Exploration and Exploitation in Innovation: Reframing the Interpretation

EXPLORATION AND EXPLOITATION IN INNOVATION

107

Exploration and Exploitation in Innovation: Reframing the Interpretation Ying Li, Wim Vanhaverbeke and Wilfred Schoenmakers There has been a burgeoning literature about exploitation and exploration since March’s seminal article in 1991. However, in reviewing the extant literature we find different interpretations of both concepts leading to ambiguity and even some inconsistency. This paper focuses in particular on the interpretation of exploration and exploitation in the literature on technological innovation. It addresses two critical research questions. First, what are the different interpretations of exploitation and exploration? Second, how can we set up a framework that reconciles these differences and reduces the ambiguity that we find in the literature? To answer these two questions, we first explain what the root causes of these different viewpoints are. Second, we provide a theoretical framework that integrates the different perspectives, sets up a new typology to define exploration and exploitation, identifies white spaces in the current research and provides guidance for future research.

Introduction

T

he notion of exploration and exploitation has been widely used in studies on organizational learning, strategic renewal and technological innovation. March (1991) introduced the two concepts as follows: ‘exploration includes things captured by terms such as search, variation, risk taking, experimentation, flexibility, discovery, and innovation. Exploitation includes such things as refinement, choice, production, efficiency, selection, implementation, and execution’ (March, 1991, p. 71). Exploration is variation-seeking, risk-taking and experimentation oriented. Exploitation is variety-reducing and efficiency oriented (March, 1991). These two concepts require different structures, processes, strategies, capabilities and cultures, and may have different impacts on an organization’s performance. Following the seminal work of March in 1991, a great number of researchers have studied the notion of exploitation and exploration from different perspectives. Although the existing literature about exploration and exploitation has greatly contributed to our understanding of technological innovation, © 2008 The Authors Journal compilation © 2008 Blackwell Publishing

organizational learning and strategic renewal, these studies displayed an amount of inconsistency in the interpretation of exploitation and exploration. First, this lack of consistency makes it difficult to compare research findings from different researchers. Second, as a lack of consistency causes greater ambiguity, it may intensify the fuzzy landscape of research on exploitation and exploration, which eventually leads to more problems in future research. Almost two decades after March’s 1991 paper, we believe it is time to look back at what we know so far and find out what the main gaps and challenges are in the research on exploitation and exploration. The large number of studies on exploration and exploitation in different research disciplines make it difficult to review all the articles in all disciplines through a single theoretical review. Therefore, we confine our attention to the literature on one single domain of organizational research – technological innovation – because technological innovation is considered as a critical competitive capability for growth and adaptation (Schumpeter, 1934), and it demonstrates a firm’s capability of effective organizational learning (Von Hippel, 1994).

Volume 17

Number 2

2008

doi:10.1111/j.1467-8691.2008.00477.x

CREATIVITY AND INNOVATION MANAGEMENT

108

The purpose of this paper is two-fold. First, we review the existing literature on technological innovation that has been published since March’s work (1991) and display the different interpretations of exploration and exploitation from different researchers. Second, based on the literature review, we try to reduce the ambiguity in the existing literature by setting up a framework that reconciles these differences. Instead of constructing a universal definition for exploration and exploitation, we argue that it is critical to distinguish two knowledge domains where different types of exploration and exploitation take place. On the one hand, exploration and exploitation may occur within the ‘function domain’ that crosses various functions along the value chain (Lavie & Rosenkopf, 2006). For instance, a manufacturing firm can form an alliance with an R&D institute in order to explore new technological opportunities. On the other hand, within each value chain function, firms may access new knowledge by local search or distant search along different dimensions. The key question is whether the new knowledge is familiar or unfamiliar, compared to a firm’s existing knowledge base. We label this as the ‘knowledge distance domain’. Reframing exploration and exploitation in such a way allows researchers to clearly understand the conceptualization of exploration and exploitation in the literature at different levels of analysis and to identify blind spots for future research. This paper is organized as follows. First, we introduce our research approach and explain how we selected the literature to review. Second, we display the ambiguity and inconsistency in the studies on exploitation and exploration in the existing innovation literature. Third, we propose a framework that may integrate the differences and reduce the conceptual ambiguity in the existing literature. Finally, we sum up with discussions and conclusions.

Research Approach In order to conduct a systematic review of studies on exploration and exploitation in the literature on technological innovation, we carried out a systematic literature search. We focused on articles published in different academic journals since 1991, when March published his seminal work, up to the present (December 2007). We applied three basic selection criteria: first, the article must focus on the notion of exploration and exploitation; second, the theme of the article must be closely related to technological innovation; third, we excluded theoretical review papers.1 A research

Volume 17

Number 2

2008

approach similar to Knoben and Oerlemans (2006) has been used for this literature search. We used the WileyInterScience database, the EBSCOHost Research database and the Web of Science database to perform a literature search. Given the differences among the search engines of these databases, we used slightly different search techniques for each of the three databases, though the underlying selection criteria remained the same. In the WileyInterScience database, the search is preliminary based on the keywords: (1) ‘Business’ as journal discipline; (2) ‘March’ in references; and (3) ‘Innovation, technology, exploration, exploitation’ in full text or abstracts.2 This search yielded 91 papers. In the EBSCOHost Research database, the search is preliminary based on the keywords: (1) ‘exploration, exploitation’ in abstracts, and (2) ‘Innovation, technology’ in full text.3 This search yielded 37 papers. In the Web of Science database, the search is based on the keywords: ‘exploration’, ‘exploitation’ and ‘innovation’ in the topic. This search yielded 46 papers. Thus, the first round of selection yielded 174 papers in total. To narrow down this list of articles, we carefully read the abstracts and full text for each paper, and eliminated those papers that only mentioned exploration and exploitation as a relevant theoretical background but did not specifically focus on them. We also eliminated those articles that were not relevant to technological innovation. This manual process reduced the number of articles further to 37. We also realized that such a search method has its disadvantages. Articles that are not listed in these three databases will not be found by this method. Therefore, we also employed a complementary source for the literature search. First, as we have been interested in this research topic for years, we have accumulated a list of papers from various journals that fall within our basic selection criteria. Second, we also consulted many researchers in this research field and asked them to recommend published papers. By carefully reading the articles from this complementary source, we finally agreed to add six papers to the list of articles. Hence, the total number of articles under review for this study is 43. Table 1 lists the selected papers in alphabetical order together with their interpretations of exploration and exploitation.

Different Interpretations of Exploration and Exploitation: A Presentation Theoretical constructs evolve as authors adapt them over time to their research needs. The © 2008 The Authors Journal compilation © 2008 Blackwell Publishing

EXPLORATION AND EXPLOITATION IN INNOVATION

109

Table 1. Alphabetical List of the Selected Literature with the Interpretations of Exploration and Exploitation Article Ahuja & Katila (2004) Ahuja & Lampert (2001)

Argyres (1996) Atuahene-Gima (2005) Audia & Goncalo (2007) Benner & Tushman (2002) Bierly & Daly (2007) Cantwell & Mudambi (2005) Cesaroni, Minin & Piccaluga (2005) Danneels (2002)

Danneels (2007) Dittrich & Duysters (2007) Dittrich, Duysters & de Man (2007) Dowell & Swaminathan (2006) Faems, Van Looy & Debackere (2005) Garcia et al. (2003)

Definition/Interpretation Path-creating search is exploration. Thus, the more diversified the search is, the greater degree of exploration (measure search diversity). Knowledge search in science and across different market dimensions. It defines exploration and exploitation based on knowledge distance, but it goes beyond local vs. distant search. It defines three levels of distant search. Novelty is categorized as ‘new to firm’, ‘new to industry’, and ‘new to world’. It only discusses the knowledge search within the technology function along the technical dimension. Exploration as technological capability broadening; exploitation as technological capability deepening. Exploration is to invest resources to refine and extend its existing product innovation knowledge, skills and processes. Exploitation is to invest resources to acquire entirely new knowledge, skills and processes. Exploration and exploitation as different types of creativity of individuals. It uses number of new subclasses and number of new citations of an inventor as indicators. It only discusses the technology knowledge. Defines exploration and exploitation in terms of technology search activities. Local search is exploitation, distant search is exploration. Exploration is experiment with radical new ideas or ways of doing things. Exploitation involves refining and leveraging existing knowledge and focus on efficiency of current practices. Competence-creating subsidiary as exploration, competence-exploiting subsidiary as exploitation. They have different nature and level of R&D. Investing in a firm’s main operations and establishing alliances to secure complementary assets are exploitation. Investing in R&D in new technology is exploration. Exploration and exploitation are defined by two dimensions of competence used in product innovation: technology and market. Exploration is to develop new technology to serve new customers, and exploitation is to strengthen existing technology to serve existing customers. Same as Danneels (2002). Exploration is non-equity alliances with new partners, who have different technologies. Exploitation is equity alliances with existing partners, who have similar technologies. Same as Dittrich & Duysters (2007).

Exploration is defined as a large variety of technology trajectories ever since a firm’s initial choice of technology. Exploratory collaboration is to create new competences such as those with universities and research institutes, while exploitative collaboration focuses on complementarities between technologies and products, such as those with customers and suppliers. Exploration is to conduct research projects, and exploitation is to conduct product development projects.

