The emperical concepttualization of innovation : a systematic review of highly ranked journal publications 2005-2015

(1)

THE EMPIRICAL CONCEPTUALIZATION OF INNOVATION:

A Systematic Review of Highly Ranked Journal Publications 2005-2015

Name:

J.B. Tieks

Student Number:

10672532

Email:

jb. tieks@gmail.com

Thesis Supervisor:

W. van der Aa

Second Reader:

B. Kuijken

Faculty:

FEB

Date:

30-08-2015

Qualification:

MSc. Business Administration

(2)

2

Statement of Originality

This document is written by Joëlle Barbara Tieks who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(4)

4

Abstract

This meta-analysis reviews how studies on innovation have approached the concept of innovation in their empirical design. Existing theories about innovation research are mapped out in order to create a theoretical foundation for this review before a sample of 206 top-tier journal publications on innovation is analyzed based on 6 focal points. The results indicate that innovation researchers tend to approach their subject a one-dimensional way although it is clearly a multi-dimensional process. At the same time firms suffer from a lopsided focus as well – they are mainly preoccupied with innovation of their offerings. These results are linked with existing literature to establish theoretical implications. Current analysis implies that top-tier publications on innovation predominantly focus on offerings innovation, based on the use of limited varied research methods and with a singular level of analysis. Implications on a managerial level indicate that a multi-dimensional approach of innovation by firms is unrealistic as long as science employs knowledge that is based on routinized research. It is therefore suggested academia and business strengthen their ties to reduce knowledge gaps and further develop the knowledge base.

(5)

5

1 Introduction

Explaining why innovation is crucial to a firm’s (competitive) success seems rather unnecessary in this context. However, emphasizing the importance and advancement of innovation research seems more prudent. The field of innovation research has continued to expand, develop and advance itself over the last years (Shafique, 2013). Studies from different disciplines have made their own distinct contribution in terms of economic, social or technological perspectives. This continuous growth also calls for an examination of the status quo in innovation research and the intellectual structure of the field. Mapping out its characteristics, patterns, traditions and knowledge bases facilitates the construction of an overview of existing results and theories. Such an overview could help pave the way for further development of the field, as well as recognizing opportunities for potential future research. Up to now, studies concerning the content of innovation as a field of research are scarce. Therefore, this paper aims to make a contribution to the literature on this topic that does exist by focusing on the empirical foundation of innovation studies. The use of a systematic research approach resulted in a sample of 206 high-rated journal publications1_{. Each of}

these 206 articles was assessed on six points of focus in order to create an overview that provides sufficient insights into the characteristics of current innovation research. By linking these results with existing theory, it is possible to arrive at conclusions concerning the current body of research. This study aims to explicitly contribute to existing research by advancing the theories and observations made with regard to innovation research. The scope of is analysis is limited to the empirical conceptualization of high rated journal papers. Decomposing the empirical foundation of innovation studies provides detailed information on how innovation is represented and carried out in research. This conceptualization of innovation is crucial to revealing patterns across studies and potential gaps in the current knowledge base. In terms of practical contributions, this study ultimately details the implications of the state of the existing research, as well as suggestions and directions for future research and management processes.

(6)

6

2 Research Design

The upcoming chapter puts forth the design and objectives of this study. Additionally, the structure of the paper is mapped out.

2.1 Aims and Objectives

The main purpose of this paper is to provide a review of existing literature with regard to the conceptualization of innovation and its dimensions, and how these dimensions are translated into empirical studies. The literature review is conducted in the form of a meta-analysis, which analyzes and maps the existing body of knowledge on innovation. A meta-analysis allows for a structured approach to assessing existing knowledge and discovering potential relationships between certain phenomena and patterns within the literature. The meta-analysis is not limited to a summary of accumulated knowledge, but rather develops this body of knowledge further by specifying a research question . Thereupon, it forms a scientific contribution on its own.

The central research question of this meta-analysis is as follows:

2.2 Structure

To ensure the validity and credibility of the literature study a high quality review process is required. A high quality review process requires a transparent and reproducible procedure and structured methodology (Tranfield et. Al, 2003). This paper is structured along a number of key themes that facilitate mapping out the theoretical field and provide considerable insights and implications. The first section elaborates on the methodology used in this meta-analysis, describing details of the selection and how the final data set was established. The second section provides a theoretical framework in order to construct a foundation of previously conducted research and theories.

“In what way is the dimensional configuration of business innovation included in the empirical foundations of studies published in leading management journals?”

(7)

7

The third section contains an analysis of the sample publications and lays out their essence, categorization and empirical foundation. Additionally, section three elaborates on the results of this analysis with statistics as well as with representative examples from the selection of articles. Section four follows by presenting implications by linking existing literature to the findings of the current study. The fifth and last section concludes this paper with a discussion and conclusion part.

(8)

8

3 Methodology

This chapter explains the choice for the concerning research method and characteristics of the data collection process. Further, it illustrates some prominent components of the data sample.

3.1 Method

A meta-analysis embodies a powerful tool for reviewing an already well-investigated research field because it allows the use of a structured approach to analyze existing literature and is often able to detect relationships within and between these studies ( Cooper & Schindler, 2005). In order to satisfactorily answer the central research question of this thesis, there is a need to gain a deeper understanding of several specific aspects of the existing literature. Initiating a systematic review approach such as a meta-analysis limits the subjectivity of data collection by using predefined selection criteria.

3.2 Compiling the Dataset

The central research question of this paper already indicates the domain of the analysis, namely the “leading scientific journals” or A* Journals. Objectivity is a challenge here, so the selection of these journals is based on the SCImago Journal & Country Rank (SJR). The SJR ranks scientific journals according to their indicator, which in itself is a measure of a journal’s impact, influence and prestige. It expresses the average number of weighted citations received in the selected year by the documents published in the journal over the past three years.

Considering the perspective of this meta-analysis (i.e. Business Administration) the journal titles were selected from the subject category “Strategy and Management”. Given the limited size of this research, its scope consists of the ten highest ranked (Q1) journal titles.

(9)

9

Figure 1: list of source titles and their rankings2.

