Luigi M.

De Luca & Kwaku Atuahene-Gima

Market Knowledge Dimensions and

Cross-Functional Collaboration:
Examining the Different Routes to
Product Innovation Performance
There is consensus in the marketing literature that market knowledge and cross-functional collaboration are two
fundamental resources for successful product innovation. However, few studies examine the dimensions or
characteristics of market knowledge and how and why these resources influence product innovation performance.
Drawing on contingency theory and the knowledge-based view of the firm, the authors argue that knowledge
integration mechanisms may account for the effects of market knowledge dimensions (i.e., breadth, depth,
tacitness, and specificity) and cross-functional collaboration on product innovation performance. They find that
market knowledge specificity and cross-functional collaboration affect product innovation performance through
knowledge integration mechanisms. In contrast, whereas the effect of market knowledge depth is partially
mediated, market knowledge breadth has a direct, unmediated effect on product innovation performance. A test of
an alternative moderating perspective shows that the effects of market knowledge depth and cross-functional
collaboration on product innovation are negatively moderated by knowledge integration mechanisms. By showing
the differential effects of market knowledge dimensions on product innovation performance, the authors provide a
more refined understanding of the interplay among market knowledge, its integration, and the firms performance
in product innovation. The authors also conclude that by overlooking the role of knowledge integration mechanisms,
previous research may have provided an overly optimistic view of the value of cross-functional collaboration in
product innovation.

he marketing literature has established that product Schmidt 1997). Knowledge integration mechanisms refer to

T innovation performance is enhanced by three distinct

yet highly complementary factors: market knowledge
(Atuahene-Gima 1995, 2005; Day 1994; Li and Calantone
the formal processes and structures that ensure the capture,
analysis, interpretation, and integration of market and other
types of knowledge among different functional units within
1998), cross-functional collaboration (e.g., Griffin and the firm (Olson, Walker, and Ruekert 1995; Zahra, Ireland,
Hauser 1996; Kahn and Mentzer 1998, Song and Parry and Hitt 2000). The inability of firms to manage the inter-
1997), and knowledge integration mechanisms (hereinafter play of these factors lies at the root of many failures in
KIMs; see also Madhavan and Grover 1998; Maltz and product innovation. For example, Fisher, Maltz, and
Kohli 2000; Ruekert and Walker 1987). Market knowledge Jaworski (1997, p. 54) report several examples, such as
refers to the firms knowledge about its customers and com- Texas Instrumentss unsuccessful early entry into the desk-
petitors (Day 1994; Kohli and Jaworski 1990; Narver and top personal computer business, and Hewlett-Packards
Slater 1990). Cross-functional collaboration refers to the early entry into the laptop business with a technology-
degree of cooperation and the extent of representation by oriented 23-pound laptop.
marketing, research and development (R&D), and other Although the contributions of previous studies are sub-
functional units in the product innovation process (Kahn stantial, extant research is lacking in three respects. First,
1996; Li and Calantone 1998; Song, Montoya-Weiss, and the knowledge-based view (hereinafter KBV) of the firm
underscores the importance of several different dimensions
of market knowledge in product innovation: breadth, depth,
Luigi M. De Luca is a doctoral candidate, Management Department, Boc- specificity, and tacitness (see Galunic and Rodan 1998).
coni University (e-mail: luigi.deluca@unibocconi.it). Kwaku Atuahene-
Gima is Professor of Marketing and Innovation Management, China However, although studies in marketing emphasize the
Europe International Business School (CEIBS) (e-mail: kwaku@ceibs. importance of broad market knowledge as reflected in the
edu). The authors thank the three anonymous JM reviewers for their con- concept of market orientation (e.g., Atuahene-Gima 2005;
structive feedback. The work described in this article was supported by a Atuahene-Gima, Slater, and Olson 2005; Jaworski and
grant from the Research Grants Council of the Hong Kong Special Admin- Kohli 1993; Li and Calantone 1998), there is little or no
istrative Region, China (No. CityU 1263/03H) awarded to the second insight into the relative importance of the different dimen-
author.
sions of market knowledge as drivers of product innovation
To read and contribute to reader and author dialogue on JM, visit performance. Given the strategic importance of market
http://www.marketingpower.com/jmblog. knowledge, an approach that considers its dimensions and
parses out their distinct contributions seems appropriate if

