Innovation research : trends and influences – a bibliometric analysis

In this paper, a bibliometric analysis about open innovation research is developed, covering the period of 2003-2013 (using the Scopus database) and carried out in three steps: 1) characterization of the research on the main trends of open innovation; 2) analysis of the theoretical influence on the open innovation research; 3) analysis of the influence of open innovation literature on other research areas and disciplines. The main conclusions are: open innovation research is mostly focused on the analysis of the U.S.A. and European countries reality; analysis by time periods shows an increase on the number of target countries and regions of open innovation research; the origins of open innovation were influenced by several areas of economics and management, developed over the last decades; there is a lack of research regarding open innovation outside the firm environment, such as in clusters/networks, innovation systems, public policies or at individual level; open innovation research is influencing a growing number of areas outside business, management and engineering; new research methodologies should be used by open innovation scholars in order to deepen the existing knowledge.


Introduction
The topic of open innovation has been one of the most discussed ones within the innovation management literature, receiving increasing attention in areas such as economics, psychology, sociology and culture (Huizingh, 2011). Existing literature on innovation has shown that changes in global economies in the recent decades (e.g. global competition, costs and risks associated to innovation activities, more qualified human resources, greater mobility of knowledge, etc.) have changed the way organizations manage and develop their innovation process (Chesbrough, 2003(Chesbrough, , 2006. In the last decades, scholars have acknowledged the growing importance of the collaboration of firms with external partners, the access to external networks, the incorporation of ideas and knowledge from various sources (internal and external), as well as the development of business models better suited for a more open reality in the innovation process (Chesbrough, 2006;Teece, 2007). Furthermore, the exploitation of technologies and ideas in the market and a more open and flexible approach of intellectual property are trends that companies and other organizations should look at, with more attention, in order to create value, taking advantage of the opportunities offered by the global markets of research and innovation (Helfat and Quinn, 2006). In 2003, based on the analysis of the economies' main trends, on a global level and also in innovation practices and processes of some firms, Chesbrough advanced with the concept of open innovation: "Open innovation means that valuable ideas can come from inside or outside the company and can go to market from inside or outside the associated with SMEs. The analysis of Chesbrough and Bogers (2014)

