You are on page 1of 15

Technological Forecasting & Social Change 155 (2020) 119990

Contents lists available at ScienceDirect

Technological Forecasting & Social Change


journal homepage: www.elsevier.com/locate/techfore

How does the ambidexterity of technological learning routine affect firm T


innovation performance within industrial clusters? The moderating effects
of knowledge attributes

Jingjing Guoa,b, Bin Guoc, , Jianghua Zhoud, Xiaobo Wue
a
Institutes of Science and Development, Chinese Academy of Sciences, Beijing 100190, China
b
School of Public Policy and Management, University of Chinese Academy of Sciences, Beijing 100049, China
c
School of Management, Zhejiang University, Hangzhou 310058, China
d
Business School, Beijing Normal University, Beijing 100875, China
e
School of Management, Zhejiang University, Hangzhou 310058, China

A R T I C LE I N FO A B S T R A C T

Keywords: This paper focuses on the influence of the ambidexterity of technological learning routine on firm innovation
Ambidexterity performance within industrial clusters. Specifically, this study seeks to examine how the combined dimension
Technological learning routine and the balance dimension of the ambidexterity of technological learning routine affect innovation performance
Knowledge tacitness of cluster firms. In addition, this study examines the moderating effect of knowledge tacitness and knowledge
Knowledge heterogeneity
heterogeneity on the relationship between the ambidexterity of technological learning routine and innovation
Innovation performance
performance. Survey data from 217 industrial cluster firms in China reveals that the combined ambidexterity of
technological learning routine promotes firm innovation performance, whereas the effect of the balance ambi-
dexterity of technological learning routine on innovation performance is not supported. Besides, knowledge
tacitness moderates the relationship between the ambidexterity of technological learning routine and firm in-
novation performance within industrial clusters, while the moderating effects of knowledge heterogeneity on the
relationship between combined or balance ambidexterity of technological learning routine and innovation
performance are not supported. This paper brings greater theoretical insights into the notion of technological
learning routine by applying the ambidexterity perspective, and advances the understanding of the contextual
factors between the routine ambidexterity-innovation performance relationship through elucidating the mod-
erating role of knowledge attributes.

1. Introduction Feldman and Pentland, 2003; Feldman et al., 2016; Pentland et al.,
2011; Koumakhov and Daoud, 2017; Salvato and Rerup, 2011;
Organization routines are regular and repetitive patterns of inter- Winter, 2017). Despite the considerable conceptual literature, the re-
dependent organizational actions that are critical to the evolution of quirements for greater clarity about the domain and operationalization
firms and their ability to innovate and achieve sustainable competitive of the routine construct still prevail today (Guo et al., 2018), and re-
advantage (Aime et al., 2010; Becker, 2005; Eisenhardt and search on routines is hampered by a lack of operational content and
Martin, 2000; Hales and Tidd, 2009; Nelson and Winter, 1982; empirical measurement for validation of the theoretical construct
Parmigiani and Howard-Grenville, 2011). As a central concept to un- (Becker, 2005; Lewin et al., 2011; Pavitt, 2002). To heed the call to
derstand how organizations accomplish their work (Parmigiani and elaborate and augment the understanding of organizational routines by
Howard-Grenville, 2011; Salvato and Rerup, 2011), organization rou- studying routines empirically (Becker, 2005; Parmigiani and Howard-
tines have been attracting the increasing research attention during past Grenville, 2011; Pavitt, 2002), this study advances the conceptualiza-
decades, which emphasizes routines are the basic components of or- tion of technological learning routine based on the absorptive capacity
ganization behavior and repositories of organisational capabilities and perspective. The absorptive capacity lens has emphasized the view of
solutions to inefficiency problems within organisations (Aroles and routines as the basis for organization learning and demonstrated that
McLean, 2016; Becker et al., 2005; Nelson and Winter, 1982; routines exist through which firms acquire, assimilate, transform and


Corresponding author.
E-mail addresses: guojingjing@casipm.ac.cn (J. Guo), guob@zju.edu.cn (B. Guo), zhoujh@bnu.edu.cn (J. Zhou), xbwu@zju.edu.cn (X. Wu).

https://doi.org/10.1016/j.techfore.2020.119990
Received 4 March 2019; Received in revised form 9 January 2020; Accepted 2 March 2020
0040-1625/ © 2020 Elsevier Inc. All rights reserved.
J. Guo, et al. Technological Forecasting & Social Change 155 (2020) 119990

exploit knowledge (Lane and Lubatkin, 1998; Lane and Koka, 2006; theoretical lens of absorptive capacity. The absorptive capacity lens has
Lavie et al., 2010; Lewin et al., 2011; Zahra and George, 2002). Spe- postulated that the performance outcome of learning activities and
cifically, this study defines the technological learning routine as the process is influenced by the characteristics of knowledge an organiza-
recurrent and interactive technological learning behavioural pattern tion possesses and seeks (Cohen and Levinthal, 1990; Lane et al., 2006;
involving multiple actors within organizations by which organizations Solís-Molina et al., 2018). As such, knowledge tacitness and knowledge
acquire, maintain, reactivate and transform knowledge (Carlile and heterogeneity have drawn much attention (Cohen and Levinthal, 1990;
Rebentisch, 2003; Feldman and Pentland, 2003; García-Muiña and Lane et al., 2006; Parmigiani and Howard-Grenville, 2011;
González-Sánchez, 2017; Garud and Nayyar, 1994; Guo et al., 2018; Simonin, 1999; Szulanski, 1996). Building on the absorptive capacity
Zahra and George, 2002). perspective, we argue that knowledge attributes influence how easily
Intensity and variety have been identified as the two key char- knowledge transfers within and across firm boundaries, how much of it
acteristics of organization routines grounded in the evolutionary theory is retained, and the rate at which knowledge is reactivated and trans-
of economics (Becker, 2005; Pentland, 2003a, 2003b; Guo et al., 2018). formed (Argote et al., 2003), and thereby may guide a firm in different
This study builds up around these two dimensions and considers them ways to coordinate the configuration of the technological learning
as the important features of the technological learning routine. More routine for innovation performance (Choo and Bontis, 2002; van den
specifically, under turbulent and complex technological environment, Berg, 2013).
firms confront in essence the challenge of carrying these two non-sub- In sum, this study seeks to examine the influence of the ambi-
stitutable set of recurrent technological learning behavioral patterns. dexterity of technological learning routine on innovation performance
On the one hand, intense technological learning routine is helpful to and how knowledge attributes moderate the impact of the ambi-
ensure constant collective learning and advance the coordination of dexterity of technological learning routine. In doing so, this study
technological learning activities (Nelson and Winter, 1982; contributes to an enriched understanding of the relationship between
Szulanski, 1996). On the other hand, diverse technological learning the ambidexterity of technological learning routine and innovation
routine facilitates firms to accommodate dynamic environment and performance, and thus responds to the call to empirically validate the
promotes flexibility and change within organizations based on complex performance impact of technological learning routine (Becker, 2004;
and extensive repetitions (Aroles and McLean, 2016; Feldman, 2000; Hales and Tidd, 2009). More specifically, in line with Cao et al. (2009),
Hargadon and Sutton, 1997; Pentland and Rueter, 1994). As this study unpacks the ambidexterity of technological learning routine
O'Reilly and Tushman (2008) and Rothaermel and Alexandre (2009) into balance and combined dimensions. Following Carlile and
suggest, when a firm pursues both the intensity and the variety of the Rebentisch (2003) and Carlo et al. (2012), we also distinguish knowl-
technological learning routine simultaneously, we define that the firm edge tacitness and knowledge heterogeneity as the two key knowledge
exhibits ambidexterity in technological learning routine. attributes. In addition, we choose industrial clusters in China as our
However, prior studies have paid little attention to the ambi- research setting. First, the pervasive nature of knowledge transfer and
dexterity of a firm's technological learning routine, as well as the per- the significance of the ambidexterity of technological learning routine
formance effect of routine ambidexterity on innovation outcomes. On to the sustainable success of firms within industrial cluster provide a
the one hand, some recent work postulates that routines are ambidex- rich research setting to explore the performance effect of learning
trous by nature (Parmigiani and Howard-Grenville, 2011) and presents routine ambidexterity (Breschi and Malerba, 2001; Giuliani and
some caveats with regarding to potential boundary conditions to Bell, 2005; Guo et al., 2018). Second, industrial clusters in China de-
achieve the ambidexterity of routines (Adler et al., 1999; Collinson and velop very rapidly in terms of both number and scale (Guo and
Wilson, 2006; Howard-Grenville, 2005). Nonetheless, a more holistic Guo, 2013), and the success of clusters in China is featured by the
understanding of the ambidexterity of technological learning routine is crucial role of technological learning and knowledge spillover of main
still missing in this field. On the other hand, the existing research actors (Guo and Guo, 2011). These make industrial clusters in Chinese
concerns more with the impacts of routines on organizational perfor- context suitable settings to examine the moderation of knowledge at-
mance (Aime et al., 2010; Knott, 2003; Peng et al., 2008; Zollo et al., tributes on the performance effect of technological learning routine
2002), whereas the link between routine ambidexterity and perfor- ambidexterity. We test our hypotheses with a sample of 217 firms
mance outcomes of organizations remains rather implicit so far. For within industrial clusters in China.
instance, Peng et al. (2008) link routines to improvement and innova- The remainder of the paper is structured as follows. Section 2 pos-
tion capabilities and further to performance measures using a plant- tulates the hypotheses. Section 3 presents the research method and
level survey. Zollo et al. (2002) conduct a survey of biotech and Section 4 reports the empirical results. Discussions and contributions
pharmaceutical alliances and find that inter-organizational routines are presented in Section 5.
improve alliance performance.
In addition, among the organizational ambidexterity literature, 2. Theoretical framework and hypotheses
previous research displays contradictory empirical evidence regarding
the relationship between organizational ambidexterity and innovation 2.1. The ambidexterity of technological learning routine
performance in spite of the theoretical impact of organizational ambi-
dexterity on organizational performance and long-term survival Technological learning routines are regular and predictable beha-
(Cao et al., 2009; He and Wong, 2004; March, 1991; Tushman and vioral patterns conducted by multiple actors during the technological
O'Reilly, 1996). One stream of research demonstrates the positive learning process of firms (Feldman and Pentland, 2003; García-
performance effect of organizational ambidexterity (Derbyshire, 2014; Muiña and González-Sánchez, 2017; Nelson and Winter, 1982). Such
He and Wong, 2004; Lin et al., 2013; Lubatkin et al., 2006; behavioral regularities are critical to the evolution of firms and their
Tushman and O'Reilly, 1996), whereas the other stream of research competence to innovate (Hales and Tidd, 2009; Nelson and
argues that ambidexterity may lead to inferior performance because of Winter, 1982), and enable firms to properly confront the opportunities
the lack of internal consistency and munificent resource and challenges of a dynamic environment to compete and survive
(Kauppila, 2015; Rothaermel and Alexandre, 2009; Stettner and (García-Muiña and González-Sánchez, 2017; Lane et al., 2006). Based
Lavie, 2014; Van Looy et al., 2005; Venkatraman et al., 2007; on the absorptive capacity perspective, this study operationalizes
Wernerfelt and Montgomery, 1988). This study attempts to reconcile technological learning routine on the basis of four distinct types of
the mix findings through introducing the contingent role of knowledge knowledge processing activities, i.e., knowledge acquisition, knowledge
attributes in shaping the relationship between ambidexterity in tech- maintenance, knowledge reactivation and knowledge transformation
nological learning routine and innovation performance through the (Carlile and Rebentisch, 2003; Cohen and Levinthal, 1990; Garud and

