology clustering strategy could result in drawing an unrealistic image
of policies' effectiveness, overestimating it for some technologies and
underestimating it for others. To reduce the risks of an inaccurate policy
assessment, we would recommend increasing the granularity of the
patent data. We also believe international institutions should critically
reect on their classication of green patents to ensure alignment with
current technological developments and offer researchers taxonomies
that in turn fuel policymaking.
In this study, we have used the widely-used logistic and Gompert
models to t and normalize the data. In order to increase their studies'
validity, scholars may consider using additional, more complex mathe-
matical models (Meade and Islam, 2006). One example is the Loglet
function (Meyer et al., 1999), which allows for the decomposition of a
growth or diffusion curve into several S-shaped logistic components,
representing consecutive waves of innovation. Furthermore, scholars
may consider using alternative measures of maturity, such as citation
network analysis (Ogawa and Kajikawa, 2015) or other indicators of
technological emergence (Porter et al., 2019), to provide a more
nuanced view of the quality and impact of patents. Doing so will help
improve our understanding of how environmental innovations emerge
and grow.
CRediT authorship contribution statement
Maryam Mazaheri: Writing – review & editing, Writing – original
draft, Software, Methodology, Investigation, Formal analysis,
Fig. 3. Patenting activity across the six clusters. The horizontal bars represent
the number of patents in each subcluster. The dotted line represents the growth
in the last ve years of each subcluster. The solid vertical lines represent the
average growth in the last ve years of each cluster.
M. Mazaheri et al.
Technological Forecasting & Social Change 209 (2024) 123813
6
Conceptualization. Jaime Bonnin Roca: Writing – review & editing,
Writing – original draft, Validation, Supervision, Methodology,
Conceptualization. Arjan Markus: Writing – review & editing, Writing –
original draft, Supervision, Conceptualization. Elena M. Tur: Writing –
review & editing, Writing – original draft, Validation, Data curation.
Bob Walrave: Writing – review & editing, Writing – original draft,
Validation, Supervision, Conceptualization.
Appendix A. Number of patents, and 5-year growth, of each subcluster in our dataset
Cluster Subcluster Total no. of
patents
Weight in database
(%)
Weight in cluster
(%)
Growth
Cluster
1
1.1.2.1 Post-combustion technologies (Emissions abatement from mobile sources) 658,811 17.65 % 33.25 % 16 %
Cluster
1
1.1.3.1 Post-combustion technologies (Not elsewhere classied) 606,825 16.26 % 30.63 % 56 %
Cluster
1
1.1.1.1 Postcombustion technologies (Emissions abatement from stationary sources) 218,813 5.86 % 11.04 % 57 %
Cluster
1
1.3.2. Material recovery, recycling and re-use 151,955 4.07 % 7.67 % 46 %
Cluster
1
1.3.4. Incineration and energy recovery 109,772 2.94 % 5.54 % 37 %
Cluster
1
1.3.6. Waste management – Not elsewhere classied 104,053 2.79 % 5.25 % 35 %
Cluster
1
1.3.3. Fertilizers from waste 42,260 1.13 % 2.13 % 80 %
Cluster
1
1.4. soil remediation 31,443 0.84 % 1.59 % 66 %
Cluster
1
1.2.2. Fertilizers from wastewater 29,131 0.78 % 1.47 % 47 %
Cluster
1
1.2.1. Water and wastewater treatment 20,341 0.55 % 1.03 % 24 %
Cluster
1
1.5. environmental monitoring 3898 0.10 % 0.20 % 139 %
Cluster
1
1.1.3.2 Integrated technologies 2014 0.05 % 0.10 % 30 %
Cluster
1
1.3.1. Solid waste collection 1953 0.05 % 0.10 % 8 %
Cluster
2
2.1.1.1 Faucets and showers 4164 0.11 % 6.52 % 14 %
Cluster
2
2.1.1.2 Aeration of water 5224 0.14 % 8.19 % 56 %
Cluster
2
2.1.1.2 Control of watering 2497 0.07 % 3.91 % 149 %
(continued on next page)
Fig. 4. Distribution of patent classes by maturity level, per cluster. The length of each bar represents the percentage of patent families which are in each maturity
level (0–25 %, 25–50 %, 50–75 %, 75–100 %). Clusters have been sorted according to the percentage of technologies in the rst stage of their technology lifecycle.