2007, American Marketing Association Journal of Marketing

ISSN: 0022-2429 (print), 1547-7185 (electronic) 95 Vol. 71 (January 2007), 95112
we are to examine how market knowledge matters in prod- agers may need to reevaluate their stance toward the role of
uct innovation performance. market knowledge and cross-functional collaboration per se
Second, previous studies have focused almost entirely in product innovation. The findings also shed light on what
and separately on the effects of market knowledge (e.g., level of importance researchers and managers need to place
Atuahene-Gima 1995, 2005; Li and Calantone 1998) and on the inherent value of both resources. Figure 1 presents
cross-functional collaboration (e.g., Kahn 1996; Song, our conceptual model.
Montoya-Weiss, and Schmidt 1997) on product innovation
outcomes. No detailed explanations are offered as to how
and why KIMs matter in these relationships. Yet it is often Theory Development
implicitly assumed that KIMs are salient factors in trans- Contingency theory has been one of the major strands of
forming market knowledge and cross-functional activities thinking about firms and their structures and strategic
into product innovation performance (Griffin and Hauser actions (Galbraith 1973). Drazin and Van de Ven (1985)
1996; Hoopes and Postrel 1999). The current state of the lit- note two fundamental strands of contingency theory. The
erature is lacking and inconsistent because a key assump- first is the fit-as-mediation view (Venkatraman 1989),
tion of the KBV is that it is not knowledge itself but rather which posits that managers choose or adopt organizational
its integration among functional units in the firm that drives structures, processes, and strategies that reflect the particu-
sustainable competitive advantage (Grant 1996a). For this lar circumstances of their organizations (Galbraith 1973, p.
reason, product innovation is often characterized as a 2). In particular, because the organization is essentially an
process by which a firm transforms knowledge embedded information-processing network the objective of organi-
in cross-functional teams into new products (Madhavan and zational design is to achieve an efficient correspondence
Grover 1998, p. 2). Understanding how market knowledge between the information-processing requirements of its
and cross-functional collaboration are transformed into strategic contingencies and the information-processing
innovation outcomes through KIMs may shed light on the capabilities of its integration mechanisms (Galbraith 1973,
salience of this factor, which the KBV suggests is at the p. 6). Thus, this theory asserts that by increasing the
root of the firms competitive advantage. information-processing demands in the firm, strategic inter-
Third, KIMs have traditionally been positioned as mod- dependence or collaboration among functions dictate the
erators in the relationship between sources of information type and degree of organizational integration mechanisms
and the breadth and depth of the firms knowledge (Zahra, adopted to transfer knowledge within the firm (see Kumar
Ireland, and Hitt 2000) and between internal and external and Seth 1998; Thompson 1967).
capabilities and innovation outcomes (Zahra and Nielsen The KBV suggests that knowledge is sticky; in other
2002). Still, there is another view based on structural con- words, its characteristics make it difficult, costly, and uncer-
tingency theory that suggests that because knowledge char- tain to transfer and recombine within the firm (e.g., Galunic
acteristics (e.g., complexity) make communication flow dif- and Rodan 1998; Grant 1996a; Szulanski 1996). For this
ficult among functional units, knowledge becomes a key reason, Birkinshaw, Nobel, and Ridderstrle (2002) and
contingency factor that determines the nature of integration others (e.g., Germain and Droge 1997) argue that the char-
designs (Thompson 1967). Furthermore, the increased acteristics of knowledge complicate the process of transfer
information-processing requirements resulting from inter- by increasing uncertainty and ambiguity and, through this
dependence among functional units can be met only by cor-
responding increases in information-processing capacity FIGURE 1
through the design of coordination mechanisms (e.g., Gal- Conceptual Model of the Role of Cross-Functional
braith 1973; Thompson 1967). This view implies that mar- Collaboration, Market Knowledge Dimensions,
ket knowledge and cross-functional collaboration affect and KIMs in Product Innovation
performance through their effects on the design of KIMs.
Extant research in marketing has paid little attention to
these different perspectives of the role of KIMs. The moder-
ating view suggests that both market knowledge and cross-
functional collaboration are inherently valuable such that
KIMs determine the strength of their effect on innovation
performance. In contrast, the mediating view suggests that
these resources are not inherently valuable and that they
could affect innovation performance through their effects
on KIMs. Because marketing theory has placed consider-
able weight on the value of market knowledge and cross-
functional collaboration for effective product innovation
and because of the widely held belief among managers that
value creation involves knowledge integration among func-
tions, examining these perspectives is particularly important
in advancing this research stream. If the active role of
KIMs is as a mediator rather than as a moderator or if it
plays both mediator and moderator roles, marketing man-