2
J. Guo, et al. Technological Forecasting & Social Change 155 (2020) 119990

Nayyar, 1994; Zahra and George, 2002). On the one hand, the ab- efficiency through choice and implementation rooted in variance-de-
sorptive capacity literature highlights that routines can provide the creasing activities (March, 1991; Lavie et al., 2010; Smith and
microfoundations and key mechanisms for knowledge accumulation Tushman, 2005). In this vein, the intensity of the technological learning
and technological learning, and posits that routines exist during the routine captures the nature of exploitation in that it reflects variance
technological learning process of organizations (Lane and Koka, 2006; reduction as to the regular behavioral patterns during the technological
Lavie et al., 2010; Lewin et al., 2011; Zahra and George, 2002). On the learning process of firms, while the variety of the technological learning
other hand, these four types of learning activities have garnered much routine can be aligned to exploration since it indicates variance in-
focus among the absorptive capacity literature and formulate a com- crease in the technological learning behavioral patterns of firms.
prehensive framework of technological learning process within orga- Second, scholars have been increasingly recognized that in reality
nizations (Carlile, 2004; Carlile and Rebentisch, 2003; Guo et al., 2018; ambidexterity is not constrained to a certain organizational level, or a
Hargadon and Sutton, 1997). For example, Hargadon and Sutton (1997) specific function area (O'Cass et al., 2014), and it can be extended by
postulate that firms solve current problems by drawing on past ex- broadly associating exploration with organizational diversity and var-
periences through three fundamental processes as acquisition, retention iation, as well as relating exploitation to organizational focus and
and retrieval of knowledge and information from the past. variance reduction (Lavie et al., 2010). For instance, Medlin and
Carlile (2004) presents the knowledge transformation cycle across the Törnroos (2015) demonstrate how ambidexterity of exploration and
stages of knowledge storage, knowledge retrieval and knowledge exploitation is managed in a dynamic business network, and elaborate
transformation. exploration and exploitation as three processes shared between actors
With regard to the four types of knowledge processing activities, that include exploring the present business network to find partners,
knowledge acquisition denotes recognizing and obtaining knowledge exploiting the emerging network and exploring and finding a network-
from external sources that is crucial to firms’ operation (Zahra and technology fit. O'Cass et al. (2014) assert that firms become ambidex-
George, 2002); knowledge maintenance involves adding knowledge to trous when corporate-level exploratory and exploitative strategies have
the existing knowledge base to facilitate its future retrieval and use interactions with operational-level exploratory and exploitative cap-
(Carlile and Rebentisch, 2003; Garud and Nayyar, 1994); knowledge abilities among different functional areas. The above noted studies
reactivation refers to accessing and bringing into use all of the indis- argue that prior ambidexterity literature mostly focuses on the internal
pensible facets of maintained knowledge once the potential for ex- capabilities of firms, while ignoring the organization process between
ploiting maintained knowledge vectors is recognized (Garud and or within firms (Medlin and Törnroos, 2015; O'Cass et al., 2014). In line
Nayyar, 1994); and knowledge transformation pertains to integrating with their argument, we use the intensity-variety lens to illustrate the
retrieved knowledge with other knowledge vectors to exploit new characteristics of organization process and behavioral patterns of
business and product opportunities for current market needs technological learning. Heeding the call by Rothaermel and
(Carlile and Rebentisch, 2003; Fernhaber and Patel, 2012). Alexandre (2009), this study purposively extends organizational am-
Intensity and variety are viewed as key features of the technological bidexterity to describe a firm's ability to pursue the intensity and the
learning routine in the evolutionary theory of economics (Cohen et al., variety of technological learning routine concurrently.
1996; Pentland, 2003a, 2003b). First, the evolutionary economics re- In a dynamic environment, the ambidexterity of technological
search identifies recurrent interaction patterns by recurrence, which learning routine is critical to firm innovation performance in that it not
can be most simply measured as intensity (Becker, 2005; Cohen et al., only mitigates the risks from overcommitting either the intensity or the
1996; Pentland and Rueter, 1994; Betsch et al., 2001). Second, the variety of technological learning routine, but also develop a wide range
evolutionary economics research highlights that variety is a critical of capabilities to be leveraged between the intensity and the variety of
feature of routines and reflects variety in either the action sequences or technological learning routine (Cao et al., 2009; Wei et al., 2014).
the content of activity (Becker, 2005; Pentland, 2003a, 2003b). More There are two different dimensions to understand ambidexterity.
specifically, the intensity of technological learning routine is defined as For instance, He and Wong (2004) use both an interaction term and a
the extent to which continuing efforts to carry out the same interaction difference term to measure ambidexterity and examine the effects of
behavioral patterns actually take place during the technological ambidexterity on sale growth revenue. Cao et al. (2009) point out the
learning process within firms, whereas the variety of technological balance dimension of ambidexterity and the combined dimension of
learning routine pertains to the extent to which the firm's technological ambidexterity as two dimensions of ambidexterity and illustrate that
learning routine contains diverse interactive behavioral patterns concurrent high level of these two dimensions yield synergistic benefits
(Becker, 2005; Figueiredo, 2003). Both intensity and variety of tech- to improve firm performance. Wei et al. (2014) categorize ambi-
nological learning routine are critical to firms’ innovation and long-run dexterity into a relative exploratory dimension and an interactive di-
survival (Pentland, 2003a, 2003b). The intensity of technological mension and further investigate the different influence of these two
learning routine makes firms become efficient structures for collective dimensions on new product development. In this study, to capture a
learning (Anand et al., 2003; March and Simon 1958) and the refine- more holistic and integrative picture of the construct, we unpack the
ment of process and product improvements (van Wijk et al., 2012), ambidexterity of technological learning routine into two distinct di-
while diverse technological learning routines are indispensible for firms mensions termed as the balance dimension and the combined dimen-
to adapt a range of responses to fit current circumstances and enable sion, respectively. The balance dimension pertains to the efforts to
firms to accommodate dynamic technological environment balance the relative magnitude of the intensity and the variety of
(Hargadon and Sutton, 1997). technological learning routine to find the optimal structure, whereas
This study applies the ambidexterity perspective to enrich our un- the combined dimension focuses on the absolute magnitude of the in-
derstanding of technological learning routine based on the intensity- tensity and the variety of technological learning routine to increase
variety lens. Specifically, we associate the intensity of the technological their complementary effects (Cao et al., 2009; Wei et al., 2014). These
learning routine with exploitation and the variety of the technological two dimensions are conceptually different. Taking an example that is
learning routine with exploration. The reasons of using such dichotomy similarly used in Cao et al. (2009), Firm A has a score of 8 on the variety
of technological learning routine to capture the feature of ex- of technological learning routine and 4 on the intensity of technological
ploration–exploitation are as follows. First, it has been acknowledged learning routine, while Firm B has a score of 4 on both the intensity and
that exploration and exploitation are related to different organizational the variety of technological learning routine. For Firm A, the assess-
processes (Tushman and O'Reilly, 1996), for which exploration involves ment of balance ambidexterity is lower and that of combined ambi-
individuals and organizations in search, trial and error, and variation dexterity is higher compared to Firm B. As to Firm B, the assessment of
rooted in variance-increasing activities, whereas exploitation improves balance ambidexterity is higher and that of combined ambidexterity is

3
J. Guo, et al. Technological Forecasting & Social Change 155 (2020) 119990

lower in comparison to Firm A. In this sense, even if both firms carry contribution to innovation performance by appropriately controlling
out both the intensity and the variety of technological learning routine, potential performance risk. On the one hand, the overemphasis on the
the differences in the relative magnitude and the absolute magnitude of intensity of technological learning routine may lead to the risk of or-
the intensity and the variety make combined ambidexterity con- ganization inertia and path dependence (Pentland, 1995). The intensity
ceptually different from balance ambidexterity. of technological learning routine is critical to innovation performance
in that it ensures the continuous acquisition of external knowledge
2.2. The ambidexterity of technological learning routine and innovation (Leonard-Barton, 1992), enables firms to draw on a pool of prior sti-
performance mulus-response information (Hargadon and Sutton, 1997), and facil-
itates the combination of prior knowledge and current knowledge
2.2.1. The combined ambidexterity and innovation performance (Carlile, 2002). However, when a firm's magnitude of the intensity of
Intensity and variety of technological learning routine may function technological learning routine well exceeds that of the variety of
in complementary domains and have mutual beneficial effects technological learning routine, the above mentioned risk may emerge
(Cao et al., 2009; Gupta et al., 2006; Wei et al., 2014). A high degree of and limit the expansion of knowledge scope in some degree
the intensity of technological learning routine often equips firms with (Argyris and Schön, 1978), constrain the firm by competence traps
the necessary capability to conduct diverse technological learning (March, 1994), and make the firm take action slowly to renew knowl-
routines that promote new product development and innovation per- edge base and upgrade product design under the circumstances of dy-
formance. Specifically speaking, intense technological learning routine namic technological and market change (Cao et al., 2009; Wei et al.,
is more likely to lead to adequate and effective knowledge acquisition, 2014). As such, the adaptability to new opportunities may be decreased
assimilation and integration for innovation purpose (Figueiredo, 2003), and firms will be limited to the intrinsic performance of that technology
generate learning by doing effects (Arrow, 1962; Becker, 2005), and trajectory due to concentrating on the same knowledge process activ-
enable firms to understand deeply the functionality of repeatedly used ities (Christensen, 1997; Dosi, 1988; Gibson and Birkinshaw, 2004;
recurrent behavioral patterns during innovation process (Cao et al., Wei et al., 2014).
2009). Such deeper understanding ensures firms the capability to better In contrast, when a firm overemphasizes the variety of technological
reconfigure existing recurrent behavioral patterns and carry out diverse learning routine, the firm will be subject to the risk of failing to accu-
technological learning routine to develop new products and promote mulate adequate technological capabilities from its costly knowledge
organizational innovation (Cao et al., 2009; Cohen and Levinthal, 1990; processing activities (Cao et al., 2009; Lee et al., 2013; Levinthal and
Wei et al., 2014; Zahra and George 2002). Besides, the intensity of March, 1993). Although the variety of technological learning routine is
technological learning routine helps firms to economize on bounded helpful to obtain diverse knowledge, enrich organization memory
cognitive resources by eliminating the need to deliberate over proper (Carlile and Rebentisch, 2003; Walsh and Ungson, 1991), and provide
action (Anand et al., 2003; Cyert and March, 1963) and dealing with solutions timely for new unstructured problems when technological
repetitive decisions through semi-conscious mechanisms environment is dynamic (Glazer, 1991), excessively diverse technolo-
(Figueiredo, 2003; Nelson and Winter, 1982). In this sense, cognitive gical learning routines possibly lead to continuous drain on organiza-
resources within organizations can be released in some degree, and tional resources and lower efficiency as the acquisition, storage and
then be used to concentrate on more kinds of knowledge processing reactivation of external knowledge and its integration with existing
activities (March and Simon, 1958). knowledge stock in a diverse manner are often costly and time con-
Diverse technological learning routine enables firms to build up suming (Lee et al., 2013). Moreover, it may be a constraining influence
technological capability (Dosi, 1988), which is helpful to improve the for organizations to make decisions that have a higher mutual fit and
usefulness of the intensity of technological learning routine. When a frustrate coordination between different parts of the firm
firm pursues a wide variety of recurrent behavioral patterns during (Becker, 2005; Simon, 1947). And also it cannot ensure enough
knowledge acquisition, maintenance, reactivation and transformation knowledge assimilation and integration since a stable baseline against
process, it obtains greater understanding of different technology do- which to assess feedback, compare and draw implication cannot be
mains involved in the process of new product development and the established (Becker, 2005; Tyre and Orlikowski, 1996).
principles and rules entailed through the whole knowledge transfor- Taken together, the conceptualization of the combined ambi-
mation process (Becker et al., 2005; Huber, 1996). In this regard, the dexterity of technological learning routines may also suggest that the
firm's capability to conduct some kind of technological learning routine intensity and the variety could compete each other sometimes. Building
intensely will be enhanced because it is able to select more appropriate on this logic, we have elaborated that failing to achieve a high level of
technological learning routine for innovation purpose and perform the balanced ambidexterity of technological learning routines can result in
relevant activities smoothly based on the accumulated technological either the risk of organization inertia due to path dependence or the risk
capability (Rumelt, 1995). of failing to accumulate adequate technological capabilities. In contrast,
In sum, the central idea in the conceptualization of the combined we expect that firms keeping a balance between the intensity and the
ambidexterity of technological learning routines is that the intensity variety of technological learning routine are able to achieve superior
and the variety of technological learning routines are not necessarily innovation performance due to effectively avoiding the above noted
competing each other (Cao et al., 2009; Gupta et al., 2006), and firms risks than firms that overemphasize one at the expense of the other
may shift between the intensity and the variety (Brown and (Levinthal and March, 1993; Tushman and O'Reilly, 1996).
Eisenhardt, 1997). In this vein, we have postulated that the intensity Hypothesis 2. The balance ambidexterity of technological learning routines
and the variety of technological learning routines can complement each is positively related to innovation performance of cluster firm.
other and further improve innovation performance since organizational
capabilities can be leveraged across both behavioral patterns
(Cao et al., 2009). 2.3. The moderating effect of knowledge tacitness
Hypothesis 1. The combined ambidexterity of technological learning
routines is positively related to innovation performance of cluster firm. Knowledge tacitness refers to the extent to which the unspecifiable
process of knowing-in-action that is unlikely to articulate and com-
municate to others (Polanyi, 1967). In a tacit (as opposed to explicit)
2.2.2. The balance ambidexterity and innovation performance knowledge context, it is difficult to understand the complicated rela-
We expect that matching the relative magnitude of the intensity and tions of the elements comprising technological knowledge, creates
the variety of technological learning routine more closely makes barriers to knowledge acquisition, reactivation and transformation