After establishing the titles used, a thorough search for publications related to the study of innovation was next. There are several approaches to establish a data set, and this paper opted to search by using keywords. Each online journal archive was accessed with a search query for articles whose publication titles contained either “innovation” or “innovate”. As a lot of research is dedicated to innovation (management) and this study is of limited size, a custom range regarding publication years was added, meaning that only articles published between 2005 and 2015 were taken into consideration. The online archives also allow for the specification of the type of document desired. Only document types “articles” and “reviews” were selected. The final sample based on SJR ranking resulted in the use of 9 source titles, as Organizational Research Methods did not produce a single title based on the above mentioned search procedure.

2_{Source: (www.scimagojr.com): based on the 2013 ranking}

RANK SJR SOURCE TITEL

1 10,001 Academy of Management Review

2 9,661 Journal of Financial Economics

3 7,886 Academy of Management Journal

4 6,914 Strategic Management Journal

5 6,783 Journal of Management

6 6,398 Organization Science

7 5,216 Journal of Operations Management

8 4,549 Long Range Planning

9 4,500 Organizational Research Methods

(10)

10

Figure 2: Breakdown of article selection by source title

As laid out in Figure 2, there is a somewhat unequal division between the contributions of each source title. The Strategic Management Journal accounts for 25 per cent of the publications, whereas four other titles (Journal of International Business Studies, Journal of Management, Academy of Management Review and Journal of Operations Management) account for 6 per cent each. Nonetheless, in light of ensuring a systematic and reproducible procedure, there is no room to balance these outcomes and adjust the final sample of articles.

3.3 Categorization of Publications

The outlined approach yielded a final sample of 206 articles. During this process, however, a number of articles were excluded from the selection due to several reasons. The articles were either editorial notes, inaccessible or removed from the internet. See Appendix 1 for a list of excluded titles and the reasons for their exclusion.

The aim of this research is to construct a comprehensive analysis of the existing literature and the attributes of its empirical infrastructure. After the assembly of the final sample, each article was studied according the following components of its research design:

6% 10% 9% 6% 6% 6% 13% 19% 25% Academy of Management Review Academy of Management Journal Journal of Financial Economics Journal of International Business Studies Journal of Management Journal of Operations Management Long Range Planning Organization Science Strategic Management Journal

Organizational Research Methods

(11)

11

Figure 3: Focal Points

The focal points are based both on the Innovation Radar as described by Sawhney, Wolcott and Arroniz (2006) as well as the multi-dimensional framework put forth by Crossan & Apaydin (2009). Assessing each publication article based on these criteria produces an overview that displays how research on innovation in top leading journals has been approached and positioned so far. Moreover, it should provide insights on and potential implications for innovation research and its empirical body of knowledge.

Innovation dimensions refer to the dimensions as described in the Innovation Radar of Sawhney et al. and represent the aspects of a business model in which a particular type of innovation can occur. It concerns the following main dimensions: offerings, customers, processes and presence. The level of analysis describes the level on which the research took place, e.g. individual, organizational or geographical level. The referent of the study is solely applicable when the article is based on an empirical study which adopted a specific sample. This concerns, for example, a number of North-American Consultancy firms, or Chinese Semi-Conductor production companies. The industry a paper focuses on is only specified if the article elaborates on the research setting and describes the industry in which the study took place. The domain of the study categorizes the independent variable or phenomenon of the empirical method that was incorporated in the research. This could be related to the organizational theoretical nature, the innovation management practices or a more unrelated topics such as financially structured products or the role of the government. The method of study simply categorizes how the studies (qualitative as well as quantitative) were carried out, be it through surveys, interviews, panel data or literature review.

(12)

12

3.4 Reproducible Research Approach

The previous section provides a detailed outline of the methodology used in this study. A systematic approach minimizes bias and error and helps collecting “high quality” evidence. According to Tranfield et al. (2003), management researchers are biased and they tend to select studies that appear relevant or interesting. Precise inclusion/exclusion criteria of these studies are often not formally agreed to, applied, recorded or monitored. In order to limit the risk of such biased decisions regarding data collection, this study has applied predefined selection criteria to its search. As the studies were selected and compiled manually (each journal title needed an individual computer-based search), a reproduction of the approach to secure its objectivity, transparency and validity was required. After initially organizing the studies and compiling the final dataset, the process was repeated yet again. Reproducing the search and selection procedure reduces the risk that some studies are mistakenly excluded from the sample. This simultaneously reduces the risk of including titles that do not meet the selection criteria. Performing the search using the Wiley Online Library did not allow for a specified search within the publications of the Strategic Management Journal, but displayed all of the journal’s publications on innovation. Manually selecting the relevant studies from this query is rather risky in terms of default and bias, - another reason why reproducing the search procedure was deemed appropriate.

.

(13)

13

4 Theoretical Framework

The upcoming section describes and explains several interrelated concepts related to innovation research. This framework serves as a guidance and foundation for the current paper.

For many organizations, innovation is an important component of their competitive strategy as it enables them to develop (sustainable) advantages. In order to make innovation management a priority, firms must be able to advance technology, exploit these capabilities and gain the market’s acceptance of new ideas, concepts and production requirements (Lengnick-Hall, 1992). Sawhney et al. confirm this organizational need for innovation while noting that many companies tend to have a restricted view on business innovation. A restricted view could lead to a one-dimensional focus on, for example, product or process innovation. By focusing on a single element of their business system, companies create a blind spot for opportunities and other potential dimensions they could include in their innovation management practices (and benefit from). This limits a company’s ability to establish a competitive advantage and leaves them more vulnerable to competitors who have adopted a broader perspective. Innovation management adds significant and durable value to competitive advantages if the possibility of replication by rivals is limited (Lengnick-Hall, 1992)

From a configurational perspective (Miller & Mintzberg, 1984), another risk of one-dimensional innovation is the lack of a configurational fit with other components of a company’s (innovation) strategy. If innovation management does not suit other elements within the configuration, it could result in serious imbalance and disturbed synergies. Innovation should be undertaken in such a way that it complements and opportunistically exploits distinct competencies that arise from the firm’s configuration and subsequently lead to a competitive advantage (Lengnick-Hall, 1992). Sawhney et al. (2006) affirm that successful business innovation requires a consideration of all dimensions of the business system. They have come up with a tool that defines these dimensions of business innovation and how they relate to one another. Their so-called ‘innovation radar’ presents the dimensions firms may use to look for innovation opportunities.