Methodology
For this bibliometric study, data was extracted on January 29 th 2014 from the Scopus database-the main worldwide database with regards to the number of scientific publications 2 -and the analysis held on the 30 th and 31 st January 2014 3 . The rationale behind the choice of Scopus as a means of data extraction is twofold: 1) Scopus is the largest academic database, integrating nearly 21,000 titles from more than 5,000 publishers, of which 20,000 are peer-reviewed journals ( based on the analysis of the references contained in those 339 articles (9,357 references found), through a) the identification of the most frequent words, which appear in the references' abstracts and b) title analysis of the articles from authors with more references (articles) cited by the 339 articles; • Step 3) analysis of the influence of open innovation literature, taking into account a) the characterization of the articles that cite those 339 articles (2807 documents found) and b) the analysis of the most "influential articles" ("seminal" articles, i.e., open innovation related articles most cited by other articles). Regarding the main trends' analysis and characterization of the research on open innovation (step 1), research was carried out in Scopus articles that contained the term "open innovation" in their title (339 articles found), of which: 321 were published in Journals; 10, in specialized publications (trade publications); five, in periodical collections (Book Series); and two, presented in conferences (conference proceedings). The target of this research is the analysis of academic papers that clearly focus on the theme "open innovation", avoiding papers that could focus on parallel areas of open innovation (this means that papers that have "open innovation" as keywords, are not included unless they also have "open innovation" in their title. For the same reason, books were not included in this analysis-only academic papers). Of the 339 articles, 327 were validated for the current analysis (of the step 1), since it was not possible to have access to the full text or abstracts of 12 articles. Of the 327 valid articles, a full download of 202 articles (60 % of total) was performed, then proceeding to the analysis of the remaining 125 articles' abstracts (since it was not possible to carry out the full download of these articles). Following a similar methodology, as the one used in other bibliometric studies (e.g. Cruz, 2007;Silva, 2008;Silva and Teixeira, 2009;Teixeira, 2014), 327 articles were analyzed and classified according to the topic of the research analysis, the technological intensity of the object studied and the size of the organizations analyzed (Table 1). The purpose of this classification is to verify changing patterns of open innovation research over time, namely if the research has been conducted in a certain direction or if there are distinct or emerging research directions. In particular, this classification provides information regarding: if the topic of analysis focuses on firm level or if there are other relevant or emerging areas of analysis outside the firm environment (e.g. clusters, innovation systems or public policies); if there has been a greater or lesser research orientation according to the technological-intensity of the object studied; if the size of organizations (including companies) is a relevant factor in the analysis that has been performed and if there are changes in the main thematic areas analyzed. The 327 articles were also classified according to the type of methodology used in their analysis (Table 1). The classification proposed in this paper is similar to the one used by Silva (2008) and Silva and Teixeira (2009), that follows the work of Nelson and Winter (1982), who proposed "formal theorizing" (development of a logic and structured theoretical argument or the usage of mathematical models) and "appreciative theorizing" (based more on explanations and concepts and not mathematically-based) as research methodologies in economics. Thus, following Nelson and Winter (1982) and based on the classification used by Silva (2008) and Silva and Teixeira (2009), the articles are going to be classified as "appreciative/survey" (critical reviews, theoretical or conceptual analysis, literature review), "empirical" (construction of an empirical evidence, through the usage of qualitative analysis-including case studies-and quantitative analysis-including the existence of statistical tests or econometric analysis) and "formal/empirical" (usage of mathematical models/simulations, with empirical construction through data analysis/econometric tests). references. To that respect, the references were first exported from Scopus in RIS format (Research Information Systems) and then software programs provided by Leydesdorff 7 (the "Scopus.exe" and "Scop2WOS.exe" programs) were used to extract the abstracts from the 9,357 references. The extracted abstracts were analyzed based on their most frequent words, using the CiteSpace software 8 , one of the most used in bibliometric analysis (Chen, 2006). In order to have a more detailed analysis, the titles of these references were also analyzed, concerning their main researched thematic areas. In this case and due to the complexity of the analysis and the large amount of data (9,357 references), it was necessary to limit the analysis to a reasonable number of data (Remneland-Wikhamn and Wikhamn, 2013). Thus, the analysis focused on the titles of references of those authors with more references (articles) cited by the 339 open innovation related articles-authors with 17 or more articles cited were chosen (35 authors found), producing a database of 957 articles (titles from the 957 articles copied from Scopus to the Excel program by the "copy-paste" procedure).
Concerning the analysis of the open innovation research influence (step 3), it was carried out in two phases. First, taking into account the identification of the "most influential articles" (open innovation related articles with more citations) 9 , by analyzing research areas, authors, affiliation, date and geographic areas. For this analysis 32 articles with at least 32 citations were identified, i.e., with an h index of 32 (Hirsch, 2005), which represent 72% of the total citations made to the 339 open innovation articles. Second, and to obtain a deeper understanding on the influence of open innovation articles, articles citing the 339 open innovation related articles (2087 articles found) were also analyzed, concerning their authors, affiliation, journal, research areas and geographic areas (results from Scopus).