4
J. Guo, et al. Technological Forecasting & Social Change 155 (2020) 119990

(Kogut and Zander, 1992; Nonaka, 1994; Simonin, 1999; Urgal et al., In contrast, when much of the knowledge is tacit, achieving an
2013), and permits indirect appropriation (Grant, 1996). Thus, in a appropriate balance between the intensity and the variety of techno-
context with high levels of knowledge tacitness, it is not only difficult logical learning routines to mitigate either the risk of organization in-
for firms to modify and adjust the extent and the range of recurrent ertia and path dependence or the risk of failure to accumulate adequate
behavioral patterns during technological learning process, but also hard technological capabilities turns to be very crucial. In such circum-
for firms to get access to external knowledge and resources. stances, when firms overemphasize either the intensity or the variety of
The combined ambidexterity of technological learning routines in- technological learning routine, it will be difficult for them to adjust the
dicates the necessity and efforts to conduct high levels of the absolute relative magnitude of the intensity and the variety of technological
magnitude of the intensity and the variety of technological learning learning routine in a timely manner because tacit knowledge must be
routine (Cao et al., 2009; Wei et al., 2014). The manner in which firms transferred and acquired through experience in a costly and slow pro-
implement the combined ambidexterity of technological learning rou- cess (Grant, 2002) and requires a greater degree of intimacy and per-
tines is critical to capability building and innovation, and entails manence to be transformed (Hedlund, 1994). Besides, when the im-
varying organizing logics, coordination rules and collective pattern of balance risks happen, it is less likely for firms to get to know where the
interaction (Figueiredo, 2003; Helfat and Peteraf, 2003; Wei et al., necessary resources are and how to get in touch with the resources
2014). In a context featured with high levels of knowledge tacitness, quickly due to the appropriability of value from tacit knowledge
these organizing logics and coordination rules are hard to be replicated (Grant, 1996). Therefore, we propose that the balance ambidexterity of
by competitors within industrial clusters as they are highly context- technological learning routines is more important for firms in a context
dependent (McIver et al., 2013) and ambiguous to competitors featured with high levels of knowledge tacitness.
(Reed and DeFillippi, 1990). Consequently, competitors’ efforts to re- Hypothesis 4. The relationship between balance ambidexterity of
plicate the way the combined ambidexterity of technological learning technological learning routines and innovation performance is moderated
routines work and how it is organized will be frustrated, and the firm's by knowledge tacitness. The higher the level of knowledge tacitness, the
performance advantage is then prolonged (Bierly et al., 2009; greater the positive effect of balance ambidexterity on innovation
McEvily and Chakravarthy, 2002). With sustained market performance, performance.
firms will be more market-oriented; that is, they emphasize monitoring
market conditions, keeping track of market trends, and analyzing cus-
2.4. The moderating effect of knowledge heterogeneity
tomer needs and competitor actions (Wang et al., 2013). Hence they
will invest more on innovation based on their expectations on market,
Knowledge heterogeneity reflects how firms differ in the knowl-
strive to present novel ideas and better solutions to respond to market
edge, know-how, and expertise with respect to their current product
change, and embrace and enhance innovations that are tailored to
portfolios, equipments and manufacturing processes within industrial
customer needs and market change (Amir and Ahad, 2018;
clusters (Rodan and Galunic, 2004; Tsai et al., 2014). In a hetero-
Dekoulou et al., 2017; Liu and Atuahene-Gima, 2018; Pekovic et al.,
geneous knowledge context, it may lead to inaccurate communication
2016; Song et al., 2015). In addition, knowledge tacitness not only
and greater processing complexity (Pelled et al., 1999; Pitt and
protects the accumulated knowledge and capability from imitation, but
McVaugh, 2008; Tsai, 2018). As heterogeneity in the amount and type
also fosters creativity due to less formalisation and increased R&D au-
of domain-specific knowledge increases, it will require more effort and
tonomy (Hedlund, 1994). At high levels of knowledge tacitness, the
energy to acquire, store, reactivate and transform knowledge
accumulated knowledge and capability arising from combining the in-
(Carlile, 2004). Hence, the feasibility to modify and adjust the extent
tensity and the variety of technological learning routine are more likely
and the range of different dimensions of technological learning routine
to stimulate unique solutions and discovery to market and technological
will be greatly decreased.
problems (Berman et al., 2002; Kim et al., 2013), and contribute largely
In a heterogeneous knowledge context, a firm is more likely to ap-
to the speeding-up of innovation process (Blümm, 2002).
propriate and sustain innovation performance advantage from the
Hypothesis 3. The relationship between combined ambidexterity of
pursuit of combined ambidexterity of technological learning routines,
technological learning routines and innovation performance is moderated
because in this case those organizing logics and coordination rules
by knowledge tacitness. The higher the level of knowledge tacitness, the
taken by this firm may be incompatible with other firms within in-
greater the positive effect of combined ambidexterity on innovation
dustrial clusters even if they are spillovered. Besides, at high levels of
performance.
knowledge heterogeneity, the accumulated knowledge and capability
Firms are likely to be subject to the risks due to the imbalance be- arising from combined ambidexterity of technological learning routines
tween the intensity and the variety of technological learning routine. are more likely to create novel linkages (Carlo et al., 2012), stimulate
However, in a context of low knowledge tacitness, failing to maintain creativity and multiple idea generation (Amabile, 1996; Smith et al.,
an appropriate balance between the intensity and the variety of tech- 2005), and further promote innovation performance.
nological learning routine may not be necessary to affect innovation Hypothesis 5a. The relationship between combined ambidexterity of
performance negatively. This is because explicit knowledge is easily technological learning routines and innovation performance is moderated
transferred, replicated and integrated due to its public and non-riv- by knowledge heterogeneity. The higher the level of knowledge heterogeneity,
alrous nature (Heiman and Nickerson, 2004; Langlois and the greater the positive effect of combined ambidexterity on innovation
Robertson, 1996). Thus, it is more feasible for firms to adjust the re- performance.
lative magnitude of the intensity and the variety of technological
On the other hand, firms may be more inclined to pursue intense
learning routine to achieve a close match when they realize the over-
technological learning routines in a heterogeneous knowledge context,
emphasis on either the intensity or the variety of technological learning
because such practices can effectively save cognitive resources
routine. As a result, the imbalance risks will be mitigated in time and
(Gersick and Hackman, 1990) and further reduce the cost of learning.
effectively when they emerge. Moreover, considering that codified
Besides, firms with heterogeneous knowledge domains have more
knowledge cannot ensure the appropriability of value (van den
chance to target differentiated niche markets in the given industry and
Berg, 2013), firms are easier to obtain an understanding of how to
they often have an enriched understanding of the technological
access necessary external resources and further acquire these resources
learning routines they need (Guo et al., 2018). In this sense, these firms
to provide a better buffer to mitigate risks arising from the imbalance
may benefit more from undertaking appropriate routines intensely ra-
between the intensity and the variety of technological learning routine
ther than broadly. As a consequence, the combined ambidexterity of
(Cao et al., 2009).
technological learning routines can not satisfy what firms truly need

5
J. Guo, et al. Technological Forecasting & Social Change 155 (2020) 119990

when knowledge heterogeneity is high and increased combined ambi- 2007.


dexterity may lead to decreased innovation performance due to misfit The questionnaire was designed in three phases. At the initial stage,
problem. Therefore, we propose the following competing hypothesis: the measurement scales were constructed based on previous studies
Hypothesis 5b. (competing hypothesis): The relationship between gathered from the review of literature. In the second phase, to ensure
combined ambidexterity of technological learning routines and innovation the appropriateness of the developed questionnaire, we pretested the
performance is moderated by knowledge heterogeneity. The higher the level questionnaire with five academics in the field of technological learning
of knowledge heterogeneity, the lower the positive effect of combined and innovation and five senior managers of R&D department in cluster
ambidexterity on innovation performance. firms. They were asked to give suggestions on the structure and content
of scales and provide clarity to the measurement. We improved or
A critical potential benefit of balance ambidexterity of technological
eliminated the items on the basis of their comments and assessment. In
learning routines comes from the alleviation of either the risk of or-
the third phase, we further pretested the revised questionnaire by col-
ganization inertia and path dependence due to overemphasis on some
lecting 15 questionnaires for a further feedback. Then we conducted a
specific technological learning routine without a commensurate com-
formal survey by using the finalized questionnaire.
mitment to the variety of technological learning routine, or the risk of
We used a manufacturing firm database developed at Zhejiang
the failure to build up adequate organizational capabilities as a result of
University which contains manufacturing firms’ information in the YRD
overemphasis on a wide range of technological learning routine without
region. After selecting firm samples according to the location and the
a corresponding level of focusing on some technological learning rou-
industry of typical clusters, we considered the managers of R&D de-
tine intensely (Cao et al., 2009). Such overemphasis risks will be less
partment in this database as appropriate respondents, as they are most
threatening in a more homogeneous knowledge context. It is easier for
likely the core drivers and executors of technological learning routine
them to adjust the relative magnitude of the intensity and the variety of
within organizations and familiar with this issue (Lee and Wong, 2015;
technological learning routine to mitigate the risks in a timely and ef-
Zhang et al., 2006). In addition, following De Clercq et al. (2013),
fective manner when firms realize overemphasis problem, because
Oerlemans et al. (2013) and Cruz-González et al. (2015), to minimize
knowledge homogeneity releases the difficulty of adjustment by facil-
concerns on endogeneity and reverse causality, it was explicitly de-
itating firms to trust and understand with each other, reducing search
monstrated in the questionnaire the specific year to which each part of
costs, and promoting knowledge acquisition and further knowledge
questions was referred. Data collection was pursued according to the
processing activities (Brown and Duguid, 1991; Corsaro et al., 2012;
following procedure. First, the electronic questionnaire was sent to the
Powell et al., 1996). For example, if firms intend to increase the in-
respondents by mail together with a covering letter explaining the
tensity of technological learning routine to achieve the balance,
purpose of the survey. Second, two months later, a follow-up procedure
common knowledge and prior experience arising from knowledge
was carried out with the same questionnaire and covering letter for the
homogeneity will facilitate them to select appropriate external sources
non-respondents (Abebe and Angriawan, 2014).
to obtain knowledge. Meanwhile, since external knowledge is more
A total of 482 questionnaires were sent out and 253 complete re-
likely to be compatible with internal knowledge in this case, the need
sponses were received, with a response rate of 52.49%. After matching
and effort to examine the use of the acquired knowledge, make storage
both the location and the dominant industry of the samples with the
decisions, and explore ways to integrate reactivated knowledge with
relevant industrial clusters, 36 samples were excluded and a total of
existing knowledge will also be decreased (Larraneta et al., 2012).
217 valid responses remained. Table 1 shows the demographics for the
Conversely, in a more heterogeneous knowledge context, it is less
217 respondent companies. Approximately, the main industries re-
likely for firms to modify the relative magnitude of the intensity and the
presented were textile (32.7%), communication, computer, other elec-
variety of technological learning routine to mitigate the risks in time
tronic equipment (14.3%), electric equipment and machinery (7.8%)
once they realize the imbalance between these two dimensions, because
and general purpose machinery (7.4%). The sample comprised 71%
knowledge heterogeneity may result in a lack of mutual understanding
private-owned firms and 9.7% sino-foreign joint ventures. Of them,
across working areas, uncertainty and ambiguity in the variety of
27.2% of the respondent firms employed less than 150 employees,
knowledge, and intra-group communication and coordination conflicts
25.3% between 151–350, 39.6% between 351 and 500 and 7.8% more
(Pitt and McVaugh, 2008; Schmickl and Kieser, 2008; Tsai et al., 2014).
than 500 employees. The average age of the samples was 18.23 years
As a result, the adjustment between different dimensions of technolo-
(s.d. = 11.639) and the average firm size was 310 employees
gical learning routine requires multiple interdependent knowledge
(s.d. = 181.278).
vectors to facilitate decision-making process, and relevant processes
To examine potential response bias, the early 30% samples were
and mechanisms that are helpful to create shared meaning and re-
compared with the late 30% samples (Mohr and Spekman, 1994). A t-
concile discrepancies in meaning (Dougherty, 1992; McIver et al., 2013;
test showed no significant differences between the two groups based on
Nonaka and Takeuchi, 1995).
firm age and firm size (tvalue = 0.488 and 0.014, respectively), sug-
Hypothesis 6. The relationship between balance ambidexterity of
gesting response bias was unlikely to be an issue in this study.
technological learning routines and innovation performance is moderated
by knowledge heterogeneity. The higher the level of knowledge heterogeneity,
3.2. Measures
the greater the positive effect of balance ambidexterity on innovation
performance.
All measures originally designed in English were translated into
The conceptual model is displayed here.(See Fig. 1.) Chinese according to the back-translation procedure to ensure the
consistent accuracy (Brislin, 1970). Unless otherwise noted, all items
3. Research method were validated on a seven-point Likert scale anchored by 1 = “strongly
disagree” and 7= “strongly agree”.
3.1. Sample and data collection Innovation performance. We measured innovation performance with
a five-item scale based on previous works of Rothaermel and
The empirical work was based on a survey that mainly examined Alexandre (2009), Zhang and Batrol (2010) and Zheng et al. (2013).
firms within industrial clusters in Yangtze River Delta (YRD) region in Respondents were asked to rate their firm's performance over the past
China. The YRD is one of the most developed industrial regions in China three year in terms of the following items:
(Zheng et al., 2011). It accounted for 20.54% of the national gross
domestic product in 2016 (National Bureau of Statistics of China, 2017) 1 sales revenue of new products in our firm keeps increasing;
and owned more than one third of all industrial clusters in China in 2 new product profit of our firm keeps increasing;

