You are on page 1of 17

Copyright eContent Management Pty Ltd. Innovation: Management, policy & practice (2011) 13: 311326.

Innovation systems as patent


networks: The Netherlands, India
and nanotech
WILFRED DOLFSMA
School of Economics and Business, Universiteit van Groningen, Groningen, The Netherlands

LOET LEYDESDORFF
Amsterdam School of Communications Research (ASCoR), Universiteit van Amsterdam,
Amsterdam, The Netherlands

ABSTRACT
Research in the domain of Innovation Studies has been claimed to allow for the study of how technology
will develop in the future. Some suggest that the National and Sectoral Innovation Systems literature
has become bogged down, however, into case studies of how specic institutions affect innovation in a
specic country. A useful notion for policy makers in particular, Balzat and Hanusch (2004 Recent
trends in the research on national innovation systems Journal of Evolutionary Economics 14(2):
197210) argued that there is a need for NIS studies to develop complementary and also quantitative
methods in order to generate new insights that are comparable across national borders. We use data for
patents granted by the World Intellectual Property Organization (WIPO) to map innovation systems.
Groupings of patents into primary and secondary classes (co-classication) can be used as relational
indicators. Knowledge from one class may be more easily used in another class when a co-classication
relation exists. Using social network analysis, we map the co-classication of patents among classes and
thus indicate what characterizes an innovation system.
A main contribution of this paper is methodological, adding to the repertoire of methods NIS
studies use and using information from patents in a different way. Policy makers may also nd
benets in the social network analysis of the complete set of patents granted by the WIPO to rms
and individuals in a country. Social network analysis indicates what innovation activity occurs in
a countries and which elds of technology are likely to give rise to innovative products in the near
future. We offer such analysis for the Dutch and Indian Innovation Systems. This social network
analysis could also be done for a sector innovation system, and we do so for nanotech to determine
empirically the knowledge eld relevant for this emerging scientic domain.

Keywords: innovation systems; networks; patent networks; national innovation system

there is much more information derivable


from the patent documents than just simply their
aggregated numbers in a particular year or for a
B oth academics and policy makers have found
the National Innovation Systems (NIS) litera-
ture to contribute useful insights. Some, however,
particular rm. (Griliches 1990, p. 1664) believe that the NIS literature (Lundvall 1992;

Volume 13, Issue 3, December 2011 INNOVATION: MANAGEMENT, POLICY & PRACTICE 311
Wilfred Dolfsma and Loet Leydesdorff

Nelson 1993; Edquist 2004) has become bogged technological elds, which may allow for theoreti-
down into case studies of how specic institutions cal integration between NIS, regional innovation
affect innovation in a specic country. As Balzat systems (Cantner & Graf 2006) and sector inno-
and Hanusch (2004) argue: there is a need for vation systems (Dittrich & Duysters 2007; Storz
NIS studies to develop complementary and also 2008) where a similar network approach can be
quantitative methods in order to generate new adopted. Analysis of industrial production in terms
insights that are comparable across national bor- of inputoutput matrices adopts or could adopt
ders. In this paper we use data on patents granted similar methods (Verspagen 1997; Lenocini &
by the World Intellectual Property Organization Montresor 2000; Lotti & Santarelli 2001).
(WIPO), a UN organization, to map innovation The empirical analysis of NIS as patent net-
systems. Applying for a patent at the WIPO is works may thus open possibilities for theoretical
relatively easy and cheap, and allows the appli- integration of NIS to adjoining elds of academic
cant to both apply in relevant markets afterwards, research. Yet, the main contribution of this paper
and establish their position vis--vis competitors. may perhaps be methodological as it adds to the
Rather than taking counts of the number of pat- repertoire of methods NIS studies use, but also as
ents granted, by eld, company, region or coun- a different kind of information on patents is used.
try, which would meet with all the drawbacks In addition, policy makers are interested in know-
that patents have as an indicator for innovation ing how knowledge development in an innova-
(Kleinknecht et al. 2002), we use different infor- tion system is interrelated, and thus obtain an
mation that can be drawn from patents. understanding of how production structures may
Patents are grouped into a primary class evolve in the near future. It might also indicate
and secondary classes by patent examiners. which policy domains may emerge as important
Co-classication of a patent in two classes signies issues. For instance, using social network analy-
a relation between these classes that is signicant sis of the complete set of 3,287 patents granted
from the point of view of knowledge develop- by the WIPO to Dutch rms and individuals in
ment and thus for studying a knowledge-based 2006, we nd that biotech, pharmaceutical and
innovation system. Using social network analysis, chemical technology, with applications in food
one can map these co-classications among pat- and medication may be overtaking the tradition-
ent classes and thus characterize a NIS. Such an ally dominant position of electronics/computer
analysis of a NIS focuses on what nation specic technology. Given that these technological elds
components and relationships in a system, each and their associated industries show high propen-
with their characteristics and attributes (Carlsson sities to patent, the dependence of the Dutch NIS
et al. 2002), actually produce. It does so in a man- on patent law thus increases. We also perform
ner that indicates the relations between knowl- such an analysis for emerging economy India,
edge elds as well as, to some degree, the nature and separately for the technological domain of
of such relations as part of the larger (socio-cog- nanotechnology.
nitive) network conguration. In doing so, the
results of an analysis of (national) innovation sys- INNOVATION SYSTEMS LITERATURE
tems becomes amenable for comparison (between In terms of both direction and success rate, inno-
nations; Liu & White 2001). vation performance differs widely across rms
One may argue that this approach ignores the and organizations more generally grouped by
idiosyncracies of national systems, but one may regions, sectors or specically nations. At these
as well argue that such an analysis may enable us aggregated levels, a systems approach has been
to understand these idiosyncracies. In addition, popular since at least the early 1990s to under-
the analysis can focus on regions as well as specic stand emergent phenomena (Lundvall 1988;