Open innovation research: overview and main trends
The evolution of the publication of articles on open innovation has had a growing trend since the early work by Chesbrough (2003) This reflects the novelty of open innovation as an area of research, given the relatively small number of published articles, but also shows the academic community's growing interest in the theme within the last decade, especially after 2009 (Huizingh, 2011;Silva, 2008;Wang et al., 2012). We reached the same conclusion as we analyzed the evolution of the annual weight of open innovation related articles in relation to the total number of articles about innovation, published between 2003-2013 ( Figure 1): the percentage of articles that include the term "open innovation" in their title, in relation to the total number of articles that have the term "innovation" in their title, has more than tripled between 2004 and 2009 (an increase from 0.3 % to 1.1 %) and has increased 2,5 times between 2009 and 2013 (increasing from 1.1 % to 2.7 %). So, we can say that there is a more productive period where open innovation published articles is concerned (2010-2013) and a less productive period (2003)(2004)(2005)(2006)(2007)(2008)(2009)  In terms of research methodology employed by the 327 articles ( Figure 4), mostly they were empirical in nature (249 articles; 76 % of total), through the use of case studies at the firm level, sector of activity or entities from the scientific and academic system, and also using questionnaires/surveys (mainly in the recent years), with data treatment using statistical and/or econometric analysis. Formalization through mathematical models is still scarce, present only in 10 articles (3 %). It should also be noted that the component of literature review and/or theoretic/conceptual analysis is deeply present in most articles (139, or 43 %), as shown in Figure

Influence on open innovation research
The In the period 2003-2013, the 339 articles from Scopus, containing the term "open innovation" in their title have cited more than 9,000 references (more precisely, 9,386), of which 48% belong to publications presented in Scopus and 52% are reference lists associated to Scopus (data generated by Scopus). About 30% of the references (2,815) have been published in only 35 journals (which have published 33 or more references). Journals with the most published references are: "Research Policy" (399 articles), the "Strategic Management Journal" (215), "Technovation" (164) and "Organization Science" (138), which jointly represent about 10% of all cited references. As we saw in section 3, although they are not mainstream journals in economics, they are very important for the areas of innovation, namely in innovation management, corporate innovation, innovation research and evolutionary economics. Through the analysis of the references' abstracts, the most frequent words used can be identified (Table 2). In the 9,386 references published between the years of 1767 and 2013, some of the most common words are "innovation", "knowledge", "research", "firms", "technology/technological", "product", "development", "performance", "processes", "management", "business", "model", "industry", "market" or "organizational" ( Table 2, 5 th column, highlighted in green). Most part of those words can also be interconnected, highlighting some influential themes on open innovation: "research and development", "product innovation", "organizational innovation", "technological innovation", "product management", "product development", "knowledge management", "business model", "innovation management" or "technology management". These are major themes in the open innovation literature: e.g. "knowledge management" is directly linked to knowledge flows, external sources and the inbound and outbound open innovation processes (Dahlander and Gann, 2010;West and Bogers, 2013); "business model", "innovation management" and "organizational innovation" requiring a proper organizational structure to develop and adopt open innovation activities (Chesbrough and Schwartz, 2007;Rodríguez and Lorenzo, 2011); "research and development" with the R&D externalization or the firms' acquisition strategies (Chesbrough, 2006). Considering the references' publication time periods (using decades as time periods), it`s possible to analyze some trends over time: there is a common base of the most frequent words in all periods (in bold, frequently from a certain period (highlighted in green, Table 2). In all the reference periods, the most cited references are the ones dealing with innovation, research, firms, product, technology, process, market or development (most frequent words appearing in all the reference periods). The word "firm" is the 1 st or 2 nd most frequent in all the referred periods, except in the 2000`s (ranking in 4 th , with "innovation" as 1 st ). The word "knowledge" gained more attention over the last two decades, being the 4 th most cited in the 1990`s and the 2 nd after 2000 (although it has also appeared in references published before the 1980`s, but less frequently). The same happens with the words "performance" and "management", which were more frequent in the 1990`s and after 2000. After 2000, "open" and "business" appear in the list of the most frequent words in the references' abstracts (11 th and 12 th place, respectively), being also the period in which the concept of open innovation appeared. As we can see in Table 2, most of the cited references were published after 2000, (68% of the total), followed by the decade of 1990. This means that part of the influence on open innovation research derives from very recent literature and, probably, from authors who also write about open innovation.
To that respect, and based on the most frequent terms on the references` abstracts, one can say that major theoretical influences on open innovation seem to derive from the literature on innovation, management and the firm, regardless of the references' publication period. There is a clear focus on the firm`s reality-"firms" is always in the top 4 of the most frequent words, together with other words such as "product" or "business". However, there has been a growing importance of references focusing on knowledge, management and business in the last two decades-"knowledge management" and "business models" are core areas within the open innovation research (Chesbrough, 2003;Chesbrough and Bogers, 2014).  (Table 3).  Most part of the references cited by the 339 articles were published in the last two decades, notably in the 2000s and in the 1990s, which clearly portrays this research's burgeoning character on some of these issues-such as user innovation, innovation systems, open source or the debate around intellectual property and knowledge commercialization. As expected, most part of these references (

