行政院國家科學委員會專題研究計畫 成果報告
促成組織雙歧創新管理的方法績效效應: 台灣與中國大陸 企業的過程長期實證研究
研究成果報告(精簡版)
計 畫 類 別 ： 個別型
計 畫 編 號 ： NSC 100-2410-H-011-001-
執 行 期 間 ： 100 年 03 月 01 日至 101 年 05 月 31 日 執 行 單 位 ： 國立臺灣科技大學企業管理系
計 畫 主 持 人 ： 張譯尹
計畫參與人員： 碩士班研究生-兼任助理人員：林育聖 碩士班研究生-兼任助理人員：蔡欣哲 大專生-兼任助理人員：林逸萍
報 告 附 件 ： 赴大陸地區研究心得報告
出席國際會議研究心得報告及發表論文
公 開 資 訊 ： 本計畫涉及專利或其他智慧財產權，1 年後可公開查詢
中 華 民 國 101 年 11 月 23 日
中 文 摘 要 ： 本研究透過探討組織雙歧的前因變項為組織結構、組織情境 因素以及領導者的領導風格對促成組織雙歧與組織績效之關 係。以 265 家台灣與大陸的中小企業為研究對象，研究結果 發現正確的組織架構以及採取有效的領導風格就有效促成一 個平衡型的組織雙歧創新。研究發現亦指出平衡型的組織雙 歧創新會有效中介組織結構、組織情境因素以及領導者的領 導風與組織績效之關係
中文關鍵詞： 組織雙歧, 組織結構、組織情境因素, 領導者的領導風格 英 文 摘 要 ：
英文關鍵詞： organizational ambidexterity, structural antecedents, contextual antecedents, leadership styles
行政院國家科學委員會補助專題研究 計畫期末報告
促成組織雙歧創新管理的方法績效效應(精簡版)
計畫類別：個別型計畫
計畫編號：NSC 100-2410-H-011 -001
執行期間：100 年 3 月 1 日至 101 年 5 月 31 日 執行機構及系所：國立臺灣科技大學企業管理系
計畫主持人：張譯尹
計畫參與人員：碩士班研究生-兼任助理人員：林育聖 蔡欣哲 大學部兼任助理人員：林逸萍
處理方式：本計畫可公開查詢
中 華 民 國 101 年 7 月 6 日
Introduction
The best firms are increasingly believed to be those who can simultaneously balance explorative
innovation with exploitative innovation in an ambidextrous fashion (He and Wong 2004; Morgan and
Berthon 2008; Raisch and Birkinshaw 2008; Raisch et al. 2009). However, to be ambidextrous, firms
must reconcile the inherent tensions that exist between acts of exploration and exploitation (March
1991). These tensions are brought about by conflicting task demands (Raisch and Birkinshaw 2008)
and competing firm design requirements (March 1991; Tushman and O’Reilly 1996). Although these
problems were initially thought of as insurmountable trade-offs forcing firms to choose either
explorative or exploitative innovation pathways (Raisch and Birkinshaw 2008), scholars have recently
put forward a series of business solutions to resolve the ambidexterity problem. One solution in
particular is that firms can shape an appropriate organizational context supportive of both innovation
types (Gibson and Birkinshaw 2004).
Structural, contextual and leadership solutions to create ambidexterity between both types of
innovation have been proposed (see Raisch and Birkinshaw 2008, for a detailed review). Structural
solutions advocate the spatial separation of explorative and exploitative innovations into separate
business units to be coordinated by integration mechanisms (Jansen et al., 2006; Raisch and
Birkinshaw 2008; Tushman and O’Reilly 1996). This is based on the assumption at the origin of
ambidexterity theory about the absolute incompatibility of explorative and exploitative activities
(March, 1991). However, recent studies have proposed that both innovations can occur within single
firms so long as the organizational context is properly specified.
Gibson and Birkinshaw (2004) suggest that contextual ambidexterity between both innovations
can be created by identifying and implementing conditions complementary to both, reducing in turn
the risk that one innovation type will self-replicate systems and processes destructive to the other (e.g.,
Hughes et al., 2007; March, 1991). Indeed, proponents of the structural separation view have accepted
that achieving ambidexterity is not simply a matter of the spatial separation of conflicting innovation
activities. For example, O’Reilly and Tushman (2007), and Tushman (2005), and Tushman and
O’Reilly (1996) highlight over-arching vision and values, flexibility and culture as conditions
supportive of ambidexterity. It is on this basis that Gibson and Birkinshaw (2004) put forward
organizational context as a route to contextual ambidexterity, validating a set of internal firm
conditions such as cooperation, autonomy and rewards in the process.
Studies into structural ambidexterity and contextual ambidexterity have also proposed that
leadership may be a critical factor in enabling innovation ambidexterity. For example, O’Reilly and
Tushman (2007), and Tushman (2005), Birkinshaw and Gibson (2004), Gibson and Birkinshaw (2004),
and Tushman and O’Reilly (1996) all suggest that supportive leaders, flexible managers and an
aligned top management team are important antecedents underpinning any form of ambidexterity. In
turn, recent studies have extended the leadership theme present in Tushman and O’Reilly’s (1996)
original thesis to suggest that leaders are essential in the ambidextrous coordination of explorative and
exploitative innovation activities (for example, Lubatkin et al., 2006; Mom et al., 2007).
So far, structural, contextual and leadership solutions are all presented as solutions to the
ambidexterity problem (Raisch and Birkinshaw 2008). However, given points raised by Tushman and
O’Reilly (1996) and several authors since, it appears increasingly apparent that these pathways
overlap. As such, our understanding of how ambidexterity is achieved is incomplete until we consider
how these conditions come together (Raisch et al., 2009; Raisch and Birkinshaw 2008). Raisch and
Birkinshaw (2008) in their review of the ‘state of the art’ propose that organizational ambidexterity
theory needs development by viewing these paths and their associated variables as complementary
rather than competing. Yet, so far, no study has brought all three strands together, leaving an important
gap in our knowledge of the theory a practice of ambidexterity.
A second important problem in the theory of ambidexterity is that, so far, almost all of the
prescriptions put forward by conceptual and empirical works are designed for large, multiunit firms.