312 INNOVATION: MANAGEMENT, POLICY & PRACTICE Volume 13, Issue 3, December 2011
Innovation systems as patent networks

Nelson 1993). A NIS approach assumes that instance bankruptcy are other institutions that
differences among the innovation performance play a role in knowledge diffusion.
across countries are due to their specicities and While a useful approach in many ways, the
idiosyncrasies that will not (simply) disappear promise of comparability across national sys-
due to market processes (Dolfsma et al. 2008). tems has been largely unmet (Liu & White 2001;
Players and other components of a system each Edquist 2004). This may be due to the hetero-
have their characteristics, while relationships geneous nature of the concept of NIS or its con-
among them affect both how they will contribute stituent parts (Bergek et al. 2008), but may as
to system outcomes as well as how each evolves well be due to the empirical and case-study based
over time (Carlsson et al. 2002). approach taken as is evident in history-friendly
An institutional perspective is often invoked analyses such as in Nelson (1993). Quite a few
(Liu & White 2001), claiming that both informal studies inspired by a NIS idea have focused on a
and formal institutions can be persistently differ- limited number of institutions, or even a single
ent among countries in a way that affects innova- one, to study their effect on innovation direc-
tion patterns and outcomes. Actors and networks tion and performance. The advantage of the
are also referred to in this respect (Carlsson & approach an awareness of idiosyncrasies may
Stankiewicz 1991), their workings perhaps best then become a drawback since from up close
understood in terms of institutions and what they the differences between systems stand out more
legitimately allow or prescribe (Bergek et al. 2008; than the similarities. It can thus be challenging
Dolfsma & Verburg 2008) In this respect, and to point to causes for the differences or the simi-
contra Bergek et al. (2008), a system can be con- larities when comparing between systems. The
sidered as more than an analytical abstract: one can choice of institutions and the choice regarding
formulate as an empirical question whether or not the aggregation level for analysis can differ sub-
at the national level one can perceive coherence stantially between studies, resulting in a situation
in the institutional structure with regard to inno- where some believe that the approach has come
vation.1 The outcomes of the largely unplanned to be stranded due to the case study approach
workings of an innovation system may thus be dif- adopted.
cult to predict precisely, but can be approached. We concur with Balzat and Hanusch (2004)
The NIS approach is an attractive starting point that it is possible to salvage a NIS analysis by devel-
because of its coherence and usefulness for policy. oping additional, complementary approaches to
Using the nation as units of analysis, authors the study of national innovation patterns. These
often refer to its institutions that help create new include, but may not be restricted to, quantitative
knowledge, such as strong universities, an attrac- methods. One may have to sacrice to some extent
tive climate for private research institutes, possi- the attractive feature of a rich or, as anthropolo-
bilities for migrant knowledge workers to enter gist Geertz (1973) called it, thick description as
a country and a patent system. Many such insti- one focuses less on the workings of a system rather
tutions also play a role in knowledge diffusion.2 than the outcome, but one potentially gains as
A well-functioning education system, peoples comparability and rigor is enhanced.
attitude towards taking the risk of setting up a We propose to use patent data to study the
new rm, or a countrys laws with regard to for structural characteristics for the outcomes of an

1
Using a different indicator, Leydesdorff and Fritsch (2006) found that Germany cannot be considered integrated
nationally as an innovation system, while The Netherlands can (Leydesdorff et al. 2006).
2
The NIS literature may be more focused on knowledge creation, while knowledge diffusion may be of greater importance
to the knowledge economy (Leydesdorff et al. 2006).

Volume 13, Issue 3, December 2011 INNOVATION: MANAGEMENT, POLICY & PRACTICE 313
Wilfred Dolfsma and Loet Leydesdorff

innovation system. As a result of the interplay and its relevant institutions. This approach to the
among actors within an innovation system, behav- studying of innovation systems would seem to
iour which is determined by extant institutions, be comprehensive in the sense of using a full set
both the direction of technological development of data that gives some indication of the relevant
as well as the robustness of that pattern emerge output on an innovation system. We suggest that
(Rip & Kemp 1998; Carlsson et al. 2002). While this approach is complementary to other method-
analysis of a particular NIS or of a particular set of ological approaches developed to analyse innova-
players within a NIS allows for detailed analysis of tion systems.
the dynamics of a NIS over time (Storz 2008), the
approach we opt for also allows for comparison DATA AND METHOD
over time of the way in which elements and func- Patents
tions of a NIS (Liu & White 2001; Bergek et al. Patents have been a widely used type of data for
2008) produce innovations. This holds, we would innovation studies, in part because of their avail-
argue, for both an analysis of NIS as well as for ability. Patents granted in the US, for many sec-
an analysis of a sector or technological innovation tors the most important single market, are easily
system (Malerba & Orsenigo 1997; Storz 2008). downloadable from the USPTO website. Such
While patent data offers a quantitative mea- US data may not be relevant for the character-
sure, they are often used in a rather unimaginative ization of, for instance, a European country
way. Patents granted are aggregated to the level (Leydesdorff 2004; Criscuolo 2006). Patent data
of rms, regions, sectors or countries to deter- as a measure of innovativeness of a country, sector
mine the respective signicance of innovative or rm has more generally come under increased
activity for an entity at such an aggregated level. discussion. Patents as an output measure of inno-
While patent data is shaped by institutions, and vation is problematic many of them do not have
reects information about applications that is the any commercial value for rms (Carlsson et al.
result of institutional congurations, they are not 2002; Kleinknecht et al. 2002). As a result, the
dened by institutions a priori. Institutions them- propensity to patent differs widely across indus-
selves are embedded in knowledge infrastructures, tries (Arundel & Kabla 1998).
providing the technological opportunities that Of all patents granted in the US, 5575%
have to be interfaced with market positions and lapse and become a part of the public domain
expected demand that agents can act upon. through failure to pay maintenance fees; if litiga-
From this perspective, patent data offer a tion against a patents validity is a sign of commer-
vastly more informative source of information, cial value of that patent, the fact that only 1.5%
for instance about actual or potential knowledge of patents are litigated and only 0.1% litigated
ows in a system. We analyse patents granted to trial does not bode well (Lemley & Shapiro
as sediment of substantive-technical efforts by 2005; Dolfsma 2008). Many patents thus are
actors to develop new knowledge or nd new applied for only strategic reasons (Griliches 1990;
(non-obvious) applications for existing knowl- Granstrand 2000; Dolfsma 2011). Small rms
edge. Classes of patent applications, and particu- are thus reluctant to patent in the technological
larly co-classications, thus may be taken as an or economic (market) neighborhood of patents
important indicator of a mutual knowledge basis held by large companies for fear of expensive law
within the boundaries of a system (Breschi et al. suits (Lerner 1995; Lanjouw & Schankermann
2003; Leydesdorff 2008a). The network of co- 2004). In addition to this, the propensity to pat-
classications for patents, drawing on a unied ent differs between product and process innova-
and harmonized database, thus indicates the tions, and by sector (Arundel & Kabla 1998).
workings of an innovation system such as NIS Fewer process innovations are patented, since