M. Mazaheri et al.
Technological Forecasting & Social Change 209 (2024) 123813
7
(continued)
Cluster Subcluster Total no. of
patents
Weight in database
(%)
Weight in cluster
(%)
Growth
Cluster
2
2.1.1.3 Drought-resistant crops 2149 0.06 % 3.37 % 65 %
Cluster
2
2.1.1.3 Sanitation (dual-ush toilets, dry toilets, Closed-circuit toilets) 7773 0.21 % 12.18 % 17 %
Cluster
2
2.1.1.4 Greywater 2862 0.08 % 4.48 % 95 %
Cluster
2
2.1.2.1 Drip irrigation 4810 0.13 % 7.54 % 207 %
Cluster
2
2.1.3. Water conservation in thermoelectric power production 15,807 0.42 % 24.77 % 50 %
Cluster
2
2.1.4.1 Piping – reducing leakage and leakage monitoring 2980 0.08 % 4.67 % 84 %
Cluster
2
2.2.1.1 Underground water collection 2638 0.07 % 4.13 % 74 %
Cluster
2
2.2.1.2 Surface water collection 5827 0.16 % 9.13 % 41 %
Cluster
2
2.2.1.3 Rainwater water collection 4401 0.12 % 6.90 % 156 %
(continued on next page)
Fig. 5. Distribution of subclusters by maturity level, across the six clusters. The length of each bar represents the percentage of patent classes in a subcluster which
are at each of the four stages of the technology lifecycle.
M. Mazaheri et al.
Technological Forecasting & Social Change 209 (2024) 123813
8
(continued)
Cluster Subcluster Total no. of
patents
Weight in database
(%)
Weight in cluster
(%)
Growth
Cluster
2
2.2.2. Water storage 2688 0.07 % 4.21 % 99 %
Cluster
3
3.1.1. Wind energy 95,584 2.56 % 12.15 % 43 %
Cluster
3
3.1.2. Solar thermal energy 108,552 2.91 % 13.80 % 26 %
Cluster
3
3.1.3. Solar photovoltaic (PV) energy 139,509 3.74 % 17.73 % 35 %
Cluster
3
3.1.4. Solar thermal-PV hybrids 3452 0.09 % 0.44 % 64 %
Cluster
3
3.1.5. Geothermal energy 6582 0.18 % 0.84 % 38 %
Cluster
3
3.1.6. Marine energy 14,038 0.38 % 1.78 % 35 %
Cluster
3
3.1.7. Hydro energy 56,763 1.52 % 7.22 % 25 %
Cluster
3
3.2.1. Biofuels 32,556 0.87 % 4.14 % 33 %
Cluster
3
3.2.2. Fuel from waste 34,972 0.94 % 4.45 % 41 %
(continued on next page)
Fig. 5. (continued).
M. Mazaheri et al.
Technological Forecasting & Social Change 209 (2024) 123813
9
(continued)
Cluster Subcluster Total no. of
patents
Weight in database
(%)
Weight in cluster
(%)
Growth
(continued on next page)
Fig. 5. (continued).
Fig. 6. Number of patents vs computed maturity, divided per cluster. Each dot represents a single patent class. We found no connection between the number of
patents in a patent class, and its maturity.
M. Mazaheri et al.
Technological Forecasting & Social Change 209 (2024) 123813
10
(continued)
Cluster Subcluster Total no. of
patents
Weight in database
(%)
Weight in cluster
(%)
Growth
Cluster
3
3.3.1. Technologies for improved output efciency (Combined heat and power,
combined cycles, etc.)
17,792 0.48 % 2.26 % 21 %
Cluster
3
3.3.2. Technologies for improved input efciency (Efcient combustion or heat usage) 22,123 0.59 % 2.81 % 44 %
Cluster
3
3.4.1. Nuclear fusion reactors 4829 0.13 % 0.61 % 9 %
Cluster
3
3.4.2. Nuclear ssion reactors 63,311 1.70 % 8.05 % 8 %
Cluster
3
3.5.1. Superconducting electric elements or equipment 39,631 1.06 % 5.04 % 37 %
Cluster
3
3.6.1.1. Batteries 30,815 0.83 % 3.92 % 40 %
Cluster
3
3.6.1.2. Capacitors 17,721 0.47 % 2.25 % 51 %
Cluster
3
3.6.1.3. Thermal storage 17,729 0.48 % 2.25 % 31 %
Cluster
3
3.6.1.4. Pressurised uid storage 1870 0.05 % 0.24 % 45 %
Cluster
3
3.6.1.5. Mechanical storage 3150 0.08 % 0.40 % 36 %
Cluster
3
3.6.1.6. Pumped storage 2640 0.07 % 0.34 % 34 %
Cluster
3
3.6.2. Hydrogen technology 26,762 0.72 % 3.40 % 33 %
Cluster
3
3.6.3. Fuel cells 20,465 0.55 % 2.60 % 24 %
Cluster
3
3.6.4. Smart grids in the energy sector 17,200 0.46 % 2.19 % 76 %
Cluster
3
3.7. other energy conversion or management systems reducing ghg emissions 8631 0.23 % 1.10 % 63 %
Cluster
4
4.1. Co2 capture or storage (ccs) 13,679 0.37 % 77.06 % 28 %
Cluster
4
4.2. Capture or disposal of greenhouse gases other than CO2 4071 0.11 % 22.94 % 23 %
Cluster
5
5.1.1. Conventional vehicles (based on internal combustion engine) 217,408 5.83 % 35.56 % 20 %
Cluster
5
5.1.2 Hybrid vehicles 51,253 1.37 % 8.38 % 26 %
Cluster
5
5.1.3 Electric vehicles 159,772 4.28 % 26.14 % 41 %
Cluster
5
5.1.4. Electric energy management in electromobility 42,060 1.13 % 6.88 % 33 %
Cluster
5
5.1.5. Fuel efciency-improving vehicle design (common to all road vehicles) 16,063 0.43 % 2.63 % 41 %
Cluster
5
5.2. Railtransport 5846 0.16 % 0.96 % 43 %
Cluster
5
5.3. Air transport 41,026 1.10 % 6.71 % 26 %
Cluster
5
5.4. maritime or waterways transport 11,356 0.30 % 1.86 % 29 %
Cluster
5
5.5.1. Electric vehicle charging 58,575 1.57 % 9.58 % 46 %
Cluster
5
5.5.2. Application of fuel cell and hydrogen technology to transportation 7967 0.21 % 1.30 % 37 %
Cluster
6
6.1. integration of renewable energy sources in buildings 65,569 1.76 % 24.16 % 41 %
(continued on next page)
Table 3
Highest Total no. of patents in level 1 maturity for each subcluster.