The influence of open innovation research
The 339 articles about open innovation published on Scopus were cited 4805 times by other studies (data up to December 31, 2013). However, 36.2 % (123 articles) never had any quote, while 63.8 % had at least one quote. It should be noted that relatively few articles have been cited many times, with only 4 % cited at least 100 times and 6 % at least 50 times. The 10 most cited articles account for 43.7 % of the total 4805 citations. If we consider only those "seminal" or "most influential" articles 22 , i.e., the 32 articles cited at least 32 times by other articles, represent 72 % (3479) of the total citations (Table 4). Of these 32 articles, 21 analyzed companies or entities in a given sector/cluster or comparison between sectors, half the articles analyze http://www.open-jim.org 152 large/multinational companies, and 41 % analyze technology-intensive companies/sectors. The analysis of SME (13 %) and mature or less intensive technology sectors (25 %) receives the attention from a smaller number of articles, following the trend of the research done by the majority of articles on open innovation, as seen in section 3. From amongst the 32 seminal papers (Table 4) Based on Table 4, we can construct a table of the most influential authors (Table 5), i.e. the ones with the highest number of articles cited among the 32 seminal articles. The most influential author is Henry Chesbrough, with 4 articles (2 of them in coauthorship), with the articles "The Era of Open Innovation" (Chesbrough is the only author) and "Beyond high tech: Early adopters of open innovation in other industries" (Chesbrough and Crowther) being the most cited articles by other articles (Table 5).
http://www.open-jim.org 155 Ulrich Lichtenthaler 24 also has 4 articles among the most cited, but with fewer citations than Chesbrough's. Chiaroni, Chiesa and Frattini (3 articles) and Henkel (2 articles) are the following. Research organizations with the most cited authors affiliated are located in the U.S. and Europe (Germany and Italy), namely the Haas School of Business at the University of Berkley (which welcomes authors such as Henry Chesbrough and Andrew Garman) and Politecnico di Milano, affiliating of authors like Chiaroni, Chiesa and Frattini (Table 5).  ), but also a higher weight of areas such as "Engineering", "Social Sciences", "Biochemistry, Genetics and Molecular Biology", "Environmental Science", "Agricultural and Biological Sciences", "Psychology", "Medicine", "Pharmacology, Toxicology and Pharmaceutics", "Arts and Humanities", "Chemistry", "Energy". higher number of entities per author`s affiliation. This is interesting, because in section 3 we saw that there were no articles analyzing the reality of some regions (e.g. Africa) and just a few articles about other regions (e.g. Latin America). So, it could mean that many of those authors are publishing articles (as authors or co-authors) without any particular region or country as focus, being a question that should be better perceived in further analysis. Open innovation literature is also extending its influence over other research areas. Although the main research areas citing open innovation are related to "Business, Management and Engineering", there are many others, including new areas (such as "Nursing") and emerging areas (such as "Phycology" or "Energy"). The question is whether this is a trend to continue in the coming years or if this is due to the novelty and curiosity about the field of open innovation. To be discussed, in the next section.