© 2008 The Authors Journal compilation © 2008 Blackwell Publishing

Volume 17

Number 2

2008

CREATIVITY AND INNOVATION MANAGEMENT

110

Table 1. continued Article Geiger & Makri (2006) Gilsing & Nooteboom (2006) Greve (2007) Hagedoorn & Duysters (2002) He & Wong (2004) Holmqvist (2004)

Jansen, Van Den Bosch & Volberda (2006) Jayanthi & Sinha (1998) Katila & Ahuja (2002)

Lavie & Rosenkopf (2006)

Lee & Ryu (2002) Lin, Yang & Demirkan (2007) McGrath (2001) Mom, Van Den Bosch & Volberda (2007) Nerkar (2003) Nerkar & Roberts (2004)

Volume 17

Number 2

2008

Definition/Interpretation Exploration as science search, exploitation as technology search. Exploration is searching and recombining technology and science; exploitation is search market knowledge. Searching new technology as exploration (explicit), searching new market as exploitation (implicit). Explorative alliances are usually established in order to explore new technological opportunities. Exploitative alliances enable firms to commercialize the technology gained through exploration. Technological innovation activities aimed at emerging new product market is exploration, and those aimed at improving existing product market is exploitation. It examines how exploration and exploitation build upon each other at firm level. The key is dissatisfaction with performance. Firms search for proximate/distant knowledge or do extensive recombination under certain incentives, either from dissatisfaction or from slack. It defines exploration and exploitation with respect to searching new or existing knowledge on the customers/markets.

Exploration as the technology search that aims at meeting future market demand; exploitation as the technology search that aims at meeting current market demand. The degree of exploration is indicated by ‘search scope’, which is how broad knowledge a firm searches. The degree of exploitation is indicated by ‘search depth’, which describes how deeply a firm reuses its existing knowledge. It defines exploration and exploitation in alliances with respect to the ‘function domain’, ‘structure domain’ and ‘attribute domain’. ‘Function domain’ refers to the nature of value chain functions. ‘Structure domain’ refers to whether to ally with a new partner without prior ties. ‘Attribute domain’ describes to what extent the new partner’s organizational attributes are different from those of prior partners. Investment in unknown technological opportunities is exploration, and investment in existing technology is exploitation. Search new knowledge through new alliance partners as exploration, while consolidating a firm’s existing partner networks is exploitation. It uses a multi-item scale to measure exploration and exploitation. It emphasizes the search for new knowledge in technology and market. Managers’ exploration activities include searching for new possibilities with respect to product, service, process or markets, which require learning of new skills and knowledge. Managers’ exploitation activities include serving existing customers with existing products/services, which requires present knowledge and accumulation of experiences. For technology search, ‘temporal recency’ is exploitation, and ‘temporal spread’ is exploration. It defines exploration and exploitation with respect to search technology and market. Distal experience in technology and market is exploration, and proximate experience in technology and market is exploitation. © 2008 The Authors Journal compilation © 2008 Blackwell Publishing

EXPLORATION AND EXPLOITATION IN INNOVATION

111

Table 1. continued Article Perretti & Negro (2007) Phene et al. (2006) Rosenkopf & Nerkar (2001) Rothaermel (2001) Rothaermel & Deeds (2004) Rothaermel, Hagedoorn & Roijakkers (2004) Sidhu, Commandeur & Volberda (2007) Sidhu, Volberda & Commandeur (2004) Vanhaverbeke & Peeters (2005) Van Looy, Martens & Debackere (2005) Vassolo, Anand, & Folta (2004)

Definition/Interpretation Recombining old, reuse existing, leverage prior knowledge is exploitation, and recombining old and new (hiring new employees) is exploration. It defines exploration and exploitation with respect to the combination of local or distant search in technical knowledge dimension and geographic/spatial dimension. It defines exploration and exploitation with respect to the combination of search across the organization boundary (spatial) and the technological boundary (technical). The activities of exploitative alliances include manufacturing, marketing or supply agreements, which are typical product market knowledge. Motivation differs in alliances: R&D alliances is exploration (technology-side), commercialization alliances is exploitation (product market knowledge). Similar to Rothaermel & Deeds (2004)

Three dimensions: 1. technology dimension (supply-side), 2. market dimension (demand-side), 3. spatial side. Supply side may involve both science and technology. It defines exploration in terms of the scope of external information acquisition (thus a search view). External information acquisition is examined through the supply-side, demand-side and geographic side. It mixes up the value chain and knowledge dimensions. Exploration is interpreted as the corporate venturing/NBD, which is related to ‘structural ambidextrous’ organization. Exploitation is to invest in the lucrative part of the technology life cycle, and exploration is to invest in various stages of the technology life cycle.

The alliances set up by big pharmaceutical companies with biotech companies are exploratory alliances, mainly to explore technological advantage.

meaning of theoretical constructs evolves as they circulate. Ambiguity often emerges along with the evolution of constructs. To reduce the ambiguity of theoretical constructs, it is important to trace the source of variation during the evolution. Therefore, we reviewed the selected articles published since 1991 and identified two main sources that cause variation in the interpretation of exploration and exploitation. First, different levels of analysis cause variety. Second, there are substantial differences in the understanding of exploration and exploitation among researchers, which is fundamentally related to what exploration and exploitation is within a particular level of © 2008 The Authors Journal compilation © 2008 Blackwell Publishing

analysis. The purpose of this section is limited to presenting these two sources of inconsistency and ambiguity in the existing literature.

Level of Analysis Scholars interpret the notion of exploration and exploitation differently because they conducted their research at different levels of analysis. These different levels of analysis constrain the focus of researchers and yield great variety in interpretation of constructs. For instance, at the individual level, exploration and exploitation are considered as two different types of creative idea generation (Audia &

Volume 17

Number 2

2008

CREATIVITY AND INNOVATION MANAGEMENT

112

Goncalo, 2007). At the project level, exploration manifests itself in the newness of a project (McGrath, 2001), or the composition of project development teams indicates the degree of exploration (Perretti & Negro, 2007). At the firm level, some scholars interpret exploration as distant knowledge search and exploitation as proximate knowledge search (Benner & Tushman, 2002; Katila & Ahuja, 2002; Nerkar & Roberts, 2004; Sidhu, Commandeur & Volberda, 2007). At the corporate group level, exploration and exploitation are usually considered in terms of corporate strategy for venturing (Cantwell & Mudambi, 2005; Vanhaverbeke & Peeters, 2005). At the alliances level, exploration and exploitation are usually seen as different motivations to enter interfirm collaboration (Hagedoorn & Duysters, 2002; Rothaermel & Deeds, 2004; Rothaermel, Hagedoorn & Roijakkers, 2004). At the industry level, exploitation and exploration build upon each other and form a dynamic ‘cycle of discovery’ (Gilsing & Nooteboom, 2006). Comparison between studies at different levels of analysis is sometimes not easy and straightforward. It, thus, requires a comprehensive understanding of the notion of exploration and exploitation at different levels of analysis. Therefore, it is important to summarize and synthesize the substantial differences in interpreting exploration and exploitation, which will be discussed later. Table 2 summarizes the selected literature according to their level of analysis. Since the majority of papers are studies on exploration and exploitation at the firm level, in the following section we discuss the substantial differences in the definition and interpretation of exploration and exploitation mostly by referring to the articles that are studies at the firm level. Articles at other levels of analysis will be referred to as related issues are raised.

Substantial Differences in Interpretation The second source of variation in the literature comes from the substantial differences in the understanding of exploration and exploitation. Although all authors agree that exploration is the search for new knowledge, technology, competences, markets or relations, and that exploitation is the further development of existing ones, their interpretation of these constructs differs substantially. Although the definitions and interpretations take very different forms, Gupta, Smith and Shalley (2006) suggested analysing the ambiguity in the definition of exploration and exploitation through the lens of the type or amount of learning. Following this suggestion, therefore, we illustrate the substantial differences in the

Volume 17

Number 2

2008

interpretation of exploration and exploitation by discussing three distinct topics, from which the ambiguity and inconsistency vividly emerge. First, it is not clear in the existing literature on which function of the value chain (science, technology or product market) learning should be considered as exploration or exploitation. In this case, exploration and exploitation are regarded as dichotomous concepts that are usually linked to a specific pair of functions along the value chain. This issue presents the different interpretations of exploration and exploitation based on the type of learning. Second, exploration and exploitation are sometimes interpreted in terms of the different dimensions of knowledge search. The degree of exploration/exploitation thus depends on how local or distant the knowledge search is along the cognitive, temporal or spatial dimension of the knowledge space. In this case, the interpretation of exploration and exploitation is not only subject to the type of learning on the one hand, because local search represents exploitation and distant search represents exploration, but also to the amount of learning on the other hand, because the degree of local or distant knowledge search can be measured on a continuous scale along different dimensions. Finally, it is also unclear whether exploration and exploitation should be seen as a learning process or as an innovation outcome. Table 3 categorizes the selected literature according to these three topics,4 which are explored further in the next section of this paper.