With few exceptions (for example, Lubatkin et al., 2006), work on ambidexterity has failed to account
for SMEs. SMEs may operate differently and exhibit different operating conditions and characteristics
to large, multiunit firms such that generalising current prescriptions for ambidexterity into innovation
strategies for these firms might prove incorrect, inappropriate or dangerous. Prior studies found that
SMEs tend to use different types innovation ambidexterity compared to larger firms (Cao et al., 2009;
Ebben and Johnson, 2005). This is because SMEs differ from larger firms in terms of available
resources such as human resources capital and financial capital (Cooper et al., 1994; Forbes and
Milliken, 1999). In addition, SMEs may differ in the pursuit of innovation strategies from larger firms
because SMEs have limited managerial expertise (Pissarides, 1999; Forbes and Millken, 1999) to
effectively manage the difficult internal and external environments (Ebben and Johnson, 2005). Cao et
al. (2009) revealed that SMEs as relatively resource constrained firms are beneficial to a balance
dimension of innovation ambidexterity (BD) but larger firms are beneficial to combined dimension of
innovation ambidexterity (CD). Moreover, SMEs differ from larger firms in terms of innovation
strategies due to the fact that SMEs have less bureaucratic, structured, diversified than larger firms
(Forbes and Milliken, 1999) and fewer formal systems and procedures and fewer planning activities to
be performed (Busenitz and Barney, 1997). Consequently, SMEs faced greater challenges in managing
tensions and tradeoffs associated with explorative and exploitative innovations than larger firms
(Bierly and Daly, 2007).
Concerns also exist about the lack of slack resources needed to create and benefit from innovation
ambidexterity in these firms. Accordingly, SMEs might seek for balanced dimension of ambidexterity
(BD) owing to limited resources available to them (Cao et al., 2009). This is because SMEs can
enhance business performance through ‘reducing the performance damaging effects of
overengagement in exploitation to the detriment of exploration or it vice versa (Cao et al., 2009: 782).
Given that fact that SMEs differ from larger firms in terms of organizational structures, leadership
styles, reactions to the environments, available resources, and internal contexts they are operating
(Chen and Hambrick, 1995; Ebben and Johnson, 2005; Man et al., 2002; Ragone, 1999), we expect
that achieving BD in these firms will likely require a response to Raisch and Birkinshaw’s (2008)
concern that structural, contextual and leadership conditions should be explored together to
understand how SMEs might balance the contradictory nature of exploration and exploitation. More
specifically, we expect that structural, contextual and leadership characteristics could be examined
together to investigate how innovation ambidexterity in SMEs is likely to be different from larger
firms. However, to date no such has emerged.
The objective of this paper then is to resolve this gap in current research into innovation
ambidexterity by studying how the role of structural, contextual and leadership conditions together
shape BD in SMEs. In doing so, this study offers several contributions. First, and to the best of our
knowledge, this is the first paper to attempt to understand how structural, contextual and leadership
conditions might create and maintain BD in SMEs. In turn, the study is a response to calls by Raisch
and Birkinshaw (2008) for multilevel and multifaceted research into organizational ambidexterity.
Second, and to the best of our knowledge, this is one of only a few papers to examine innovation
ambidexterity in SMEs. In turn, the study is a response to calls by Lubatkin et al. (2006) to extend and
validate research into the antecedents and consequences of BD in SMEs.
Theory and hypotheses
Explorative Innovation, Exploitative Innovation, and Balanced Dimension of Ambidexterity
A product innovation is typically classified by its closeness to one or a confluence (as appropriate)
of the following conditions: new or existing technologies; new or existing product features and
functions; or new and existing customers, market segments and routes to market (Abernathy and Clark
1985; Benner and Tushman 2003; Danneels 2002; He and Wong 2004; Jansen et al., 2006; Smith and
Tushman 2005). Therein, explorative product innovations meet new or emerging customer needs in new
or emerging markets with new technologies, features and functions materially different to existing
products; exploitative product innovations on the other hand meet the existing needs of customers in
existing markets with improvements in existing technologies, features and functions that incrementally
differentiate it beyond competitor products (Danneels 2002; Jansen et al., 2006; Morgan and Berthon
2008; Raisch and Birkinshaw 2008; Smith and Tushman 2005). Explorative innovations depend on new
knowledge and creative insights developed through acts of play, experimentation and discovery
whereas exploitative innovations build on existing knowledge through acts of refinement and gradual
improvement (He and Wong 2004; Levinthal and March 1993; March 1991; Smith and Tushman 2005).
Explorative and exploitative innovations are interdependent activities. March (1993) suggested that
maintaining the balance between explorative and exploitative innovations is crucial to firm survival.