(14)

14

Scientific literature has noted the importance of innovation management and its prominent role within business strategies. Consequently, there exists a wide array of empirical research, carried out in various ways and with differing purposes.

However, Damanpour (2014) noted that research has primarily examined technology-based product and process innovation. As a result, the majority of studies consider manufacturing industries as they are the primary sources of technological innovation. Limiting research to this technological domain results in the fact that most theories are shaped by that particular setting (i.e. manufacturing sectors) (Damanpour & Aravind, 2012) and are not able to include the real-world heterogeneity of organizations’ innovative behaviors. Their observation is shared by Armbruster, Bikfalvib, Kinkela and Laya (2008), who argue that in the current scientific debate, innovation is predominantly linked to Research & Development activities and expenditures, which should “naturally” lead to innovative products in turn that enable firms to gain competitive advantages. One of the most prominent reasons for this over-emphasis on technological based innovation is its high observability; product innovations refer to embodied technology and are an outcome an sich (Damanpour, 2014).

As a counterpart of popular measurable product innovation, the research paper of Armbruster et al. focuses on organizational innovation. Their results propose different approaches to measuring and monitoring organizational innovations. Damanpour centralizes management innovation (also referred to as organizational innovation) in several of his studies in order to “complement existing theories of innovation derived from the studies of technological innovations and thus help extend and advance theory and research on innovation processes and outcomes in organizations.” (Damanpour & Aravind, 2012). Management Innovation is defined as the introduction of new processes to organize firm activities and coordinate human resources that essentially have no technical components (Damanpour, 2014). The Academy of Management Review published an article titled “Management Innovation” by Birkinshaw, Hamel and Mol (2008) on management innovation in order to contribute to this under-researched form of innovation. They provide a theoretical framework as a foundation for the concept of organizational innovation and its processes.

It is but one example that attempts to increase awareness of innovation types other than product and process innovation. In his “Footnotes to Research on Management Innovation” (2014), Damanpour argues that new developments in innovation literature (should) highlight the

(15)

15

combinative effects of multiple types of innovation and their interdependence. Most studies of innovation presume that organizational behavior and actions are determined by technological development, and prioritize technological innovations in order to enable growth. As a consequence, non-technological innovation is of lesser concern and merely introduced by firms to facilitate the enactment of its technological innovations (Damanpour, 2014). Organizations that have developed certain capabilities and knowledge are likely to channel this body of knowledge into specific types of innovation, as they have prior experience with it (Bierly et al. 2009). This logic causes an imbalance between the (research) attention given to technological innovations and that to non-technological innovation.

Although the above mentioned publications and the work of Sawhney, Wolcott and Arroniz are formulated and operationalized in different ways, both emphasize the dimensional configuration of innovation. Organizations run the risk of assessing innovation as a single domain practice, whereas these authors actually argue that innovation is a multi-dimensional phenomenon. So far, however, innovation studies tend to strongly focus on one dimension as well; being technological innovation. Crossan and Apaydan (2010) have conducted a meta-analysis on innovation research and reported that 38 per cent of the papers in their sample concerned (technological) product and service innovation. The authors concluded that a large amount of innovation research is too narrowly focused, which impedes theoretical strengthening of the field. Their literature review found that research on innovation is rather inconsistent as multiple levels of analysis and dimensions are used, which in turn leads to mixed empirical results. Therefore, they aimed to gain a more detailed understanding of innovation management by putting forth an overarching framework based on the highest cited literature. Through synthesis of the collected data, Crossan and Apaydan constructed a multi-dimensional framework that enables a larger picture of innovation.

A comparable study by Anderson, De Dreu and Nijstad (2004) has analyzed a selection of studies in order to review the state-of-the-science of innovation research. This review states that innovation research has expanded considerably and the focus has become increasingly multi-level. Innovation researchers therefore seem to be creative in their attempts to undertake studies of societal, organizational and psychological value. However, partly as a result of their content analysis, Anderson et al. argue that, ironically, innovation research has become less radically innovative. The methods used are routinized and cross-sectional, and this routine approach is less

(16)

16

likely to uncover new characteristics and consequences of innovation. Most of the cases within their sample are replication-extension studies, which suggests that research is preoccupied with searching for regularities. This is unfavorable for the number of studies that focus on irregular events and exceptions to the rule. Researchers are required to re-conceptualize innovation as a phenomenon that could cause multiple outcomes and different levels of analysis. As a concluding section to their findings, Anderson et al. propose several pathways for future innovation research. Shafique (2008) has also recognized the routine nature of current innovation research but defined this phenomenon as so-called “research traditions”. His primary purpose is to establish a global view of the intellectual structure (i.e. constituent research traditions, disciplinary composition, the topics they address, and the pattern of their inter-relationships within a specific scientific domain) of innovation research. Bibliometrics were used to map out the patterns and customs that have dominated innovation research between 1988 and 2008. Results of this study produce distinct research traditions during each period, and relevant for the current meta-analysis is the timeframe of 2002-2008 (in absence of a category that concerns the years after 2008). Shafique reports that research on innovation where management is the main contributor continues to grow. There seems to be a shift of scope from the external environment to internal elements such as capability development, knowledge creation and learning. Another prominent tradition related to organization theory is the heavy focus on product innovation management and the development of a social structural perspective on innovation and competition. The research conducted in these years is also characterized by a persistent fragmentation between the two major disciplines of management and economics, where both are rather inwardly focused and seldom apply knowledge of outside its disciplinary boundaries.

This theoretical framework provides a solid foundation for the upcoming sections of this meta-analysis. Even though reviews and analyses on existing innovation research are scarce, there is a need to pool and assess observations and theories that might be of value to this paper. With the studies mentioned above as its starting point, the current study aims to elaborate on the existing literature concerning innovation management research. The actual focus lies on reviewing how exactly business innovation has been conceptualized and how that is translated into empirical studies. More specifically, what are the dimensions of innovation that were included and analyzed in those studies? It will examine how the so called “narrow scope” discussed in previously mentioned articles, is reflected in empirical research.