Patent class Maturity
level
Growth Total no.
patents
Cluster Subcluster
Y02E 10/50,
cpc
0–25 % 27 % 63,396 Climate change mitigation technologies related to energy generation,
transmission of distribution
3.1.3. Solar photovoltaic (PV) energy
Y02T 10/
7005,cpc
0–25 % 79 % 51,134 Climate change mitigation technologies related to transportation 5.1.3 Electric vehicles
Y02B 10/20,
cpc
0–25 % 26 % 21,706 Climate change mitigation technologies related to buildings 6.1. Integration of renewable energy
sources in buildings
Y02C 10/08,
cpc
0–25 % 35 % 4094 Capture, storage, sequestration or disposal of greenhouse gases 4.1. CO2 capture or storage (ccs)
F16L 55/07,
ipc
0–25 % 61 % 3905 Water-related adaptation technologies 2.1.1.2 Aeration of water
M. Mazaheri et al.
Technological Forecasting & Social Change 209 (2024) 123813
11
(continued)
Cluster Subcluster Total no. of
patents
Weight in database
(%)
Weight in cluster
(%)
Growth
Cluster
6
6.2.1. Lighting 76,266 2.04 % 28.10 % 74 %
Cluster
6
6.2.2. Heating, ventilation or air conditioning [HVAC] 61,535 1.65 % 22.67 % 35 %
Cluster
6
6.2.3. Home appliances 13,896 0.37 % 5.12 % 45 %
Cluster
6
6.2.4. Elevators, escalators and moving walkways 1412 0.04 % 0.52 % 33 %
Cluster
6
6.2.6. End-user side 31,845 0.85 % 11.73 % 45 %
Cluster
6
6.3. architectural or constructional elements improving the thermal performance of
buildings
7358 0.20 % 2.71 % 36 %
Cluster
6
6.4. enabling technologies in buildings 13,548 0.36 % 4.99 % 38 %
Sum 3,732,271 100.00 %
Appendix B. PATSTAT search query
/* before running this code:
1) create a table called green_ipc and populate it with all the ipc codes to be selected
2) create a table called green_cpc and populate it with all the cpc codes to be selected
*/
if object_id('green_data', 'U') IS NOT NULL
drop table green_data
;
create table green_data (
class_symbol varchar(19),
type_symbol varchar(3),
priority_yr int,
num_fams int
)
;
insert into green_data
(class_symbol, type_symbol, priority_yr, num_fams)
select green.ipc_class_symbol as class_symbol, 'ipc' as type_symbol,
app.earliest_filing_year as priority_yr, count(distinct docdb_family_id) as
num_fams
from green_ipc green /* this is the table with the ipc codes, that you created before running
the code */
join tls209_appln_ipc ipc on ipc.ipc_class_symbol = green.ipc_class_symbol
join tls201_appln app on app.appln_id = ipc.appln_id
group by green.ipc_class_symbol, app.earliest_filing_year
;
insert into green_data
(class_symbol, type_symbol, priority_yr, num_fams)
select green.cpc_class_symbol as class_symbol, 'cpc' as type_symbol,
app.earliest_filing_year as priority_yr, count(distinct docdb_family_id) as
num_fams
from green_cpc green /* this is the table with the cpc codes, that you created before running
the code */
join tls224_appln_cpc cpc on cpc.cpc_class_symbol = green.cpc_class_symbol
join tls201_appln app on app.appln_id = cpc.appln_id
group by green.cpc_class_symbol, app.earliest_filing_year
Appendix C. Supplementary data
Supplementary data to this article can be found online at https://doi.org/10.1016/j.techfore.2024.123813.