96 / Journal of Marketing, January 2007

complication, increase the likelihood that firms will develop Knowledge about technology and other environmental
appropriate KIMs. In this view, knowledge characteristics properties is important, but only to the extent that it
are the fundamental quality to which managers adapt orga- enhances the understanding of customers and competitors
nizational integration structures. Consistent with the KBV, behavior. For this reason, Srinivasan, Lilien, and Ran-
market orientation theory indicates that market knowledge gaswamy (2002, p. 49) show that the firms ability to sense
generation is an outside-in process and that firms use inter- and respond to technology developments per se is different
nal processes to integrate the acquired knowledge (Day from its market orientation, that is, the ability to sense and
1994; Jaworski and Kohli 1993). If the KBV tenet is respond to customers and competitors. Thus, we define
acceptedthat is, it is not knowledge per se but rather its market knowledge as the firms knowledge of its customers
integration that affects competitive advantage (Grant 1991, behaviors and needs as well as its competitors behavior. We
1996b)it suggests that KIMs are mediators of the link examine the effects of four characteristics of market knowl-
among market knowledge dimensions, cross-functional col- edge on KIMs: breadth, depth, tacitness, and specificity.1
laboration, and firm performance. Knowledge breadth refers to the number of different
The second strand of contingency theory is the inter- knowledge domains with which the firm is familiar (Bierly
active fit argument, or fit-as-moderation view (Venkatra- and Chakrabarti 1996). Prabhu, Chandy, and Ellis (2005, p.
man 1989). This view proposes that the firms performance 116) offer a similar view, referring to knowledge breadth as
is attributable to a match between its strategic behaviors the range of fields over which the firm has knowledge. We
and the internal and external environment conditions define market knowledge breadth as the firms under-
(Atuahene-Gima and Murray 2004; Zeithaml, Varadarjan, standing of a wide range of diverse customer and competi-
and Zeithaml 1988). This suggests that KIMs moderate the tor types and factors that describe them. In other words, a
effect of market knowledge and cross-functional collabora- firm is said to have broad market knowledge if it has knowl-
tion on performance. Given the limited attention the mediat- edge of a wide variety of current and potential customer
ing view of KIMs has received, we focus on this perspec- segments and competitors and also uses a diverse set of
tive, though we also empirically examine the moderating parameters related to customers (e.g., needs, behaviors,
view. characteristics) and competitors (e.g., products, markets,
strategies) to describe and evaluate them (e.g., Zahra, Ire-
Effect of Market Knowledge Dimensions on KIMs land, and Hitt 2000). This concept highlights the broad
KIMs. According to Barki and Pinsonneault (2005, p. understanding of customers and competitors that is deeply
166), integration captures the extent to which distinct but rooted in the market orientation construct.
interdependent functional units, departments, and resources Firms with a broad knowledge base have a greater
are coordinated or made to constitute a unified whole. The potential to recombine different elements of the knowledge
task of integration or coordinating interfunctional inter- to improve opportunity recognition and creative potential
actions is often accomplished by the use of structures and (Kogut and Zander 1992). However, knowledge breadth
processes (see Griffin and Hauser 1996; Sobek, Liker, and might hamper knowledge recombination because the high
Ward 1998). As Olson, Walker, and Ruekert (1995, p. 49, degree of heterogeneity of knowledge elements contributes
emphasis in original) note, integration mechanisms are lat- to the complexity of transfer across functional units (Galu-
eral linkage devices or structural coordination nic and Rodan 1998). In contrast, limited market knowledge
mechanisms that firms use to coordinate cross-functional can be easily disseminated to and internalized by members
interactions (see also Griffin and Hauser 1996, p. 203). Fol- of different departments. Furthermore, because customer
lowing Zahra, Ireland, and Hitt (2000, p. 930), we define segments and competitors strategies change over time,
KIMs as structures and processes, such as the use of docu- broad market knowledge may involve frequent and numer-
mentation, information-sharing meetings, analysis of suc- ous variations that further increase the difficulties of cross-
cessful and failed projects, project reviews, and briefings by functional transfer. In addition, when market knowledge is
external experts and consultants, that ensure the capture, broad, not all of it can be usefully employed in every inno-
analysis, interpretation, and combination of knowledge vation project. Bringing in marginally useful information or
within the firm. These structures and processes enable man- leaving out relevant information might be detrimental to
agers to determine systematically and understand what has performance (Leonard-Barton 1992). Thus, the increased
been learned in the product innovation process, to interpret difficulties in sharing broad knowledge lead managers to
and articulate the importance of the learning, and to devise develop KIMs to provide the necessary information-
ways to exploit the knowledge competitively (Zahra and processing capacity for the organization. Empirically, Ger-
Nielsen 2002). main and Droge (1997) find that the breadth of the firms
knowledge about a task is an antecedent to processes for
Market knowledge dimensions. Information about the knowledge integration because these processes are the nec-
market environment, particularly about customers and com-
petitors, is the source of stimulation for the firms knowl-
edge (Day 1994; Nonaka 1994, p. 27) and the driver of a
1Galunic and Rodan (1998) identify another characteristic of
market-oriented strategy (Day and Nedungadi 1994, p. 32).
knowledge, namely, knowledge dispersion. We contend that this
This implies that a firm that correctly identifies, collects, characteristic is reflected in the construct of market knowledge
and uses information about customer and competitor condi- breadth because it captures different elements underlying market
tions is deemed to be knowledgeable about the market. knowledge (Turner and Makhija 2006).