Learning in Science, Technology and Market along the Value Chain Exploration and exploitation are different types of learning. Exploration is associated with terms such as search, variation, risk taking, experimentation and discovery, while exploitation is associated with refinement, production, efficiency, selection, implementation and execution, for example (March, 1991). These two different types of learning activities are sometimes linked with unique functions in the value chain of a firm, where learning takes place. To simplify the issue, we focus on three main functions along the value chain: science (fundamental research), technology (product development) and product market (manufacturing and marketing). In the existing literature, there is no consensus on which function of the value chain is associated with exploration or exploitation. Hereby, we illustrate the different interpretations in the literature. First, some researchers distinguish exploration from exploitation by highlighting the © 2008 The Authors Journal compilation © 2008 Blackwell Publishing

© 2008 The Authors Journal compilation © 2008 Blackwell Publishing

Industry level

Corporate group level Alliances level

Project/project team level Firm level

Individual level

Level of analysis

Gilsing & Nooteboom (2006)

Rothaermel & Deeds (2004), Lavie & Rosenkopf (2006), Lin, Yang & Demirkan (2007), Dittrich, Duysters & de Man (2007), Dittrich & Duysters (2007)

Rothaermel, Hagedoorn & Roijakkers (2004), Hagedoorn & Duysters (2002), Faems, Van Looy & Debackere (2005)

Vanhaverbeke & Peeters (2005)

Cantwell & Mudambi (2005)

Rothaermel (2001), Vassolo, Anand & Folta (2004)

Sidhu, Volberda & Commandeur (2004), Holmqvist (2004)

Rosenkopf & Nerkar (2001), Greve (2007), Ahuja & Lampert (2001), Benner & Tushman (2002), He & Wong (2004), Katila & Ahuja (2002), Nerkar (2003), Nerkar & Roberts (2004), Garcia et al. (2003) Lee & Ryu (2002), Atuahene-Gima (2005), Bierly & Daly (2007)

Jayanthi & Sinha (1998), Cesaroni, Minin & Piccaluga (2005), Geiger & Makri (2006), Sidhu, Commandeur & Volberda (2007), Jansen, Van Den Bosch & Volberda (2006), Van Looy, Martens & Debackere (2005)

Complementary sources

Argyres (1996), Danneels (2002), Ahuja & Katila (2004), Dowell & Swaminathan (2006), Phene et al. (2006), Danneels (2007)

Mom, Van Den Bosch & Volberda (2007)

Web of Science

McGrath (2001)

Audia & Goncalo (2007)

EBSCOHost Research

Source

Perretti & Negro (2007)

WileyInterScience

Table 2. List of Selected Literature and Level of Analysis

EXPLORATION AND EXPLOITATION IN INNOVATION 113

Volume 17

Number 2

2008

CREATIVITY AND INNOVATION MANAGEMENT

114

Table 3. Categorizing Selected Literature According to the Four Discussed Issues Issues Search science, technology and product-market knowledge

Knowledge distance in cognitive, temporal or spatial dimension

Process vs. Outcome

Ahuja & Lampert (2001), Benner & Tushman (2002), Katila & Ahuja (2002), Nerkar (2003), Nerkar & Roberts (2004), Rosenkopf & Nerkar (2001), Sidhu, Volberda & Commandeur (2004), Rothaermel (2001), Ahuja & Katila (2004), Phene et al. (2006), Geiger & Makri (2006), Gilsing & Nooteboom (2006), Greve (2007), Sidhu, Commandeur & Volberda (2007), Audia & Goncalo (2007), Argyres (1996), Danneels (2002), Dowell & Swaminathan (2006), Danneels (2007), Cantwell & Mudambi (2005), Vassolo, Anand & Folta (2004), Jayanthi & Sinha (1998), Jansen, Van Den Bosch & Volberda (2006), McGrath (2001), He & Wong (2004), Hagedoorn & Duysters (2002), Rothaermel & Deeds (2004), Vanhaverbeke & Peeters (2005), Garcia et al. (2003), Van Looy, Martens & Debackere (2005), Faems, Van Looy & Debackere (2005)

Phene et al. (2006), Nerkar (2003), Sidhu, Volberda & Commandeur (2004), Sidhu, Commandeur & Volberda (2007), Rosenkopf & Nerkar (2001), Katila & Ahuja (2002), Ahuja & Katila (2004), Lavie & Rosenkopf (2006) Ahuja & Lampert (2001), Benner & Tushman (2002), Nerkar & Roberts (2004), Rosenkopf & Nerkar (2001), Rothaermel (2001), Geiger & Makri (2006), Gilsing & Nooteboom (2006), Greve (2007), Perretti & Negro (2007), Bierly & Daly (2007), Atuahene-Gima (2005), Mom, Van Den Bosch & Volberda (2007), Dittrich & Duysters (2007), Dittrich, Duysters & de Man (2007), Lin, Yang & Demirkan (2007)

Benner & Tushman (2002), He & Wong (2004), Holmqvist (2004), Katila & Ahuja (2002), Nerkar (2003), Nerkar & Roberts (2004), Argyres (1996), Jayanthi & Sinha (1998), Danneels (2002), Dowell & Swaminathan (2006), Phene et al. (2006), Jansen, Van Den Bosch & Volberda (2006), Geiger & Makri (2006), Gilsing & Nooteboom (2006), Greve (2007), Sidhu, Commandeur & Volberda (2007), Audia & Goncalo (2007), Lavie & Rosenkopf (2006) Van Looy, Martens & Debackere (2005), Lee & Ryu (2002), Faems, Van Looy & Debackere (2005)

distinction between science and technology. Science refers to knowledge concerning general theories about the relationships associated with natural and social phenomena, and technology refers to theoretical and practical knowledge, skills and experiences that are of use to develop products or services (Geiger & Makri, 2006). Science search, thus, is related to fundamental research, which is exploratory and often driven by the researcher’s curiosity, interest or intuition. It is conducted without any practical end in mind, although it may have unexpected results pointing to practical applications. Technology search is related to applied research, which is exploitative and often

Volume 17

Number 2

2008

driven by the motivation of solving a particular practical problem. Science and technology are supposed to play different roles in innovation. Science can provide a guide for new technology search (Fleming & Sorenson, 2004). Given their very different natures, the uncertain science search is argued to be exploration, and the technology search is argued to be exploitation (Ahuja & Katila, 2004; Geiger & Makri, 2006). Some examples manifest this interpretation of exploration and exploitation based on the distinction between science and technology. For instance, with respect to R&D projects, some researchers define research projects as exploration and development © 2008 The Authors Journal compilation © 2008 Blackwell Publishing

EXPLORATION AND EXPLOITATION IN INNOVATION

projects as exploitation (Garcia, Calantone & Levine, 2003). In a recent study on the biotechnology industry, Gilsing and Nooteboom (2006) describe that, for the firms in the biotechnology industry, the collaboration between a biotechnology firm and academic institutes aiming for scientific research is exploration for the biotechnology firm. Nevertheless, the search for science and technology is not sufficient to achieve successful innovations. A successful innovation also requires searching for product market knowledge gained from customers, suppliers and even competitors as a complementary source to scientific and technological knowledge (Von Hippel, 1988; Chesbrough & Rosenbloom, 2002). A commonly accepted product innovation typology defines exploration and exploitation according to the interplay between technology search and product market knowledge search (Danneels, 2002; Nerkar & Roberts, 2004). Sidhu, Volberda and Commandeur (2004) and Sidhu, Commandeur and Volberda (2007) argue that the notion of exploration and exploitation should be understood as external information acquisition. They use the term ‘supply-side’ to label technology search and the term ‘demand-side’ to label the search for product market knowledge. Jayanthi and Sinha (1998) also consider the interplay of technology and product market knowledge, but their definition of exploration and exploitation is different from that of other researchers. They define exploration as the technology search that aims at meeting future market demand, and exploitation as the technology search that aims at meeting current market demand. Several studies at the alliance level also consider technology and product market knowledge in defining exploratory or exploitative alliances (Rothaermel, 2001; Hagedoorn & Duysters, 2002; Rothaermel & Deeds, 2004; Rothaermel, Hagedoorn & Roijakkers, 2004; Lavie & Rosenkopf, 2006). Whether an alliance is exploratory or exploitative depends on the main activities and the motivation to enter an alliance. On the one hand, explorative alliances are usually established in order to explore new technological opportunities (technology search). Therefore, these alliances inevitably have an R&D component (Rothaermel & Deeds, 2004). On the other hand, exploitative alliances are those that leverage complementary competencies across the alliance partners. Exploitative alliances enable firms to commercialize the technology gained through exploration. Therefore, the activities of this type of alliance include manufacturing, marketing or supply agreements, which are typical product market knowledge (Rothaermel, 2001). © 2008 The Authors Journal compilation © 2008 Blackwell Publishing

115

The interpretation of exploration and exploitation based on the functions of the value chain has also been recognized by Lavie and Rosenkopf (2006). They argued that the ‘function domain’ is one way to define exploration and exploitation. The rationale here is that the nature of a specific function on the value chain determines ex facto whether the learning is exploratory or exploitative within the function. For each pair of functions, i.e., science vs. technology and technology vs. product market knowledge, the earlier function is exploration, based on which further exploitation takes places in the following function. While this is one way to interpret exploration and exploitation in terms of the type of learning, another track of studies does so based on the amount of learning (knowledge search).