M. Mazaheri et al.
Technological Forecasting & Social Change 209 (2024) 123813
12
Data availability
Patent data is publicly available. We have shared a supplementary
le with the results of our statistical analysis.
References
Adamuthe, A.C., Thampi, G.T., 2019. Technology forecasting: a case study of
computational technologies. Technol. Forecast. Soc. Change 143, 181–189. https://
doi.org/10.1016/j.techfore.2019.03.002.
Afeltra, G., Alerasoul, S.A., Strozzi, F., 2021. The evolution of sustainable innovation:
from the past to the future. Eur. J. Innov. Manag. 26, 386–421. https://doi.org/
10.1108/EJIM-02-2021-0113.
Albitar, K., Al-Shaer, H., Liu, Y.S., 2023. Corporate commitment to climate change: the
effect of eco-innovation and climate governance. Res. Policy 52, 104697. https://
doi.org/10.1016/j.respol.2022.104697.
Alkemade, F., Suurs, R.A.A., 2012. Patterns of expectations for emerging sustainable
technologies. Technol. Forecast. Soc. Change 79, 448–456. https://doi.org/10.1016/
j.techfore.2011.08.014.
Altuntas, S., Dereli, T., Kusiak, A., 2015. Forecasting technology success based on patent
data. Technol. Forecast. Soc. Change 96, 202–214. https://doi.org/10.1016/j.
techfore.2015.03.011.
Azam, M., 2016. Does environmental degradation shackle economic growth? A panel
data investigation on 11 Asian countries. Renew. Sust. Energ. Rev. 65, 175–182.
https://doi.org/10.1016/j.rser.2016.06.087.
Bai, Y., Song, S., Jiao, J., Yang, R., 2019. The impacts of government R&D subsidies on
green innovation: evidence from Chinese energy-intensive rms. J. Clean. Prod. 233,
819–829. https://doi.org/10.1016/j.jclepro.2019.06.107.
Barbieri, N., Marzucchi, A., Rizzo, U., 2020. Knowledge sources and impacts on
subsequent inventions: do green technologies differ from non-green ones? Res.
Policy 49, 103901. https://doi.org/10.1016/j.respol.2019.103901.
Bekamiri, H., Hain, D.S., Jurowetzki, R., 2024. PatentSBERTa: a deep NLP based hybrid
model for patent distance and classication using augmented SBERT. Technol.
Forecast. Soc. Change 206, 123536. https://doi.org/10.1016/j.
techfore.2024.123536.
Borghesi, S., Cainelli, G., Mazzanti, M., 2015. Linking emission trading to environmental
innovation: evidence from the Italian manufacturing industry. Res. Policy 44,
669–683. https://doi.org/10.1016/j.respol.2014.10.014.
Castellani, D., Marin, G., Montresor, S., Zanfei, A., 2022. Greeneld foreign direct
investments and regional environmental technologies. Res. Policy 51, 104405.
https://doi.org/10.1016/j.respol.2021.104405.
Cecere, G., Corrocher, N., Gossart, C., Ozman, M., 2014. Technological pervasiveness and
variety of innovators in Green ICT: a patent-based analysis. Res. Policy 43,
1827–1839. https://doi.org/10.1016/j.respol.2014.06.004.
Chen, T., Wong, Y.D., Wang, X., Li, D., 2022. Investigating consumers’ intention of using
contactless logistics technology in COVID-19 pandemic: a Copula-Bayesian Network
approach. Int J Log Res Appl 0, 1–21. https://doi.org/10.1080/
13675567.2022.2116410.
Cheng, C.C., Shiu, E.C., 2012. Validation of a proposed instrument for measuring eco-
innovation: an implementation perspective. Technovation 32, 329–344. https://doi.
org/10.1016/j.technovation.2012.02.001.
Cohen, L., Gurun, U.G., Nguyen, Q.H., 2020. The ESG-Innovation Disconnect: Evidence
From Green Patenting. In: Working Paper Series. https://doi.org/10.3386/w27990.
Corrocher, N., Mancusi, M.L., 2021. International collaborations in green energy
technologies: what is the role of distance in environmental policy stringency? Energy
Policy 156, 112470. https://doi.org/10.1016/j.enpol.2021.112470.
Costantini, V., Crespi, F., Martini, C., Pennacchio, L., 2015. Demand-pull and technology-
push public support for eco-innovation: the case of the biofuels sector. Res. Policy
44, 577–595. https://doi.org/10.1016/j.respol.2014.12.011.
Costantini, V., Crespi, F., Marin, G., Paglialunga, E., 2017a. Eco-innovation, sustainable
supply chains and environmental performance in European industries. In: J. Clean.
Prod., Making, Buying and Collaborating for More Sustainable Production and
Consumption, 155, pp. 141–154. https://doi.org/10.1016/j.jclepro.2016.09.038.
Costantini, V., Crespi, F., Palma, A., 2017b. Characterizing the policy mix and its impact
on eco-innovation: a patent analysis of energy-efcient technologies. Res. Policy 46,
799–819. https://doi.org/10.1016/j.respol.2017.02.004.