314 INNOVATION: MANAGEMENT, POLICY & PRACTICE Volume 13, Issue 3, December 2011
Innovation systems as patent networks

secrecy as an alternative appropriability measure the rst inventor. Patent data accessibility provides
is more feasible. Propensity to patent also differs a degree of certainty to the innovator applying for
by sector. While this may somehow bias a pic- a patent as she can browse the patents applied for
ture of a NIS that is otherwise relatively unbiased or granted to determine whether the knowledge
because of the rather objective information used embodied in a patent is already legally protected.
and the systematic way in which to collect that In the US inventors who had not applied for a
information. Patents are becoming increasingly patent may challenge a patent granted if they can
important as a way to protect innovative knowl- convincingly show that they had been earlier to
edge, which means that the approach to innova- invent. On the other hand, in the US applicant
tion systems research that we suggest in this paper only needs to publish the information contained
will over time face fewer objections. in a patent after the patent is granted, while pub-
Patent law tries to nd a balance between the lication is required in Europe when a patent is
public interest of stimulating development of new applied for.
and the wide diffusion of existing knowledge, on As a patent application can be rejected, the
the one hand, and the private interest of return on inventor thus runs the risk of diffusing her knowl-
investment, on the other. The institution of pat- edge without receiving the legal right to exclusive
ent law, then, both seeks to facilitate two functions commercial exploitation in return. This clause
of an innovation system: knowledge development in European patent legislation favors the public
and diffusion (Liu & White 2001). Whether or interest more. This obviously opens possibili-
not it is true that patent law stimulates knowledge ties for strategic behavior for parties (Granstrand
development and diffusion, to the same degree, 2000). However this may be, the important meth-
remains an empirical issue (Dolfsma 2008; Lerner odological point to make is that the level of analy-
2009) but innovation is believed to be stimulated sis chosen for our analysis is that of the knowledge
if innovators have a legal right to exclusively eld (Carlsson et al. 2002). Specication of the
exploit the results of innovative efforts. boundaries of a knowledge eld is left in the
However, the balance is struck differently in hands of patent applicants and patent ofcers.
different countries (Organization for Economic
Cooperation and Development [OECD] 1997), WIPO patents
and in noticeably different ways. An important The WIPO (www.wipo.org), an organization
characteristic of the US patent law, for instance, residing under the UN based in Geneva, offers
which makes it unique, concerns who is deemed the possibility to easily and cheaply apply for a
to have the right in an invention: the one who rst patent. WIPO staff assists in drafting the patent
les a patent application with the patent ofce or application, which means that expensive addi-
the one who is able to prove he was the rst to tional technical and legal services that might be
invent. In the US the administratively less trac- required for a European or US application need
table rst-to-invent may claim the rights. Parties not be hired. The application for a patent sub-
who have been rst to invent may decide to come mitted at WIPO can subsequently be submitted
forward with their invention only after another in other countries or jurisdictions as well, within
party is granted a patent by the patent ofce. a specic time frame, if commercially attractive.
This can give rise to legal conicts that could pre- The legal systems of other countries recognize
vent or limit the commercial use of the knowl- WIPO applications technically and legally. In
edge involved (cf Bittlingmayer 1988). A rst to addition, WIPO patents are part of the prior art
le system of administrating patent applications that patent ofcers need to consult in case they
seems to be an example of public interests of pro- receive an application from a different party on a
viding clarity outweighing the private interests of related technical invention. Such an application

Volume 13, Issue 3, December 2011 INNOVATION: MANAGEMENT, POLICY & PRACTICE 315
Wilfred Dolfsma and Loet Leydesdorff

may then have to be rejected, or can relatively indicating its technological eld. Based on
easily and cheaply be challenged in court by the an understanding of the substantive technical
patentee of a WIPO patent. In addition, due to knowledge in the patent, it is determined in what
the the low cost of applying for a patent at the technological domain or paradigm (Dosi 1982)
WIPO as well, a larger number of small rms and it belongs and on what previous knowledge it
applications from low-income countries apply for draws. In a substantial number of cases patents
a patent at WIPO. draw on knowledge previously developed in
World Intellectual Property Organization different paradigms. In addition, patent exam-
patent protection is thus an accessible means to iners provide co-classications as well. The clas-
obtain legal protection for an invention that may sication and co-classication indicates actual or
have industrial applications. Especially for par- potential knowledge transfer between different
ties that lack nancial means, this makes apply- technological elds (Verspagen 2006). The 1994
ing for a WIPO patent attractive. Relatively small OECD Manual (OECD 1994, p. 52) mentions
rms and parties from developing or emerging patent co-classications as a potential indicator of
economies may nd applying for a WIPO patent linkages among technologies. Based on the pat-
particularly attractive. Such parties may also have ents granted to a particular entity, it can be estab-
defensive motives to apply for a patent, prevent- lished how the knowledge base of the entity can
ing others to le for a patent in the same area. The be characterized empirically (Engelsman & van
tendency for smaller rms to shy away from R&D Raan 1994; Breschi et al. 2003).
deployment in areas where larger rms already Together with the European Patent Ofce
have a patent position for fear of being sued (EPO), a.o.,3 WIPO invests substantial resources in
by these rms may thus diminish (Lanjouw & developing the International Patent Classication
Schankermann 2004). (IPC). Currently the eighth edition is in use, using
Litigation in patent law, specically in the US, a 12-digit coding system covering some 70,000
has grown increasingly rife, where especially large patent classes. Because of the standardized nature
rms reserve substantial funds to legally defend of the data presented in the WIPO database, pat-
their patent position even if technically their posi- ents registered there are a good source for data on
tion might not seem particularly strong. Much innovation. Comparison across countries, or an
patenting, again in the US in particular, is thus of analysis of a specic sector across country bound-
a offensively strategic nature (Lemley and Shapiro aries is also possible using this data. However,
2005). Needless to say, the number of patents patent data from WIPO are not perfect as an indi-
applied for has increased substantially in recent cator. As noted earlier, the propensity to patent,
years, at the USPTO as well as at WIPO where for example, is known to differ substantially across
strategic patenting is less dominant. We may con- sectors. Also, on average only 35% of product and
clude that, as a source of information about tech- 25% of process innovations are patented (Arundel
nological development and innovation, WIPO & Kabla 1998). The propensity to patent prod-
data are more valuable. uct innovation ranges between 8 and 80%. Other
means may be deployed, and be deemed more
2006 WIPO patent data important, to protect a rms intellectual property.
Patent databases are a rich source of information Secrecy is one of these (Levin et al. 1987).
(Griliches 1990). One can do more than count Controlling for the number of citations to a
the number of patents for each country or rm. patent, sometimes advocated as a way to control
Each patent is, for instance, given a classication for the quality of patents in an analysis, may not