6
J. Guo, et al. Technological Forecasting & Social Change 155 (2020) 119990

Fig. 1. Research model.

Table 1 3 the number of patent application in our firm keeps increasing;


Sample demographics. 4 the new product profit has met our goals; and
Sample characteristics Frequency Percentage
5 we develop new products more quickly than competitors in the in-
dustry.
Ownership
Private-owned 154 71% Knowledge tacitness. Four measures were used to capture knowledge
Sino-foreign joint venture 21 9.7%
State-owned 19 8.8%
tacitness adapted from Bierly et al. (2009), Guo and Yin (2008),
Collective-owned 4 1.8% Simonin (1999) and Zander and Kogut (1995), including:
Others 19 8.8%
Industry 1 knowledge is easily to be acquired in our firm by instructions,
Textile 71 32.7%
blueprints, formulas, etc.;
Communication, computer, other electronic equipment 31 14.3%
Electric equipment and machinery 17 7.8% 2 knowledge is easily to be acquired in our firm by interacting with
General purpose machinery 16 7.4% experts who have rich experience;
Pharmaceutical 14 6.5% 3 our R&D staffs rely on experience during the process of product
Transport equipment 11 5.1% development in a high degree; and
Artwork and other manufacture 11 5.1%
Textile wearing apparel, footwear and caps 6 2.8%
4 extensive documentation and manuals describing critical parts of
Plastic products 5 2.3% the manufacturing process exist in our firm.
Processing of food from agricultural products 3 1.4%
Others 32 14.7% Knowledge heterogeneity. Drawing on the work of Tsai et al. (2014)
Firm age
and Guo et al. (2018), knowledge heterogeneity was assessed with a
0–10 years 54 24.9%
11–15 years 51 23.5% three-item scale that captures the extent to which the focal firm is
16–20 years 49 22.6% different from other firms in the cluster in terms of product portfolio,
21–30 years 44 20.3% manufacturing equipment, and manufacturing processes.
More than 30 years 19 8.8% Technological learning routine. Parallel instruments were developed
Firm size(number of employees)
0–150(>0 and≤150) 59 27.2%
to measure technological learning routine during four knowledge pro-
151–350 55 25.3% cessing stages. Technological learning routine during the knowledge
351–500 86 39.6% acquisition stage was measured by a fifteen-item scale that was taken
More than 500 17 7.8% from previous work of Guo and Guo (2011), Figueiredo (2003) and
Sales (million RMB yuan)
Harabi (1997). The instrument captures the degree to which the firm
<5 62 28.6%
5–30 82 37.8% could acquire external technological knowledge through hiring en-
30–100 42 19.4% gineers with rich experience and newly graduated engineers, technical
More than 100 31 14.3% publications, patent disclosure, open technical meetings, industrial
trade shows, training, on-site visit, education support, collaborative
development, and interactions with various partners including users,
suppliers, universities and research institutes, etc. Drawing on the work

7
J. Guo, et al. Technological Forecasting & Social Change 155 (2020) 119990

of Garud and Nayyar (1994), Technological learning routine during the Balance ambidexterity of technological learning routine. Balance am-
knowledge maintenance stage was captured by an eight-item measure bidexterity of technological learning routine refers to the relative
which reflects a series of knowledge storage activity sets within firms. magnitude of the intensity and the variety of technological learning
Specific items include: “gather information about market and tech- routine. Following previous treatments adapted by He and
nology dynamics”, “make choice decision involving business and re- Wong (2004) and Cao et al. (2009), balance ambidexterity of techno-
search departments”, “use rich media for communication”, “retain staff logical learning routine was measured by subtracting the absolute dif-
with key knowledge”, “store and shelve documents about product de- ference score between the intensity and the variety from 5. In so doing,
velopment”, “catalog shelved technological knowledge”, and “reward a higher value indicated greater balance ambidexterity of technological
maintenance”. Technological learning routine during the knowledge learning routine.
reactivation stage was measured by a six-item measure which was also Combined ambidexterity of technological learning routine. In line with
adapted from Garud and Nayyar (1994). This measure examines how He and Wong (2004), Cao et al. (2009) and Wei et al. (2014), combined
important it is for firms to seek opportunities to utilize maintained ambidexterity of technological learning routine was operationalized by
knowledge through encouraging engineers to move around among computing the multiplicative interaction between the intensity and the
various product teams, organizing symposia to facilitate knowledge variety of technological learning routine. We mean-centered the in-
sharing, and publicizing topics under research within organizations, tensity and the variety of technological learning routine scales before
retrieve and evaluate maintained knowledge through reviewing the obtaining their product to mitigate the potential of multicollinearity.
catalog of maintained knowledge and assessing the reliability and va- Control variables. Five variables (firm size, firm age, industry type,
lidity of shelved technologies, and provide incentives for knowledge ownership and network centrality) were controlled to eliminate the
reactivation during the process of technology and product develop- effects they might have had on innovation performance. Specifically,
ment. Technological learning routine during the knowledge transfor- firm size indicates that a firm possesses more resource and capability to
mation stage was measured by five items modified from Carlile and improve innovation performance (Acs and Audretsch, 1987), and it was
Rebentisch's (2003) comprehensive description of knowledge transfor- controlled and estimated by the log values of the number of employees.
mation stage within the knowledge transformation cycle. This measure Second, firm age, which may affect a firm's incentive to devote into
reflects the importance of building teams to reuse maintained knowl- innovation (Sørensen and Stuart, 2000), was also treated as a control
edge and seeking opportunities for knowledge integration, and facil- variable. It was measured in years since a firm was established. Third,
itating knowledge flow across specialization and within specialization following Wang et al. (2017), this study controlled for industry type by
based on meeting and discussion, technical experts who can promote introducing industry dummies, in which 0 indicated labor-intensive
informal communication among different functional departments, and industries and 1 indicated knowledge-intensive industries. Fourth, as
the adoption of simulation, product models and drawings during the firms with state ownership may receive more government support and
research and development process. leverage more resources for innovation (Piotroski and Zhang, 2014),
The respondents answered the questions on the basis of a 7-point this study incorporated ownership and measured it by a dummy vari-
Likert scale (1 = not at all important; 4 = moderately important; able, in which 0 indicated the presence of state ownership and 1 in-
7 = very important) for each of the above technological learning dicated the absence of state ownership (Chen et al., 2018; Wang et al.,
routines. The basic question was: For each of the following activity, 2017). Finally, given the importance of network centrality to the ac-
please indicate the importance you attach to it during the process of quisition and assimilation of external knowledge and firm innovation
technology and product development. The raw data of these assess- (Dong and Yang, 2016; Tsai et al., 2019), this study controlled for
ments are represented as IMPj, where j represents the behavioral pat- network centrality with a four-item scale based on Wang (2007) and
tern during each knowledge processing stage and IMP represents the Giuliani (2007). These items are as follows: (1) local firms often seek for
degree of importance of specific behavioral pattern to a firm. Besides, help from our firm when they need technical advices or technical
respondents were required to indicate whether each behavioral pattern support; (2) most cluster firms are familiar with the technology level
was systematically conducted within firms on the basis of a 0–1 re- and products of our firm; (3) it is easy for local firms to communicate
sponse format, 1 being systematicness and 0 being no systematicness. with our firm about technology or experience; (4) local firms often
Knowledge processing activity with systematicness means that firms communicate with each other about technology or experience through
have clear rules, procedures and methods to undertake such activity. In our firm.
this study, we regarded the behavioral pattern with systematicness as
routine within organizations. 4. Results
Based on the rating results for each behavioral pattern, we formed
two indicators, the intensity and the variety of technological learning 4.1. Reliability and validity
routine. As in the case of the intensity of the technological learning
routine, it was measured as a combination of the above noted 34 be- Construct reliability was assessed with Cronbach's Alpha coefficient
havioral patterns. The variety of the technological learning routine and composite reliability (CR) measure. All constructs surpassed the
refers to the opposite of the concentration degree in the use of tech- cut-off point of 0.7 for both the Cronbach's alpha value and the CR
nological learning routine during technology and product development value (Nunnally and Bernstein, 1994). Convergent validity was verified
process. We measured the variety of technological learning routine with by examining factor loadings and average variance extracted (AVE)
the entropy index (Jacquemin and Berry, 1979) as follows: values (Hair et al., al.,2010). Most loadings of items were above the
recommended threshold of 0.70 (Gefen et al., 2000). All AVE values
n
n ⎛ IMPj ⎞ ⎛ ∑j IMPj ⎞ IMPj exceeded 0.50 (Fornell and Larcker, 1981), indicating adequate con-
Variety= ∑ ∑ ∑ IMP
⎜ ∑ IMPj ⎟ In ⎜ IMPj ⎟,
= 1, i = 1, 2, 3, …, n
i=1
j vergent validity. We evaluated discriminant validity by comparing the
⎝ j ⎠ ⎝ ⎠ i=1 j
squared AVE of a construct with the correlations between the construct
The ambidexterity of technological learning routine. The ambidexterity with other constructs(Chin, 1998; Fornell and Larcker, 1981). As
of technological learning routine is an integrative construct of the in- Table 2 displays, all constructs had adequate discriminant validity be-
tensity and the variety of technological learning routine, and therefore cause the square roots of AVE were greater than the corresponding
balance ambidexterity and combined ambidexterity of technological inter-construct correlations.
learning routine are operationalized on the basis of the instruments of To minimize the effects of common method bias, we ensured the
the intensity and the variety of technological learning routine respondents the anonymity when collecting data to eliminate socially
(Cao et al., 2009; Wei et al., 2014). desirable answers (Podsakoff et al., al.,2003). In addition, the