Cvijanovi´
c, Vladimir, Hollanders, Hugo, Reid, Alasdair, 2021. How to Measure
Environmental Innovation. European Commission.
De Marchi, V., 2012. Environmental innovation and R&D cooperation: empirical
evidence from Spanish manufacturing rms. Res. Policy 41, 614–623.
de Rassenfosse, G., van Pottelsberghe de la Potterie, B., 2009. A policy insight into the
R&D–patent relationship. Res. Policy 38, 779–792. https://doi.org/10.1016/j.
respol.2008.12.013.
Fabrizi, A., Guarini, G., Meliciani, V., 2018. Green patents, regulatory policies and
research network policies. Res. Policy 47, 1018–1031. https://doi.org/10.1016/j.
respol.2018.03.005.
Flostrand, A., Pitt, L., Bridson, S., 2020. The Delphi technique in forecasting– a 42-year
bibliographic analysis (1975–2017). Technol. Forecast. Soc. Change 150, 119773.
https://doi.org/10.1016/j.techfore.2019.119773.
Fusillo, F., 2023. Green technologies and diversity in the knowledge search and output
phases: evidence from European Patents. Res. Policy 52, 104727. https://doi.org/
10.1016/j.respol.2023.104727.
Ganguly, I., Koebel, C.T., Cantrell, R.A., 2010. A categorical modeling approach to
analyzing new product adoption and usage in the context of the building-materials
industry. Technol. Forecast. Soc. Change 77, 662–677. https://doi.org/10.1016/j.
techfore.2009.10.011.
Gao, L., Porter, A.L., Wang, J., Fang, S., Zhang, X., Ma, T., Wang, W., Huang, L., 2013.
Technology life cycle analysis method based on patent documents. In: Technol.
Forecast. Soc. Change, Future-oriented Technology Analysis, 80, pp. 398–407.
https://doi.org/10.1016/j.techfore.2012.10.003.
García-Granero, E.M., Piedra-Mu˜
noz, L., Galdeano-G´
omez, E., 2020. Measuring eco-
innovation dimensions: the role of environmental corporate culture and commercial
orientation. Res. Policy 49, 104028. https://doi.org/10.1016/j.respol.2020.104028.
Ghisetti, C., Quatraro, F., 2017. Green technologies and environmental productivity: a
cross-sectoral analysis of direct and indirect effects in Italian regions. Ecol. Econ.
132, 1–13. https://doi.org/10.1016/j.ecolecon.2016.10.003.
Ghisetti, C., Montresor, S., Vezzani, A., 2021. Design and environmental technologies:
does ‘green-matching’ actually help? Res. Policy 50, 104208. https://doi.org/
10.1016/j.respol.2021.104208.
Girod, B., Stucki, T., Woerter, M., 2017. How do policies for efcient energy use in the
household sector induce energy-efciency innovation? An evaluation of European
countries. Energy Policy 103, 223–237. https://doi.org/10.1016/j.
enpol.2016.12.054.
Griliches, Z., 1990. Patent statistics as economic indicators: a survey. J. Econ. Lit. 28,
1661–1707.
Gupta, R., Jain, K., 2012. Diffusion of mobile telephony in India: an empirical study. In:
Technol. Forecast. Soc. Change, Contains Special Section: Innovation in India, 79,
pp. 709–715. https://doi.org/10.1016/j.techfore.2011.08.003.
Haessler, P., Giones, F., Brem, A., 2023. The who and how of commercializing emerging
technologies: a technology-focused review. Technovation 121, 102637. https://doi.
org/10.1016/j.technovation.2022.102637.
Haˇ
sˇ
ciˇ
c, I., Migotto, M., 2015. Measuring Environmental Innovation Using Patent Data.
Hinge, G., Surampalli, R.Y., Goyal, M.K., Gupta, B.B., Chang, X., 2021. Soil carbon and its
associate resilience using big data analytics: for food security and environmental
management. Technol. Forecast. Soc. Change 169, 120823. https://doi.org/
10.1016/j.techfore.2021.120823.
Hoang, T.-H.-V., Przychodzen, W., Przychodzen, J., Segbotangni, E.A., 2020. Does it pay
to be green? A disaggregated analysis of U.S. rms with green patents. Bus. Strateg.
Environ. 29, 1331–1361. https://doi.org/10.1002/bse.2437.
Horbach, J., 2008. Determinants of environmental innovation—new evidence from
German panel data sources. Res. Policy 37, 163–173. https://doi.org/10.1016/j.
respol.2007.08.006.
Huang, Y., Li, R., Zou, F., Jiang, L., Porter, A.L., Zhang, L., 2022. Technology life cycle
analysis: from the dynamic perspective of patent citation networks. Technol.
Forecast. Soc. Change 181, 121760. https://doi.org/10.1016/j.
techfore.2022.121760.