Knowledge Search and its Three Dimensions While March (1991, p. 71) simply associates ‘search’ with exploration, many researchers later extended the idea of knowledge search in explaining exploration and exploitation. Most studies employed the idea of local or distant knowledge search to interpret exploration and exploitation. They interpreted exploitation as activities that search for familiar, mature, current or proximate knowledge; and exploration as consisting of activities that search for unfamiliar, distant and remote knowledge (Ahuja & Lampert, 2001; Rosenkopf & Nerkar, 2001; Benner & Tushman, 2002; Katila & Ahuja, 2002; Nerkar, 2003). Particularly in technological innovation, exploitation involves local search that builds on a firm’s existing technological capabilities, while exploration involves more distant search for new capabilities. Local search provides a firm with advantages in making incremental innovations, while distant search might bring opportunities for a firm to achieve radical innovations (Nerkar & Roberts, 2004). ‘Innovation is increasingly exploratory the more it departs from knowledge used in prior innovation efforts and, conversely, increasingly exploitative the more deeply anchored it is in existing firm knowledge’ (Benner & Tushman, 2002, p. 679). In the existing literature on technological innovation, distant or local knowledge search is a matter of different dimensions. We summarize three independent dimensions that construct the knowledge space. We label the first dimension as the cognitive dimension. It measures the degree of familiarity between the newly searched knowledge and a firm’s existing knowledge base in term of the cognitive distance. It is a matter of substantial content of knowledge. For instance, it is exploration for a firm that is specialized in electronic technology

Volume 17

Number 2

2008

CREATIVITY AND INNOVATION MANAGEMENT

116

to get access to technology from the pharmaceutical industry, and it is exploitation for the firm to search for new technology within the electronics industry. In the literature on technological innovation, a large number of studies define exploration and exploitation from such a perspective (Argyres, 1996; Ahuja & Lampert, 2001; Rosenkopf & Nerkar, 2001; Benner & Tushman, 2002; Katila & Ahuja, 2002; Nerkar, 2003; Dowell & Swaminathan, 2006). Some define the degree of exploration as the variety of technological trajectories developed by a firm since its initial choice of technology (Dowell & Swaminathan, 2006). Some look at the degree of novelty of the technology that a firm searches for. The ‘new to the world’ is most exploratory and ‘new to the firm only’ is least exploratory (Ahuja & Lampert, 2001). Another interpretation is that firms search for new technology within or outside their organizational boundary or technology field. In that case, exploration is considered as technology search in a new technology field outside the firm and exploitation is the search in a firm’s existing technology field within the firm (Rosenkopf & Nerkar, 2001). Most of the studies usually measure how local or distant the knowledge search is along the cognitive dimension by means of patent classification. The patent classes represent the differences between two patents with respect to their cognitive distance. Second, knowledge search also crosses the temporal dimension, which is independent from the cognitive dimension. The temporal dimension of knowledge search examines the role of time and the tension between exploitation and exploration (Katila, 2002; Nerkar, 2003). Temporal exploitation is the creation of new knowledge through searching for recent knowledge (recency) and temporal exploration is the creation of new knowledge through searching for knowledge remote in time (time spread) (Nerkar, 2003). On the one hand, owing to bounded rationality and path dependence, firms tend to look to most recent knowledge to solve current problems. Excessive temporal exploitation may lead to temporal myopia (Miller, 2002). On the other hand, older knowledge is valuable for exploration for two reasons. First, individuals and firms tend to choose the path close to the neighbourhood of their current expertise along the knowledge development path. As a result, some valuable choices might be missed in this process. Second, some of those missed opportunities were not useful because complementary knowledge was not available in the company at a particular point in time. Knowledge that was useless in the past may nevertheless have a high potential in the future because the necessary complementary knowledge and institu-

Volume 17

Number 2

2008

tions become available. The time lags between the emerging technological opportunities and complementary markets and technologies require a firm’s capability to explore the storehouse of technology over time (Garud & Nayyar, 1994). For example, large pharmaceutical firms now hire specialty biotech companies to put their previously discarded experimental compounds, some of which failed clinical trails as long as 20 years ago, through series of new tests. They hope that a then useless compound intended for one treatment may be highly useful treating something entirely different nowadays (Simons, 2006). The third dimension is the spatial (geographical) dimension, which refers to the knowledge search crossing physical space. There are different reasons why this spatial dimension matters. First, the availability of common resources within a region is related to agglomeration economies (Saxenian, 1994). Second, since ‘knowledge’ is more tacit than ‘information’, knowledge, and particularly stickyknowledge (Von Hippel, 1994; Szulanski, 1996), is more likely to be transmitted within a small geographical area where organizations have sufficient interactions and joint practices (Asheim & Isaksen, 2002). Third, the geographical dimension is usually tightly related to the institutional and cultural dimension (Knoben & Oerlemans, 2006). Given the differences among countries with respect to culture, customs and regulations, learning tends to be more difficult and the return to learning might be more uncertain across different institutional regimes than within the same institutional regime. At the macro level, the spatial dimension is a matter of national difference because particular countries develop relatively stable and distinct trajectories of technological specialization and display different patterns in R&D (Le Bas & Sierra, 2002). For example, Phene, Fladmoe-Lindquist and Marsh (2006) distinguish knowledge sources between ‘international’ and ‘national’ origin. The search for proximate technology from a national origin is analogous to exploitation, and the search for distant technology from an international origin is analogous to exploration.5 In sum, from the knowledge search perspective, exploration and exploitation are defined according to the knowledge distance between the new knowledge and the existing knowledge along any of the three dimensions of the knowledge space. Search locally is exploitation and search distantly is exploration. Since this perspective concerns the amount of learning, exploration and exploitation can be operationalized as a continuous measure along any of the three dimensions of the knowledge space, while the value chain function perspective © 2008 The Authors Journal compilation © 2008 Blackwell Publishing

EXPLORATION AND EXPLOITATION IN INNOVATION

usually treats exploration and exploitation as a dichotomous measure. The main challenge in defining exploration and exploitation is how to integrate the value chain function perspective and the knowledge distance perspective, taking the level of analysis into account.

Innovation Process vs. Innovative Outcome The third issue concerning the ambiguity and inconsistency in the interpretation of exploration and exploitation in the literature is the tension in regarding exploration and exploitation as either features of the innovation process or the innovative outcome itself. The differences in the literature on this issue are rather subtle but important. Some researchers investigate exploration and exploitation in terms of the innovation process, which involves learning activities, behaviour, investment and strategies (e.g., Jayanthi & Sinha, 1998; Nerkar, 2003; He & Wong, 2004; Nerkar & Roberts, 2004; Van Looy, Martens & Debackere, 2005; Phene, Fladmoe-Lindquist & Marsh, 2006; Sidhu, Commandeur & Volberda, 2007). These researchers regard exploration and exploitation as different forms of the learning process through which innovations come forth. He and Wong (2004, p. 485) explicitly assert that: We did not use scales related to radical versus incremental innovation because exploration and exploitation should be used with reference to a firm’s ex-ante strategic objectives in pursuing innovation, whereas the radical versus incremental innovation is often used in an ex-post outcome sense. Studies on technology search usually use patent data to measure to what extent the search is distant or local. Patents, here, are considered as indicators of technology search rather than the outcome of innovation (e.g., Argyres, 1996; Katila & Ahuja, 2002; Nerkar & Roberts, 2004). Studies on product market knowledge search usually use multi-item measurement to capture the attributes of the search process (McGrath, 2001, Sidhu, Volberda & Commandeur, 2004; Sidhu, Commandeur & Volberda, 2007). Studies at the alliance level and the industry level also interpret exploration and exploitation as organizational learning from a firm’s upstream or downstream partners (Rothaermel, 2001; Vassolo, Anand & Folta, 2004; Gilsing and Nooteboom, 2006). Others relate exploration and exploitation directly to innovative outcomes, which are the products or services (Dowell & Swaminathan, 2006; Jansen, Van Den Bosch & Volberda, 2006; Greve, 2007). In such case, exploration and exploitation are usually used as synonymous © 2008 The Authors Journal compilation © 2008 Blackwell Publishing

117

with ‘radical innovation’ and ‘incremental innovation’, respectively (Benner & Tushman, 2003; Jansen, Van Den Bosch & Volberda, 2006). For example, Greve (2007) measured exploration as the number of innovations that involved the development of new technology that is ‘new to the firm’, and exploitation as all other types of innovations. Similarly, in a study on the bicycle industry, Dowell and Swaminathan (2006) identified four types of bicycles in history. Exploration is measured by the relative numbers of the types of bicycle introduced by a firm before it finally introduces the most modern type. What counts in their study is still the innovative outcome. More interestingly, as a vivid example of how the ambiguity in the literature distorts researchers’ interpretation and makes them take the meaning of the constructs for granted, Jansen, Van Den Bosch and Volberda (2006) in their theoretical argument clearly refer to exploration and exploitation as innovative outcome, but the design of their questionnaire, measures and items take rather a process perspective. To the best of our knowledge, there are only two studies in the selected literature that examine the relationship between exploration and radical innovation, and between exploitation and incremental innovation. First, Atuahene-Gima (2005) found that exploration is positively related to radical innovation and exploitation is positively related to incremental innovation based on a study on 500 firms in a province in China. Second, Faems, Van Looy and Debackere (2005) provided empirical evidence that exploitative collaboration with suppliers and customers has a positive impact on incremental innovation, while explorative collaboration with research institutes has a positive impact on radical innovation. To summarize, the ambiguity in the definition of exploration and exploitation lies mainly between the value chain function perspective and the knowledge distance perspective, and between the innovation process viewpoint and the innovative outcome viewpoint. It seems unlikely that we can achieve a universal definition of exploration and exploitation, but it is worthwhile reconciling the different definitions in the literature within an integrated framework. Such an effort not only reduces the ambiguity in the interpretation of exploration and exploitation, but also provides guidance for future research.