3
Inpadoc in Vienna, Austria.

316 INNOVATION: MANAGEMENT, POLICY & PRACTICE Volume 13, Issue 3, December 2011
Innovation systems as patent networks

present a better picture of the importance of or (US). If the number of patents granted to parties
value for a patent as, at least for patents granted in a system is very low, a patent network may not
by the EPO, there is only a very tenuous rela- present a useful view of an innovation system.
tion between patent value and number of cita- We analyze at the four-digit level patent data.
tions (Gambardella et al. 2008).4 This may be Circles or nodes in these gures are patent classes.
due to the fact that many citations in an appli- Size of the nodes could be drawn such that they
cation are included at the behest of the patent reect the number of patents which have received
ofcer (Griliches 1990, p. 1689). Nelson (2009) such a (co-) classication. If we would have opted
concurs, suggesting that licenses of a patent to to do this, it would reect a raw count of the
other rms are a better indicator. Patents that are number of patents in such a class. Given that our
not licensed to other rms might, however, be aim is to provide indication of the way in which
important for the knowledge base of an innova- technological domains or paradigms relate, we
tion system nonetheless. Given these consider- focus on the ties connecting the nodes. Lines
ations we believe that the concerns raised about between classes indicate co-classication, while
the use of patents in the context of analysis of thickness of lines indicates the number of co-
innovation systems (Carlsson et al. 2002, p. 241) classications. The thicker the lines the more often
are properly addressed when co-classication co-classications occur in the data. Thickness of
information from patent data are used only and the lines reect something different from a raw
have decided to use such data in our analysis. patent count. While the number of patents in a
class applied for by actors in a country reects the
Network analysis research strengths of that countrys innovation sys-
From a network point of view, whenever co- tem in a way, we believe that the relations between
classication between patent classes occurs, the classes provide insights that are complementary.
classes thus connected may be said to be connected Since clustering of research strengths can be easily
or tied together. The more such co-classications traced and given that science increasingly devel-
occur, the stronger the tie between the classes. We ops where scientists from different disciplines
use classications and co-classications for patents interact and collaborate (Wuchty et al. 2007), we
to analyze the innovation system as a network of believe that the insights provided by the networks
related technological classes. For any of the 132 of co-classied patents offer a fuller understand-
countries where individuals or organizations are ing. As these databases containing information of
located that had been granted one of the 138,751 patents granted can be publically searched, this
patents by the WIPO in 2006 a network analysis means more knowledge is actually or potentially
of the NIS as a patent network can be prepared. exchanged between classes. WIPO recognizes 624
We observe a spread in terms of the number of different major patent classes. The average num-
patents granted to parties in countries from 2 ber of classications per patent is 2.4.
(Brunei Darussalam, Burundi, Cayman Islands, Social network analysis allows one to construct
Cte DIvoire, Guatamala, San Marino, Seychelles, a visual presentation for the NIS (De Nooy et al.
Tuvalu, Uzbekistan, Virgin Islands) to 48,190 2005). To enhance readibility, weaker relations

4
Using US data, Hall et al. (2005) nd that the number of citations to patents correlates with the value of the rm holding
the patents measured in terms of Tobins Q. Since the US patent ofce, the USPTO is known to be quite lenient when
granting patents and given the size and importance of the US market for most rms, the signal of owning a patent granted
by the USPTO might need to be complemented with additional information. They nd self-citations, i.e., citations to
other patents owned by the same rm, in particular to correlate to the market value of the rm. For the purposes of the
analysis here, the focus is less on the specic rm that owns patents, nor their monetary value, but rather the focus is how
knowledge develops and may subsequently be exchanged in an innovation system.

Volume 13, Issue 3, December 2011 INNOVATION: MANAGEMENT, POLICY & PRACTICE 317
Wilfred Dolfsma and Loet Leydesdorff

between classes can be excluded from a picture by The Dutch innovation system
imposing a threshold of a minimum number of A factor analysis to nd out if a cluster of patent
co-classications before a relation is included in a classes among the 3,287 patents granted to par-
gure (either by using a k-core or the very much ties in the Netherlands can be determined based
similar m-slice threshold level to trace clusters or on the extent to which they co-classify (available
cliques of more strongly interconnected sets of upon request from the authors) provides little
patent classes; cf. De Nooy et al. 2005). clues as to which patent classes may be combined
to form factors that help explain variance in the
INNOVATION SYSTEMS data. The factor analysis is explored as well at the
To give indication of the usefulness of the three-digit aggregation level. Only a small frac-
approach suggested, in what follows, we compare tion of the variance (11.8%) is explained cluster-
two NIS. For purposes of interpretation of the ing nine factors. This indicates that the Dutch
results we take the Netherlands as a developed innovation system is quite dispersed and broadly
country and India as an emerging country. The based, which may be a sign of its relative maturity.
Netherlands is a highly developed economy that Social network analysis offers as a highly attrac-
also turns out to be a mature innovation system. tive and informative possibility the option of
On the other hand, India is both an emerging visualization. Given the visualization of Figure 1,
economy as well as an emerging innovation it is clear that the innovation complex around
system. One element which shows a relatively Eindhoven and the north of the Limburg province
strong presence in India is nanotechnology. The is strongly present. Some 50% of R&D formally
sector innovation system for nanotechnology, spent by Dutch rms is spent in these two NUTS
that we also present to indicate the potential of three regions (Leydesdorff et al. 2006). This is
the kind of analysis we propose, shows a regime the region where electronics, computer and infor-
in development. mation processing technology, and optics cluster

FIGURE 1: PATENT CLASSIFICATION CATEGORIES AND CO-CLASSIFICATION RELATIONS; CORE NETWORK FOR INTERNATIONALLY
REGISTERED DUTCH PATENTS (K-CORE = 10; 2006; N = 3,287). SOURCE: WIPO