(17)

17

5 Results

This section elaborates on the analysis and results deduced from the collected data. In order to provide the reader with a coherent framework for these results, the descriptive analysis part discusses every focal point (as discussed in the methodology section) separately. Subsequently,

more detailed results are used to construct an answer to the central research question of this meta-analysis.

5.1 Descriptive Analysis

5.1.1 Level of Analysis

The aim of this literature review is to establish a complete overview of the selected research conducted with regard to innovation. As stated in the methodology section, the assessment of the selected articles was based on different criteria. One criterion is the level at which the analysis of the empirical foundation took place. In the sample, which consisted of 206 publications on innovation, the analysis levels occur in the following frequencies:

Figure 4: Breakdown of article selection by level of analysis

46% 3% 3% 5% 5% 2% 7% 4% 2% 3% 12% 8% Firm level Geographic level Industry level Individual level Team level Department level Product level Network level Article level Project level Multi-Level Undefined

(18)

18

A substantial amount of empirical research on innovation focuses on the firm level (46 per cent). This result is in line with the findings of Anderson et al. (2004) who explain that the larger part of innovation research aims at both the individual as well as the organizational level. Within this sample, individual level analysis and product level analysis represent another 5 per cent and 7 per cent, respectively.

Team level analysis and network level analysis account for 5 per cent and 4 percent, respectively. Frequently, articles based on literature review and theory building do not specify the level of analysis. Additionally, there is a minority of empirical testing models that do not explicitly state their level of analysis. Due to (particularly) this share of literature studies the graph contains an 8 per cent share designated “Undefined”. The number of studies operating on the geographic level, project level, department level or article level is limited.

As argued by Gupta et al. (2007), innovation is a multi-level phenomenon and should be treated as such. The modest 12 per cent of multi-level studies in this sample, however, indicate that empirical researchers of innovation have not (yet) adopted this viewpoint. The lack of true multi-level research may be due to the absence of well-developed theoretical frameworks with regard to the innovation subject. Anderson et al. also emphasize the importance of multi-level research by stating that different variables affect innovative behavior on different levels. In order to gain more understanding about innovative behavior, it is crucial to move beyond a single level of analysis so it becomes possible to recognize multi-level linkages and create a more complete perspective on innovation (management). A publication by the Journal of International Business Studies titled “An international multilevel analysis of product innovation” (D. Lederman, 2010) incorporated such a multi-level approach. The research was conducted within 68 countries, covering 25,000 firms across 8 different manufacturing industries. The author aims to explore the (international) multi-level determinants of product innovation by implementing a model that considers three multi-levels of analysis: firm characteristics, industry characteristics, and the national context. Data was collected from longitudinal surveys, but as a secondary source since these surveys were conducted by the World Bank. Findings of the empirical testing pointed out that firms that are involved with global business activities, be it through adopting foreign technologies, foreign investments or import of capital goods, are more likely to introduce new products than other firms. Industry level trade policy misuse or foreign ownership did not turn out to be significantly related to product innovation. Firm size does correlate with product innovation rates, but evidence suggest that this

(19)

19

effect is rather limited and declines with firm size. Lederman’s article is one example of a small pool of studies that has a multi-level character.

5.1.2 Domain

Statistics have shown that management and economics are two major contributors in the field of innovation, where the share of innovation research published by the management discipline is steadily increasing (Shafique, 2013). As the journal titles in this meta-analysis were selected from the Strategy and Management discipline, it was to be expected that most of these themes are related to organization theory or management topics. Nearly all the articles in the sample revolve around a central theme related to innovation or innovation management practices. These central themes (variables) are assigned to overarching domains that represent different subject within organization theory and management studies. The categorization of the domains is based on themes that continuously re-occur throughout the articles. These themes include: leadership style, organizational culture, team composition, knowledge networks, learning mechanisms, open innovation, strategic out-source decision making, strategic alliances, governmental influence, corporate governance structures, and innovation search.

The domain labeled “other” contains themes that did not re-occur in other publications (e.g.: expected stock returns and local banking development).

Figure 5: Breakdown of article selection by domain

19% 17% 9% 8% 12% 13% 19% 3% Organizational Structure Inter-firm Collaboration Knowledge Structures and Mechanisms Inter-firm Network Innovation Process Strategic Management Other Institutional Forces

(20)

20

Figure 5 displays how leading articles have included different domains in their empirical research. Organizational structure and Innovation Process are both accountable for 19 per cent of presence. Strategic Management and Inter-Firm collaboration represent 13 per cent and 17 per cent, respectively. Institutional forces describes governmental action, corporate governance and regulation. Its under-representation might be related to the fact that most publications involve research conducted in the private sector and concerns different (commercial) business activities and management-structures.

5.1.3 Industry Type

Another element of this literature review is the documentation of the industries in which the research was performed in or applies to. A majority of the articles that involve empirical testing describe their research setting and the type of firms used in the sample. Analysis of the nature of the industry identifies certain contextual conditions and characteristics that are likely to affect the choice of innovation dimensions. Contextual conditions refers to the competitive intensity, enforcement of intellectual property laws, the speed and uncertainty of technological change and the fragmentation of suppliers, buyers and consumers (S.D. James et al. 2013).

Figure 6: Breakdown of article selection by industry type 31% 3% 4% 4% 14% 6% 4% 11% 4% 7% 12% Across Industries Financial Services and Institutions

Transportation Industry Pharmaceutical Industry Technology Intensive Industry

Services (including health centers) Design/Fashion Industry Manufacturing Industries (general) Semi-Conductor Industry Other Unspecified

(21)

21

It was rather complicated to categorize the research characteristic into specific industries. The circle graph demonstrates that 31 per cent of the selected journal publications do not refer to a particular industry or sector, because a lot of empirical models include firms from varying industries. Their sample contains data extracted from databases such as National Bureau of Economic Research (NBER) Patent Citation database, Institutional Brokers' Estimate System (I/B/E/S) database, CompuStat, Dow Jones Venture Source and USPTO Database. This data is often selected based on criteria other than industry type, such as firm size, patent applications or sales (revenues), and therefore explain why these samples move across industries. D.J Miller et al. (2007), have written an article published by the Academy of Management Journal titled “The Use of Knowledge for Technical Innovation within Diversified Firms”. The research studies the effects of interdivisional knowledge on the impact of an invention. Firms in their sample move across industries because they believe that the theory of distant search applies in a variety of settings. Most of the data was obtained from the National Bureau of Economic Research Patent Citations Data File and processed in the negative binomial model for testing. The results indicate that using interdivisional knowledge has a significant, positive effect on the impact of an invention. A remarkable aspect of this study is that it does not focus on knowledge sharing outside of an organization’s boundaries, as the authors consider this to be unfeasible and subjected to limitations. Instead, they focus on the alternative: knowledge sourcing within the same firm, but across divisions. This article by D.J. Miller et al. represents a common way of sampling in which there is no specific industry addressed, but rather a multitude of sectors.