Reconciling the Variety: An Integrated Framework The lack of consistency and its accompanying ambiguity in the interpretation of exploration

Volume 17

Number 2

2008

CREATIVITY AND INNOVATION MANAGEMENT

118

and exploitation call for a framework that reconciles these different perspectives. In the first place, exploration and exploitation comprise various types of organizational learning. Therefore, it is not necessary to find a universal definition for exploration and exploitation. Instead, it is logical to define and interpret the constructs from different perspectives. However, different ways of interpretation need to be organized and synthesized based on the established theories. We suggest that exploration and exploitation can be defined in two different domains. First, the ‘function domain’ defines exploration and exploitation according to the unique nature of a specific value chain function (Lavie & Rosenkopf, 2006). Second, the ‘knowledge distance domain’ takes a knowledge searching perspective and defines exploration and exploitation according to the relative distance between the new knowledge and the existing knowledge base of a firm. These two domains, respectively, correspond to the suggestion by Gupta, Smith and Shalley (2006) that exploration and exploitation should be defined according to the type or amount of learning. Further, these two domains are highly related to and embedded within each other. We propose a framework which does not develop a new theory but clarifies the relationships between these two domains of exploration and exploitation. Such a framework may be applicable to different levels of analysis. In the following, we first introduce the ‘function domain’ and ‘knowledge distance domain’ to define exploration and exploitation. Second, we explain how to link these two domains to study the balance of exploration and exploitation. Finally, by comparing the different perspectives, this framework also identifies the white spaces in the existing literature and provides guidance for future research on this topic.

‘Function Domain’ and ‘Knowledge Distance Domain’ with Three Dimensions To define exploration and exploitation, some researchers examine the search for new scientific knowledge, some focus on technology and others are interested in product market knowledge. Meanwhile, other researchers argue that the local or distant knowledge search is the key to defining exploration and exploitation. They also note that knowledge search takes place along different dimensions. However, a comprehensive picture is missing. That is because these studies define exploration and exploitation within different domains. Our framework explicitly distinguishes two domains to define exploration and exploitation.

Volume 17

Number 2

2008

First, the ‘function domain’ regards each function on the value chain as unique in its type of learning. Science, technology and product market knowledge correspond to the sequence along a firm’s value chain. In the early stages, firms invest in fundamental research to gain scientific knowledge. In the middle stages, firms conduct applied research to develop new technology. Finally, firms look for new knowledge that commercializes the products and services. Such a distinction of knowledge along the value chain may also be understood through the lens of the technology life cycle (Van Looy, Martens & Debackere, 2005). At the outset of a technology life cycle is the seed stage, when fundamental research of scientific knowledge is crucial. Activities at this seed stage are highly experimental and exploratory. At the growth stage, applied research and technological knowledge play important roles. Both the seed and growth stages are highly related to creativity. In contrast, at the mature and decline stage of a technology life cycle, productivity and commercialization are more important than creativity, where activities are highly exploitative. In this way, exploration and exploitation must be interpreted as the comparative attributes of different activities along the value chain. The logic is that for each pair of functions, i.e., science vs. technology and technology vs. product market knowledge, the earlier function is comparatively exploratory, which provides input for the next function to exploit. However, the ‘function domain’ is not sufficient to fully comprehend exploration and exploitation. Exploration and exploitation can also be defined within the ‘knowledge distance domain’, which distinguishes exploration from exploitation based on the distance between the new knowledge that a firm searches and its existing knowledge base. Local knowledge search approximates exploitation, while distant knowledge search approximates exploration. The decomposition of knowledge search into a three-dimensional space allows measuring knowledge distance along the cognitive dimension, temporal dimension and spatial dimension. The critical linkage between the ‘function domain’ and the ‘knowledge distance domain’ is that the threedimensional knowledge search may occur at any function of the value chain. Figure 1 depicts the framework that integrates the ‘function domain’ and ‘knowledge distance domain’. At each function of the value chain, knowledge can be further decomposed into three dimensions. The first dimension is the temporal dimension. It holds for the knowledge search at each function along the value chain. © 2008 The Authors Journal compilation © 2008 Blackwell Publishing

EXPLORATION AND EXPLOITATION IN INNOVATION

Temporal

Temporal Spatial

Science

combine

119

Temporal Spatial

Technology

combine

Spatial

Innovative Outcome

Productmarket

(Products/Services) Incremental or Radical

Organizational boundary Cognitive (Disciplinary)

Cognitive (Technical)

Cognitive (Segment)

Temporal

Familiar (Exploitation)

Spatial

Cognitive Unfamiliar (Exploration)

Figure 1. An Integrated Framework for Studying Exploration and Exploitation from Different Perspectives

No matter what type of knowledge a firm searches, one can always investigate how old the knowledge is and how frequent one uses this knowledge. The second dimension is the spatial dimension, which also holds for all value chain functions because one can usually trace where geographically the knowledge originates, regardless of the scientific, technological and product market knowledge. The introduction of the third dimension is based on the substantial cognitive differences between social actors, which is independent of time and space. At the science (fundamental research) function, it refers to the differences between scientific disciplines: we call it the disciplinary dimension. Scientific knowledge search may cross disciplines such as biology, chemistry, physics, electronics, etc. As to the technology function, the cognitive dimension refers to the difference between technology fields: we call it the technical dimension.6 This dimension involves skills and practices such as chemical compounds development, semiconductor material, software coding, motion engineering, etc. Finally, the cognitive dimension at the product market knowledge function refers to the differences between market segments: we label it the segmental dimension. Examples for this dimension are the different marketing practices across different industries. The best practices, well known in one market, © 2008 The Authors Journal compilation © 2008 Blackwell Publishing

can be surprisingly new in another. In sum, based on the extant literature, we identify that knowledge search differs along the value chain and takes place in different dimensions.7

Definition of Exploration and Exploitation Given the distinction and the linkage between the ‘function domain’ and the ‘knowledge distance domain’, as depicted in Figure 1, exploration and exploitation can be defined by combining both domains and taking into account whether a firm has all value chain functions. On the one hand, within a single function of the value chain, firms exploit by search for knowledge within the organizational boundary and knowledge that is local to their existing knowledge base and explore by searching distant knowledge that is unfamiliar. The local or distant search may occur along cognitive, temporal and spatial dimensions. Thus, at each value chain function, exploration and exploitation can be specified by the type of knowledge based on the ‘knowledge distance domain’. We label them scientific exploration and exploitation, technological exploration and exploitation, and product market exploration and exploitation, which are within-functional. On the other hand, firms may allocate learning activities on different value chain functions. Learning activities at the upstream of the value chain

Volume 17

Number 2

2008

CREATIVITY AND INNOVATION MANAGEMENT

120

Value chain domain

Science

Scientific exploitation

Scientific exploration

Within-functional Cross-functional exploitation

Cross-functional exploration Technological exploitation

Technological exploration

Technology Within-functional Cross-functional exploration

ProductMarket knowledge

Cross-functional exploitation

Product-market exploration

Product-market exploitation

Within-functional

Local

Distant

Knowledge distance domain

Figure 2. The Typology for Defining Within-Functional and Cross-Functional Exploration and Exploitation

are more exploratory than those at the downstream of the value chain. The rationale here is that the upstream value chain activities are more research oriented, and the downstream value chain activities are more commercially oriented with a sharp focus on seeking profitability. We label them as cross-functional exploration and exploitation (see Figure 2). Based on the typology in Figure 2, exploration and exploitation can be clearly defined by taking into account whether or not a firm is involved in all value chain functions. First, for firms that are involved in all value chain functions, exploration and exploitation can be defined either as within-functional or crossfunctional. Second, firms that are not involved in all value chain activities may either search knowledge within the value chain function in which they are active, or search knowledge from the complementary value chain functions externally. The former ones may carry on scientific, technological or product market exploration and exploitation, depending on which value chain functions they belong to. The latter ones may not only conduct withinfunctional but also cross-functional exploration and exploitation. The typology of withinfunctional and cross-functional exploration and exploitation is not only applicable at the firm level of analysis but also at the project team and corporate levels of analysis, because project teams and corporations are also capable of searching knowledge along differ-

Volume 17

Number 2

2008

ent dimensions within a single function and across different functions. It is less likely for an individual to explore different types of knowledge. Therefore, the within-functional exploration and exploitation is more suitable for the analysis of individual learning.