318 INNOVATION: MANAGEMENT, POLICY & PRACTICE Volume 13, Issue 3, December 2011
Innovation systems as patent networks

related to such rms as Philips, Oc, ASML, and FrieslandCampina). Despite its small size and
supplying rms. A second large cluster, however, high population density, the Netherlands remains
is that of chemical technology, biotech, and phar- the second largest exporter of agricultural pro-
maceutical technology, especially with applica- duce, to some extend undoubtedly founded in its
tions in medication and functional foods. Even knowledge base in the area.
though rms in these sectors are more likely to Innovation policy as developed by the Dutch
apply for patents in case of an innovation than government (Innovatieplatform 2004, 2009)
rms in different sectors, the cluster is larger and largely emphasizes the technological areas that
more closely knit than expected. This interre- our analysis shows the Netherlands to be strong
lated set of classes (chemical technology, biotech, in already. Chemistry, food and owers are clearly
and pharmaceutical technology, especially with present in Figure 1. Almost off necessity, the cre-
applications in medication and functional foods) ative industries are not, since patents to works
is not generally recognized as important in the of art are not generally granted. This obviously
Dutch innovation cluster, and its link to clusters reects on the approach advocated here, but
that are recognized as key may go largely unno- might equally give a clue as to what such a sector
ticed (Innovatieplatform 2004, 2009; Ministry may be expected to contribute. Other domains
of Economic Affairs 2005). It is certainly a rel- such as electronics and computer technology are
atively younger cluster in terms of innovation not a focus of innovation policy, while they seem
focus, while the link to functional food indicates to t with the idea on which this policy is based
a relation to an traditional strength of the Dutch of backing winners (Nooteboom & Stam 2008).
economy. A new high-tech stronghold seems to Our analysis thus suggests possible oversights
be developing. of important technical domains, as well as pos-
In addition to these larger clusters of techno- sible (future) connections by the comprehensive
economic activities, smaller clusters can be appre- empirical approach (Dolfsma 2009).
ciated in the visualization. Chemical technology
related to application in paints is visible in the The national innovation system of
network representation of the Dutch NIS despite India
the high (k-core) threshold applied. Multinational Nine hundred and thirty six patents were granted
companies such as DSM and AkzoNobel are in 2006 to parties located in India. Without
active here. So is oil rening, with companies such throwing up a threshold of number of co-
as Royal Dutch/Shell, despite the focus on process classications any tie between two classes should
innovation and thus a lower propensity to patent have in order to enter the picture, the picture
if only for this reason in this mature industry. becomes quite difcult to understand, let alone
Given the recent change in strategic emphasis of interpret (as an illustration included as Figure A1
the Dutch industrial behemoth Philips, the clus- in the appendix).
ter indicating medical diagnostic equipment may Introducing such a threshold of, for example,
change too, and possibly move in the direction of a m-slice of two (2), produces Figure 2. Here one
the electronics/computing cluster. Packaging, for nds that the Indian innovation systems main
example of food stuffs, is on the verge of being strengths seem to be in chemistry, a relatively
included. It is a sector that develops new prod- mature industry and knowledge eld. Chemistry
ucts that year-upon-year are perceived by experts is a well-connected eld in the Indian innovation
to be highly innovative and valuable. Consumers system with that nds applications in food, clean-
have been impressed more by the functional food ing and medical elds. Medical applications seem
mostly by dairy industrialists such as Campina important, which may be where Indias known
and Friesland Foods (recently merged to form strengths in production of generic medication

Volume 13, Issue 3, December 2011 INNOVATION: MANAGEMENT, POLICY & PRACTICE 319
Wilfred Dolfsma and Loet Leydesdorff

FIGURE 2: PATENT CLASSIFICATION CATEGORIES AND CO-CLASSIFICATION RELATIONS; CORE NETWORK FOR INTERNATIONALLY
REGISTERED INDIAN PATENTS (M-SLICE = 2; 2006; N = 936). SOURCE: WIPO

shows (Chittoor et al. 2009). The set of compa- but these are ltered out very quickly for India
nies involved in this industry, including Ranbaxy when applying a k-core threshold. This is no
and Dr. Reddys Laboratories, is reported to move surprise as nanotechnology is only recognized
into the phase of developing medication them- as a separate class recently. For India, it can be
selves, rather than copying medical innovations observed that nanotechnology only has a single
developed elsewhere. The industry is known to tie with another patent class (Soil working in
show a high propensity to patent, and so India agriculture forestry etc.).
can be expected to see its presence in patent data- Nanotechnology indeed is a separate eld of
bases enhanced in the years to come. knowledge development, at least from the per-
What is striking is that another strong segment spective of application also when taking a sector
in the Indian economy, IT, is quite visibly pres- innovation systems approach: an increasing num-
ent in the upper left and bottom right corners. ber of patents are being granted. The approach to
Despite the fact that IT in India is mostly focused understanding innovation systems developed in
on services aspects, for which, of course, possibili- this paper suggests how the eld relates to other
ties to apply for patents in most countries except technological domains. Conceptually, thus, the
for the US is restricted. approach advocated relates to the discussion of
technological regimes (Dosi 1982) as it allows
Nanotechnology for a better understanding a domain through an
Patents have been granted to parties in India in analysis of its connections with adjacent domains.
the eld of nanotechnology (patent tag Y01N),5 Using ideas developed in this paper for a sector or
5
Nanotechnology was recently added as an additional tag to the existing database for all nano-technologies (Scheu et al. 2006).