8
J. Guo, et al. Technological Forecasting & Social Change 155 (2020) 119990

Table 2 variables were included to examine their influences on the dependent


Results of reliability and validity analysis. variable (Model 4 and Model 5). In addition, referring to Larsen (2016)
Construct Items Loadings AVE CR Cronbach's and Wang and Chen (2020), the empirical analyses are conducted with
alpha robust standard errors to control for heteroscedasticity.
Table 4 indicates the steps conducted to investigate the hypotheses.
Network centrality (NC) NC1 0.596 0.633 0.871 0.827
Model 1 just included the control variables and the moderators. We
NC2 0.812
NC3 0.851
added combined ambidexterity of technological learning routine in
NC4 0.891 model 2. Furthermore, we entered balance ambidexterity of technolo-
Knowledge tacitness (KT) KT1 0.831 0.623 0.866 0.827 gical learning routine in model 3. After introducing the interaction
KT2 0.878 between knowledge tacitness and the ambidexterity of technological
KT3 0.593
learning routine in model 4, we added the interaction of knowledge
KT4 0.823
Knowledge heterogeneity KH1 0.694 0.608 0.822 0.843 heterogeneity and the routine ambidexterity of technological learning
(KH) in model 5. To reduce potential multicollinearity, all interaction vari-
KH2 0.796 ables were mean-centered (Aiken and West, 1991). We calculated
KH3 0.841
variance inflation factors (VIFs) for each variable to check for potential
Innovation performance IP1 0.878 0.700 0.921 0.939
(IP)
multicollinearity. All VIF values ranged from 1.16 to 1.88, which were
IP2 0.796 below the threshold of 3.33 (Diamantopoulos and Winklhofer, 2001),
IP3 0.836 indicating that multicollinearity was not a problem in our study.
IP4 0.810 The results presented that combined ambidexterity of technological
IP5 0.861
learning routine adds significant prediction for innovation performance
(β = 1.526, p < 0.001), whereas balance ambidexterity of technolo-
questionnaire was designed with reversed wording and changed item gical learning routine does not show significant relationship with in-
ordering to avoid the stylistic answers (Conway and Lance, 2010). novation performance. Thus, H1 is supported and H2 is not supported.
Further, we performed Harman's one-factor test to assess common The argument of Hypothesis 3 is that knowledge tacitness interacts
method bias (Podsakoff et al., 2003). A principal component analysis significantly with combined ambidexterity of technological learning
that incorporated all measurement items was conducted, and the ro- routine to influence innovation performance. The results supported this
tation sum of squared loadings for the largest factor explained 32.659% argument (β = 0.934, p < 0.01). According to Aiken and West (1991),
of the variance, indicating that common method bias was not a major we plotted the high and low levels of each variable (one standard de-
concern in the present study (Podsakoff and Organ, 1986). viation above and below the mean). The plot (see Fig. 2) suggests that
the positive impact of combined ambidexterity of technological
learning routine on innovation performance is higher when knowledge
4.2. Hypotheses tests tacitness is high than when it is low.
Thus, Hypothesis 3 is supported. Consistent with Hypothesis 4, the
Table 3 displays the mean and standard deviations and provides interaction between balance ambidexterity of technological learning
correlations of the variables. The results presents that combined am- routine and knowledge tacitness has a significantly positive relationship
bidexterity of technological learning routine is positively related with with innovation performance (β = 0.213, p < 0.05). The plot, pre-
innovation performance (r = 0.309, p < 0.01), while balance ambi- sented in Fig. 3, suggests that, while a higher level of balance ambi-
dexterity of technological learning routine is insignificantly and nega- dexterity of technological learning routine is associated with higher
tively correlated with innovation performance (r = −0.127). innovation performance, balance ambidexterity is likely to be more
Hierarchical multiple regression analysis and the moderated method effective in influencing innovation performance when knowledge ta-
(Baron and Kenny, 1986) were used to examine the relationship be- citness is higher. Therefore, Hypothesis 4 is supported.
tween the ambidexterity of technological learning routine, knowledge As to Hypothesis 5a and 5b, the interaction effect of combined
tacitness and heterogeneity, and innovation performance. The hier- ambidexterity of technological learning routine and knowledge het-
archical feature allows the gradual establishment of separate, but re- erogeneity on innovation performance is negative and not significant.
lated, models that add more explanatory variables (Larsen, 2016). The Therefore, Hypothesis 5a and 5b are not supported. With Hypothesis 6,
test for moderation is the significance of the interaction between the the results showed that the interaction effects of balance ambidexterity
independent variable and the potential moderating variable (Baron and of technological learning routine and knowledge heterogeneity on in-
Kenny, 1986; Simmonds et al., 2019). Specifically, first, the dependent novation performance is not significant. Therefore, Hypotheses 6 is not
variable was regressed on control variables and moderating variables supported.
(Model 1). Second, the direct effects of the two independent variables
were tested on the dependent variable in Model 2 and Model 3. In the
third step, according to Baron and Kenny (1986)’s recommended pro-
cedure, the interactions of independent variables and moderating

Table 3
Descriptive statistics and correlations.
Variables Mean s.d. 1 2 3 4 5 6 7 8

1 Firm size 5.510 0.763 n.a


2 Firm age 18.230 11.639 0.187** n.a
3 Network centrality 4.515 1.005 0.317** 0.104 0.909
4 Knowledge tacitness 5.123 0.884 0.162* 0.179** 0.328** 0.807
5 Knowledge heterogeneity 3.665 1.108 −0.062 0.027 −0.207** −0.123 0.835
6 Combined ambidexterity of technological learning routine 0.092 0.259 −0.160* 0.027 0.000 0.041 0.163* n.a
7 Balance ambidexterity of technological learning routine 0.822 0.598 −0.103 −0.088 −0.019 −0.250** −0.051 0.132 n.a
8 Innovation performance 4.557 1.032 0.293** 0.259** 0.074 0.214** 0.458** 0.309** −0.127 0.870

Notes: The italic diagonal elements are the square root of the AVE values; n = 217. * p < 0.05; ** p < 0.01; *** p < 0.001.

9
J. Guo, et al. Technological Forecasting & Social Change 155 (2020) 119990

Table 4
Results of regression analysis.
Variable Model 1 Model 2 Model 3 Model 4 Model 5

† † †
Constant −0.171(0.075)* −0.126(0.074) −0.133(0.075) −0.126(0.074) −0.116(0.076)
Industry type 0.305(0.140)* 0.229(0.132) † 0.246(0.135) † 0.267(0.131)* 0.265(0.131)*
Firm size 0.302(0.092)** 0.379(0.088)*** 0.371(0.089)*** 0.343(0.080)*** 0.346(0.077)***
Firm age 0.009(0.006) 0.010(0.006) † 0.010(0.006) † 0.007(0.005) 0.007(0.005)
Ownership 0.238(0.232) 0.156(0.233) 0.135(0.224) 0.151(0.216) 0.171(0.217)
Network Centrality −0.013(0.070) −0.025(0.067) −0.020(0.068) −0.059(0.070) −0.077(0.072)
Knowledge tacitness(KT) 0.180(0.073)* 0.175(0.071)* 0.151(0.073)* 0.114(0.073) 0.129(0.075) †
Knowledge heterogeneity(KH) 0.438(0.065)*** 0.403(0.061)*** 0.397(0.062)*** 0.378(0.061)*** 0.374(0.062)***
Combined ambidexterity of technological learning routine(CA) 1.021(0.239)*** 1.058(0.222)*** 1.480(0.249)*** 1.526(0.279)***
Balance ambidexterity of technological learning routine(BA) −0.111(0.080) 0.006(0.094) 0.046(0.109)
CA*KT 0.921(0.285)** 0.934(0.287)**
BA*KT 0.208(0.102)* 0.213(0.105)*
CA*KH −0.176(0.156)
BA*KH 0.070(0.122)
N 217 217 217 217 217
R2 0.409 0.469 0.473 0.507 0.509
F-values 23.66*** 26.99*** 25.52*** 22.23*** 19.70***

Notes: Robust standard errors reported in parentheses. n = 217. * p < 0.05; ** p < 0.01; *** p < 0.001.

5. Discussion and implications performance.


The present study also sheds new light on in what circumstances
5.1. Discussion and conclusion routine ambidexterity can be more influential. The findings indicate
that when knowledge embedded within product and process develop-
This study applies an ambidexterity perspective to the performance ment is more tacit, the role of both combined and balance ambi-
effect of technological learning routine and examines the moderating dexterity of technological learning routines would be more important to
effects of knowledge tacitness and knowledge heterogeneity on the firm innovation performance. However, the moderating effects of
relationship between the ambidexterity of technological learning rou- knowledge heterogeneity on the relationship between combined or
tine and innovation performance. Using a sample of 217 firms within balance ambidexterity of technological learning routine and innovation
industrial clusters in China, the empirical findings reveal that combined performance are not supported. This can be explained as follows. On
ambidexterity of technological learning routine positively influences one hand, knowledge heterogeneity may lead to mutual conflicts and
firm innovation performance, whereas the direct effect of balance am- ambiguity in knowledge processing activities within organizations
bidexterity on innovation performance is not significant. One possible (Tsai et al., 2014), and the adjustment of the relative magnitude be-
explanation is that most firms within industrial clusters do not possess tween the intensity and the variety of technological learning routine is
sufficient dynamic resource management capabilities (Sirmon et al., likely to be more difficult. Firms within industrial clusters are lack of
2011; Wei et al., 2014), which may affect the way firms reconfigure abundant resources to deal with such adjustment when knowledge
resource portfolios to leverage the relative magnitude of the intensity heterogeneity is at a high level (Cao et al., 2009). Therefore, knowledge
and the variety of technological learning routine for innovation and heterogeneity may not significantly impact the relationship between
hence influence the impact of the balance ambidexterity on innovation the balance dimension and innovation performance. On the other hand,

Fig. 2. Interaction plot of knowledge tacitness and combined ambidexterity of technological learning routine on innovation performance.

10
J. Guo, et al. Technological Forecasting & Social Change 155 (2020) 119990

Fig. 3. Interaction plot of knowledge tacitness and balance ambidexterity of technological learning routine on innovation performance.

knowledge heterogeneity may result in intercommunication troubles, technological learning routine by providing greater operational content
decision disagreement or even debates (Cegarra-Navarro et al., 2019), for the theoretical construct (Becker, 2004; Cohen et al., 1996;
and thus firms are likely to undertake differing featured technological Hales and Tidd, 2009; Pavitt, 2002; Reynaud, 1998).
learning routines to suit for market and technology changes in a het- Second, this study contributes to reconciling the mixed findings
erogeneous knowledge context (Guo et al., 2018). In this regard, the about whether routine ambidexterity facilitates or frustrates innovation
link between the combined dimension and innovation performance may by examining the moderating roles of knowledge attributes in shaping
be negatively but not significantly moderated by knowledge hetero- the ambidexterity-innovation performance relationship. Previous re-
geneity. search has postulated that theory on the contextual factors that affect
the ambidexterity–innovation performance relationship remains im-
5.2. Theoretical implications poverished (Raisch and Birkinshaw, 2008; Simsek et al., 2009). Some
recent research examines how specific attributes of organizational and
This study contributes to the literature as follows. First, this study environmental contexts can shape the relationship between ambi-
extends the understanding of technological learning routine by ap- dexterity and subsequent performance (Cao et al., 2009; De Clercq
plying an ambidexterity perspective and addresses the performance et al., 2013; Jansen et al., 2012; Rothaermel and Alexandre, 2009;
turn in routine research by studying routines empirically. In line with Wei et al., 2014). Since the outcomes of either the intensity or the
Rothaermel and Alexandre (2009), we concur that the exploitation- variety of technological learning routine are based on how easily
exploration lens is but one manifestation of organizational ambi- knowledge can be acquired, maintained, reactivated and integrated
dexterity and organizational ambidexterity can also be extended to (Argote et al., 2003; Guo et al., 2018), the specific attributes of
describe other organizational phenomenon. The present study applies knowledge contexts should be one of the essential contingencies.
the ambidexterity perspective to depict a firm's ability to simulta- Nonetheless, the contingent roles of knowledge attributes are less ex-
neously pursue the intensity and the variety of technological learning plored in previous research. By addressing this issue, we find that firms
routine. The theoretical link and differentiation between the exploita- with high levels of knowledge tacitness achieve commensurately higher
tion-exploration lens and the intensity-variety lens are as follows. On innovation performance from both combined and balance ambi-
one hand, the intensity-variety lens is related to the exploitation-ex- dexterity of technological learning routines. Such findings extend the
ploration lens in that both intensity and exploitation reflect variance argument of Lee and Huang (2012) and Lin and Chang (2015) that
reduction while both variety and exploration indicate variance increase knowledge representation may have a crucial role in leveraging the
as to learning activities within firms (March, 1991; Lavie et al., 2010). ambidexterity for innovation.
On the other hand, the intensity-variety lens is to a certain extent dif-
ferent from the exploitation-exploration lens in that the former captures 5.3. Managerial implications
the characteristics of organization process and behavioral patterns un-
derlying technological learning, whereas the latter focuses on the scale Our findings have several practical implications. First, since the
and scope of knowledge created or acquired relative to a firm's existing balance ambidexterity of technological learning routine does not fa-
knowledge base (Lavie et al., 2010; Levinthal and March, 1993), ig- cilitate firm innovation performance directly, managers have to assess
noring the recurrent organization process of technological learning carefully when and whether to carry out balance ambidexterity within
within firms. The empirical results indicate that combined ambi- industrial clusters. However, the finding pertaining to the critical role
dexterity of technological learning routine is critical to firm innovation of combined ambidexterity of technological learning routine highlights
performance. This finding not only resonates with previous research the need to take measures to improve the absolute magnitude of the
examining ambidexterity hypotheses in different contexts intensity and the variety of technological learning routine to increase
(Lubatkin et al., 2006; Rothaermel and Alexandre, 2009; Sidhu et al., their complementary effects for innovation performance. In particular,
2007), but also answers the call to explore the performance effect of for cluster firms in developed countries, managers can deploy units that