As Levinthal and March (1993) argued that “The basic problem confronting an organization is to
engage in sufficient exploitation to ensure its current viability and, at the same time, to devote enough
energy to exploration to ensure its future viability” (p.105). Achieving a high level of BD of
ambidexterity could contribute to firm performance through more structural control of performance
risk (Cao et al., 2009) but failure to do so can result in the firm being mediocre at both types of
innovation, and suffer the performance consequences of this mediocrity in turn (March, 1991). More
specifically, firms are most likely to suffer the risk of obsolescence if a firm overemphasise the extent
of exploration innovation rather than exploitation innovation (Cao et al., 2009). By contrast, firms
may face the risk of losing to retain their long term success due to deliberately focusing on exploiting
existing products and services (Tushman and Anerson, 1986). This is because existing competencies
holding by firms will be obsolete and thus becoming inflexible that hinder the firm’s learning and
revitalization (Cao et al., 2009; Leonard-Barton, 1992). The need for an appropriate balance between
exploration and exploitation has been emphasised by Tushman and O’Reilly (1996). An ambidextrous
firm has the management and organizational capability to both compete in a mature market (where the
cost, efficiency, and exploitation innovation are crucial) and to expand new products and services in
an emerging market (where exploration innovation, speed, and flexibility are critical) (Tushman and
O’Reilly, 1996). More specially, firms need to be able to balance exploration and exploitation
innovations simultaneously to achieve better performance (He and Wong, 2004; Tushman and
O’Reilly, 1996). More recently, Cao et al. (2009) distinguished BD and combined dimension of
ambidexterity (CD) owing to ‘BD reduces the performance demoing effects of overengagement in
exploitation to the detriment of exploration or it vice versa, CD enhances firm performance through
the generation of a greater pool of complementary resources that may be leveraged across both’
(p.782). More specifically, they argued that BD is more beneficial to relatively resource-constrained
firms due to their relatively small size or scarce operating environments (Cao et al., 2009). This is
because a close balance of exploration and exploitation (i.e., BD) can enhance firms’ performance
such as SMEs through ease the risks arising from overcommitment to exploration or exploitation
innovation (Cao et al., 2009). Andriopoulos and Lewis (2009) found that firms can promote BD by
managing nested paradoxes of innovation across level and using corresponding tactics, i.e., integration
and differentiation. Integration emphasizes on interdependence between seeming opposites and enable
coordination that help actors share and connect divergent knowledge; differentiation focuses efforts on
either exploitative or exploratory qualities of the paradox (Andriopoulos and Lewis, 2009: 702). In
summary, prior researchers argued that a balance between exploration and exploitation is vital to the
concept of ambidexterity. O’Reilly and Tushman (2007) propose that achieving a high level of BD of
ambidexterity between exploitation and exploration activities requires both to be integrated around a
common set of values and a shared vision set out by top managers coupled with an over-arching
structure, complementary context and a proper governance process. Prior studies also argued that top
management need to produce supportive structures, structures, and context to facilitate the ideal type
of organizational ambidexterity, i.e., BD (Gibson and Birkinshaw, 2004; Smith and Tushman, 2005;
Tushman and O’Reilly, 1995). In turn, Raisch and Birkinshaw (2008) propose that the theory of
innovation ambidexterity requires extension to consider the simultaneous effects of structural,
contextual and leadership characteristics in achieving BD. We examine these in turn.
Structural Characteristics
Theory has so far associated explorative activity with organic structures and loosely-coupled systems
that support path-breaking behavior, and exploitative activity with mechanistic structures and
tightly-coupled systems that support path-refining behavior (He and Wong, 2004). Exploitative activity
then appears to thrive from mechanistic structures in which standardised rules, procedures and routines
exist to efficiently coordinate the actions of individuals; explorative activity instead appears to thrive on
simple organic structures with limited routines that offer only priorities, vision and boundary conditions
to inform the actions of individuals (Kang and Snell, 2009). The firm’s structure may then influence the
firm’s ability to pursue each type of innovation. Current prescriptions put forward two structural
conditions, formalisation and connectedness, as pertinent to this debate (for example, Jansen et al.,
2006).
Formalisation captures the extent to which the firm’s structure exhibits mechanistic properties, and is
defined as the degree to which rules, procedures, job instructions and communications are formalised,
written down or have records kept of (Jansen et al., 2006; Khandwalla, 1977). Standardised processes
and structures, detailed routines, and written rules tend to reinforce efficiency and the refinement and
improvement of existing activities by establishing ingrained patterns of behavior (Kang and Snell,
2009). Organizational learning under these conditions tends to focus on refining and improving existing
knowledge (Kang and Snell 2009), which is reflective of an act of exploitation (March, 1991) and
supportive of exploitative innovation (He and Wong, 2004).
Some firms can be successful at generating explorative innovations even when the firm exhibits a
level of formalisation, however. Informal mechanisms can cause firms to fail to gain the full benefits of
their explorative activity because their structural conditions do not effectively integrate these
innovations into the firm’s existing activities (Zahra and Nielsen, 2002). Also, in a test of large multiunit
firms, Jansen et al. (2006) found no evidence of a negative effect between formalisation and explorative
innovation. Kang and Snell (2009) also offer support for the view that mechanistic structures can
support the use of entrepreneurial capital. In sum, these studies suggest that formal mechanisms do not
necessarily prevent explorative innovation despite initial theoretical expectations to the contrary (March,
1991). One would expect formalisation to positively affect exploitative innovation but the state of
evidence is such that one could not predicted beyond a ‘no effect’ relationship with explorative
innovation.
Organic or informal structural conditions support simplified routines and are more loosely
connected to rules and traditional expectations about work and its outputs, which in turn should provide
individuals with opportunities for autonomy to experiment with the way they work and the way they
organize that work (Kang and Snell, 2009). Interdepartmental connectedness is one feature of such
organic or informal structural conditions. Connectedness increases opportunities for informal
knowledge sharing by exposing an individual to pockets of knowledge from across the firm
(Atuahene-Gima 2005; Jansen et al., 2006; Jaworski and Kohli 1993). Connectedness allows
individuals to combine unrelated matrices of knowledge in ways that may encourage explorative
learning (March, 1991) and explorative innovation in turn (He and Wong 2004; Ireland et al., 2003).
Connecting unrelated parts of the firm together through a structure that encourages informal
communication and knowledge sharing can enable the firm to consistently search and absorb novel
information; but it can also enable the integration of disparate pockets of knowledge to improve the
overall knowledge base (Kang and Snell, 2009). In turn, connectedness can help individuals to acquire
knowledge that refines their current understanding of existing technologies, product features and
functions, and fuel exploitative innovation as a consequence (Jansen et al., 2006). Connectedness can
enable a firm to synthesise, assimilate and apply exploitative and explorative knowledge to shape
explorative and exploitative innovations (Atuahene-Gima, 2005).
Some concern still exists over the informality attribute of connectedness, however. Jansen et al.
(2006) found connectedness to be positively associated with explorative and exploitative innovations in
large multiunit firms, but an SME lacks the amount of slack resources needed to cope with profuse
autonomous experimentation by individuals (confer, Lubatkin et al., 2006).
SMEs faced disadvantages in the markets due to the fact that they are constrained to access to capital,
acquire managerial expertise, lack of bargaining power with suppliers and buyers, lack of experiences
in curve effects (Forbes and Milliken, 1999; Holes and Dunstan, 1994; Pissarides, 1999; Rajan and
Zingales, 1995). For instance, SMEs are short of managerial expertise to manage know-how owning
by the entrepreneur or existing in the firms from other sources (Cooper et al., 1994). Kyriakopoulos
and Moorman (2004) and Raisch and Birkinshaw (2008) suggest that achieving organizational
ambidexterity may be contingent on the availability of sufficient resources particularly as operating
complexity grows. Thus:
Hypothesis 1: In SMEs, the structural characteristics - co-existence of formalization and connectedness - are positively related to the appearance of innovation ambidexterity (absolute difference of exploration and exploitation innovation i.e., BD).