320 INNOVATION: MANAGEMENT, POLICY & PRACTICE Volume 13, Issue 3, December 2011
Innovation systems as patent networks

technological innovation system approach, it is pos- development in the area of nanotechnology at


sible to indicate which technological areas are close present largely plays out in the scientic journals,
to nanotechnology eld globally. When patents in where its dynamics might be of a very different
any particular technological eld are co-classied nature (Leydesdorff 2008b).
with another eld one may expect knowledge to be
exchanged between these classes. By repeating this DISCUSSION AND CONCLUSIONS
analysis for the course of a number of years, the The literature on innovation systems has a ten-
changing position of nanotechnology within the dency to get bogged down in case studies of the
larger knowledge eld can be indicated. effects of specic institutions, for example on
Figure 3 pictures the Nanotech Innovation innovation in a specic countries, regions or spe-
System without imposing an additional (k-core) cic sectors and technologies (Balzat & Hanusch
threshold. One can thus conclude that the area 2004). Analyzing co-classication relations of
of nanotechnology is rather loosely connected to patents granted by WIPO opens up avenues for
other technological regimes or elds, even though research in the analysis of Systems of Innovation
a process of fusion might be under way to become that promise to provide a more comprehensive
visible in patent applications some time from now perspective that is complementary with studies
(Islam & Miyazaki 2009). Based on our analy- that have adopted a qualitative approach.
sis, the way in which the technology is related to From a methodological point of view, acknowl-
other elds of research does not show any struc- edging the drawbacks of patent data, discussed at
turally specic shape yet. While this may change length above, the approach suggested allows for a
in the future, it thus does appear to constitute a great deal of exibility and rigor. From national
separate regime. Knowledge dynamics shaping point of view, which technological domains
this eld has not yet developed to the extent that within a country are relatively stronger, as least in
many industrial applications are to be expected, terms of patents applied for, becomes apparent at
making such knowledge patentable. Knowledge a glance. In addition, the way in which domains

FIGURE 3: TECHNOLOGICAL NEIGHBORHOOD FOR NANOTECHNOLOGY, Y01N (N = 762; K-CORE = 1; VISUALIZATION


BASED ON THE ALGORITHM OF KAMADA AND KAWAI (1989). SOURCE: LEYDESDORFF (2008b)

Volume 13, Issue 3, December 2011 INNOVATION: MANAGEMENT, POLICY & PRACTICE 321
Wilfred Dolfsma and Loet Leydesdorff

may interrelate in ways that may not be expected somewhat of a rmer base. These may be input
can be indicated as well. One may easily shift from for a better understanding of how specic insti-
a focus on NIS to regional innovation systems. tutions affect different aspects of innovation
A social network approach using patent data can systems. Address details in patent data can allow
also provide insights into sectoral or technical for a better understanding of how the relations
innovation systems. It shows empirically where between actors help shape innovation systems
technical domains draw their knowledge from (Cantner & Graf 2006).
and where its knowledge is used. Developments The picture that emerges for the Dutch inno-
in a specic technical domain may be plotted vation system, for example, is both familiar and
geographically as well, and can be animated over somewhat surprising. What is to be expected is
time (Leydesdorff & Rafols 2011). The approach the strong presence of the electronics, computer,
suggested could thus be history friendly (Nelson and optical clusters. Internationally well-rec-
1993) as well. ognized and established industrial rms such as
In this paper we have mainly done so in rela- electronics giant Philips, world leader in semi-
tion to NIS, but similar analyses can be under- conductor productions ASML, or producer of
taken for sector and regional innovation systems. copy machines Oc feature in this corner of the
Making use of patent data in such a way has not innovation system. What is more surprising is the
been suggested before in innovation studies and strongly intertwined chemical, biotechnical and
so we offer three distinct contributions in this pharmaceutical cluster, especially with applica-
paper. tion for (veterinary) medication and (functional)
First, we offer a possible avenue for future food. The presence of this element in the Dutch
research in the area of innovation studies and NIS is not generally recognized, and, given the
the eld of NIS in particular that allows for high propensity to patent in the related industries
comparison of innovation systems as well as a (Arundel & Kabla 1998), would suggest that the
more comprehensive analysis of the develop- Dutch innovation becomes increasingly depen-
ment of a system over time.6 Secondly, we show dent on intellectual property law. For India, the
that widely available data on patents can be put picture may be more surprising even. IT, which in
to a different and more comprehensive use than India is known to be a strong sector, is noticeably
has hitherto been done. This is a methodologi- absent, while chemistry and pharmacy are most
cal advance of particular importance for innova- strongly present. Looking at nanotech, globally,
tion and industry studies. Thirdly, drawing on taking a sector innovation systems approach, it
this, we offer empirical insights into important appear that it is very much a technology in devel-
aspects of particular innovation systems, the opment, currently rather loosely connected to the
Dutch and Indian innovation systems as well as broader set of technological elds as currently
the sector innovation system for nanotechnol- recognized.
ogy, that might suggest some important policy
implications as well. These insights can be rep- ACKNOWLEDGEMENTS
licated for any other nation or sector for which This paper was presented at a number of semi-
comparable data is available. Especially when nars and conferences. We would like to thank
studying developments over time, and taking all participants at EAEPE conference (Rome),
into account the phase in a life cycle or stage Innovatieplatform (The Hague), ISID (New
of development that a region, country or sec- Delhi) and UNAM (Mexico City). The usual dis-
tor is in, comparisons between these may have claimer applies.

6
Compare the analysis of scientic journals over time (Leydesdorff & Schank 2008).

322 INNOVATION: MANAGEMENT, POLICY & PRACTICE Volume 13, Issue 3, December 2011
Innovation systems as patent networks

References Dolfsma, W. (2008) Knowledge economies.