A number of articles were categorized as “unspecified” as these articles did not outline an industry of focus. That is justified by the fact that 17 per cent of the publications are based on empirical theory building or theoretical modelling. Such theoretical models are generally applicable and are not necessarily associated with specific industries.

Technology Intensive Industries account for 14 per cent of the publications. Technology Intensive Industries include Biotechnology firms, High-Tech firms, IT-firms and Nano-technology. Manufacturing industries represent 11 per cent of the papers on innovation research and involve a combined variety of industries such as chemicals, consumer goods, electronics and industrial machinery. Some overlap of domain may exist between Manufacturing Industries as a whole and categories such as Pharmaceutical Industry, Semi-conductor Industry and Transportation Industry, as one could consider these categories also to be part on the manufacturing sector. However, the

(22)

22

publications have only been assigned to these categories if the research setting was focused on this specific industry and that was also explicitly named in the article.

The research setting of an empirical analysis also includes the geographical scope applied. Statistics indicate that 31 per cent of the articles contains research that took place in the US, 18 per cent was performed in European countries, 16 per cent of the research is globally dispersed and 6 per cent was carried out in Asian countries. As some of the articles do not apply an empirical testing model, 17 per cent accounts for theoretical modelling.

5.1.4 Research Methods

Figure 7 illustrates the number of studies that applied a certain research method. The sample of selected articles contains papers based on empirical theory testing, empirical theory building and literature studies. A significant number of the papers are field studies which are performed outside of a laboratory environment. Data collection in these field studies was based on conducting interviews (15 per cent), administering surveys (20 per cent), observation (3 per cent) or a combination of one of these methods complemented by panel data (i.e. longitudinal as well as archival data). A mixed method research design which combines primary data (e.g. interviews, surveys or observation) and secondary sources such a panel data are widely used across the publications (24 per cent). An example representative of such a multitude research method study is “Top Management Attention to Innovation: The Role of Search Selection and Intensity in New Product Introductions”, an article by Q. Li et al. (2013) published in the Academy of Management Journal. This study’s central issue concerns the extent to which Top Management Teams affect a firm’s innovation level, more specifically; TMT search process.3_{Their clearly stated hypotheses were}

tested by interviewing 92 CEO’s, conducting surveys among top management employees and collecting archival data on the companies in question from COMPUSTAT. The hypotheses were tested through the use of negative binomial regression analyses. Results from these analyses indicate that both selection and intensity of search significantly affect the introduction of new products. Additionally, their findings state that achieving high levels of production requires TMT to find a fit between search selection and search intensity.

3_{In this article, search is defined as “… the controlled and proactive process of attending to, examining, and evaluating}

(23)

23

The authors motivate the use of the three methods by arguing that such a combination contributes to the strength of the study, as it overcomes common methods biases and validates the data.

That argument is supported by Anderson et al. (2004), who plead for a triangulation of research methods in order to construct a more complete picture of the innovation process. Further, combining different methods enables higher levels of internal as well as external validity of the research.

Another advantage of combining field studies with experimentation is that experiments are an efficient and effective substitute for labor-intensive longitudinal research. Experiments also enable the researcher to use certain levels of variables that are sometimes impossible to examine in a field setting (Anderson et al., 2004).

Even though the empirical basis of a significant number of publications is formed by primary data, the majority of studies extracted their data from secondary sources. Over 17 per cent of the studies in this meta-analysis’ sample consists of studies which are based on the review of existing literature. These studies contribute to research by establishing empirical theory, constructing models and/or developing a theoretical framework. A publication found in Long Range Planning titled “Business Models, Business Strategy and Innovation” by David J. Teece (2010) aims to link the design of business models to innovation strategy. The author attempts to resolve his central research questions through the creation of a theoretical foundation of relevant literature and by referring to real-life examples. Not only business model’s relation to innovation is explained, also business model innovation itself receives attention as the author emphasizes the importance of new organizational forms and new organizational methods to society. Due to the lack of research on this topic, the main goal of this article is to advance the understanding of the significance of business models by providing building on existing literature.

Empirical theory testing based on longitudinal and/or archival data is the most commonly used method among the studies. Such a (secondary) data collection approach accounts for 38 per cent of the publications, where analysis of longitudinal data occurs more frequently than its archival counterpart. This striking number is mainly the result of a pattern that exists within the selection of articles, as assessing and measuring innovation output by analyzing patent application and citation data is a prevalent research method. Nearly 25 per cent of the studies employ patent data as a

(24)

24

measure of (product) innovation output, of which 60 per cent is U.S. oriented. Patents are good indicators of new technology creation, but do house certain flaws in their reliability when it comes to measuring innovation activities. These flaws include, among others, the fact that not all new inventions are patented, and that patents differ greatly in their economic impact (Z.J. Acs et al., 2002). The next section of this meta-analysis discusses the dominating pattern of product innovation and patent data in more detail.

Figure 7: Breakdown of article selection by Research Methods

5.1.4 Dimensions of Innovation

In order to proceed with the essential part of the analysis it might be useful to recap the purpose of this research paper. A dataset of high rated journal articles was established with the goal of discovering in what way dimensions of innovation were part of their empirical foundation. Sawhney et al. argue that there exists a one-dimensional view on innovation among organizations that causes a tendency to innovate along the same dimensions as other competitors in the industry. This literature review aims to explore whether the above described one-dimensional view is also reflected in the selection of innovation publications.