Market Knowledge Dimensions and Cross-Functional Collaboration / 97

essary mechanisms to combat excessive compartmentaliza- projects, and frequent advice from experts (Galunic and
tion of the diverse knowledge. Thus: Rodan 1998; Szulanski 1996). As an embedded knowledge,
H1: The broader the market knowledge, the greater is the use tacit knowledge requires the development of KIMs to
of knowledge integration mechanisms. unearth its potential value (Madhavan and Grover 1998).
Birkinshaw, Nobel, and Ridderstrle (2002) find that the
Prabhu, Chandy, and Ellis (2005) describe technical degree of observability (i.e., explicitness) of technology
knowledge depth as the amount of within-field knowledge knowledge is negatively related to knowledge integration
the firm possesses. McEvily and Chakravarthy (2002) also structures. Thus:
argue that complex knowledge reflects the degree to which
H3: The more tacit the market knowledge, the greater is the
knowledge consists of many different, unique, and inter-
use of KIMs.
dependent elements, such that knowing how one element
works reveals little about how the different elements work Market knowledge specificity refers to the extent to
together. Consequently, we define market knowledge which the firms knowledge is tailored to the requirements
depth as the level of sophistication and complexity of a of specific contexts in which it is maximally effective but
firms knowledge of its customers and competitors. It cap- loses its value in other contexts (Galunic and Rodan 1998).
tures the level of sophistication with which the firm is able For example, a firms knowledge can be related to the atti-
to connect the unique and interdependent relationships tudes of a specific customer segment toward a specific
among the factors that describe key issues about customers product or strategy. Specific competitor knowledge may
and competitors. Knowledge of the interdependencies of result from an in-depth analysis of the behavior, products,
elements such as customers needs, behaviors, and prefer- and strategies of a particular competitor. In both cases, the
ences and competitors products and strategies indicates knowledge acquired is valuable only in the context of the
that a firm has a deep understanding of its market. Thus, focal firm. For example, Lenovos knowledge of the com-
whereas breadth captures the horizontal dimension of puter market was so specific to the Chinese context that it
knowledge, depth captures the vertical dimension. needed to acquire IBMs personal computer division to
There are two reasons market knowledge depth should increase its ability to compete on the global level.
affect KIMs. First, deep market knowledge implies high and Specific market knowledge is likely to be acquired and
complex interdependencies among the knowledge elements used by experienced people and experts in specific market
(McEvily and Chakravarthy 2002). The transfer of such domains. Thus, it could be argued that the transfer of such
knowledge is error prone and involves a greater risk of mis- knowledge is unproblematic even in the absence of KIMs.
interpretation and misapplication in product innovation However, we have three reasons to predict a positive effect
(Galunic and Rodan 1998). This is because deep functional of specific market knowledge on KIMs. First, because spe-
knowledge limits the firms ability to draw new conclusions cific market knowledge is valuable in the context in which it
and find new links among diverse pieces of knowledge. has been generated, its value is highly time sensitive. Subra-
Second, deep market knowledge implies differential func- mani and Venkatraman (2003) show that because specific
tional expertise in collecting and disseminating market market knowledge must be exploited in a timely manner, it
information. This leads to different thought worlds engenders strong norms of joint decision making, a key
(Leonard-Barton 1992), which further increase the uncer- aspect of KIMs. Second, specific market knowledge dimin-
tainty and ambiguity of knowledge transfer (Szulanski ishes knowledge recombination, impairing timely and
1996). For these reasons, deep knowledge tends to lead to effective contextual use (Galunic and Rodan 1998, p. 1197).
rigidities in knowledge transfer and sharing, hindering the Third, specific market knowledge reflects the use of idio-
firms ability to assimilate knowledge. Consequently, firms syncratic routines in the collection and use of market infor-
develop coordinating structures to provide functional units mation by different functions. As Galunic and Rodan
with the means to understand and develop confidence in (1998) argue, by producing specific knowledge, these rou-
one anothers analysis and interpretation (Hoopes and tines make the transfer of knowledge to other functions dif-
Postrel 1999). Thus: ficult. These properties of specific market knowledge
H2: The deeper the market knowledge, the greater is the use of
increase the implementation of KIMs to ensure early settle-
KIMs. ment of communication difficulties and enable timely flow
and recombination of knowledge from different functional
Market knowledge tacitness is the extent to which units in the product innovation process.
market knowledge is not explicit but rather is difficult to
H4: The more specific the market knowledge, the greater is the
codify and communicate (e.g., Nonaka 1994). Market
use of KIMs.
knowledge is tacit when people and functional units find it
difficult to articulate explicitly what they know about cus-
tomers and competitors and are unable to explain effec- Effect of Cross-Functional Collaboration on KIMs
tively the causal relationships between their actions and the Cross-functional collaboration refers to the degree of coop-
associated outcomes. Tacitness slows the internal transfer of eration, the extent of representation, and the contribution of
market knowledge because tacit knowledge cannot be fully marketing, R&D, and other functional units to the product
codified and articulated even by an expert. It can be trans- innovation process (Kahn 1996; Li and Calantone 1998;
ferred from one person to another only through a long Ruekert and Walker 1987; Song, Montoya-Weiss, and
process of apprenticeship, which necessarily involves face- Schmidt 1997). Cross-functional collaboration is intangible,
to-face interactions, review of successful and unsuccessful volitional, and unstructured in that it reflects only the recog-