Contextual Characteristics
Theory has recently specified that ambidexterity between contradictory activities may be found
when the managers of a business develop a supportive firm context that enables individuals to make
their own judgments on how best to manage conflicting task demands (Gibson and Birkinshaw, 2004).
In this instance, although ambidexterity is a characteristic of the firm, it manifests itself in the actions of
individuals across the firm (Gibson and Birkinshaw, 2004). Individuals must constantly choose how to
allocate their time and effort but the presence of fixed instructions and specific incentives can direct
individuals’ towards acts of exploitation or acts of exploration (Gibson and Birkinshaw 2004; Griffin et
al., 2007). In turn, by shaping a set of systems and processes that define a context that allows
exploration and exploitation to take place, individuals can be directed to act innovate ambidextrously
(Birkinshaw and Gibson 2004; Gibson and Birkinshaw 2004). We examine two contextual conditions:
social context and performance management (Denison et al., 1995; Gibson and Birkinshaw 2004).
Knowledge underpins acts of explorative and exploitative innovation (He and Wong 2004; March
1991; Morgan and Berthon 2008). But knowledge possesses a social component which renders acts of
exploration and exploitation subject to the social context of the firm (Kogut and Zander, 1992). Social
context contributes to knowledge processing activities, particularly by shaping a common
communication system within interpersonal social relationships (Verona, 1999), improving a firm’s
ability to acquire, assimilate, transform and leverage new knowledge over time (Jansen et al., 2005).
A firm’s social context enables ties to form among individuals from different functional
backgrounds (De Luca and Atuahene-Gima, 2007; Tsai and Ghoshal, 1998). As ties form among
individuals in the firm, new channels for knowledge flows emerge (Tsai and Ghoshal, 1998). These
channels enable individuals to gain access to the knowledge stocks of other individuals across the firm.
Accordingly, social context conditions underlie a firm’s ability to effectively combine knowledge that is
embedded across different functional areas of the firm (De Luca and Atuahene-Gima, 2007). Two
effects are then likely. First, the cross-pollination of knowledge across individuals at different points in
the firm’s structure enables individuals to creatively combine unrelated matrices of knowledge (Ireland
et al., 2003), which should in turn promote the entrepreneurial pursuit of exploratory innovation (De
Clercq et al., 2009). Social context conditions can potentially increase the conversion rate of
individuals’ explorative ideas into explorative innovations (Nonaka, 1994). Second, as social ties
among individuals across the firm increase, both the volume and quality of knowledge unlocked
increases as well (Tsai and Ghoshal, 1998), and so does the firm’s ability to take advantage of
exploitative knowledge to refine and improve existing products (Atuahene-Gima, 2005; Brown and
Duguid, 2001; De Luca and Atuahene-Gima, 2007; March 1991).
Social context conditions supplant the need for more formal, restrictive structural mechanisms to
generate knowledge flows within the firm. In turn, a firm context that encourages socialisation among
individuals can increase the firm’s ability to benefit from extensive and high-quality internal knowledge
exchange (Gibson and Birkinshaw, 2004), which ought to be amenable to both types of innovation
given their dependence on knowledge.
Adler et al. (1999) posited that the tension between refinement-led behavior and exploration-led
behavior can be reconciled by enabling individual employees to make their own choices on
systematising the creative process and by managers enabling workers to become more innovative and
flexible in their day-to-day tasks. Such task enrichment can be framed through performance
management (Gibson and Birkinshaw, 2004).
Performance management is concerned with stimulating individuals within a firm to deliver
high-quality outcomes, and stimulating a sense of responsibility for the achievement of those outcomes
(Birkinshaw and Gibson, 2004). Contemporary performance management acknowledges the growing
interdependence and uncertainty among work activities that result from complex business activities,
such as the firm's innovation initiatives (Griffin et al., 2007). A performance management regime must
therefore account for a broader range of individuals’ behaviors to understand a firm’s effectiveness at
innovation (Griffin et al., 2007; Birkinshaw and Gibson, 2004). A rounded performance management
system will seek to account for the behaviors that contribute to the firm's adaptability to new market
opportunities and its gradual improvement of products for short-term value creation (Birkinshaw and
Gibson, 2004; Crant, 2000; Parker et al., 2006; Pulakos et al., 2000); doing so ought to facilitate
explorative and exploitative innovations to appear (Gibson and Birkinshaw, 2004). Thus, performance
management should affect the extent to which both types of innovation occur within the firm.
Gibson and Birkinshaw (2004) and Ghoshal and Bartlett (1994) argue that a context promoting
discipline, stretch, support and trust is necessary to direct employees’ behaviors towards explorative and
exploitative tasks. Such a set of conditions involves setting clear standards of performance and behavior,
a system of open and rapid feedback and consistent reward and sanction to reinforce performance. Also,
individuals should be involved in the goal-setting process and encouraged to set increasingly ambitious
goals. Mechanisms should then be in place to allow individuals to access resources to pursue these goals
(Gibson and Birkinshaw, 2004). Importantly, such implementation of the performance management
system is designed to create a supportive environment that encourages individuals to take ownership in
delivering results, not instead to dictate specific types of action (Ghoshal and Bartlett, 1998; Gibson and
Birkinshaw, 2004).
The commitment of employees towards achieving managers’ desired outcomes depends on those
individuals seeing the strategic relevance of their actions (Fletcher and Williams, 1996). When
individuals are set and self-set challenging goals, the absence of strategic relevance would weaken the
likelihood that specific innovation outcomes occur. However, contextualising performance
management around strategically-relevant goals can inspire behavior and ownership towards those
goals (Mathieu and Zajac, 1990). In turn, managers that encourage employees to self-set challenging
goals and managers that issue creative challenges to employees instead of narrowly defined tasks would
be expected to see a consequent increase in the innovation behavior of employees (Birkinshaw and
Gibson, 2004; Gibson and Birkinshaw, 2004). Meta-analyses have also shown that the impact of such
performance management interventions is greater in small firms (Guzzo et al., 1985). This is because
accomplishing top managers’ goals in SMEs such as obtaining independence or developing new ideas
is primary motive to operate SME (Douglas and Shepherd. 2000). In addition, prior studies found that
goal orientation performance management approach was critical success factor for SMEs (Chawla et
al., 1997). More specifically, top managers tend to use a goal-based approach to manage their
employees to keep their competitiveness (Covin and Slein, 1989). Thus, a well-designed performance
management system should encourage explorative and exploitative innovations to appear:
Hypothesis 2: In SMEs, contextual characteristics -the co-existence of social context and performance management- are positively related to the appearance of innovation ambidexterity (i.e., absolute difference of exploration and exploitation innovation i.e., BD).