Arundel, A. and Kabla I. (1998) What percentage Routledge, London.
of innovations are patented? Empirical estimates Dolfsma, W. (2009) Kennis delen: Micro en macro.
for European rms, Research Policy 27: 127141. In: J.-P. van den Toren (Ed.), Connecting global
Balzat M. and Hanusch H. (2004) Recent trends ambitions: De Nederlandse manier van samenwerken.
in the research on national innovation Systems Den Haag, The Netherlands, Innovatieplatform.
Journal of Evolutionary Economics 14(2): Dolfsma, W. (2011) Patent strategizing, Journal of
197210. Intellectual Capital 12(2): 168178.
Bergek, A., Jacobsson, S., Carlson, B., Lindmark, Dolfsma, W., Chaminade, C. and Vang, J. (Eds.)
S. and Rickne, A. (2008) Analyzing the func- (2008) Knowledge dynamics out of balance.
tional dynamics of technological innovation Research Policy 37(10): 16571658.
systems: A scheme of analysis, Research Policy Dolfsma, W. and Verburg, R. (2008) Structure,
37: 407429. agency and the role of values in processes of
Bittlingmayer, G. (1988) Property rights, progress, institutional change, Journal of Economic Issues
and the aircraft patent agreement, Journal of 42(4): 10311054.
Law and Economics 31(1): 227248. Dosi, G. (1982) Technological paradigms and tech-
Breschi, S., Lissoni, F. and Malerba, F. (2003) nological trajectories, Research Policy 11: 147162.
Knowledge-relatedness in rm technological Edquist, C. (2004) Systems of innovation:
diversication, Research Policy 32(1): 6987. Perspectives and challenges. In: J. Fagerberg, D.
Cantner, U. and Graf, H. (2006) The Network of C. Mowery and R. R. Nelson (Eds.), The Oxford
Innovators in Jena: An application of social net- handbook of innovation. Oxford University Press,
work analysis, Research Policy 35(4): 463480. Oxford.
Carlsson, B., Jacobsson, S., Holmn, M. and Engelsman, E. C. and van Raan, A. F. J. (1994) A
Rickne, A. (2002) Innovation Systems: patent-based cartography of technology, Research
Analytical and methodological issues, Research Policy 23(1): 126.
Policy 31: 233245. Gambardella, A., Harhoff, D. and Verspagen, B.
Carlsson, B. and Stankiewicz, R. (1991) On the (2008) The value of European patents, European
nature, function, and composition of technolog- Management Review 5: 6984.
ical systems, Journal of Evolutionary Economics 1: Geertz, C. (1973) The interpretation of cultures.
93118. Basic Books, New York.
Chittoor, R., Sarkar, M. B., Ray, S. and Aulakh, Granstrand, O. (2000) The economics and man-
P. S. (2009) Thirdworld copycats to emerg- agement of intellectual property. Edward Elgar,
ing multinationals: Institutional changes and Cheltenham, VIC.
organizational transformation in the Indian Griliches, Z. (1990) Patent statistics as economic
pharmaceutical industry, Organization Science indicators, Journal of Economic Literature 28(4):
20(1): 187205. 16611707.
Criscuolo, P. (2006) The home advantage effect Hall, B. H., Jaffe, A. and Trajtenberg, M. (2005)
and patent families. A comparison of OECD Market value and patent citations, RAND
triadic patents, the USTPTO and EPO, Journal of Economics 36(1): 1638.
Scientometrics 66(1): 2341. Innovatieplatform (2004) Voorstellen
De Nooy, W., Mrvar, A. and Batagelj, V. (2005) Sleutelgebieden-aanpak. Den Haag, The
Exploratory social network analysis with Pajek. Netherlands, Innovatieplatform.
Cambridge University Press, Cambridge. Innovatieplatform (2009) Voortgang Sleutelgebieden
Dittrich, K. and Duysters, G. (2007) Networking en tussentijdse evaluatie Sleutelgebieden-aanpak.
as a means to strategic change: The case of Den Haag, The Netherlands, Innovatieplatform.
open innovation in mobile telephony, Journal Islam, N. and Miyazaki, K. (2009) Nanotechnology
of Product Innovation Management 24(6): innovation system: Understanding hidden
510521. dynamics of nanoscience fusion trajectories,

Volume 13, Issue 3, December 2011 INNOVATION: MANAGEMENT, POLICY & PRACTICE 323
Wilfred Dolfsma and Loet Leydesdorff

Technological Forecasting and Social Change 76: systems in Germany in terms of a triple helix
128140. dynamics Research Policy 35(10): 15381553.
Kamada, T. and Kawai, S. (1989) An algorithm for Leydesdorff, L. and Rafols, I. (2011) Local emergence
drawing general undirected graphs, Information and global diffusion of research technologies: An
Processing Letters 31(1): 715. exploration of patterns of network formation,
Kleinknecht, A., van Montfort, K. and Brouwer, E. Journal of the American Society for Information
(2002) The non-trivial choice between innova- Science and Technology 62(5): 846860.
tion indicators, Economics of Innovation and Leydesdorff, L. and Schank, T. (2008) Dynamic
New Technology 11: 109121. animations of journal maps: Indicators of struc-
Lanjouw, J. O. and Schankermann, M. (2004) tural changes and interdisciplinary developments,
Protecting intellectual property rights: Are Journal of the American Society for Information
small rms handicapped? Journal of Law and Science and Technology 59(11): 18101818.
Economics 47: 4574. Liu, X. and White, S. (2001) Comparing innova-
Lemley, M. A. and Shapiro, C. (2005) Probabilistic tion systems: A framework and application to
patents, Journal of Economic Perspectives 19(2): Chinas transitional context, Research Policy 30:
7598. 10911114.
Lenocini, R. and Montresor, S. (2000) Network anal- Lotti, F. and Santarelli, E. (2001) Linking knowledge
ysis of eight technological systems, International to productivity: A GermanItaly comparison
Review of Applied Economics 14(2): 213234. using the CIS database, Empirica 28: 293317.
Lerner, J. (1995) Patenting in the shadow of competi- Lundvall, B.-. (1988) Innovation as an interac-
tors, Journal of Law and Economics 38: 463495. tive process From userproducer interaction
Lerner, J. (2009) The empirical impact of intellectual to national systems of innovation. In: G. Dosi,
property rights on innovation: Puzzles and clues, C. Freeman, R. Nelson, G. Silverberg and L.
American Economic Review 99(2): 343348. Soete (Eds.), Technical change and economic
Levin, R. C., Klevorick, A. K., Nelson, R. R. and theory. Pinter Publishers, London.
Winter, S. G. (1987) Appropriating the returns Lundvall, B.-. (Ed.) (1992) National systems of
from industrial research and development, innovation. Pinter Publishers, London.
Brookings Papers on Economic Activity 3: 783832. Malerba, F. and Orsenigo, L. (1997) Technological
Leydesdorff, L. (2004) The universityindustry regimes and sectoral patterns of innovative activi-
knowledge relationship: Analyzing patents and ties, Industrial and Corporate Change 6: 83117.
the science base of technologies, Journal of the Ministry of Economic Affairs (2005) Onderscheidend
American Society for Information Science and vermogen Sleutelgebieden-aanpak: Samen werken
Technology 55(11): 9911001. aan innovatie op kansrijke gebieden. The Hague,
Leydesdorff, L. (2008a) Patent classications as The Netherlands, Ministry of Economic Affairs.
indicators of intellectual organization, Journal of Nelson, A. J. (2009) Measuring knowledge spill-
the American Society for Information Science and overs: What patents, licenses and publications
Technology 59(10): 15821597. reveal about innovation diffusion, Research Policy
Leydesdorff, L. (2008b) The delineation of 38: 9941005.
nanoscience and nanotechnology in terms of Nelson, R. R. (Ed.) (1993) National innova-
journals and patents: A most recent update, tion systems: A comparative analysis. Oxford
Scientometrics 76(1): 159167. University Press, New York.
Leydesdorff, L., Dolfsma, W. and van der Panne, Nooteboom, B. and Stam, E. (Eds.) (2008) Micro-
G. (2006) Measuring the knowledge base foundations for innovation policy. Amsterdam
of an economy in terms of triple-helix rela- University Press, Amsterdam, The Netherlands.
tions among Technology, Organization, and Organization for Economic Cooperation and
Territory Research Policy 35(2): 181199. Development [OECD] (1994) The measurement
Leydesdorff, L. and Fritsch, M. (2006) Measuring of scientic and technological activities: Using pat-
the knowledge base of regional innovation ent data as science and technology indicators (Vol.