11
J. Guo, et al. Technological Forecasting & Social Change 155 (2020) 119990

are dedicated to conduct diverse technological learning routine and Data curation. Xiaobo Wu: Investigation, Supervision, Project admin-
separate them from those devoted to carry out intense technological istration.
learning routine (Cao et al., 2009; Christensen, 1997; Tushman and
O'Reilly, 1996). On the contrary, for cluster firms in emerging econo- Supplementary materials
mies, managers may use sequential attention or rhythmic pacing to shift
between the intensity and the variety of technological learning routine Supplementary material associated with this article can be found, in
to pursue the combined ambidexterity due to possible insufficiency in the online version, at doi:10.1016/j.techfore.2020.119990.
resource stock (Brown and Eisenhardt, 1997).
Second, the present study emphasizes the importance of knowledge References
tacitness in conditioning the ambidexterity-innovation performance
relationship. Although the balance dimension does not significantly Abebe, M.A., Angriawan, A., 2014. Organizational and competitive influences of ex-
influence firm innovation performance, knowledge tacitness still play a ploration and exploitation activities in small firms. J. Bus. Res. 67 (3), 339–345.
Acs, Z.J., Audretsch, D.B., 1987. Innovation, market structure, and firm size. Rev. Econ.
complementary role in moderating the effect of balance ambidexterity. Stat. 69 (4), 567–574.
As such, apart from environmental munificence (Cao et al., 2009), co- Adler, P.S., Goldaftas, B., Levine, D.I., 1999. Flexibility versus efficiency: a case study of
ordination flexibility (Wei et al., 2014), resource munificence model changeovers in the Toyota production system. Org. Sci. 10, 43–68.
Aiken, L.S., West, S.G., 1991. Multiple regression: Testing and Interpreting Interactions.
(Jansen et al., 2012), absorptive capacity (Rothaermel and Sage, Newbury Park, CA.
Alexandre, 2009), and internal competitive environment (De Clercq Aime, F., Johenson, S., Ridge, J.W., Hill, A.D., 2010. The routine may be stable but the
et al., 2013), managers should also attentively take into account the advantage is not: competitive implications of key employee mobility. Strat. Manag. J.
31 (1), 75–87.
level of knowledge tacitness embedded within their product and pro- Amabile, T., 1996. Creativity in Context. Westview Press, Boulder, CO.
cess development, and leverage both the balance and the combined Amir, A., Ahad, Z.R., 2018. How market orientation contributes to innovation and market
dimension of the ambidexterity of technological learning routine for performance: the roles of business analytics and flexible IT infrastructure. J. Bus.
Indus. Mark. 33 (7), 970–983.
higher innovation performance at high levels of knowledge tacitness.
Anand, V., Clark, M., Zellmer-Bruhn, M., 2003. Team knowledge structures: matching
task to information environment. J. Manag. Issues 15 (1), 15–33.
5.4. Limitations and future directions Argote, L., McEvily, B., Reagans, R., 2003. Managing knowledge in organizations: an
integrative framework and review of emerging themes. Manage. Sci. 49 (4), 571–582.
Argyris, C., Schön, D.A., 1978. Organizational learning: a theory of action perspective.
This study also has several limitations which suggest directions for Reis 10 (77/78), 345–348.
future research. First, the sample studied here is based in China as an Aroles, J., McLean, C., 2016. Rethinking stability and change in the study of organiza-
emerging economy. These sample firms may not possess a large stock of tional routines: difference and repetition in a newspaper-printing factory. Org. Sci. 27
(3), 535–550.
resource as firms in developed countries do. The generalizability of our Arrow, K., 1962. The economic implications of learning by doing. Rev. Econ. Stud. 29 (3),
findings should be examined with diverse samples from developed 155–173.
countries in future research. Second, this study is designed in cross- Baron, R.M., Kenny, D.A., 1986. The moderator–mediator variable distinction in social
psychological research: conceptual, strategic, and statistical considerations. J. Pers.
sectional nature, which cannot explicate how the contingent role of Soc. Psychol. 51 (6), 1173–1182.
knowledge attributes in shaping routine ambidexterity-innovation Becker, M.C., 2004. Organizational routines: a review of the literature. Indus. Corp.
performance relationship evolves over time. In this situation, there is a Change 13 (4), 643–678.
Becker, M.C., 2005. A framework for applying organizational routines in empirical re-
compelling need to conduct longitudinal studies to explore how the search: linking antecedents, characteristics and performance outcomes of recurrent
above relationship dynamically changes over time. Third, future re- interaction patterns. Indus. Corp. Change 14 (5), 817–846.
search exploring how other specific attributes of knowledge domain Becker, M.C., Lazaric, N., Nelson, R.R., Winter, S.G., 2005. Applying organizational
routines in understanding organizational change. Indus. Corp. Change 14 (5),
may affect the performance effect of the ambidexterity of technological
775–791.
learning routine can shed more light on ambidexterity research. Fourth, van den Berg, H.A., 2013. Three shapes of organisational knowledge. J. Knowl. Manag. 17
prior research on ambidexterity has highlighted that the ambidexterity- (2), 159–174.
innovation performance link is contingent on organizational and en- Berman, S.L., Down, J., Hill, C.W., 2002. Tacit knowledge as a source of competitive
advantage in the national basketball association. Acad. Manag. J. 45 (1), 13–31.
vironmental contexts (Cao et al., 2009; De Clercq et al., 2013; Betsch, T.S., Haberstroh, A., Glöckner, T.H., Fiedler, K., 2001. The effects of routine
Jansen et al., 2012; Rothaermel and Alexandre, 2009; Wei et al., 2014). strength on adaptation and information search in recurrent decision making. Org.
It is worthy of validating whether organizational and environmental Behav. Human Decis. Processes 84 (1), 23–53.
Bierly III, P.E., Damanpour, F., Santoro, M.D., 2009. The application of external knowl-
contexts attenuate or amplify the moderating effect of knowledge at- edge: organizational conditions for exploration and exploitation. J. Manag. Stud. 46
tributes on the relationship between routine ambidexterity and in- (3), 481–509.
novation performance. Blümm, C. (2002) Die Bedeutung impliziten Wissens im Innovationsprozess, Gabler,
Wiesbaden.
Breschi, S., Malerba, F., 2001. The geography of innovation and economic clustering:
Acknowledgements some introductory notes. Indus. Corp. Change 10 (4), 817–833.
Brislin, R., 1970. Back-translation for cross-cultural research. J. Cross Cult. Psychol. 1 (3),
185–216.
This study was supported by the National Natural Science
Brown, J., Duguid, P., 1991. Organizational learning and communities-of-practice: to-
Foundation of China [Grant no. 71672185]; the Youth Innovation ward a unified view of working, learning, and innovation. Org. Sci. 2 (1), 40–57.
Promotion Association, Chinese Academy of Sciences [Grant no. Brown, S.L., Eisenhardt, K.M., 1997. The art of continuous change: linking complexity
theory and time-paced evolution in relentlessly shifting organizations. Adm. Sci. Q.
2017201]; the Major project of the National Social Science Fund of
42, 1–34.
China [Grant no. 17ZDA050]; the National Natural Science Foundation Cao, Q., Gedajlovic, E., Zhang, H., 2009. Unpacking organizational ambidexterity: di-
of China [Grant no. 71772014]; and the Dean's Youth Fund of Institutes mensions, contingencies, and synergistic effects. Org. Sci. 20 (4), 1–16.
of Science and Development, Chinese Academy of Sciences [Grant Carlile, P., 2002. A pragmatic view of knowledge and boundaries: boundary objects in
new product development. Org. Sci. 13 (4), 442–455.
no.Y7X1061Q01]. Carlile, P.R., Rebentisch, E.S., 2003. Into the black box: the knowledge transformation
cycle. Manage. Sci. 49 (9), 1180–1195.
CRediT authorship contribution statement Carlile, P.R., 2004. Transferring, translating, and transforming: an integrative framework
for managing knowledge across boundaries. Org. Sci. 15 (5), 555–568.
Carlo, J.L., Lyytinen, K., Rose, G.M., 2012. A knowledge-based model of radical in-
Jingjing Guo: Conceptualization, Methodology, Software, Formal novation in small software firms. Manag. Inf. Syst. Q. 36 (3), 865–895.
analysis, Investigation, Writing - original draft, Writing - review & Cegarra-Navarro, J.G., Ruiz, F.J.A., Martínez-Caro, E., Garcia-Perez, A., 2019. Turning
heterogeneity into improved research outputs in international R&D teams. J. Bus.
editing, Funding acquisition. Bin Guo: Conceptualization, Res. https://doi.org/10.1016/j.jbusres.2019.05.023.
Methodology, Writing - original draft, Writing - review & editing, Chen, J., Heng, C.S., Tan, B.C.Y., Lin, Z.J., 2018. The distinct signaling effects of R&D
Visualization. Jianghua Zhou: Validation, Investigation, Resources, subsidy and non-R&D subsidy on IPO performance of IT entrepreneurial firms in