98 / Journal of Marketing, January 2007

nition by functional units of their strategic interdependence [must] evolve to help reduce the uncertainty and ambiguity
and their need to cooperate for the benefit of the organiza- of resource, work, and assistance flows. Similarly, Garud
tion (Galunic and Rodan 1998, p. 1198; Kahn 1996; Kahn and Nayyar (1994, p. 372) point out that firms develop inte-
and Mentzer 1998; Olson et al. 2001). It ensures the align- gration mechanisms because of the uncertainty and ambigu-
ment of goals among functional units. In contrast, KIMs ity in translating embedded collective knowledge into
permit regular patterns of interactions that enable the trans- knowledge embodied in the new product (see also Griffin
fer, recombination, and use of knowledge from different and Hauser 1996, p. 209). In brief, we argue that increased
functions. Therefore, KIMs respond to the problem of coor- functional collaboration leads to the greater use of KIMs to
dinating different knowledge elements to execute complex regulate communication flow and learning in new product
organizational tasks. projects.2
Anecdotal evidence supports the distinction between H5: The greater the cross-functional collaboration, the greater
cross-functional collaboration and KIMs. First, despite the is the use of KIMs.
high degree of cooperation among its functional units, Toy-
ota established coordinating mechanisms, including stan- Market Knowledge, Cross-Functional
dardized reporting and documentation, formalized work Collaboration, and Product Innovation
processes (e.g., project reviews), problem-solving meetings, Performance
and integrative leaders, to ensure knowledge sharing and
integration among its different units (Sobek, Liker, and Market knowledge. Extant research considers market
Ward 1998). Second, a study of the most innovative firms in knowledge the fundamental driver of product innovation
the world indicates that two such firms, Southwest and performance (Atuahene-Gima 1995, 2005; Li and Calan-
BMW, have adopted mechanisms, such as colocation, face- tone 1998; Moorman and Miner 1997). However, few stud-
to-face meetings, and standard documentation, to integrate ies have provided detailed insights into the effects of differ-
the knowledge among members of their cross-functional ent market knowledge characteristics on product innovation
teams, despite their high degree of cooperation proclivity performance. Market knowledge breadth engenders product
(BusinessWeek 2006). The distinction of the two constructs innovation performance because it increases the firms abil-
is salient because coordinating knowledgeparticularly, ity to make connections among disparate market informa-
specialized and complex knowledgeamong different units tion, ideas, and concepts to gain broader and insightful per-
is problematic for firms, even when perfect goal congruence spectives (Reed and DeFillippi 1990). This logic underpins
(or full collaboration) is established among them (Grant the general idea about the positive role of market orienta-
1996b, p. 114). tion in product innovation (Atuahene-Gima 1995, 2005; Li
Contingency theory offers two principal reasons that and Calantone 1998). A firm with broad market knowledge
cross-functional collaboration predicts KIMs. First, cross- has heterogeneous information and understanding of cus-
functional collaboration implies increased resource depen- tomers and competitors, enabling it to design products that
dency among functional units and, thus, a greater need for match the diverse needs of its customer segments. By
enhanced information-processing capability to coordinate enhancing the firms purview of the market, broad market
the acquired knowledge. As the need for sharing knowledge knowledge also increases the firms ability to implement
and other resources among functional units increases, their and execute complex tasks quickly in product innovation
interdependence grows along with the volume of resource processes (Kogut and Zander 1992). It does this by enhanc-
flows and increases the use of coordinating mechanisms ing the chances of what Prabhu, Chandy, and Ellis (2005)
(see Olson, Walker, and Ruekert 1995, p. 53). Thus, call happy accidents, in which concepts from different
increased collaboration represents a critical strategic contin- knowledge domains are applied in unexpected ways.
gency for the design of coordination mechanisms (Galbraith A new product based on deep market knowledge limits
1973; Kumar and Seth 1998, p. 581). As Kahn and Mentzer a competitors ability to observe and understand the whole
(1998) suggest, collaboration reflects the willingness of dif- set of distinct and interdependent knowledge elements that
ferent functional units to cooperate; yet firms need to pro- underlies it (Reed and DeFillippi 1990). This is because
vide structural mechanisms to put such willingness into market knowledge depth reflects a complex understanding
action. Indeed, Birkinshaw, Nobel, and Ridderstrle (2002) of the causal interdependencies among customer problems
find that functional interdependence influences the design and requirements and potential competitor strengths and
of interunit KIMs in the firm.
2The direction of causality can be questioned here. However, in
Second, the transfer and flow of knowledge among
addition to structural contingency theory, several other arguments
interdependent units is often costly, ambiguous, and uncer- support our interpretation. First, rational managers would invest in
tain because of the diversity of functional information, costly KIM processes only when they are satisfied that different
backgrounds, experiences, and thought worlds. This com- functions are willing and able to collaborate. Second, rational
plicates analysis and interpretation and hinders the likeli- managers know that structures such as KIMs may be ineffective in
hood of novel recombinations of the firms knowledge imposing collaboration, which is more of a human and social con-
(Galunic and Rodan 1998). For this reason, Madhavan and dition. Indeed, prior research suggests that an attempt to impose
cross-functional collaboration through an early implementation of
Grover (1998) argue that the collective knowledge of func-
KIMs can yield the opposite result (Kahn 1996). Xie, Song, and
tional units constitutes only potential knowledge. As a Stringfellow (2003, p. 233) also note that merely imposing a
result, Ruekert and Walker (1987, p. 6) argue that to obtain cross-functional structure for product innovation cannot ensure
value from cross-functional interactions, mechanisms that different functional units will collaborate.