Combination of Exploration and Exploitation and Organizational Ambidexterity Recognizing the legitimacy of distinguishing the ‘function domain’ and the ‘knowledge distance domain’ and their linkage provides insights into how knowledge at different functions and along different dimensions can be combined and how organizations can achieve ambidexterity by combining exploration and exploitation. Given the three functions along the value chain and their three unique dimensions, one can identify two different opportunities to combine knowledge (see Figure 1). First, old and new knowledge elements gained from different dimensions can be combined within a single knowledge domain of the value chain. For instance, for technological knowledge, some studies have explored the interplay between the technical dimension and the spatial dimension within the technological domain of knowledge (Rosenkopf & Nerkar, 2001; Phene, Fladmoe-Lindquist & Marsh, 2006). Second, knowledge can also be combined across the functions along the value chain. Marketing literature gives a good © 2008 The Authors Journal compilation © 2008 Blackwell Publishing

EXPLORATION AND EXPLOITATION IN INNOVATION

example of how technological knowledge and product market knowledge can be recombined across different functions of the value chain. Keegan (1996) argues that in order to introduce a product into a new international market, firms may simply apply the same product technology and same advertising and promotion in all markets. Firms may also either change product technology or transform product image by tailor-made marketing tools. The most expensive strategy is dual adaptation, which means that firms decide to change both technological and product market knowledge in every single market. The possibilities to combine knowledge within and across value chain functions provide different opportunities for organizations to achieve ambidexterity by combining exploration and exploitation. Organizational ambidexterity in the innovation management literature is defined as the ability to simultaneously pursue both incremental and discontinuous innovation and change by exploration and exploitation (Tushman & O’Reilly, 1996). Gibson and Birkinshaw (2004) define two types of organizational ambidexterity. First, structural ambidexterity refers to balancing exploitation and exploration by allocating conflicting activities at different units at various levels within an organization; second, contextual ambidexterity refers to the behavioural capacity to simultaneously demonstrate alignment and adaptability across an entire business unit (Gibson & Birkinshaw, 2004). One way to realize contextual ambidexterity is to combine exploration and exploitation across domains. Lavie and Rosenkopf (2006, p. 804) identified three distinct domains in defining the exploration and exploitation at the alliance level of analysis. They argued that firms can combine exploration and exploitation across the function, structure and attribute domains, so that the tendency to explore (exploit) in one domain will be compensated by the tendency to exploit (explore) in some other domains. Although we introduced two different domains – ‘function domain’ and ‘knowledge distance domain’ – to define exploration and exploitation, the logic of their combination within and across domains is similar to Lavie and Rosenkopf (2006). First, within a single value chain, one can combine the exploration and exploitation from the ‘knowledge distance domain’ perspective by searching distant knowledge in some dimensions and local knowledge in other dimensions. For instance, if a firm searches knowledge from unfamiliar technical fields, it can reduce the risk of exploration by limiting its search scope within a recent time span and within a nearby geo© 2008 The Authors Journal compilation © 2008 Blackwell Publishing

121

graphic region. The second way of combining exploration and exploitation might be to search knowledge from upstream and downstream of the value chain functions simultaneously. However, this balancing strategy requires companies to get access to scientific research, manufacturing, and marketing and sales, which sometimes is too expensive to realize. Therefore, the third way of balancing is to combine the ‘function domain’ and the ‘knowledge distance domain’. For instance, one can explore the knowledge from the upstream functions of the value chain, while searching knowledge locally within its own value chain functions, or exploit the knowledge from the downstream functions of the value chain, while searching knowledge distantly within its own value chain functions. Various combinations of exploration and exploitation across the ‘function domain’ and the ‘knowledge distance domain’ are possible. We suggest that future research can examine the notion of organizational ambidexterity through the combination of exploration and exploitation based on the distinction between the ‘function domain’ and the ‘knowledge distance domain’ for exploration and exploitation. Given the ‘function domain’ and the ‘knowledge distance domain’, one can see that some studies in the existing literature either confuse these two domains or focus only on a specific part of one single domain. For instance, Sidhu, Volberda and Commandeur (2004) and Sidhu, Commandeur and Volberda (2007) propose that exploration and exploitation have their ‘supply-side’, ‘demand-side’ and ‘geographic side’, which correspond to the technology function, product market function and the spatial dimension, respectively. They mixed up two functions in the ‘function domain’ with a single dimension in the ‘knowledge distance domain’. Other researchers focus only on one type of knowledge along the value chain and usually only look at one or two dimensions. For instance, Nerkar (2003) focuses only on technological knowledge and its temporal dimension. Benner and Tushman (2002) look only at technological knowledge and its technical dimension. Geiger and Makri (2006) investigate only scientific knowledge and technological knowledge but do not recognize the three dimensions of search. And there are many more such examples. Research in the future should provide more meaningful insight by clearly defining which functions of the value chain and which dimensions are under investigation. A comprehensive investigation of knowledge search along the entire value chain including all the dimensions is a major but interesting challenge for future research.

Volume 17

Number 2

2008

CREATIVITY AND INNOVATION MANAGEMENT

122

Innovation Process vs. Innovative Outcome Finally, the difference in interpretation of exploration and exploitation as an innovation process or an innovative outcome also leaves room for interesting research in the future. Both the process view and the outcome view have their legitimacy and merits. However, we suggest that it is better to define exploration and exploitation from the innovation process perspective. After all, the reason why exploration and exploitation are so important for the adaptation and survival of organizations is because these two distinctive types of learning entail different risk taking, require different investment and resources, and eventually lead to different economic returns. Exploration and exploitation, thus, as learning activities per se, compete for managers’ attention and determine the growth of an organization. The innovative outcomes, i.e., products or services, by no means can represent the complex learning process that a firm goes through. Furthermore, there has been little insight in the existing literature into whether exploratory processes always lead to radical innovation and exploitative processes always lead to incremental innovation. In the extant literature, only a few studies look at the link between process and outcome, and, more importantly, they fail to measure how exploratory or exploitative the outcome is. Researchers usually use a count variable to measure outcome, i.e., the number of new products or innovations (Ahuja & Lampert, 2001; Katila & Ahuja, 2002; Ahuja & Katila, 2004). Sidhu, Commandeur and Volberda (2007) measure innovativeness as the total sales from new products. Nerkar and Roberts (2004) look at the initial sales level as an indicator of new product success. Consequently, readers cannot know how radical the innovations are. For those who do measure the innovativeness of the outcome, they usually pay no attention to knowledge search as a learning process (Jansen, Van Den Bosch & Volberda, 2006). Perhaps the impact of technology on the later knowledge creation can be used as an indicator of innovativeness, but researchers usually limit their focus to the technological knowledge and overlook the product market side (Rosenkopf & Nerkar, 2001; Nerkar, 2003). Hence, there is a lack of comprehensive understanding of how exploratory/exploitative learning processes lead to exploratory/exploitative innovations. Although AtuaheneGima (2005) and Faems, Van Looy and Debackere (2005) found empirical evidences that exploratory activities are associated with radical innovative outcomes and exploitative activities are associated with incremental

Volume 17

Number 2

2008

innovative outcomes, it is still questionable whether exploration in some activities may ultimately result in incremental innovations, and whether exploitation in some activities may ultimately surprise a firm with radical innovations. This doubt becomes more vivid if one considers exploration and exploitation as learning processes that occur along the value chain function and its multiple dimensions. Therefore, it might be promising to study the match between exploration and exploitation as an innovation process on the one hand, and the innovativeness of products or services as innovative outcomes on the other.

Conclusion This paper is based on a systematic review of the technological innovation literature on exploration and exploitation since the seminal paper of March (1991). We focus on how different studies define and interpret these constructs. We find that the variety in the literature comes from two sources, namely the level of analysis and the substantial differences in explanation and interpretation of the two concepts. We discuss in detail several issues from which inconsistency and ambiguity emerge in the literature. First, exploration and exploitation are interpreted differently because researchers made their analysis at different levels. There are studies at the individual level, project level, business level, corporate group level, alliance level and industry level. Studies at different levels of analysis focus on different social actors. In this case, the source of variety comes from who explores or exploits. The second source of variation in the literature comes from the substantial differences in the understanding of exploration and exploitation. Some authors define exploration and exploitation from a knowledge search perspective. While most researchers agree that exploration is the search for distant knowledge and exploitation is the search for local knowledge, there is more than one dimension along which knowledge search can take place. Ambiguity also emerges when some define exploration and exploitation based on whether a firm searches for scientific, technological or product-market knowledge within the functions of the value chain. Furthermore, it is unclear how distant and local search can be combined with the knowledge search in science, technology or market applications. Another source of ambiguity comes from whether exploration and exploitation should be interpreted as an innovation process or as the innovative outcomes. There has been little insight into whether © 2008 The Authors Journal compilation © 2008 Blackwell Publishing