Huenteler, J., Schmidt, T.S., Ossenbrink, J., Hoffmann, V.H., 2016. Technology life-
cycles in the energy sector — technological characteristics and the role of
deployment for innovation. Technol. Forecast. Soc. Change 104, 102–121. https://
doi.org/10.1016/j.techfore.2015.09.022.
Hwang, J.-A., Kim, Y., 2017. Effects of environmental regulations on trade ow in
manufacturing sectors: comparison of static and dynamic effects of environmental
regulations. Bus. Strateg. Environ. 26, 688–706. https://doi.org/10.1002/bse.1965.
Jeffrey, H., Sedgwick, J., Gerrard, G., 2014. Public funding for ocean energy: a
comparison of the UK and U.S. Technol. Forecast. Soc. Change 84, 155–170. https://
doi.org/10.1016/j.techfore.2013.08.006.
Jiang, L., Zou, F., Qiao, Y., Huang, Y., 2022. Patent analysis for generating the
technology landscape and competition situation of renewable energy. J. Clean. Prod.
378, 134264. https://doi.org/10.1016/j.jclepro.2022.134264.
Kamateri, E., Salampasis, M., Perez-Molina, E., 2024. Will AI solve the patent
classication problem? World Patent Inf. 78, 102294. https://doi.org/10.1016/j.
wpi.2024.102294.
Kim, B., 2003. Managing the transition of technology life cycle. Technovation 23,
371–381. https://doi.org/10.1016/S0166-4972(02)00168-2.
Klepper, S., 1996. Entry, exit, growth, and innovation over the product life cycle. Am.
Econ. Rev. 86, 562–583.
Kodama, F., 2004. Measuring emerging categories of innovation: modularity and
business model. In: Technol. Forecast. Soc. Change, Rethinking Technology Policy
and Innovation, 71, pp. 623–633. https://doi.org/10.1016/S0040-1625(03)00084-
2.
Lee, C., Kim, J., Kwon, O., Woo, H.-G., 2016. Stochastic technology life cycle analysis
using multiple patent indicators. Technol. Forecast. Soc. Change 106, 53–64.
https://doi.org/10.1016/j.techfore.2016.01.024.
Lee, H., Lee, S., Yoon, B., 2011. Technology clustering based on evolutionary patterns:
the case of information and communications technologies. Technol. Forecast. Soc.
Change 78, 953–967. https://doi.org/10.1016/j.techfore.2011.02.002.
Leiponen, A., Helfat, C.E., 2010. Innovation objectives, knowledge sources, and the
benets of breadth. Strateg. Manag. J. 31, 224–236. https://doi.org/10.1002/
smj.807.
Lerner, J., Seru, A., 2022. The use and misuse of patent data: issues for nance and
beyond. Rev. Financ. Stud. 35, 2667–2704. https://doi.org/10.1093/rfs/hhab084.
Lezama-Nicol´
as, R., Rodríguez-Salvador, M., Río-Belver, R., Bildosola, I., 2018.
A bibliometric method for assessing technological maturity: the case of additive
manufacturing. Scientometrics 117, 1425–1452. https://doi.org/10.1007/s11192-
018-2941-1.
Liao, Z., 2018. Environmental policy instruments, environmental innovation and the
reputation of enterprises. J. Clean. Prod. 171, 1111–1117. https://doi.org/10.1016/
j.jclepro.2017.10.126.
M. Mazaheri et al.
Technological Forecasting & Social Change 209 (2024) 123813
13
Losacker, S., 2022. ‘License to green’: regional patent licensing networks and green
technology diffusion in China. Technol. Forecast. Soc. Change 175, 121336. https://
doi.org/10.1016/j.techfore.2021.121336.
Luan, C., Deng, S., Porter, A.L., Song, B., 2024. An approach to construct technological
convergence networks across different IPC hierarchies and identify key technology
elds. IEEE Trans. Eng. Manag. 71, 346–358. https://doi.org/10.1109/
TEM.2021.3120709.
Marinakis, Y.D., 2012. Forecasting technology diffusion with the Richards model.
Technol. Forecast. Soc. Change 79, 172–179. https://doi.org/10.1016/j.
techfore.2011.02.010.
Martino, J.P., 1992. Technological Forecasting for Decision Making, 3rd edition. ed.
McGraw-Hill, New York.
Martino, J.P., 2003. A review of selected recent advances in technological forecasting. In:
Technol. Forecast. Soc. Change, TF Highlights from ISF 2002, 70, pp. 719–733.
https://doi.org/10.1016/S0040-1625(02)00375-X.
Martino, J.P. (Ed.), 2020. An Introduction to Technological Forecasting, 1st edition. ed.
Routledge.
Martins, V.W.B., Rampasso, I.S., Anholon, R., Quelhas, O.L.G., Leal Filho, W., 2019.
Knowledge management in the context of sustainability: literature review and
opportunities for future research. J. Clean. Prod. 229, 489–500. https://doi.org/
10.1016/j.jclepro.2019.04.354.