12
J. Guo, et al. Technological Forecasting & Social Change 155 (2020) 119990

China. Res. Policy 47 (1), 108–120. Guo, J.J., Yin, Q.X., 2008. Empirical study on influential factors of inter-enterprises’ tacit
Chin, W.W., 1998. The partial least squares approach to structural equation modeling. In: knowledge transfer effectiveness. Technol. Econ. 27 (7), 7–17.
Marcoulides, G.A. (Ed.), Modern Methods For Business Research. Lawrence Erlbaum Gupta, A.K., Smith, K.G., Shalley, C.E., 2006. The interplay between exploration and
Associates, Mahwah, NJ, pp. 295–336. exploitation. Acad. Manag. J. 49 (4), 693–706.
Choo, C.W., Bontis, N., 2002. The Strategic Management of Intellectual Capital and Hair, J.F., Black, W.C., Babin, B.J., Anderson, R.E., 2010. Multivariate Data Analysis: A
Organizational Knowledge. Oxford University Press, New York, NY. Global Perspective. Pearson Education, New Jersey, NJ.
Christensen, C.M., 1997. The Innovator's Dilemma. Harvard Business School Press, Hales, M., Tidd, J., 2009. The practice of routines and representations in design and
Boston. development. Indus. Corp. Change 18 (4), 551–574.
Cohen, M.D., Burkhart, R., Dosi, G., Egidi, M., Marengo, L., Warglien, M., Winter, S., Harabi, N., 1997. Channels of r&d spillovers: an empirical investigation of Swiss firms.
1996. Routines and other recurring action patterns of organizations: contemporary Technovation 17 (11–12), 627–635.
research issues. Indus. Corp. Change 5 (3), 653–698. Hargadon, A., Sutton, R.I., 1997. Technology brokering and innovation in a product
Cohen, W., Levinthal, D., 1990. Absorptive capacity: a new perspective on learning and development firm. Adm. Sci. Q. 42 (4), 716–749.
innovation. Adm. Sci. Q. 35, 128–152. He, Z.L., Wong, P.K., 2004. Exploration vs. exploitation: an empirical test of the ambi-
Collinson, S., Wilson, D.C., 2006. Inertia in Japanese organizations: knowledge man- dexterity hypothesis. Org. Sci. 15, 481–494.
agement routines and failure to innovate. Org. Stud. 27 (9), 1359–1388. Hedlund, G., 1994. A model of knowledge management and the N-form corporation.
Conway, J.M., Lance, C.E., 2010. What reviewers should expect from authors regarding Strat. Manag. J. 15 (S2), 73–90.
common method bias in organizational research. J. Bus. Psychol. 25 (3), 325–334. Heiman, B.A., Nickerson, J.A., 2004. Empirical evidence regarding the tension between
Cyert, R., March, J., 1963. A Behavioral Theory of the Firm. Prentice-Hall, Englewood knowledge sharing and knowledge expropriation in collaborations. Manag. Decis.
Cliffs, NJ. Econ. 25 (6/7), 401–420.
Corsaro, D., Cantù, C., Tunisini, A., 2012. Actors' heterogeneity in innovation networks. Helfat, C., Peteraf, M.A., 2003. The dynamic resource-based view: capability lifecycles.
Indus. Mark. Manag. 41 (5), 780–789. Strat. Manag. J. 24 (10), 997–1010.
Cruz-González, J., López-Sáez, P., Navas-López, J.E., Delgado-Verde, M., 2015. Open Howard-Grenville, J.A., 2005. The persistence of flexible organizational routines: the role
search strategies and firm performance: the different moderating role of technolo- of agency and organizational context. Org. Sci. 16 (6), 618–636.
gical environmental dynamism. Technovation 35, 32–45. Huber, G., 1996. Organizational learning: a guide for executives in technology-critical
De Clercq, D., Thongpapanl, N., Dimov, D., 2013. Shedding new light on the relationship organizations. IJTM Spec. Pub. Unlearn. Learn. 11, 821–832.
between contextual ambidexterity and firm performance: an investigation of internal Jacquemin, A., Berry, C., 1979. Entropy measure of diversification and corporate growth.
contingencies. Technovation 33 (4–5), 119–132. J. Ind. Econ. 27 (4), 359–369.
Dekoulou, P., Dekoulou, P., Trivellas, P., Trivellas, P., 2017. Organizational structure, Jansen, J.J.P., Simsek, Z., Cao, Q., 2012. Ambidexterity and performance in multiunit
innovation performance and customer relationship value in the Greek advertising and contexts: cross-level moderating effects of structural and resource attributes. Strat.
media industry. J. Bus. Indus. Mark. 32 (3), 385–397. Manag. J. 33 (11), 1286–1303.
Derbyshire, J., 2014. The impact of ambidexterity on enterprise performance: evidence Kauppila, O.P., 2015. Alliance management capability and firm performance: using re-
from 15 countries and 14 sectors. Technovation 34 (10), 574–581. source-based theory to look inside the process black box. Long. Range Plann. 48 (3),
Diamantopoulos, A., Winklhofer, H., 2001. Index construction with formative indicators: 151–167.
an alternative to scale development. J. Mark. Res. 38, 269–277. Kim, N., Im, S., Slater, S.F., 2013. Impact of knowledge type and strategic orientation on
Dong, J.Q., Yang, C.H., 2016. Being central is a double-edged sword: knowledge network new product creativity and advantage in high-technology firms. J. Prod. Innov.
centrality and new product development in U.S. pharmaceutical industry. Technol. Manag. 30 (1), 136–153.
Forecast. Soc. Change 113, 379–385. Knott, A.M., 2003. The organizational routines factor market paradox. Strat. Manag. J. 24
Dosi, G., 1988. Sources, procedures, and microeconomic effects of innovation. J. Econ. (10), 929–943.
Lit. 26 (3), 1120–1171. Kogut, B., Zander, U., 1992. Knowledge of the firm, combinative capabilities, and the
Dougherty, D., 1992. Interpretive barriers to successful product innovation in large firms. replication of technology. Org. Sci. 3 (3), 383–397.
Org. Sci. 3 (2), 179–202. Koumakhov, R., Daoud, A., 2017. Routine and reflexivity: simonian cognitivism vs
Eisenhardt, K., Martin, J., 2000. Dynamic capabilities: what are they? Strat. Manag. J. 21 practice approach. Indus. Corp. Change 26 (4), 727–743.
(10–11), 1105–1121. Lane, P.J., Koka, B.R., 2006. The reification of absorptive capacity: a critical review and
Feldman, M.S., 2000. Organizational routines as a source of continuous change. Org. Sci. rejuvenation of the construct. Acad. Manage. Rev. 31 (4), 833–863.
11 (6), 611–629. Lane, P.J., Koka, B.R., Pathak, S., 2006. The reification of absorptive capacity: a critical
Feldman, M.S., Pentland, B.T., 2003. Reconceptualizing organizational routines as a review and rejuvenation of the construct. Acad. Manage. Rev. 31 (4), 833–863.
source of flexibility and change. Adm. Sci. Q. 48 (1), 94–121. Lane, P.J., Lubatkin, M., 1998. Relative absorptive capacity and interorganizational
Feldman, M.S., Pentland, B.T., D'Adderio, L., Lazaric, N., 2016. Beyond routines as things: learning. Strat. Manag. J. 19 (5), 461–477.
introduction to the special issue on routine dynamics. Org. Sci. 27 (3), 505–513. Langlois, R.N., Robertson, P.L., 1996. Stop Crying Over Spilt knowledge: a Critical Look
Fernhaber, S.A., Patel, P.C., 2012. How do young firms manage product portfolio com- At the Theory of Spillovers and Technical Change. Department of Economics,
plexity? The role of absorptive capacity and ambidexterity. Strat. Manag. J. 33 (13), University of Connecticut, Storrs-Mansfield, CT.
1516–1539. Larraneta, B., Zahra, S.A., Gonzalez, J.L.G., 2012. Enriching strategic variety in new
Figueiredo, P.N., 2003. Learning, capability accumulation and firms differences: evidence ventures through external knowledge. J. Bus. Ventur. 27 (4), 401–413.
from latecomer steel. Indus. Corp. Change 12 (3), 607–643. Larsen, M.M., 2016. Failing to estimate the costs of offshoring: a study on process per-
Fornell, C., Larcker, D.F., 1981. Evaluating structural equation models with unobservable formance. Int. Bus. Rev. 25 (1), 307–318.
variables and measurement error. J. Mark. Res. 18 (1), 39–50. Lavie, D., Stettner, U., Tushman, M.L., 2010. Exploration and exploitation within and
García-Muiña, F.E., González-Sánchez, R., 2017. Absorptive routines and international across organizations. Acad. Manag. Ann. 4 (1), 109–155.
patent performance. Org. Sci. 20 (2), 96–111. Lee, C.S., Wong, K.Y., 2015. Development and validation of knowledge management
Garud, R., Nayyar, P.R., 1994. Transformative capacity: continual structuring by inter- performance measurement constructs for small and medium enterprises. J. Knowl.
temporal technology transfer. Strat. Manag. J. 15 (5), 365–385. Manag. 19 (4), 711–734.
Gefen, D., Straub, D., Boudreau, M., 2000. Structural equation modeling and regression: Lee, C.Y., Huang, Y.C., 2012. Knowledge stock, ambidextrous learning and firm perfor-
guidelines for research practice. Commun. Assoc. Inf. Syst. 7 (7), 1–78. mance: evidence from technologically intensive industries. Manag. Dec. 50,
Gersick, C., Hackman, J., 1990. Habitual routines in task-performing groups. Organ. 1096–1116.
Behav. Hum. Decis. Process. 47 (1), 65–97. Lee, C.Yu, Wu, H.L., Liu, C.Y., 2013. Contextual determinants of ambidextrous learning:
Gibson, C.B., Birkinshaw, J., 2004. The antecedents, consequences, and mediating role of evidence from industrial firms in four industrialized countries. IEEE Trans. Eng.
organizational ambidexterity. Acad. Manag. J. 47 (2), 209–226. Manage. 60 (3), 529–540.
Giuliani, E., 2007. The selective nature of knowledge networks in clusters: evidence from Leonard-Barton, D., 1992. The factory as a learning laboratory. Sloan. Manage. Rev. 34
the wine industry. J. Econ. Geograph. 7 (2), 139–168. 23-23.
Giuliani, E., Bell, M., 2005. The micro-determinants of meso-level learning and innova- Levinthal, D., March, J.G., 1993. The myopia of learning. Strat. Manag. J. 14 (S1),
tion: evidence from a Chilean wine cluster. Res. Policy 34 (1), 47–68. 95–112.
Glazer, R., 1991. Marketing in an information-intensive environment: strategic implica- Lewin, A.Y., Massini, S., Peeters, C., 2011. Microfoundations of internal and external
tions of knowledge as an asset. J. Mark. 55 (4), 1–19. absorptive capacity routines. Org. Sci. 22 (1), 81–98.
Grant, R.M., 1996. Toward a knowledge-based theory of the firm. Strat. Manag. J. 17, Lin, C., Chang, C.C., 2015. A patent-based study of the relationships among technological
109–122. portfolio, ambidextrous innovation, and firm performance. Technol. Anal. Strat.
Grant, R.M., 2002. The knowledge-based view of the firm. In: Choo, C.W., Bontis, N. Manag. 27 (10), 1193–1211.
(Eds.), The Strategic Management of Intellectual Capital and Organizational Lin, H.E., McDonough, E.F., Lin, S.J., Lin, C.Y.Y., 2013. Managing the exploitation/ex-
Knowledge. Oxford University Press, Oxford. ploration paradox: the role of a learning capability and innovation ambidexterity. J.
Guo, J., Guo, B., Chen, X., Du, J., 2018. The impact of knowledge attributes on techno- Prod. Innov. Manag. 30 (2), 262–278.
logical learning routine within industrial clusters. Int. J. Technol. Manage. 78 (3), Liu, W., Atuahene-Gima, K., 2018. Enhancing product innovation performance in a
234–260. dysfunctional competitive environment: the roles of competitive strategies and
Guo, B., Guo, J.J., 2011. Patterns of technological learning within the knowledge systems market-based assets. Indus. Mark. Manag. 73, 7–20.
of industrial clusters in emerging economies: evidence from China. Technovation 31 Lubatkin, M.H., Simsek, Z., Yan, L., Veiga, J.F., 2006. Ambidexterity and performance in
(2/3), 87–104. small-to medium-sized firms: the pivotal role of top management team behavioral
Guo, J.J., Guo, B., 2013. How do innovation intermediaries facilitate knowledge spil- integration. J. Manage. 32, 646–672.
lovers within industrial clusters? A knowledge-processing perspective. Asian J. March, J.G., Simon, H.A., 1958. Organizations. Wiley, NewYork.
Technol. Innov. 21 (sup2), 31–49. March, J.G., 1991. Exploration and exploitation in organizational learning. Org. Sci. 2 (1),