324 INNOVATION: MANAGEMENT, POLICY & PRACTICE Volume 13, Issue 3, December 2011
Innovation systems as patent networks

OCDE/GD(94)114). OECD, Paris. Accessed Storz, C. (2008) Dynamics in innovation systems:


http://www.oecd.org/dataoecd/33/62/2095942. Evidence from Japans game software industry,
pdf (Accessed July 2011). Research Policy 37: 14801491.
Organization for Economic Cooperation and Verspagen, B. (1997) Measuring intersectoral tech-
Development [OECD] (1997) Patents and inno- nology spillovers: Estimates from the European
vation in the international context. OECD, Paris. and US patent ofce databases, Economic
Rip, A. and Kemp, R. (1998) Technological Systems Research 9(1): 4764.
Change. In S. Rayner and L. Malone (Eds.), Verspagen, B. (2006) University research, intellectual
Human choice and climate change. Resources and property rights and European innovation systems,
Technology, Vol 2. Washington, DC, Batelle Journal of Economic Surveys 20(4): 607632.
Press, pp. 327399. Wuchty, S., Jones, B. and Uzzi, B. (2007) The
Scheu, M., Veefkind, V., Verbandt, Y., Galan, E. M., increasing dominance of teams in the produc-
Absalom, R. and Frster, W. (2006) Mapping tion of knowledge, Science 316: 10361039.
nanotechnology patents: The EPO approach,
World Patent Information 28: 204211. Received 14 March 2011 Accepted 29 July 2011

N O W AVA I L A B L E
NETWORK ANALYSIS APPLICATION PUBLIC SECTOR INNOVATION
IN INNOVATION STUDIES A special issue of Innovation: Management,
A special issue of Innovation: Management, Policy & Practice Volume 12 Issue 2
Policy & Practice Volume 12 Issue 1 ii+126 pages ISBN 978-1-921314-33-4 August 2010
120 pages ISBN 978-1-921348-32-7 April 2010 Editor: Jason Potts (School of Economics,
Editors: John Steen and Tim Kastelle (University of The University of Queensland, Queensland, Australia)
Queensland Business School, Queensland, Australia) Public sector innovation research: Whats next? Jason Potts
Introduction: Using network analysis to understand and Tim Kastelle
innovation Tim Kastelle and John Steen Innovative and entrepreneurial activity in the public sector:
The use of social network analysis in innovation studies: The changing face of public sector institutions Belinda Luke,
Mapping actors and technologies Tessa van der Valk and Martie-Louise Verreynne and Kate Kearins
Govert Gijsbers Innovative power of Dutch secondary education
Networking, entrepreneurship and productivity in universities Carla Haelermans
Alex Maritz Dreams of silence: Employee voice and innovation in a public
Investigating the structure of regional innovation system sector community of practice Francesca Gambarotto and
research through keyword co-occurrence and social network Alberto Cammozzo
analysis Pei-Chun Lee and Hsin-Ning Su The role of promoters in effecting innovation in higher
Networks for generating and for validating ideas: The social education institutions Stefan Hsig and Hans-Georg Mann
side of creativity Sandra Ohly, Robert Kase and Miha kerlavaj
Understanding Web 2.0s inuences on public e-services: A
Dynamics of a technological innovator network and its impact protection motivation perspective Cory Cromer
on technological performance Ju Liu and Cristina Chaminade
Public-private innovation: Mediating roles and ICT niches of
Are small world networks always best for innovation?
industrial research institutes Dzamila Bienkowska, Katarina
Tim Kastelle and John Steen
Larsen and Sverker Srlin
Inter-technology networks to support innovation strategy: An
analysis of Koreas new growth engines Sungjoo Lee and Innovation in IT outsourcing relationships: Where is the best
Moon-Soo Kim practice of IT outsourcing in the public sector? Junghoon
Moon, Bobby Swar, Young Chan Choe, Miri Chung and
Social capital and individual innovativeness in university
Gu Hyun Jung
research networks Cristbal Casanueva and ngeles Gallego
The enabling role of the public sector in innovation: A case
Book Reviews study of drug development in India Kavita Mehra and
Here Comes Everybody: The power of organizing without Kirti Joshi
organizations Clay Shirky Jason Potts and Kate Morrison Innovation by elimination: A proposal for negative policy
We Think: Mass innovation, not mass production experiments in the public sector Jason Potts
Charles Leadbeater & 257 other people Jason Potts and Evaluation copies available to course coordinators and
Kate Morrison collections librarians
http://www.innovation-enterprise.com/archives/vol/12/issue/1/ http://www.innovation-enterprise.com/archives/vol/12/issue/2/
marketing/ marketing/

eContent Management Pty Ltd, PO Box 1027, Maleny QLD 4552, Australia
Tel.: +61-7-5435-2900; Fax. +61-7-5435-2911;
subscriptions@e-contentmanagement.com www.e-contentmanagement.com

Volume 13, Issue 3, December 2011 INNOVATION: MANAGEMENT, POLICY & PRACTICE 325
326
APPENDIX
Wilfred Dolfsma and Loet Leydesdorff

INNOVATION: MANAGEMENT, POLICY & PRACTICE


Volume 13, Issue 3, December 2011
FIGURE A1: PATENT CLASSIFICATION CATEGORIES AND (COSINE-NORMALIZED) CO-CLASSIFICATION RELATIONS FOR INDIA, 2006 (N = 936; K-CORE = 1). Source: WIPO.
THE INPUT FILE FOR THIS FIGURE IS AVAILABLE AT HTTP://WWW.LEYDESDORFF.NET/WIPO06/IN.TXT IN THE PAJEK FORMAT
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

You might also like