EXPLORATION AND EXPLOITATION IN INNOVATION

exploratory or exploitative knowledge search processes eventually lead to radical or incremental innovative outcomes. Given the ambiguity identified in the extant literature, we propose a framework that aims not to develop new theory but rather clarifies the relationships between the different perspectives and focuses on what has been overlooked in the earlier studies. In so doing, this study contributes to the literature in several ways. First, it clarifies the ambiguity in the existing literature in interpreting exploration and exploitation. Second, it clarifies the distinction between knowledge domains along the value chain and the search dimensions, which have not been clearly defined and are usually operationalized in a deficient way in the literature. The framework points out that those activities at earlier functions of the value chain are by nature more exploratory than those at the later stages of the value chain. Third, based on the ‘function domain’ and the ‘knowledge distance domain’, we propose a typology to define within-functional exploration and exploitation and cross-functional exploration and exploitation. Next, this framework also contributes to explaining how firms can achieve organizational ambidexterity by combining exploration and exploitation within or across domains. Firms can balance exploration and exploitation not only within a value chain function along different dimensions of search but also across value chain functions. Finally, our study indicates some gaps in the current research and provides guidance for future research on exploitation and exploration. For instance, future research should clearly define which functions of the value chain and which dimensions are under investigation. A comprehensive investigation into knowledge search along the whole value chain could be promising too. Similarly, integrating different search dimensions is also highly recommended in future research. Interesting research opportunities may also arise if studies focus on knowledge search both within and across knowledge domains along the value chain. Moreover, studying the match between knowledge search as an innovation process on the one hand, and the innovativeness of products or services on the other, could also be a promising direction for future research. However, our framework was developed based on the existing literature in innovation management with a strong emphasis on the organizational learning perspective. We focused on technology-based innovations. This is reflected in the framework through the links between science and technology on the one hand and between technology and new product development on the other. Future © 2008 The Authors Journal compilation © 2008 Blackwell Publishing

123

research should broaden the scope of perspectives from which exploration and exploitation can be examined. For instance, since many technological developments are not necessarily driven by scientific insights, technological developments are not the only source of new businesses opportunities. They can be generated by new business models, customer insights, design and fashion, new social needs, etc. Furthermore, owing to the different natures of learning for exploration and exploitation, they may have different relational features. For instance, exploration and exploitation may differ in ways of building social ties and network configurations and risk taking among organizations. Thus, a relational perspective on exploration and exploitation, complementary to the learning perspective, can be a promising direction for future studies. Finally, exploration and exploitation are shown as multi-level and multi-facet concepts in the framework, which may provide practitioners with several insightful implications for innovation management. First, we have distinguished three types of knowledge along the value chain that are crucial in technological innovations. Scientific knowledge is increasingly important as the source of technological developments, which, in turn, opens opportunities to develop new products or services. Firms that want to develop new products based on science-driven technologies have to understand two processes. First, they need to understand how new technology can be developed successfully from new scientific insights. Second, they need to translate new technologies into new and profitable business models. Scientific research stands on one side of the technological innovation and commercialization stands on the other. The technological innovativeness, which is usually measured by the citation of new patents, is only an intermediate output. What really counts is the successful introduction of new products or services in the markets. Most studies have focused on the intermediate output, i.e., the patents, rather than the products or services as the final innovation outcome. The critical challenge is to manage the three functions as a whole in a coherent way. Exploration and exploitation in science and product market knowledge have different drivers and should be managed differently. Exploration of new scientific disciplines to rejuvenate technological capabilities is a quite different task from the development of new business based on the current technology in the company. The technology base itself should be continuously adapted in conjunction with new societal needs (demand pull) and scientific discoveries (technology push). The integrated framework introduced in this

Volume 17

Number 2

2008

CREATIVITY AND INNOVATION MANAGEMENT

124

study provides managers with a guide to trace the variation in organizational activities in scientific research, technology development and commercialization as a whole. For instance, intensive investment in exploratory scientific research may be best combined with focused product development projects and market plans. Firms that are specialized in one particular function of the value chain may make better use of their expertise by allying with other firms who have complementary functions within the value chain. We hope that our study not only inspires more researchers to explore the gaps concerning exploration and exploitation, but also provides a common ground to encourage the necessary dialogue between academia and practitioners.

Notes 1 We focus on empirical papers to limit the review effort. From empirical papers, we can see directly how researchers interpret exploration and exploitation, while theoretical review papers usually present the reviewer’s subjective interpretation of other authors. We exclude theoretical papers from the literature review, but still refer to them where analysis is necessary. 2 We also used ‘innovation, technology, exploring, exploiting’ or ‘innovation, technology, exploratory, exploitative’ in full text or abstracts. This made no difference to the results. The relationship between these keywords is ‘AND’, which means there is no sequence for the keywords. 3 We also used ‘exploring, exploiting’ or ‘exploratory, exploitative’ in abstracts. This made no difference to the results. The relationship between these keywords is ‘AND’, which means there is no sequence for the keywords. 4 Note that in this section, we only intend to present the substantial differences in the existing literature. Further analysis and integration will follow in the next section. 5 At the micro level, one could consider interorganizational learning as more ‘distant’ or more explorative than intra-firm learning. While reusing a firm’s own knowledge does not necessarily involve spatial distance, acquiring knowledge from other firms inevitably crosses space. For example, Rosenkopf and Nerkar (2001) define the degree of exploration depending on whether knowledge search spans organizational and technological boundaries. Search with the strongest explorative nature is search for distant technology from outside the organization. The least exploratory search is local technology search from within the organization. 6 We use ‘technological’ to label knowledge in the value chain, and ‘technical’ for its content dimension in order to avoid misunderstanding. 7 It is worth noting that studies at the alliance level consider two other domains to define exploratory alliances and exploitative alliances (Lavie & Rosenkopf, 2006). One is the ‘structure domain’,

Volume 17

Number 2

2008

which refers to whether or not to form an alliance with a partner with no prior ties. The other is the ‘attribute domain’, which refers to how different a new partner’s organizational attributes are from a firm’s other prior partners. We acknowledge the merits of these two domains. However, these two domains do not go beyond the ‘knowledge distance domain’ as we introduced in our framework. The essence behind whether or not having prior ties and any difference in attributes between old and new partners, eventually turns out to be the knowledge distance between the focal firm and the new partner. And these knowledge differences can, anyhow, be captured by the cognition, temporal and spatial dimensions. For instance, having a new partner with no prior ties implies that the new partner is from a different industry or from a foreign market, which the focal firm is considering tapping into for the first time. Lin, Yang and Demirkan (2007) also recognized that the ‘structure domain’ is highly related to new knowledge searching beyond the local boundary. The ‘attribute domain’ implies that the new partner has a very different knowledge base from other prior partners. Still, it does not go beyond the ‘knowledge distance domain’. For these reasons, we do not incorporate the ‘structure domain’ and ‘attribute domain’ into our framework.

References Ahuja, G. and Katila, R. (2004) Where Do Resources Come from? The Role of Idiosyncratic Situations. Strategic Management Journal, 25, 887–907. Ahuja, G. and Lampert, C.M. (2001) Entrepreneurship in the Large Corporation: A Longitudinal Study of How Established Firms Create Breakthrough Inventions. Strategic Management Journal, 22, 521–43. Argyres, N. (1996) Capabilities, Technological Diversification and Divisionalization. Strategic Management Journal, 17, 395–410. Asheim, B.T. and Isaksen, A. (2002) Regional Innovation Systems: The Integration of Local ‘Sticky’ and Global ‘Ubiquitous’ Knowledge. Journal of Technology Transfer, 27, 77–86. Atuahene-Gima, K. (2005) Resolving the CapabilityRigidity Paradox in New Product Innovation. Journal of Marketing, 69, 61–83. Audia, P.G. and Goncalo, J.A. (2007) Past Success and Creativity over Time: A Study of Inventors in the Hard Disk Drive Industry. Management Science, 53, 1–15. Benner, M.J. and Tushman, M.L. (2002) Process Management and Technological Innovation: A Longitudinal Study of the Photography and Paint Industries. Administrative Science Quarterly, 47, 676–706. Benner, M.J. and Tushman, M.L. (2003) Exploitation, Exploration, and Process Management: The Productivity Dilemma Revisited. Academy of Management Review, 28, 238–56. Bierly, P.E. and Daly, P.S. (2007) Alternative Knowledge Strategies, Competitive Environment, and © 2008 The Authors Journal compilation © 2008 Blackwell Publishing