All articles within the sample have been studied on the innovation dimensions (i.e. form of the innovation) they have included in its empirical foundation. This concerns the innovation outcome,

0 10 20 30 40 50 60 70

Archival Data Archival and Longitudinal Data Archival Data and Interviews Archival Data and Literature Review Archival Data and Surveys Surveys Surveys and Longitudinal Data Surveys and Interviews Interviews and Observations Interviews Interviews and Longitudinal Data Longitudinal Data Longitudinal Data and Literature Review Literature Review Simulation Model Combination ≥ 3 methods

(25)

25

rather than the innovation process or method. That is, the processes and methods are often applied as independent variables. The independent variables or methods are handled as the “Domain” focus point in this study. As a counterpart, the dependent variable is the innovation outcome, and their innovation form (or dimension) relates to this focus point.

As stated in the methodology section, these dimensions are based on both the Innovation Radar of Sawhney et al., as well as the literature review of Crossan & Apaydin. Categorization is mostly based on the Innovation Radar, meaning that the articles are classified along the twelve dimensions. The category of “Business Model Innovation” is the only exception, as it is not one of the dimensions of Sawhney’s framework. However, Crossan & Apaydin mention Business Model innovation within the same context as product and service innovation. A significant amount of articles explicitly refer to Business Model innovation as the focus of their research. Therefore, it seems justified to include this as a separate dimension instead of assigning those articles to one of the twelve dimensions that do not entirely capture the essence of its nature.

Within the sample, the Strategic Management Journal has, relatively speaking, the highest number of publications on product and service innovation, as over 84 per cent of the selected articles concern the Offerings dimension.4_{Sample wide, a remarkable 71 per cent puts its focus on the}

innovation of products and services. The Offerings dimension comprises a firm’s product and services that deliver new value to customers.

4_{As the composition of articles is unequally divided, a calculation was used in order to produce the relative proportion of}

(26)

26

Figure 8: Breakdown of article selection by Innovation Dimension

From the 71 per cent of studies classified as having focused on Offerings innovations, more than 60 per cent clearly specify that their scope is solely focused on product innovation. Such an example is the Strategic Management Journal’s publication “The Effects of Geographic and Network Ties on Exploitative and Exploratory Product Innovation”. 5_{This article by Ozer & Zhang (2015) deals with}

industrial clusters and its effect on exploitative and exploratory product innovation. They collected the required data by administering surveys among 153 Chinese firms. However, as already discussed in the previous section, product innovation is also frequently measured by patent application and citation (nearly 25 per cent of the complete sample). The specific purpose of Ozer & Zhang’s study is to investigate the moderating role of network ties, and therefore the hypotheses were tested through moderating regression models. The results from these models indicate that cluster membership positively relates to exploitative innovation, but negatively to exploratory innovation. Network ties with suppliers and buyers also strengthen the effects that industrial clusters have on exploitative innovation. As a crucial contribution to future research, this study revealed that the negative effects from industrial clusters on exploratory innovation can be diminished by network ties with buyers.

5_{Exploitative innovation builds on the firm’s existing knowledge base to improve its existing processes and products}

whereas exploratory innovation involves a shift to a different knowledge domain with the aim to adopt or create new processes and products (Ozer & Zhang, 2015)

Business Model 5% Customer Experience 1% Networking 1% Organization 10% Platform 0% Process 10% Supply Chain 1% Offerings 71% Value Stream 1% Business Model Customer Experience Networking Organization Platform Process Supply Chain Offerings Value Stream

(27)

27

Even though a large number of publications concentrate their research on novel products, only 3 out of the 147 articles are exclusively focused on service innovation. In addition to product innovation, almost 40 per cent of studies that belong to the Offerings category refer in their empirical foundation to product as well as service innovation. In many cases, the author(s) narrowed this scope to technological product and service innovation.

Both the dimensions of Processes and Organization represent 21 per cent of the empirical research. Process as an innovation outcome refers to the introduction of new production methods, new management approaches, and new technology that can be used to improve production and management processes (Crossan & Apaydin, 2010). Process innovation is an internal phenomenon and directly affects the organization itself. The Journal of Operations Management published an article titled “Innovation-supportive culture: The impact of organizational values on process innovation” by S. Khazanchi et al. (2007). As the main purpose of this paper is to study the relationship between organizational culture and the adaptation and implementation of certain manufacturing technology, it is a suitable example of a study that concentrates on process innovation. The study focuses on a specific Advanced Manufacturing Technology (AMT), namely die/mold machinery. The researchers administrated a survey among manufacturing plants that adopted this AMT. A regression analysis pointed out that the role of control values is quite complex. Innovation-supportive culture appears paradoxical because of the mechanism of flexibility versus control values.

Organizational innovation manifests itself when an organization restructures the scope of its activities or redefines the roles, responsibilities and incentives of different departments and individuals (Sawhney et al., 2006). Articles from the sample demonstrate that organizational innovation, among others, involve topics such as outsourcing activities, incentive structures and innovation culture.

Business Model innovation is merely about how to capture value from new products and services. It embodies the organizational and financial architecture of a business, and essentially refers to the conceptual model of that business (D.J. Teece, 2010). The Business Model dimension might overlap somewhat with the Organization dimension, as both target organizational structure. However, a number of articles specifically refer to the innovation of business models, such as “Business Model Innovation through Trial-and-Error Learning: The Naturhouse Case” published in Long Range Planning and written by M. Sosna et al. (2010). This single case study investigates how, through the

(28)

28

implementation of business model innovation, a company can anticipate economic recessions and increased competition. The case revolves around Naturhouse, a dietary supplement shop. The choice for this company is based on the fact that the its business model innovation allowed for a 40 per cent growth per year, for more than 10 years. In contrast to many other firms, this innovation is not related to information technology, but involves an undifferentiated product in a mature market. Through the use of interviews and archival data it is revealed how Naturhouse dealt with trial and error in their innovation strategy. Results point out that in order to overcome the barriers related to business model innovation a severe “crisis” might be necessary to awake deep enough reflection on the status quo of the business model design. 6_{Reorientation, which}

often takes place after a business catastrophe, can lead to a better search for new solutions and survival strategy. Supported by the case study, the authors describe these learning mechanisms and processes into detail.