Market Knowledge Dimensions and Cross-Functional Collaboration / 99

responses, thus increasing the likelihood of the emergence constitute factors that determine the firms integration
of new ideas that are highly unique to the firm (Galunic and mechanisms (Birkinshaw, Nobel, and Ridderstrle 2002;
Rodan 1998). As McEvily and Chakravarthy (2002) find, Galbraith 1973; Germain and Droge 1997; Kumar and Seth
competitor efforts to reconstruct such a product are likely to 1998; Thompson 1967). Indeed, the KBV suggests that it is
be undermined or lead to partial and erroneous results. not knowledge per se but rather its integration that ensures
Prabhu, Chandy, and Ellis (2005, p. 116) also contend that competitive advantage (Grant 1991, 1996a). Specifically,
knowledge depth results in superior products because it KIMs ensure better performance because they enhance the
allows for the cross-fertilization of knowledge elements that likelihood of finding solutions in the product innovation
increases flexibility in generating complex patterns of new process by infusing the functional units with collective
knowledge. learning capability. Implicit in this argument is the notion
Tacit market knowledge leads to better new product per- that KIMs connect knowledge to performance. Because
formance because it allows for differences in cross- KIMs are mandated processes for learning, they provide
functional logics (Galunic and Rodan 1998). Prahalad and milestones that ensure a sense of order in cross-functional
Bettis (1986) suggest that such differential logics ensure knowledge sharing, use, and learning. By providing a for-
thoughtful deliberations and generate new perspectives, mal structure for knowledge integration, KIMs also provide
novel strategic alternatives, analyses, and interpretations a common forum for periodic feedback, which ensures
that enhance the efficiency of the firms innovation activi- quality decision making and completeness of a project
ties. Furthermore, tacit knowledge develops in a commu- teams activities. In so doing, KIMs reduce wasteful explo-
nity of interaction (Nonaka 1994) and thus is highly rations and errors in product innovation (Sheremata 2000).
embedded in the firms social system. Competitors find it Thus, dimensions of market knowledge and cross-
difficult to imitate the social context within which the focal functional collaboration affect product innovation perfor-
firm develops its new products (Reed and DeFillippi 1990). mance indirectly through their effects on KIMs. This dis-
Finally, specific market knowledge enhances innovation cussion suggests the following hypotheses:
performance because it ensures a long-term relationship H8: The greater the KIMs, the better is the product innova-
with specific contexts (e.g., customer segment) that gener- tion performance.
ates highly idiosyncratic insights for product innovation. H9: KIMs mediate the effects of market knowledge (a)
This protects the new products the firm develops from imi- breadth, (b) depth, (c) tacitness, and (d) specificity on
tation because competitors lack the contextual customer product innovation performance.
knowledge that went into the process. Furthermore, a new H10: KIMs mediate the effect of cross-functional collaboration
product based on specific market knowledge is more likely on product innovation performance.
to be differentiated to suit the specific customer and com-
petitor conditions (McEvily and Chakravarthy 2002).
Research Methods
Cross-functional collaboration. The positive effect of
cross-functional collaboration on product innovation perfor- Sample and Data Collection
mance is well documented in the literature (see Griffin and
Hauser 1996; Luo, Slotegraaf, and Pan 2006; Song and China is an ideal context for this study. The extreme com-
Parry 1997). The logic is that cross-functional collaboration plexity and dynamism of this transitional environment
ensures that marketing, technical, and other functional means that firms must confront not only the challenges of
capabilities are combined to develop a product that satisfies new (and often dysfunctional) competition but also collaps-
customer needs. It accomplishes this by improving the effi- ing capabilities (Li and Atuahene-Gima 2001). For this rea-
ciency of knowledge use and allowing for quality decision son, cross-functional collaboration and mechanisms for
making in new product project teams (Madhavan and using market knowledge in product innovation are critical
Grover 1998). for firms to sustain innovation performance. The instrument
was prepared in English and then translated into Chinese. It
H6: The greater the market knowledge (a) breadth, (b) depth, was checked for accuracy in line with the conventional
(c) tacitness, and (d) specificity, the better is the product back-translation process. We pretested the instrument with
innovation performance.
25 managers who had at least three years of business
H7: The greater the cross-functional collaboration, the better is
the product innovation performance.
experience in China to examine the face validity and to
assess informants understanding of the survey questions.
The study used a sample of 750 firms selected randomly
Mediating Effect of KIMs on Product Innovation from a mailing list of 2500 high-technology firms provided
Performance by a local consulting firm.
The direct-effects arguments for the impacts of market As with previous studies in China (Atuahene-Gima
knowledge and cross-functional collaboration on product 2005; Li and Atuahene-Gima 2001), we collected the data
innovation performance are persuasive. However, a careful on-site. An interviewer scheduled appointments with the
inspection indicates that the arguments implicitly assume a marketing manager/director as a first informant, who then
role for KIMs. As we argued previously, both the KBV and nominated a second knowledgeable informant. The inter-
contingency theory suggest that the stickiness of market viewer presented the key informants with the survey ques-
knowledge and the increased information-processing tionnaire and collected the questionnaire on completion. We
demands that result from cross-functional collaboration motivated respondents by assuring them of confidentiality