Leadership Characteristics
Recent theory developments have suggested that the actions of top managers can engender
explorative and exploitative innovations to emerge (Lubatkin et al., 2006; Mom et al., 2007). The
important role played by top managers’ actions is due to SMEs require top managers to secure slack
resources such as human capital and financial capital to pursue a balance of explorative and
exploitative innovations in SMEs (Wiklund and Shepherd, 2005). The human capital provides top
managers with the resources and ability necessary to explore and exploit new opportunities (Ajzen,
1991; Covin and Slevin, 1997; Wiklund and Shepherd, 2003). Also, in order to generate superior firm
performance, SMEs need to secure a great amount of financial resources that involved in both
explorative and exploitative innovations (Greene and Brown, 1997; Wiklund and Shepherd, 2005). It
is plausible that lack of financial capital would limit top managers’ opportunities to move towards
(Wiklund and Shepherd, 2003). For instance, an individual manager with risk-taking leadership gets
involved with large and risky resources commitments such as investing new products and services
with new technology that requires assuring access to financial resources to alleviate the opportunity of
risky projects being deadly (Wiklund and Shepherd, 2005). Prior studies found that slack financial
resources motivate firms’ innovation ambidexterity (Zahra, 1991) through encouraging a culture of
experimentation and protecting firms from the tentative results of the risky projects (Wiklund and
Shepherd, 2005).
Proponents of structural (Tushman and O’Reilly 1997, 1996) and contextual (Gibson and Birkinshaw,
2004) views also recognize the value of supportive leaders and flexible managers as drivers of
ambidexterity. This is because supportive leaders and flexible managers can package limited resources
(Cooper et al., 1994) to invest human capital resources in SMEs to promote the occurrence of
innovation ambidexterity. Mom et al. (2007) indicate that firms’ exploitation or exploration may be a
product of the exploitation or exploration properties their individual managers exhibit. This view is
consistent with the top management literature that firms over time become reflections of their top
managers (Hambrick and Mason, 1984). We examine two leadership conditions: risk-taking and
adaptation.
Kohli and Jaworski (1990) argue that if top managers demonstrate a tolerance to take risks and to
accept the cost of possible failures, employees will then be more likely to propose and introduce new
product-service offerings in response to emerging market opportunities. However, if top managers are
risk averse and intolerant of failures, employees are less likely to generate new solutions to market
opportunities and instead will focus on gradually improving or refining existing product-service
offerings (He and Wong, 2004). Research into entrepreneurial risk taking has concluded that those
managers displaying entrepreneurial risk tolerance frame risk in different ways to non-entrepreneurial
managers, obtaining a better understanding of the opportunities available to them in turn (Janney and
Dess, 2006; Mullins and Forlani, 2005).
Risk tolerance, and a bias to take risky action, is likely to lead top managers to favour higher return,
innovation-led opportunities (Lumpkin and Dess, 1996). However, managers with a tolerance for
risk-taking can also exhibit a tendency to protect current investments and will not seek to sink their
ventures with improper decision making (Janney and Dess, 2006; Mullins and Forlani, 2005). In turn,
risk tolerance is based on an understanding of the market-based consequences of choice and
incorporates both the cost and magnitude of a failed decision as well as the opportunity cost of a
decision (Dickson and Giglierano, 1986). This balance between entrepreneurial risk and calculative risk
management should favour both explorative and exploitative innovation activities as the tolerance of
risk is sufficient to overcome the inertia brought on by the uncertainty of explorative acts, and is
sufficient to maintain the need for exploitative refinement to optimise value from existing operations as
a means of maintaining the vitality of the venture (March, 1991). Also, risk-taking engages
Individuals within firms benefit from clear signals from top managers about the importance of
specific activities (Hambrick and Mason, 1984). Thus, the actions of top managers signal to employees
the sort of behavior considered desirable among top managers (Ireland et al., 2009; Jaworski and Kohli,
1993). Managers that exhibit adaptability to new and emerging conditions in the face of the demands
placed on them by the firm’s existing operations will signal to employees the need to ambidextrously
manage innovation opportunities. To this end, Burgelman (1991) posited the existence of induced and
autonomous processes in top managers’ decision-making.
Burgelman (2002) relates induced processes to exploitation and autonomous processes to
exploration. The induced process builds initiatives that are within the scope of the firm’s current
activities and build on existing knowledge and competencies. The autonomous process concerns
initiatives that emerge outside the scope of the firm’s current activities, providing opportunities for new
knowledge and competencies to emerge. Both processes determine the firm’s strategic direction and
method of operation. The adaptability of top managers to the two sets of activities ought to instil the
value of managing current product-services in relation to existing market needs while adapting to face
new challenges (Atuahene-Gima, 2005; Birkinshaw and Gibson, 2004; Jaworski and Kohli, 1993).
Adaptability depends on the knowledge managers and individuals possess on the firm’s markets,
technologies, product-services and customers (Jaworski and Kohli, 1993). Managers that repeatedly
stress to employees the importance of adapting to market trends, being sensitive to the activities of
competitors and the need to act now to meet customers’ future needs ought to shape action to increase
individuals’ learning and knowledge about these constituencies (Jaworski and Kohli, 1993), which in
turn ought to promote both explorative and exploitative innovations to appear (Morgan and Berthon,
2008). Thus:
Hypothesis 3: Among the top managers of SMEs, top managers’ risk-taking tolerance and adaptability are positively related to the appearance of innovation ambidexterity (i.e., absolute difference of exploration and exploitation innovation i.e., BD).