As the Offerings innovation is part of a consistent pattern among the articles, dimensions such as Supply Chain, Networking, Customer Experience and Platform enjoy little attention within empirical research. Not a single title within the sample included innovation (outcome) in terms of Brand, Solutions and Customers. The scarce amount of articles on Networking and Supply Chain innovation seems rather counterintuitive, given the fact that inter-firm networks and inter-firm collaborations seem to be a prevailing topic within innovation research (see figure 3). However, even though these articles include alliances and collaboration activities in their research as a method, the main emphasis lies with innovation outcomes such as Offerings. A publication by Organization Science titled “Technology Shocks, Technological Collaboration, and Innovation Outcomes” by M.A. Schilling (2015), focuses on alliance activity and collaborative behavior of firms. The study empirical tests how alliances and network-structures affect innovation outcomes, which is measured by patent data of U.S. firms retrieved from the Delphion Database. Essentially, the outcome does not concern innovation along the dimensions of networking and collaborations but rather the innovation of products/services as a result of collaborative behavior. The panel data is analyzed with a random effects regression. Significant findings of the analyses reveal that the prominent role of the internet is a big motivator of alliances. These alliances, combined with the

6_{Barriers to business model experimentation: e.g. traditional configuration of assets or employee resistance to change}

(29)

29

technology shock and a firm network’s reach7_{all have a positive relationship with patenting}

output.

7_{A firms network reach is defined as “an outcome of the size and density of the connected component within which a}

(30)

30

6 Implications and Discussion

This section elaborates on the results that were produced by the analysis. These results are linked to existing theory in order to establish certain implications of the study.

6.1 Theoretical Implications

The aim of this meta-analysis is to establish how innovation is conceptualized given its clearly multidimensional nature (Sawhney et al., 2006). More specifically: it looks at how empirical studies of innovation published in high rated journal publications reflect this multidimensional nature. The accumulated knowledge on innovation is rather extensive, and a few authors have made attempts assessing this current body of knowledge and its foundation. This study aspires to further develop these insights by exploring the field of innovation research. A sample of 206 publications by the top 10 Strategy & Management scientific journals was categorized and analyzed. This analysis contained six focal points with regard to the empirical foundation of the studies: level of analysis, domain, industry, research method, referent of the study and innovation dimension.

Results show that a significant majority of the studies conducted involve a single level of analysis. Whereas only 12 per cent of the studies involved a multilevel research. This indicates there is a significant gap between the realities of innovation processes and the academic assessment thereof. Innovation is a multi-level occurrence and its different variables will affect innovative behavior at different levels and needs to be approached with that in mind. However, the outcome of 12 per cent does confirm the studies from Gupta et al. (2007) and Anderson et al. (2004), which both elaborate on the scarcity of multi-level analysis. However, all is not lost: Anderson et al. do foresee a positive development in the number of multi-level studies. The current analysis suggests that this may be a valid prognosis, as 15 out of 25 multilevel studies were conducted in the last five years. Over 30 per cent of the studies involved in this analysis do not focus on a single industry and are therefore ‘unspecified’. A reason for this high number of unspecified studies may be due to the fact that almost one-third of these “Across Industries” publications extracted their data from patent databases, which is an all too common measure of innovation outcome.

With regard to specified studies; most studies were performed in the technology intense industries (14 per cent) and manufacturing industries (11 per cent). As Damanpour & Aravind have argued

(31)

31

before, it does seem to be the case that most innovation theory relates to empirical studies that are rooted in the technological or manufacturing industries.

All in all, the research methods used in the innovation studies can hardly be considered varied. Over 38 per cent of the studies are solely based on longitudinal data and archival datasets as a means to test their respective hypotheses. A common component of these studies is the use of patent-databases and COMPUSTAT as a source of secondary data. Conducting interviews (15 per cent) and administering surveys (20 per cent) as a single method are also a common ways to measure innovation activities. This evidence supports Anderson’s theory that innovation research is routinized. A lack of triangulation of methods and continuous repetition dominates the field and articles in this study are no exception to the rule.

Among the studies that applied empirical testing, nearly 37 per cent originates in the U.S.A. In absolute terms, this equals 63 studies, whereas only 38 studies originate in European countries and 12 studies are Asia oriented. This cannot be considered surprising, as 5 of the 9 source titles are American publications. Whether the findings in these studies can be generalized across countries, continents and cultures is another interesting avenue of research but beyond the scope of this analysis.

The most valuable focal point of this meta-analysis is how studies have conceptualized innovation dimensions in their empirical models. Based on 206 articles published in 9 different source titles it becomes clear that a striking 71 per cent of the studies narrowed its empirical scope to product and service innovation. As Sawhney et al. have discussed, firms tend to approach innovation from a one-dimensional perspective. A one-dimensional view seems to be reflected in innovation research as well, as this meta-analysis reports that the focus lies most heavily on (technological) offering innovation. It cannot be considered a breakthrough result, since several other reviews have arrived at the same outcome (i.e., where innovation research is primarily about technological product and process innovation, see (Crossan & Apaydin, 2009; Damanpour & Aravind, 2011; Armbruster, Bikfalvi, Kinkel & Lay, 2008). However, it is remarkable that within the timeframe of the last 10 years this observation is still accurate and applicable.

Little attention is paid to other dimensions aside from offerings (e.g. brand innovation, networking innovation, business model innovation, organizational innovation), despite the fact that several studies emphasize their value (Teece, Damanpour, Birkinshaw).

(32)

32

Even though empirical testing models enjoy significant attention in this meta-analysis, theoretical papers have been included and assessed on the same focal points. Of the 33 theoretical papers, over half also focuses on Offerings (i.e. products and services) innovation. Despite the fact that literature based review and theory building are not bound to empirical testing, it seems that the majority of these studies still approach innovation on one dimension (i.e. product innovation). One exceptional observation among these articles is the few studies on Business Model Innovation, of which nearly every title is published by Long Range Planning journal.