100 / Journal of Marketing, January 2007

and by offering a summary of the research results and a free Cross-functional collaboration. On the basis of prior
workshop on the research findingsinformation that would studies (e.g., Li and Calantone 1998), we measured cross-
be meaningless to them in the absence of accurate data (Li functional collaboration with three items that reflected the
and Atuahene-Gima 2001). We received 363 usable ques- extent of cooperation among functions. For example, we
tionnaires, for a response rate of 48%. We compared a sam- asked the respondents to rate the extent to which different
ple of 50 participating firms with a sample of nonparticipat- functional units cooperate in establishing goals and priori-
ing firms for which we had data on sales, R&D expense, ties for the firms product innovation.
and number of employees. Analyses of variance indicated KIMs. Two previous studies (Zahra, Ireland, and Hitt
no significant differences between the two groups on num- 2000; Zahra and Nielsen 2002) informed the measure of
ber of employees (F = 1.01), sales in the most recent year KIMs. We used five items that addressed the extent to
(F = .89), and R&D expenditure as percentage of sales (F = which a firm uses a set of formal processes (e.g.,
.98). information-sharing meetings, formal analysis of projects)
Of the first informants, 90% were from marketing and to capture, interpret, and integrate knowledge.
sales, 4% were chief executive officers/general managers,
and 6% were product development managers. These infor- Product innovation performance. On the basis of
mants had a mean industry experience of 8.18 years and a Atuahene-Gima, Slater, and Olsons (2005) research, we
mean firm experience of 4.77 years. Of the second infor- measured product innovation performance with five items
mants, 73% were product development director/managers, that asked respondents to indicate the extent to which the
9% were from marketing/sales, 11% were chief executive firm has achieved its product development objectives, such
officers/general managers, and 7% were from R&D. These as market share and profitability.4
informants had a mean industry experience of 7.76 years Control variables. In testing our hypotheses, we con-
and a mean firm experience of 4.55 years. We also exam- trolled for firm size and slack, which reflect the firms
ined the quality of the informants by asking them to indi- resources and market power to exploit existing competen-
cate on a ten-point scale their degree of knowledge of (1 = cies, build new ones, and develop innovations (Chandy and
not at all knowledgeable, and 10 = extremely knowl- Tellis 1998). To prevent skewness, we measured firm size
edgeable) and involvement in (1 = no involvement, and with the logarithm of the number of employees. We mea-
10 = very high involvement) product development issues sured slack with three items that tapped the availability of
in the firm. The means for the first and second informants, excess resources to fund new projects. We controlled for
respectively, were 7.55 and 7.40 for the first item and 8.00 environmental uncertainty because the KBV suggests that
and 7.78 for the second item. Using a multigroup analysis, knowledge integration becomes even more critical in uncer-
we obtained the following results: unconstrained confirma- tain and competitive environments (Grant 1996a). Further-
tory factor analysis (CFA) model for the marketing and more, functional units tend to perceive higher inter-
nonmarketing groups: 2 = 2225.7, d.f. = 1440; constrained dependence and therefore are motivated to implement
model for marketing and nonmarketing groups with equal- integration mechanisms in uncertain environments (Ruekert
ity constraints for structural paths, variances, and covari-
ances: 2 = 2307.6, d.f. = 1540; 2 = 81.9, d.f. = 100; p > approach to the concept of market knowledge. Because the charac-
.10. Given these results of tests of invariance, we pooled the teristics of knowledge we describe here are the same for customer
and competitor knowledge, we did not have any theoretical ratio-
data for analysis. nale to expect differential effects of customer and competitor
knowledge on KIMs. Thus, to reduce the complexity of the model
Measures of Constructs and the burden on our informants, we chose a more aggregated
Market knowledge dimensions. We measured market measure of market knowledge.
4We took steps to ensure that respondents did not confuse this
knowledge breadth with four items that asked respondents,
for example, to evaluate their firms market knowledge on a measure of the achievement of the firms product innovation
objectives with its overall performance. We measured overall firm
continuum from limited to wide ranging (Zahra, Ire- performance relative to competitors with eight items: return of
land, and Hitt 2000). We measured market knowledge depth sales, growth in profit, return on assets, sales growth, market share
with four items that asked the respondents, for example, to growth, cash flow, overall operational efficiency, and reputation
evaluate their firms customer knowledge on a continuum for performance. We performed three discriminant tests for prod-
from basic to advanced (Zahra, Ireland, and Hitt 2000). uct innovation performance and firm performance. First, a CFA
We measured market knowledge tacitness with four items two-factor solution yielded the following: 2 = 351.20, d.f. = 64.
that addressed the degree to which a firms market knowl- When we constrained them into one factor, we obtained the fol-
lowing: 2 = 773.30, d.f. = 65. Therefore, we have a significant
edge could be formally documented, communicated, and change in chi-square of 422.10 (d.f. = 1, p < .000), suggesting dis-
learned without personal experience (Szulanski 1996). On criminant validity. Second, we performed discriminant tests using
the basis of Reed and DeFillippis (1990) work, we mea- Fornell and Larckers (1981) approach. The average variance
sured market knowledge specificity with three items that extracted for both product innovation performance (.72) and firm
reflected the extent to which market knowledge is specific performance (.71) was greater than the squared correlation
to the firms environment.3 between them (.66). Finally, the correlation between product inno-
vation performance and firm performance was .81, with a standard
error of .02. The two-standard-error confidence interval for this
3Unlike other studies (e.g., Atuahene-Gima 2005; Li and Calan- correlation estimate is between .77 and .85. Because it does not
tone 1998; Narver and Slater 1990) that differentiate between cus- include 1, discriminant validity is demonstrated (Anderson and
tomer and competitor knowledge, we take a holistic measurement Gerbing 1988).