13
J. Guo, et al. Technological Forecasting & Social Change 155 (2020) 119990

71–87. Simmonds, L., Bellman, S., Kennedy, R., Nenycz-Thiel, M., Bogomolova, S., 2019.
March, J.G., 1994. A Primer on Decision-making. Free Press, New York. Moderating effects of prior brand usage on visual attention to video advertising and
McEvily, S.K., Chakravarthy, B., 2002. The persistence of knowledge-based advantage: an recall: an eye-tracking investigation. J. Bus. Res. https://doi.org/10.1016/j.jbusres.
empirical test for product performance and technological knowledge. Strat. Manag. J. 2019.02.062.
23 (4), 285–305. Simon, H.A., 1947/1997. Administrative Behaviour. The Free Press, New York.
McIver, D., Lengnick-Hall, C.A., Lengnick-Hall, M.L., Ramachandran, I., 2013. Simonin, B.L., 1999. Ambiguity and the process of knowledge transfer in strategic alli-
Understanding work and knowledge management from a knowledge-in-practice ances. Strat. Manag. J. 20 (7), 595–623.
perspective. Acad. Manage. Rev. 38 (4), 597–620. Simsek, Z., Heavey, C., Veiga, J.A., 2009. Typology for aligning organizational ambi-
Medlin, C.J., Törnroos, Jan-Åke, 2015. Exploring and exploiting network relationships to dexterity's conceptualizations, antecedents, and outcomes. J. Manag. Stud. 46 (5),
commercialize technology: a biofuel case. Indus. Mark. Manag. 49, 42–52. 864–894.
Mohr, J., Spekman, R., 1994. Characteristics of partnership success: partnership attri- Sirmon, D.G., Hitt, M.A., Ireland, R.D., Gilbert, B.A., 2011. Resource orchestration to
butes, communication behavior, and conflict resolution techniques. Strat. Manag. J. create competitive advantage: Breadth, depth and life cycle effects. Journal of
15 (2), 135–152. Management 37 (5), 1390–1412.
Nelson, R.R., Winter, S.G., 1982. An Evolutionary Theory of Economic Change. The Smith, K.G., Collins, C.J., Clark, K.D., 2005. Existing knowledge, knowledge creation
Belknap Press of Harvard University, Cambridge, MA. capability, and the rate of new product introduction in high-technology firms. Acad.
National Bureau of Statistics of China (2017) China Statistical Yearbook 2017, China Manag. J. 48 (2), 346–357.
Statistics Press, Beijing. Smith, W.K., Tushman, M.L., 2005. Managing strategic contradictions: a top management
Nonaka, I., 1994. A dynamic theory of organizational knowledge creation. Org. Sci. 5 (1), model for managing innovation streams. Org. Sci. 16 (5), 522–536.
14–37. Solís-Molina, M., Hernández-Espallardo, M., Rodríguez-Orejuela, A., 2018. Performance
Nonaka, I., Takeuchi, H., 1995. The Knowledge Creating Company: How Japanese implications of organizational ambidexterity versus specialization in exploitation or
Companies Create the Dynasties of Innovation. Oxford University Press, New exploration: the role of absorptive capacity. J. Bus. Res. 91, 181–194.
York, NY. Song, J., Wei, Y.S., Wang, R., 2015. Market orientation and innovation performance: the
Nunnally, J.C., Bernstein, I.H., 1994. Psychometric Theory, 3ed. McGraw-Hill, New York. moderating roles of firm ownership structures. Int. J. Res. Mark. 32 (3), 319–331.
O'Cass, A., Heirati, N., Ngo, L.V., 2014. Achieving new product success via the syn- Sørensen, J.B., Stuart, T.E., 2000. Aging, obsolescence, and organizational innovation.
chronization of exploration and exploitation across multiple levels and functional Adm. Sci. Q. 45, 81–112.
areas. Indus. Mark. Manag. 43 (5), 862–872. Stettner, U., Lavie, D., 2014. Ambidexterity under scrutiny: exploration and exploitation
Oerlemans, L.A.G., Knoben, J., Pretorius, M.W., 2013. Alliance portfolio diversity, radical via internal organization, alliances, and acquisitions. Strat. Manag. J. 35 (13),
and incremental innovation: the moderating role of technology management. 1903–1929.
Technovation 33 (6–7), 234–246. Szulanski, G., 1996. Exploring internal stickiness: impediments to the transfer of best
O'Reilly, C.A., Tushman, M.L., 2008. Ambidexterity as a dynamic capability: resolving the practice within the firm. Strat. Manag. J. 17, 27–43.
innovator's dilemma. Res. Organ. Behav. 28, 185–206. Tsai, F.S., Gayle, S.B., Fang, S.C., Lin, J.L., 2014. Contingent contingency: knowledge
Parmigiani, A., Howard-Grenville, J., 2011. Routines revisited: exploring the capabilities heterogeneity and new product development performance revisited. Asia Pac. J.
and practice perspectives. Acad. Manag. Ann. 5 (1), 413–453. Manag. 31 (1), 149–169.
Pavitt, K., 2002. Innovating routines in the business firm: what corporate tasks should Tsai, F.S., 2018. Knowledge heterogeneity, social capital, and organizational innovation.
they be accomplishing. Indus. Corp. Change 11 (1), 117–133. J. Org. Change Manag. 31 (2), 304–322.
Pekovic, S., Pekovic, S., Rolland, S., Rolland, S., Gatignon, H., Gatignon, H., 2016. Tsai, L.C., Zhang, R.H., Zhao, C.F., 2019. Political connections, network centrality and
Customer orientation and organizational innovation: the case of environmental firm innovation. Financ. Res. Lett. 28, 180–184.
management practices. J. Bus. Indus. Mark. 31 (7), 835–848. Tushman, M.L., O'Reilly, C.A., 1996. The ambidextrous organization: managing evolu-
Pelled, L., Eisenhardt, K., Xin, K., 1999. Exploring the black box: an analysis of work tionary and revolutionary change. Calif. Manage. Rev. 38 (4), 1–23.
group diversity, conflict, and performance. Adm. Sci. Q. 44 (1), 1–3. Tyre, M.J., Orlikowski, W.J., 1996. The episodic process of learning by using. Int. J.
Peng, D.X., Schroeder, R.G., Shah, R., 2008. Linking routines to operations capabilities: a Technol. Manage. 11, 790–798.
new perspective. J. Oper. Manage. 26 (6), 730–748. Urgal, B., Quintas, M.A., Arevalo-Tome, R., 2013. Knowledge resources and innovation
Pentland, B.T., 1995. Grammatical models of organizational processes. Org. Sci. 6, performance: the mediation of innovation capability moderated by management
541–556. commitment. Technol. Anal. Strat. Manag. 25 (5), 543–565.
Pentland, B.T., 2003a. Sequential variety in work processes. Org. Sci. 14, 528–540. Van Looy, B., Martens, T., Debackere, K., 2005. Organizing for continuous innovation: on
Pentland, B.T., 2003b. Conceptualizing and measuring variety in organizational work the sustainability of ambidextrous organizations. Creat. Innov. Manag. 14, 208–221.
processes. Manage Sci 45, 425–443. Venkatraman, N., Lee, C.H., Iyer, B., 2007. Strategic ambidexterity and sales growth: a
Pentland, B.T., Hrem, T., Hillison, D., 2011. The (N)Ever-Changing world: stability and longitudinal test in the software sector. In: Unpublished Manuscript (earlier version
change in organizational routines. Org. Sci. 22 (6), 1369–1383. presented at the Academy of Management Meetings, 2005.
Pentland, B., Reuter, H., 1994. Organizational routines as grammars of action. Adm. Sci. Walsh, J., Ungson, G., 1991. Organizational memory. Acad. Manage. Rev. 16 (1), 57–91.
Q. 39 (3), 484–510. Wang, E.T., Hu, H.F., Hu, P.J.H., 2013. Examining the role of information technology in
Pitt, M., McVaugh, J., 2008. Knowledge management for new product development. J. cultivating firms’ dynamic marketing capabilities. Inf. Manag. 50 (6), 336–343.
Knowl. Manag. 12 (4), 101–116. Wang, F., Chen, K.H., 2020. Do product imitation and innovation require different pat-
Piotroski, J.D., Zhang, T., 2014. Politicians and the ipo decision: the impact of impending terns of organizational innovation? Evidence from Chinese firms. J. Bus. Res. 106,
political promotions on ipo activity in china. Journal of Finance Economics 111, 60–74.
111–136. Wang, X.J., 2007. Knowledge network and competitive advantage of cluster firms. Ph.D.
Podsakoff, P.M., Mackenzie, S.B., Lee, J., Podsakoff, N.P., 2003. Common method biases thesis paper. Zhejiang University.
in behavioral research: a critical review of the literature and recommended remedies. Wang, Y.B., Li, J.Z., Furman, J.L., 2017a. Firm performance and state innovation funding:
J. Appl. Psychol. 88 (5), 879–903. evidence from China's Innofund program. Res. Policy 46 (6), 1142–1161.
Podsakoff, P.M., Organ, D.W., 1986. Self-reports in organizational research: problems and Wernerfelt, B., Montgomery, C.A., 1988. Tobin's q and the importance of focus in firm
prospects. J Manage 12 (4), 531–544. performance. Am. Econ. Rev. 78, 246–251.
Polanyi, M., 1967. The Tacit Dimension. Garden City, NY. van Wijk, R., Jansen, J.J.P., Van den Bosch, F.A.J., Volberda, H.W., 2012. How firms
Powell, W.W., Koput, K.W., Smith-Doerr, L., 1996. Interorganizational collaboration and shape knowledge to explore and exploit: a study of knowledge flows, knowledge
the locus of innovation: networks of learning in biotechnology. Adm. Sci. Q. 41 (2), stocks and innovative performance across units. Technol. Anal. Strat. Manag. 24 (9),
116–145. 929–950.
Raisch, S., Birkinshaw, J., 2008. Organizational ambidexterity: antecedents, outcomes, Wei, Z., Yi, Y., Guo, H., 2014. Organizational learning ambidexterity, strategic flexibility,
and moderators. J. Manage. 34, 375–409. and new product development. J. Product Innov. Manag. 31 (4), 832–847.
Reed, R., DeFillippi, R.J., 1990. Causal ambiguity, barriers to imitation, and sustainable Winter, S.G., 2017. Pursuing the evolutionary agenda in economics and management
competitive advantage. Acad. Manage. Rev. 15 (1), 88–102. research. Cambridge J. Econ. 41 (3), 721–747.
Reynaud, B., 1998. Les propriétés des routines: outils pragmatiques de décision et modes Zahra, S.A., George, G., 2002. Absorptive capacity: a review, reconceptualization, and
de coordination collective. Sociol. Trav. 465–477. extension. Acad. Manag. Rev. 27 (2), 185–203.
Rodan, S., Galunic, D.C., 2004. More than network structure: how knowledge hetero- Zander, U., Kogut, B., 1995. Knowledge and the speed of the transfer and imitation of
geneity influences managerial performance and innovativeness. Strat. Manag. J. 25 organizational capabilities: an empirical test. Org. Sci. 6 (1), 76–92.
(6), 541–556. Zhang, M., Macpherson, A., Jones, O., 2006. Conceptualizing the learning process in
Rothaermel, F.T., Alexandre, M.T., 2009. Ambidexterity in technology sourcing: the SMEs: improving innovation through external orientation. Int. Small Bus. J. 24 (3),
moderating role of absorptive capacity. Org. Sci. 20 (4), 759–780. 299–323.
Rumelt, R.P., 1995. Inertia and transformation. In: Montgomery, C. (Ed.), Resource-based Zhang, X., Bartol, K.M., 2010. Linking empowering leadership and employee creativity:
and Evolutionary Theories of the Firm—Towards a Synthesis. Kluwer, Dordrecht, pp. the influence of psychological empowerment, intrinsic motivation, and creative
101–132. process engagement. Acad. Manag. J. 53 (1), 107–128.
Salvato, C., Rerup, C., 2011. Beyond collective entities: multilevel research on organi- Zheng, S.L., Zhang, W., Wu, X.B., Du, J., 2011. Knowledge-based dynamic capabilities
zational routines and capabilities. J. Manage. 37 (2), 468–490. and innovation in networked environments. J. Knowl. Manag. 15 (6), 1035–1051.
Schmickl, C., Kieser, A., 2008. How much do specialists have to learn from each other Zheng, S.L., Li, H.P., Wu, X.B., 2013. Network resources and the innovation performance
when they jointly develop radical product innovations? Res. Policy 37, 473–491. Evidence from Chinese manufacturing firms. Manag. Decis. 51 (6), 1207–1224.
Sidhu, J., Commandeur, H.R., Volberda, H.W., 2007. The multifaceted nature of ex- Zollo, M., Reuer, J.J., Singh, H., 2002. Interorganizational routines and performance in
ploration and exploitation: value of supply, demand, and spatial search for innova- strategic alliances. Org. Sci. 13 (6), 701–713.
tion. Org. Sci. 18, 20–38.

14
J. Guo, et al. Technological Forecasting & Social Change 155 (2020) 119990

Jianghua Zhou is an associate professor in the Business School, Beijing Normal University,
Jingjing Guo is an associate professor at Institutes of Science and Development, Chinese China. His research interests are mainly on innovation management and strategic man-
Academy of Sciences and School of Public Policy and Management, University of Chinese agement within the context of emerging markets, among others. He has published well
Academy of Sciences. Dr. Guo's research concentrates on topics such as patterns of over 30 articles in major refereed journals in business and management.
technological learning and knowledge flow, innovation management, and innovation
policy. She has published articles in journals such as Technovation, Technological
Forecasting and Social Change, International Journal of Technology Management, Asian Xiaobo Wu is the Qiushi Chair Professor of Strategy and Innovation Management, an
Journal of Technology Innovation, and International Journal of Mobile Communications. professor of School of Management, and Director of National Institute for Innovation
Management at Zhejiang University. He is also the Director of the Zhejiang University-
Cambridge University Joint Research center for Global Manufacturing and Innovation
Bin Guo is a professor at School of Management, Zhejiang University. His research in- Management, and a member of a number of governmental and professional bodies,
terests mainly cover innovation management and strategy, learning and capability providing consultancy services for the industries. His research interests include managing
building in catch-up. He has published articles in academic journals such as Research technological innovation, global manufacturing and competitive strategy. He has pub-
Policy, Technovation, Technology Analysis & Strategic Management, European lished articles in journals such as International Journal of Technology Management, Asian
Management Journal, and International Journal of Technology Management. Journal of Technology Innovation, International Journal of Mobile Communications, etc.

15

You might also like