Naturally, the characteristics and outcomes of this study are interlinked. The dimension of the innovation outcome determines its most convenient and reliable measurement, in an industry which provides the best setting. However, as Damanour has already pointed out in his publications, the analysis reports a consistent pattern across the studies, with little deviating characteristics. The innovation research field is dominated by studies that conceptualize innovation as product, service or process outcomes, often measured by patent data from high-technology and medium-high technology industries. Literature continues to emphasize that innovation is a multi-level and multi-dimensional phenomenon with a multitude of characteristics and effects. It is therefore astonishing to conclude that that observation is not reflected within the empirical context

(33)

33

6.2 Managerial Implications

In addition to adding to the academic field, this meta-analysis aims to contribute to the management discipline as well. As noted in the previous section, business and research can benefit tremendously from one another, which is why it is useful to point out the managerial value of the findings presented here.

As shown by the results of this analysis, the current state of innovation research is biased in the sense that it has the particular tendency to narrow its scope to (technological) product and service innovation. Sawhney et al. observed the same tendency among companies and their innovation management practices. A variety in dimensions researched offers the possibility to innovate in different ways, a way that fits all components of a company’s configuration. As long as research presents innovation as a one-dimensional practice with an emphasis on innovation outcomes based on product invention, it is unlikely companies would adopt deviating innovation strategies which entail innovation along “other” dimensions. Therefore, this meta-analysis pleads for a breakthrough in innovation research where innovation outcomes are measured through other innovation approaches as well (e.g. customer innovation, brand innovation or a combination of several dimensions) in order to create a foundation for changing companies’ innovation strategies. The assumption here is, of course, that academia’s accumulated knowledge and outcomes fuels managerial decisions and processes. This can be considered as valid, as there is a very real symbiotic relationship between both entities: reality informs academia and the other way around.

One of the main findings of this literature study is the field’s scarcity of both multilevel research as well as multi-method research (section 6 elaborates on these findings in more detail). In order to conduct more “innovative” research on business innovation, the design of the studies also need to be more multileveled and diverse. Reasoning from a reciprocal perspective, strengthening the relationship between science and business might enable that. Firms stand to benefit from better innovation research and can actively contribute to the advancement of the field by opening up their organizations to innovation researchers. By opening up and facilitating in innovation research, field studies and experimentation become more realistic. The current reliance on singular research methods, such as panel data or surveys, will decrease and studies are able to cover more ground. That way, firms can contribute to the same knowledge base that they benefit from.

(34)

34

7 Conclusion

Management and strategic perspectives on innovation are often narrowly restricted (Sawhney et al., 2006). In line with that observation, it is the objective of this research to map out how studies published by top-tier journals have conceptualized innovation in their research. More specifically, in what way is the dimensional configuration of business innovation included in the empirical foundations of studies published in leading management journals? Based on the thorough meta-analysis of 206 articles, a number of conclusions can be drawn.

To start there are a number of remarks to be made on the set-up of the studies assessed here. First, the majority of the studies was conducted at the firm level. Second, insofar studies mention the specific industry in which they took place, high-tech and medium high-tech industries dominate. A large number of studies do not specify the industry, however. Third, the most common research method appears to be empirical testing through the use of panel data in which the use of patent data is widely applied.

The dimensional configuration is reflected by top-tier journals in a rather one sided and fragmented way. The current body of knowledge contains many gaps which impedes a multi-dimensional approach for innovation research. Additionally, it prevents researchers as well as managers from clearly seeing interlinkages and connections between the various components that make up innovation. Organizations can innovate in a multitude of ways, but tend to prioritize technological product and service innovation over others. This is reflected in the research considered here as well: a significant majority of studies conceptualize innovation outcome as product and service innovation, and limit their research scope as such. The articles are varied in terms of domain, which means they apply different sorts of independent variables in their model. However, these non-technological innovation variables are only a means to facilitate and fully exploit the (technological) product and process innovations. Final results of this study support previously established theories that the field of innovation research, in this case represented by top rated publication titles, is itself in dire need of innovation. Currently, the field insufficiently includes a holistic view of the dimensions in which real-life innovation can occur. Breaking through the current research routine is required in order to pave the way for multi-dimensional, i.e. more realistic, innovation studies.

(35)

35

8 Limitations

This study recognizes the limitations of its research design. Even though there exists no method free of flaws, the constraints of the current paper are pointed out and defended. First and foremost, there is the choice of the sample of the 206 publications. These titles are derived from ten top-tier journals, according to their SJR ranking. The SJR ranking includes more than ten top rated journal titles, so it is hereby assumed that these ten titles compose a representative sample from the total amount of journals. Further, newer or more specialized journal may not have made it to the top ten ranking. It is quite possible that newer or more specialized journals contain important, progressive research on innovation. The current study is of limited size and therefore bound to a limit of data that can be analyzed. Choosing the first ten source titles ensures that the highest quality journals, according to an official ranking, are included. Newer, progressive journals are maybe not among those titles, but as this research has a custom range of 2005-2015 this may have led to inconsistency. The third limitation is the search for publications on innovation. The use of keywords allow for an effective approach, but this was narrowed to publication titles that contain either the words “innovation” or “innovate”. By filtering on solely the publication title, instead of both the title and the abstract, valuable articles on innovation may have been left out. Once more, this study is of limited size and therefore only relevant publication titles are collected. Additionally, if innovation is indeed the main subject of the research, one could reasonably assume the publication title includes the word ‘innovation’.

(36)

36

9 Suggestions for Future Research

The academic treatment of innovation and innovation research stands to significantly improve if it seriously reflects on the current body of work and its strengths as well as shortcomings. With regard to the latter, it is the absence of multi-level research, in itself a result of a too narrow conceptualization of innovation, that threatens to undermine real-world credibility. If scientific assessments of the world of innovation wants to inform organizations and their innovation efforts much like it already happens the other way around, academia needs to expand its concept of innovation by including multiple dimensions.

In addition to improving the focus of research by adding a multi-dimensional character, current research methods also leave room for improvement. The current reliance on, for example, patent data, skews results and leaves numerous avenues unexplored. Instead, the results of this meta-analysis show that innovation research would benefit greatly from the incorporation of the so-called triangulation of research method. Varying research methods produce complementary results, and can only lead to a strengthening of the field.

These recommendations closely follow those made in The Routinization of Innovation Research: a

constructively critical review of the state-of-the-science by Anderson, De Dreu and Nijstad (2004) as