EXPLORATION AND EXPLOITATION IN INNOVATION

Organizational Performance in Small Manufacturing Firms. Entrepreneurship Theory and Practice, 31, 493–516. Cantwell, J. and Mudambi, R. (2005) MNE Competence-Creating Subsidiary Mandates. Strategic Management Journal, 26, 1109–28. Cesaroni, F., Minin, A.D. and Piccaluga, A. (2005) Exploration and Exploitation Strategies in Industrial R&D. Creativity & Innovation Management, 14, 222–32. Chesbrough, H. and Rosenbloom, R. (2002) The Role of the Business Model in Capturing Value from Innovation: Evidence from Xerox Corporation’s Technology Spin-off Companies. Industrial and Corporate Change, 11, 529–55. Danneels, E. (2002) The Dynamics of Product Innovation and Firm Competences. Strategic Management Journal, 23, 1095–121. Danneels, E. (2007) The Process of Technological Competence Leveraging. Strategic Management Journal, 28, 511–33. Dittrich, K. and Duysters, G. (2007) Networking as a Means to Strategy Change: The Case of Open Innovation in Mobile Telephony. Journal of Product Innovation Management, 24, 510–21. Dittrich, K., Duysters, G. and de Man, A.P. (2007) Strategic Repositioning by Means of Alliance Networks: The Case of IBM. Research Policy, 36, 1496–511. Dowell, G. and Swaminathan, A. (2006) Entry Timing, Exploration, and Firm Survival in the Early U.S. Bicycle Industry. Strategic Management Journal, 27, 1159–82. Faems, D., Van Looy, B. and Debackere K. (2005) International Collaboration and Innovation: Toward a Portfolio Approach. Journal of Product Innovation Management, 22, 238–50. Fleming, L. and Sorenson, O. (2004) Science as a Map in Technological Search. Strategic Management Journal, 25, 909–28. Garcia, R., Calantone, R. and Levine, R. (2003) The Role of Knowledge in Resource Allocation to Exploration versus Exploitation in Technologically Oriented Organizations. Decision Sciences, 34, 323–49. Garud, R. and Nayyar, P.R. (1994) Transformative Capacity: Continual Structuring by Intertemporal Technology Transfer. Strategic Management Journal, 15, 365–85. Geiger, S.W. and Makri, M. (2006) Exploration and Exploitation Innovation Processes: The Role of Organizational Slack in R&D Intensive Firms. Journal of High Technology Management Research, 17, 97–108. Gibson, C.B. and Birkinshaw, J. (2004) The Antecedents, Consequences, and Mediating Role of Organizational Ambidexterity. Academy of Management Journal, 47, 209–26. Gilsing, V. and Nooteboom, B. (2006) Exploration and Exploitation in Innovation Systems: The Case of Pharmaceutical Biotechnology. Research Policy, 35, 1–23. Greve, H.R. (2007) Exploration and Exploitation in Product Innovation. Industrial and Corporate Change, May, 1–31. Gupta, A.K., Smith, K.G. and Shalley, C.E. (2006) The Interplay between Exploration and Exploi© 2008 The Authors Journal compilation © 2008 Blackwell Publishing

125

tation. Academy of Management Journal, 49, 693– 706. Hagedoorn, J. and Duysters, G. (2002) Learning in Dynamic Inter-Firm Networks: The Efficacy of Multiple Contacts. Organization Studies, 23, 525– 48. He, Z.-L. and Wong, P.-K. (2004) Exploration vs. Exploitation: An Empirical Test of the Ambidexterity Hypothesis. Organization Science, 15, 481– 94. Holmqvist, M. (2004) Experiential Learning Process of Exploitation and Exploration within and between Organizations: An Empirical Study of Product Development. Organization Science, 15, 70–81. Jansen, J.J.P., Van Den Bosch, F.A.J. and Volberda, H.W. (2006) Exploratory Innovation, Exploitative Innovation, and Performance: Effects of Organizational Antecedents and Environmental Moderators. Management Science, 52, 1661–74. Jayanthi, S. and Sinha, K.K. (1998) Innovation Implementation in High Technology Manufacturing: A Chaos-Theoretic Empirical Analysis. Journal of Operations Management, 16, 471–94. Katila, R. (2002) New Product Search over Time: Past Ideas in Their Prime? Academy of Management Journal, 45, 995–1010. Katila, R. and Ahuja, G. (2002) Something Old, Something New: A Longitudinal Study of Search Behavior and New Product Introduction. Academy of Management Journal, 45, 1183–94. Keegan, W.J. (1996) Global Marketing Management, 5th edn. Prentice-Hall, Englewood Cliffs, NJ. Knoben, J. and Oerlemans, L.A.G. (2006) Proximity and Inter-Organizational Collaboration: A Literature Review. International Journal of Management Reviews, 8, 71–89. Lavie, D. and Rosenkopf, L. (2006) Balancing Exploration and Exploitation Alliance Formation. Academy of Management Journal, 49, 797–818. Le Bas, C.B. and Sierra, C. (2002) Location versus Home Country Advantages in R&D Activities: Some Further Results on Multinationals Locational Strategies. Research Policy, 31, 589–609. Lee, J. and Ryu, Y.U. (2002) Exploration, Exploitation and Adaptive Rationality: The NeoSchumpeterian Perspective. Simulation Modeling Practice and Theory, 10, 297–321. Lin, Z., Yang, H.B. and Demirkan, I. (2007) The Performance Consequences of Ambidexterity in Strategic Alliance Formations: Empirical Investigation and Computational Theorizing. Management Science, 53, 1645–58. McGrath, R.G. (2001) Exploratory Learning, Innovative Capacity, and Managerial Oversight. Academy of Management Journal, 44, 181–31. March, J.G. (1991) Exploration and Exploitation in Organizational Learning. Organization Science, 2, 71–87. Miller, K. (2002) Knowledge Inventories and Managerial Myopia. Strategic Management Journal, 23, 689–706. Mom, T.J.M., Van den Bosch, F.A.J. and Volberda, H.W. (2007) Investigating Managers’ Exploration and Exploitation Activities: The Influence of TopDown, Bottom-Up, and Horizontal Knowledge Inflows. Journal of Management Studies, 44, 910–31.

Volume 17

Number 2

2008

CREATIVITY AND INNOVATION MANAGEMENT

126

Nerkar, A. (2003) Old is Gold? The Value of Temporal Exploration in the Creation of New Knowledge. Management Science, 49, 211–29. Nerkar, A. and Roberts, P.W. (2004) Technological and Product-Market Experience and the Success of New Product Introductions in the Pharmaceutical Industry. Strategic Management Journal, 25, 779–99. Perretti, F. and Negro, G. (2007) Mixing Genres and Matching People: A Study in Innovation and Team Composition in Hollywood. Journal of Organizational Behavior, 28, 563–86. Phene, A., Fladmoe-Lindquist, K. and Marsh, L. (2006) Breakthrough Innovations in the U.S. Biotechnology Industry: The Effects of Technological Space and Geographic Origin. Strategic Management Journal, 27, 369–88. Rosenkopf, L. and Nerkar, A. (2001) Beyond Local Search: Boundary-Spanning, Exploration, and Impact in the Optical Disk Industry. Strategic Management Journal, 22, 287–306. Rothaermel, F. (2001) Incumbent’s Advantage through Exploiting Complementary Assets via Interfirm Cooperation. Strategic Management Journal, 22, 687–99. Rothaermel, F. and Deeds, D.L. (2004) Exploration and Exploitation Alliances in Biotechnology: A System of New Product Development. Strategic Management Journal, 25, 201–21. Rothaermel, F., Hagedoorn, J. and Roijakkers, N. (2004) Technological Core Transformation through Collaboration: The Role of Exploration and Exploitation Alliances. Academy of Management Special Research Forum: Managing Exploration and Exploitation. Saxenian, A. (1994) Regional Advantage, Culture and Competition in Silicon Valley and Route 128. Harvard University Press, Cambridge, MA. Schumpeter, J. (1934) The Theory of Economic Development. Harvard University Press, Cambridge, MA. Sidhu, J.S., Volberda, H.W. and Commandeur, H.R. (2004) Exploring Exploration Orientation and its Determinants: Some Empirical Evidence. Journal of Management Studies, 41, 913–32. Sidhu, J.S., Commandeur, H.R. and Volberda, H.W. (2007) The Multifaceted Nature of Exploration and Exploitation: Value of Supply, Demand, and Spatial Search for Innovation. Organization Science, 18, 20–38.

Volume 17

Number 2

2008

Simons, J. (2006) Big Pharma’s New R&D Center: The Trash Bin. Fortune (European edn), 154, 28. Szulanski, G. (1996) Exploring Internal Stickiness: Impediments to the Transfer of Best Practice within the Firm. Strategic Management Journal, 17, 27–43. Tushman, M.L. and O’Reilly, C.A. III (1996) Ambidextrous Organizations: Managing Evolutionary and Revolutionary Change. California Management Review, 38, 8–30. Vanhaverbeke, W. and Peeters, N. (2005) Embracing Innovation as Strategy: Corporate Venturing, Competence Building and Corporate Strategy Making. Creativity and Innovation Management, 14, 246–57. Van Looy, B., Martens, T. and Debackere, K. (2005) Organizing for Continuous Innovation: On the Sustainability of Ambidextrous Organizations. Creativity and Innovation Management, 14, 208–21. Vassolo, R.S., Anand, J. and Folta, T.B. (2004) Nonadditivity in Portfolios of Exploration Activities: A Real Options-Based Analysis of Equity Alliances in Biotechnology. Strategic Management Journal, 25, 1045–61. Von Hippel, E. (1988) The Source of Innovation. Oxford University Press, New York. Von Hippel, E. (1994) Sticky Information and the Locus of Problem Solving: Implications for Innovation. Management Science, 40, 429–39.

Ying Li ([email protected]) is a researcher and PhD candidate at the Faculty of Applied Economics, Hasselt University, Belgium. Wim Vanhaverbeke is a professor at the Faculty of Applied Economics, Hasselt University, Belgium and visiting research fellow at the Eindhoven University of Technology, the Netherlands. Wilfred Schoenmakers is an assistant professor at the Faculty of Applied Economics, Hasselt University. He holds his PhD from Maastricht University, the Netherlands.

© 2008 The Authors Journal compilation © 2008 Blackwell Publishing