The Mediating Role of Innovation Ambidexterity on Business Performance
The argumentation contained in the hypotheses presented above implies that structural
characteristics, contextual conditions, leadership attributes, and innovation ambidexterity are linked
to SMEs performance. Specifically, the arguments suggest that the consequences of structural
characteristics, contextual characteristics, leadership characteristics on firm performance are due to
a balance of explorative and exploitative innovations, i.e., BD. Previous studies (e.g., Pinto et al.,
1993; Atuahene-Gima, 2003) suggest that internal organizational structural factors have indirect
effects on firm performance when firms engage in both explorative and exploitative innovations. In
addition, prior studies suggest that innovation ambidexterity has a mediation effect on the
relationship between contextual attributes and firm performance (e.g., Gibson and Birkinshaw,
2004; Birkinshaw and Gibson, 2004). More specifically, contextual attributes can create and
amplify internal tensions if they do not contribute to the simultaneous capacity for a balance of
explorative and exploitative innovation (Gibson and Birkinshaw, 2004: 215). Gibson and
Birkinshaw (2004) also argued that without combining other attributes such as formalization and
connectedness and consistent leadership behaviour such as risk-taking tolerance and adaptability,
firms have no capacity to develop innovation ambidexterity. Consequently, firms failed to adapt to
changing markets needs and thus generate superior performance. He and Wong (2004) and studies
since have reported positive firm performance returns to explorative and exploitative innovations.
However, on the basis that the hypotheses above hold true, we expect a balance of explorative
and exploitative innovations to affect SME performance by mediating the effects of structural
characteristics, contextual characteristics, and leadership characteristics. Thus,
Hypothesis 4: In SMEs, innovation ambidexterity mediates the relationship between structural characteristics (formalization and connectedness), contextual characteristics (social context and performance management), leadership characteristics (risk taking tolerance and adaptability) and firm performance.
Methodology
Sample and data collection
The sampling frame consisted of 1000 SMEs in Taiwan and China. These firms were randomly
selected from the Taiwan Economic Journal (TEJ) database. Of this sample, 243 firms (24.3%)
responded to a questionnaire survey. Manufacturing and service industries are represented by the firms
in the sample (see Table 1). Managing directors (MDs) and members of the top management team in
SMEs were selected as informants for data collection owing to their knowledge of the processes,
activities, pressures and overall identity of their businesses (Cohen and Musson, 2000). The second
informant was randomly chosen from the top management positions of the 1000 SMEs such as chief
product design manager (PDM). Together with the MD, we selected PDM as a second informant
because our sample consisted of manufacturing and services sector that compete on providing new
products and services. Thus, we rationale that PDMs will play a vital role in establishing precedence
and devising new strategies both in new and existing products and services in these SMEs. Use of key
informants in this study was due to the fact that both MDs and PDMs had complete knowledge and
information within SMEs and were able and willing to communicate (Kumar et al., 1993) about the
ways of their firms towards ambidexterity especially in SMEs context. Respondents’ participation in
this study was voluntary and all respondents were asked to complete a survey questionnaire. We used
an interrater reliability coefficient created by James et al. (1993) to inspect the intragroup reliability
(rwg) of responses. There is a sign of good agreement within a group if an rwg is greater than or equal to
0.70 (George and Bettenhausen, 1990). We computed the intragroup reliability by using James et
al.’s (1984, 1993) procedure: first, we calculated the standard deviation of items in each construct;
second, we then calculated standard deviation square value of items in each construct; third, we then
calculated the average value of items in each construct. All the three steps were conducted with syntax
in SPSS. The rwg for adaptability, risk-taking, formalization, connectedness, social context,
performance management, exploration innovation, exploitation innovation, business performance are
89, .79, .77, 78, .88, 82, 85, 88, 89 , respectively, all above .70 (George, 1990). These scores suggest a
acceptable level of agreement within the teams. These findings authorize the aggregation of individual
team member scores. In addition, we validated the data reliability through checking the
representativeness of the sample. First, respondents were divided into two subsamples (responses from
1-122 and responses from 123-243). These subsamples were compared based on the hypothesis that
those who responded late might be more similar to those who did not respond than those who
responded earlier (Armstrong and Overton, 1977). The subsamples were compared on dimensions
including firm age, profit and sales, and the number of employees. The results revealed no significant
difference (p<0.05). Also, we compared the responses of the first round and 30 of final phone call
round (Armstrong and Overton, 1977). No significant differences were found (p<0.01). The final
response rate of 24.3 percent was achieved after three rounds of attempts (two postal mailings and a
final round of phone calls). We were unable to compare the nonparticipant firms with those firms
responded to the survey on dimensions of firm age, profit and sales, and the number of employees
because we did not have information of MDs and PDMs of nonresponding or nonsurveyed firms. All
data were collected during an eight-month period from November 2008 to June 2009.
Measures
Independent variables
All items used to measure constructs were framed around seven-point Likert scales. Respondents
were asked to assess the extent to which their firm has undertaken a range of activities (1 = strongly
disagree; 7 = strongly agree). Measures for operationalizing the constructs were drawn from existing
studies. Leadership characteristics were measured with items adapted from Jaworski and Kohli (1993),
Covin and Slevin (1989), and Miller (1983). Respondents were asked questions regarding the
characteristics of top management with respect to their risk tolerance and adaptability. The
confirmatory factor analysis (CFA) result indicated that adaptability and risk-taking tolerance were
distinct from each other. The two-factor model (x2=36.26, d.f. = 4, p<0.1, RMSEA = 0.22, CFI=.82,
GFI=.92, NFI=.81) fit the data better than one factor model x2=98.07, d.f. = 5, p<0.1, RMSEA = 0.28,
CFI=.67, GFI=.86, NFI=.67). The results, therefore, supported the proposed dimensions of leadership
characteristics for innovation ambidexterity.
The measures for structural characteristics were adapted from Jaworski and Kohli (1993) and
Jansen et al. (2006). Respondents were asked questions regarding the relationship between
organizational structure in terms of the degree of formalisation and connectedness. The confirmatory
factor analysis CFA result indicated that formalization and connectedness were distinct from each
other. The two-factor model (x2=11.43, d.f. = 4, p<0.1, RMSEA = 0.09, CFI=.96, GFI=.98, NFI=.95)
fit the data better than one factor model x2=17.27, d.f. = 5, p<0.1, RMSEA = 0.11, CFI=.94, GFI=.97,
NFI=.94). The results, therefore, supported the proposed dimensions of structural characteristics for
innovation ambidexterity.