Mazaheri, M., Bonnin Roca, J., Markus, A., Walrave, B., 2022. Market-based instruments
and sustainable innovation:a systematic literature review and critique. J. Clean.
Prod. 373, 133947. https://doi.org/10.1016/j.jclepro.2022.133947.
Meade, N., Islam, T., 1998. Technological forecasting—model selection, model stability,
and combining models. Manag. Sci. 44, 1115–1130. https://doi.org/10.1287/
mnsc.44.8.1115.
Meade, N., Islam, T., 2006. Modelling and forecasting the diffusion of innovation – a 25-
year review. In: Int. J. Forecast., Twenty Five Years of Forecasting, 22, pp. 519–545.
https://doi.org/10.1016/j.ijforecast.2006.01.005.
Meyer, P.S., Yung, J.W., Ausubel, J.H., 1999. A primer on logistic growth and
substitution: the mathematics of the Loglet Lab software. Technol. Forecast. Soc.
Change 61, 247–271. https://doi.org/10.1016/S0040-1625(99)00021-9.
Morgan, M.G., 2014. Use (and abuse) of expert elicitation in support of decision making
for public policy. Proc. Natl. Acad. Sci. 111, 7176–7184. https://doi.org/10.1073/
pnas.1319946111.
Nelson, K.P., Parton, L.C., Brown, Z.S., 2022. Biofuels policy and innovation impacts:
evidence from biofuels and agricultural patent indicators. Energy Policy 162,
112767. https://doi.org/10.1016/j.enpol.2021.112767.
Ning, L., Guo, R., 2022. Technological diversication to green domains: technological
relatedness, invention impact and knowledge integration capabilities. Res. Policy 51,
104406. https://doi.org/10.1016/j.respol.2021.104406.
Ogawa, T., Kajikawa, Y., 2015. Assessing the industrial opportunity of academic research
with patent relatedness: a case study on polymer electrolyte fuel cells. Technol.
Forecast. Soc. Change 90, 469–475. https://doi.org/10.1016/j.
techfore.2014.04.002.
P´
erez, C., Soete, L., 1988. Catching up in Technology: Entry Barriers and Windows of
Opportunity.
Perruchas, F., Consoli, D., Barbieri, N., 2020. Specialisation, diversication and the
ladder of green technology development. Res. Policy 49, 103922. https://doi.org/
10.1016/j.respol.2020.103922.
Porter, A.L., Cunningham, S.W., 2004. Tech Mining: Exploiting New Technologies for
Competitive Advantage. John Wiley & Sons.
Porter, A.L., Cunningham, S.W., Banks, J., Roper, A.T., Mason, T.W., Rossini, F.A., 2011.
Forecasting and Management of Technology, 2nd edition. ed. Wiley, Hoboken, N.J.
Porter, A.L., Garner, J., Carley, S.F., Newman, N.C., 2019. Emergence scoring to identify
frontier R&D topics and key players. Technol. Forecast. Soc. Change 146, 628–643.
https://doi.org/10.1016/j.techfore.2018.04.016.
Reiffenstein, T., 2009. Specialization, centralization, and the distribution of patent
intermediaries in the USA and Japan. Reg. Stud. 43, 571–588. https://doi.org/
10.1080/00343400701874255.
Rotolo, D., Hicks, D., Martin, B.R., 2015. What is an emerging technology? Res. Policy
44, 1827–1843. https://doi.org/10.1016/j.respol.2015.06.006.
Savin, I., Winker, P., 2012. Heuristic optimization methods for dynamic panel data
model selection: application on the Russian innovative performance. Comput. Econ.
39, 337–363. https://doi.org/10.1007/s10614-010-9243-x.
Savin, I., Ott, I., Konop, C., 2022. Tracing the evolution of service robotics: insights from
a topic modeling approach. Technol. Forecast. Soc. Change 174, 121280. https://
doi.org/10.1016/j.techfore.2021.121280.
Senna, P.P., Barros, A.C., Bonnin Roca, J., Azevedo, A., 2023. Development of a digital
maturity model for Industry 4.0 based on the technology-organization-environment
framework. Comput. Ind. Eng. 185, 109645. https://doi.org/10.1016/j.
cie.2023.109645.
Sohrabpour, V., Oghazi, P., Toorajipour, R., Nazarpour, A., 2021. Export sales forecasting
using articial intelligence. Technol. Forecast. Soc. Change 163, 120480. https://
doi.org/10.1016/j.techfore.2020.120480.
Tattershall, E., Nenadic, G., Stevens, R.D., 2021. Modelling trend life cycles in scientic
research using the Logistic and Gompertz equations. Scientometrics 126,
9113–9132. https://doi.org/10.1007/s11192-021-04137-0.