Conclusion and Discussion
Open innovation is a new research field within economics and management. Having started by observing cases of multinational companies located in the U.S. (work initially conducted by Chesbrough), mainly in high-tech sectors, open innovation research has received a growing attention over the years, extending its geographical and research scope of analysis. Being a new research area, the open innovation field is still under development and consolidation, requiring a broader and deeper analysis in order to gain influence in the academic field and a higher acceptance at a managerial and business level (Remneland-Wikhamn and Wikhamn, 2013). Through bibliometric analysis, this paper aims at contributing to that end.

Exploring other levels of analysis
The bibliometric analysis developed in this paper shows that the "firm" is still open innovation's main level of analysis, which is understandable, since it was the starting point of open innovation research (in 2003) and innovation (the new or improved product, service, organizational process or marketing) is mainly conducted by companies. Looking to the first 10 years of open innovation research (as seen in section 3), we can see that a large percentage of studies had their focus on firms (on large and/or multinational companies) and on technology-intensive areas and/or sectors. However, from the economics of innovation literature standpoint, we know that: innovation is a systemic and complex phenomenon; firms cannot innovate alone; innovation can occur outside the business sector and in all type of firms; innovation can occur in high-technology as well as in mature or low-technology areas (Caraça et al., 2009;Lundvall, 1988;Nelson and Rosenberg, 1993). So, it is expected that, for the consolidation and better understanding of the importance of open innovation as a research field, other levels of analysis can be studied with further intensity and depth. For example: • How can the clusters' approach be linked with open innovation activities? Breschi and Malerba (2005) have a detailed analysis on the impact of clusters in the innovation process, analyzing cooperation networks, geographical location, agglomeration economies and network externalities, regional innovation networks and social networks. Based on Simard and West (2006), the benefits of open innovation can be enhanced in clusters (regional clusters), since the effect of innovation networks can be expanded by geographical proximity, with a positive impact on economic growth. But further analysis is necessary in order to perceive all the potentialities of clusters in promoting open innovation activities, and vice-versa. Kong), some European countries (France, Poland, Hungary, Luxembourg, Lithuania, Portugal, Greece and Norway), Latin America (Peru and Brazil), Middle East (Iran) or Canada. However, there is a lack of open innovation analysis at a regional level within those countries, as well as a lack of inter-regional or inter-countries comparisons. One of the examples comes from Padilla-Meléndez, Del Aguila-Obra and Lockett (2012), through the analysis of the social capital role in knowledge transfer and exchange in Andalucía, Southern Spain. A more recent example is given by Savitskaya, Salmi and Torkkeli (2014), who analyze outbound open innovation in China and Russia, in an innovation system approach. But further data and information at a regional or inter-regional ( We have seen in section 3 that most part of the methodology used by open innovation studies is empirical in nature, using mainly case studies and questionnaires. Being open innovation an "open" reality in a global and turbulent world, involving several actors and different types of relations between them to explore new ways to create innovation, open innovation literature could be enriched through the use of additional research methods. For example: new ways to collect information, using focus groups, online platforms or the Delphi method; new ways of predict trends and construct strategies, using prospective or forecasting methods. Experimental economics can also provide some insights to open innovation research, mainly with regards to the analysis of openness at the individual level.

Limitations of this research
As a limitation of the research, it can be pointed out the non-inclusion of articles that had "open innovation" in keywords or in abstracts, having only been considered the articles with the term "open innovation" in the title. This choice avoided the dispersion of the analysis with articles that may not have open innovation as the main focus, although they might analyze related areas (as referred in section 2) and can be important in terms of marking some research trends or other emerging research areas.
A more detailed understanding of the trend and impact of open innovation research could be reached through the inclusion of these articles. In the text analysis, other kind of software can be employed, in order to highlight other possible details. It would also be interesting to compare the results with bibliometric papers that use cocitations as a methodological tool. Nevertheless, it is expected that this research can be considered as a step further into the knowledge on the open innovation field.