Contextual characteristics were adapted from Gibson and Birkinshaw (2004). The measure for
contextual characteristics relating to social context and performance management captured the extent
to which systems encourage employees’ contributions at their level/position in the firm (Gibson and
Birkinshaw, 2004). The CFA result indicated that social context and performance management were
distinct from each other. The two-factor model (social context as one factor, performance as one factor)
(x2=505.35, d.f. = 34, p<0.1, RMSEA = 0.19, CFI=.28, GFI=.78, NFI=.28) fit the data better than one
factor model x2=625.29, d.f. = 35, p<0.1, RMSEA = 0.23, CFI=.10, GFI=.71, NFI=.11). The results,
therefore, supported the proposed dimensions of contextual characteristics for innovation
ambidexterity.
To measure exploitative innovation, a four–item measure was adapted measure from Jansen et al.
(2006) and captured the extent to which the firm builds upon existing knowledge to pursue
incremental innovations that meet the needs of existing customers. Explorative innovation was
adapted from Jansen et al. (2006), He and Wong (2004), Birkinshaw et al. (1998), and Birkinshaw
(1997) and captured the extent to which the firm departs from existing knowledge and pursues radical
innovations for emerging customers or markets. Explorative innovation was adapted from Jansen et al.
(2006), He and Wong (2004), Birkinshaw et al. (1998), and Birkinshaw (1997) and captured the extent
to which the firm departs from existing knowledge and pursues radical innovations for emerging
customers or markets. The CFA result indicated that exploration innovation and exploitation
innovation were distinct from each other. The two-factor model (x2=1192.19, d.f. = 53, p<0.1,
RMSEA = 0.23, CFI=.55, GFI=.66, NFI=.54) fit the data better than one factor model (x2=1207.59, d.f.
= 54, p<0.1, RMSEA = 0.24, CFI=.54, GFI=.64, NFI=.53). The results, therefore, supported the
proposed dimensions of exploration innovation and exploitation innovation.
Dependent variables
Performance was measured with four items adapted from Gibson and Birkinshaw (2004). These
items reflect on performance over the last five years and ask respondents to indicate the degree to
which they agree with the extent to which the firm had met certain performance criteria. These were:
(1) Our company is achieving its full potential; (2) People at all levels are satisfied with the level of
business performance; (3) Our company does a good job of satisfying our customers; (4) This
company gives me the opportunity and encouragement to do the best work I am capable of. These four
items captured the extent to which SMEs used their exploitative and exploratory innovations to
achieve business and market potential (Gibson and Birkinshaw, 2004).
To test the reliability of our subjective measure, we obtained annual sales figures (computed over
the same time period as our subjective measure) from the FAME database to correlate with this
four-item subjective performance construct. A positive and significant association (r = 0.761, p<0.001)
was found. This finding provides evidence of the convergent validity of the self-report performance
measure.
Control variables
We controlled for firm age, firm size (number of employees) and industry sector. We controlled for
firm size and firm age because studies have found these to influence firm growth (Carroll and Hannan,
2000; He and Wong, 2004) and these can be related to structural characteristics such as organizational
routines that affect organizational failure (Tushman and Romanelli, 1985). The natural log of firm age
and firm size were used to compensate for skewness. Two broad industry sectors (manufacturing and
service) were used as a third control variable. Industry sector has been associated with firms’
motivation regarding adaptation to unpredictable resource conditions and performance fluctuations
(Lubatkin et al., 2006).
Analysis methods
Factor analysis was first used to identify the underlying dimensions of leadership characteristics
(risk tolerance and adaptability), structural characteristics (formalisation and interdepartmental
connectedness), contextual characteristics (social context and goal-based performance management),
exploitative innovation, explorative innovation and business performance. Factor analysis discovers
the separate dimensions of the construct structure and then determines the extent to which each
construct is explained by each dimension (Gorsuch 1997, 1983). Robust factor structures emerge
when the component items correlate with each other to demonstrate convergent validity (Blau 2001;
Law et al., 1998). We then applied hierarchical regression to test our hypotheses. Hierarchical
regression adds controls, independent variables and joint effect terms incrementally to gauge their
relative contributions to explaining variance in the dependent variable.
Results
Factor analysis and intercorrelations
All factor analysis results are contained in Table 2. Factor analyses were conducted using principal
component extraction with varimax rotation. All of the expected constructs were formed. However,
the items for social context and performance management created additional constructs. For social
context, the first factor is devotion-based social context (Cronbach alpha 0.707; 48.824 percent
variance explained) and the second factor is facts-based social context (Cronbach alpha 0.721; 30.371
percent variance explained). The factor analysis of performance management revealed goal-based
performance management (as expected) (Cronbach alpha 0.807; 48.664 percent variance explained)
and a second factor named effort-based performance management (Cronbach alpha 0.791; 31.511
percent variance explained). Although we used existing scales by Gibson and Birkinshaw (2004),
these scales did not form their anticipated study constructs either. As such, our factor analysis adds
another layer of purification to their measures.
The internal consistency (Cronbach alpha) of the constructs was in the range of 0.698 to 0.921
(Table 2) and is comparable to that obtained in previous studies using the same measures (Gibson and
Birkinshaw, 2004; He and Wong, 2004; Jansen et al., 2006). These values exceed the minimum
suggested by Nunnally (1978) and are taken as evidence of acceptable reliability. Table 3 shows the
intercorrelations, means and standard deviations for the variables used in the regression analyses.
[Please insert Table 2, 3 here]
We deployed several post hoc tests including the Harman single-factor test, confirmatory factor
analysis (CFA) and bivariate correlations to search for common method bias (Podsakoff et al., 2003).
Exploratory factor analysis combining items from the dependent and independent variables revealed
that several factors were extracted. The first factor accounted for 22.042 percent variance with an
eigenvalue of 3.792. This offers evidence that there is no single factor emerging from these variables
to suggest common method bias in the data. Moreover, all dependent and independent variables were
loaded onto a one-factor, a two-factor, and a three-factor CFA model to examine fit. If common
method variance exists among these variables, then the one-factor CFA model will fit the data well.