Taylor, M., Taylor, A., 2012. The technology life cycle: conceptualization and managerial
implications. In: Int. J. Prod. Econ., Sustainable Development of Manufacturing and
Services, 140, pp. 541–553. https://doi.org/10.1016/j.ijpe.2012.07.006.
Tushman, M.L., Anderson, P., 1986. Technological discontinuities and organizational
environments. Adm. Sci. Q. 31, 439–465.
Utterback, J.M., Abernathy, W.J., 1975. A dynamic model of process and product
innovation. Omega 3, 639–656. https://doi.org/10.1016/0305-0483(75)90068-7.
Van Sark, W., 2008. Introducing errors in progress ratios determined from experience
curves. Technol. Forecast. Soc. Change 75, 405–415.
Vik, J., Melås, A.M., Stræte, E.P., Søraa, R.A., 2021. Balanced readiness level assessment
(BRLa): a tool for exploring new and emerging technologies. Technol. Forecast. Soc.
Change 169, 120854. https://doi.org/10.1016/j.techfore.2021.120854.
Wang, C.-C., Sung, H.-Y., Huang, M.-H., 2016. Technological evolution seen from the
USPC reclassications. Scientometrics 107, 537–553. https://doi.org/10.1007/
s11192-016-1851-3.
Wang, H., Wei, W., 2020. Coordinating technological progress and environmental
regulation in CO2 mitigation: the optimal levels for OECD countries & emerging
economies. Energy Econ. 87, 104510. https://doi.org/10.1016/j.
eneco.2019.104510.
Wang, W., Li, Y., Lu, N., Wang, D., Jiang, H., Zhang, C., 2020. Does increasing carbon
emissions lead to accelerated eco-innovation? Empirical evidence from China.
J. Clean. Prod. 251, 119690. https://doi.org/10.1016/j.jclepro.2019.119690.
Wurlod, J.-D., Noailly, J., 2018. The impact of green innovation on energy intensity: an
empirical analysis for 14 industrial sectors in OECD countries. Energy Econ. 71,
47–61. https://doi.org/10.1016/j.eneco.2017.12.012.
Yang, S., Wang, W., Feng, D., Lu, J., 2022. Impact of pilot environmental policy on urban
eco-innovation. J. Clean. Prod. 341, 130858. https://doi.org/10.1016/j.
jclepro.2022.130858.
Yuan, X., Cai, Y., 2021. Forecasting the development trend of low emission vehicle
technologies: based on patent data. Technol. Forecast. Soc. Change 166, 120651.
https://doi.org/10.1016/j.techfore.2021.120651.
Maryam Mazaheri is a PhD candidate in the Innovation, Technology Entrepreneurship &
Marketing (ITEM) group at Eindhoven University of Technology. She holds a Master's
degree in Innovation Management. Her PhD project focuses on the institutional promotion
of sustainable innovation.
Jaime Bonnin Roca is an Assistant Professor of Innovation and Entrepreneurship in the
Innovation, Technology Entrepreneurship & Marketing (ITEM) group at Eindhoven Uni-
versity of Technology. He is interested in how public and private organizations respond to
uncertainty in the adoption of emerging technologies, with a focus on advanced
manufacturing. In addition, his research analyses how to balance safety and innovation in
the introduction of immature technologies, especially in highly regulated environments.
Arjan Markus is an Assistant Professor with tenure in the Innovation, Technology
Entrepreneurship, and Marketing (ITEM) group at Eindhoven University of Technology
(TU/e). His research focuses on networks and ecosystems in innovation. He holds a Ph.D.
from Copenhagen Business School and BSc and MSc degrees from Utrecht University.
During his Ph.D. and as a postdoc, he visited the Wharton School of the University of
Pennsylvania. Prior to joining the Eindhoven University of Technology, Arjan worked as
an Assistant Professor at Tilburg University.
Elena Mas Tur is an assistant professor at the Technology, Innovation & Society group of
the Eindhoven University of Technology (TU/e). Her main research lines include mathe-
matical models of innovation, simulation models of diffusion, complex networks and
patentometrics. Elena Mas Tur obtained her PhD in Innovation Studies from Utrecht
University, in 2016. She also holds degrees in Mathematics and Statistics as well as an MSc
in Industrial Economics. She is a member of the International Network for Social Network
Analysis and a member of the International Schumpeter Society.
Bob Walrave is an Associate Professor of Modeling Innovation Systems at the Eindhoven
University of Technology, and the chair of the Innovation, Technology Entrepreneurship,
and Marketing group. Bob Walrave holds a Ph.D. in strategic innovation management from
the same university. His research interests are centered around strategic decision making
in dynamically complex situations in the context of innovation management and entre-
preneurship. He is particularly interested in modeling innovation systems within and
across public and private organizations. Typically, he draws on longitudinal research
approaches, aiming to answer so-called ‘how’ questions—to develop process theories and
intervention mechanisms by means of system's thinking and system dynamics modeling.
M. Mazaheri et al.
Technological Forecasting & Social Change 209 (2024) 123813
14