The results of a one-factor, a two-factor, and a three-factor of CFA disclosed that the fit of a one-factor
model as the poorest containing wholly unacceptable fit statistics (χ2=585.62, d.f.=54, p=0.00,
CFI=0.62, GFI=0.72, NNFI=0.53, RMSEA=0.19). Finally, in order to more directly exclude the
common method bias in our data, we examined bivariate correlations between subjective performance
from respondents and objective performance obtained from the FAME database. These were
significantly correlated (r = 0.761, p<0.001). Thus, we conclude that common method bias does not
affect the data.
Hierarchical regression results
Tables 4 and 5 present the results of hierarchical regression analyses for leadership, structural and
contextual characteristics onto exploitative innovation and explorative innovation respectively. The
baseline model 1 contains control variables. For hypothesis 1, there appears a significantly positive
relationship between formalisation and exploitative innovation (β = 0.177, p<0.001) (model 4, Table 4)
but no relationship is found between formalisation and explorative innovation (β = 0.060) (model 4,
Table 5). Thus, hypothesis 1(a) is supported and the ‘no effect’ predicted in hypothesis 1(b) is also
supported. Regarding hypothesis 2, no significant relationship is found between connectedness and
exploitative innovation or explorative innovation. Hypothesis 2, therefore, is not supported.
For hypothesis 3, a further non-significant relationship is found between social context and
exploitative innovation and explorative innovation. Thus, hypothesis 3 is not supported. For
hypothesis 4, a positive relationship is found between goal-based performance management and
exploitative innovation (β = 0.235, p<0.001) (model 4, Table 4) as well as between goal-based
performance management and explorative innovation (β = 0.262, p<0.001) (model 4, Table 5). Thus,
hypothesis 4 is fully supported. These results are embellished by the fact that the second factor
revealed in the factor analysis of performance management, effort-based performance management,
has a significant negative effect on both exploitative and explorative innovations. Accordingly, an
emphasis on job completion over the setting and completion of challenging creative goals will damage
the extent to which the firm can pursue innovation. This finding neatly complements the important of
goal-based performance management.
Regarding hypothesis 5, a significant positive relationship is found between risk-taking tolerance
and exploitative innovation (β = 0.122, p<0.05) (model 4, Table 4) and between risk-taking tolerance
and exploratory innovation (β = 0.366, p<0.001) (model 4, Table 5). Thus, hypothesis 5 receives full
support. For hypothesis 6, a further positive and significant relationship is found between adaptability
and exploitative innovation (β = 0.297, p<0.01) (model 4, Table 4) and between adaptability and
exploratory innovation (β = 0.197, p<0.01) (model 4, Table 5). Thus, hypothesis 6 receives full
support.
Finally, Table 6 contains the regression results for exploitative innovation, explorative innovation
and ambidexterity on business performance. The results reveal that a positive and significant
relationship exists between the interaction of exploitative innovation and explorative innovation
(innovation ambidexterity) on business performance (β = 0.968, p<0.01) (model 2, Table 6). Thus,
hypothesis 7 receives support. Moreover, the findings also show that separately, exploitative
innovation and explorative innovation have no effect on business performance, which lends further
support to hypothesis 7. We also carried out an additional robustness test by examining the effect of
exploitative innovation, explorative innovation and ambidexterity on the objective financial
performance measure we collected. The pattern of results is identical with innovation ambidexterity
again positively and significantly influencing performance (β = 0.408, p<0.01) and both exploitative
and explorative innovations again show no significant relationship with performance. In conclusion,
hypothesis 7 is fully supported by these results.
[Please insert Table 4, 5,6 here]
Discussion, contributions and implications
The intentions of this study were to first determine the relevance to SMEs of structural,
contextual and leadership conditions put forward by theory to shape explorative and exploitative
innovations, given that the research into conditions underpinning ambidexterity are so far almost
exclusively aimed at large, multiunit firms; and then second to test whether innovation ambidexterity
is in fact a relevant strategy to enable superior performance in SMEs.
Our findings contribute to the debate surrounding how to manage and organize for explorative
innovation and exploitative innovations. Our results contrast with Tushman and O’Reilly’s (1996)
arguments that ambidexterity can only be created in separate organizational units, a prescription
largely irrelevant to SMEs. Instead, our findings confirm beliefs that organizational context can create
conditions by which both types of innovation are supported as a precursor to ambidexterity (Gibson
and Birkinshaw, 2004). We extend prior studies therefore (for example, Mom et al., 2007) to show the
hitherto ignored role of top managers in achieving both innovations in SMEs, and add new value by
demonstrating how leaders can use their management approach in conjunction with performance
management to shape and reinforce these innovations. In SMEs, structural conditions may de-stabilize
efforts towards both innovations and investments in social context, although fruitful to large firms,
appear meaningless to SMEs. These findings offer several important contributions to scholars and
managers.
Scholarly Contributions
First, and most strikingly, we find that only 7 of the original 12 conditions put forward by theory
are statistically relevant to SMEs. This supports our concern that prescriptions put forward for large
firms do not sufficiently represent the conditions suitable for SMEs. This conclusion validates our
argument that ignoring SMEs in the conceptual and empirical debates on the theory of ambidexterity
represents a dangerous scholarly and managerial gap. To the best of our knowledge, this is the first
paper to consider the confluence of structural, contextual and leadership conditions in an attempt to
understand how SMEs can create and maintain both innovation types. In turn, the study directly
contributes to calls by Raisch and Birkinshaw (2008) for multi-level and multi-faceted research into
explorative and exploitative innovations to understand their antecedents.
Second, we discover that comprehensive organization-wide investments into innovation
ambidexterity will yield unique and difficult to imitate performance advantages for SMEs. Prior
studies have reported that firms can gain from exploitative and explorative innovation strategies
individually and in combination through ambidexterity (for example, He and Wong 2004; Morgan and
Strong, 2008). However, we find that for SMEs, performance advantages accrue only to those that
succeed in generating ambidexterity and individual efforts towards either innovation strategy alone
will result in mediocrity. In turn, these findings support March’s (1991) contention that excellence in
individual innovation strategies is no substitute for ambidexterity. In turn, the study directly
contributes to calls by Lubatkin et al. (2006) to extend and validate research not only the antecedents