Market Knowledge Dimensions and Cross-Functional Collaboration / 101

and Walker 1987). We examine two aspects of uncertainty. invariance of items and latent variables means between the
First, market uncertainty reflects the speed of change in two groups. Following Steenkamp and Baumgartners
customer needs and preferences and in competitor actions; (1998) guidelines, we first compared the raw score means
we measured this with three items that cover the speed of for each single item with a one-way analysis of variance
change of customer needs and competitive conditions. Sec- and found no statistically significant differences between
ond, technological uncertainty reflects the speed of change the two informant groups (.22 < F < .99). Then, we esti-
and instability of the technology environment; we measured mated two CFA models in which the latent variable means
this with four items that tap the speed and unpredictability were first freely estimated and then constrained to equality
of technological changes. (Srbom 1974). The results suggested the absence of sig-
Finally, we controlled for the degree of radicalness of nificant differences in latent variable means between the
the firms product innovation activities to capture the firms two groups of respondents (unconstrained model: 2 =
learning capability, which influences propensity for knowl- 2602.04, d.f. = 1478; constrained model: 2 = 2607.84,
edge sharing and product innovation outcomes (see d.f. = 1489; 2 = 5.80, d.f. = 11; p > .10). This indicates
McGrath 2001). We measured product innovation radical- that the pattern of factor loadings and the latent variables
ness with three items that assessed the newness of the firms means, variances, and covariances do not differ signifi-
innovation activities in terms of, for example, customers cantly across the two groups. As further evidence of consis-
targeted and products offered. The Appendix presents all tency between the two informant groups, we computed the
the measures and their sources. interrater agreement index (rwg) for each measure (James,
Demaree, and Wolf 1984). For constructs measured on the
Assessing the Reliability and Validity of Measures five-point scale, the index ranges from 1 to 1, indicating
We performed an exploratory factor analysis with Varimax minimum and maximum agreement respectively.6 The low-
rotation for data we obtained from both types of informants. est rwg index for the whole set of items was .83, indicating a
Both the first and the second informants responses resulted high level of agreement between the two informant groups
in 11 factors with eigenvalues greater than 1, accounting for (see the Appendix).
68.58% and 68.50% of the total variance, respectively. For Given these results, we ran a third CFA on the com-
both informants, all the items loaded cleanly on the bined items for the two informant groups. This model per-
expected factors, without significant cross-loadings. Then, formed reasonably well (2/d.f. = 1163.2/720 = 1.62, GFI =
we tested the measurement model for both informants with .86, CFI = .94, and RMSEA = .04; see the Appendix). On
a CFA, using AMOS 6.0. All the items loaded significantly the basis of Fornell and Larckers (1981) work, we tested
on the expected constructs, indicating convergent and dis- for discriminant validity of the constructs by examining
criminant validity of the measures. The fit indexes showed whether the square root of the AVE of each construct
that the model fit the data reasonably well for both the first (shown in the diagonal in Table 1) was greater than the
(2/d.f. = 1.56, goodness-of-fit index [GFI] = .87, compara- highest correlation between latent variables involving the
tive fit index [CFI] = .94, and root mean square error of focal construct (shown above the diagonal in Table 1). Mar-
approximation [RMSEA] = .04) and the second (2/d.f. = ket knowledge breadth had slightly weak discriminant
1.43, GFI = .88, CFI = .95, and RMSEA = .03) informant. validity. As a fallback, we followed Bagozzi, Yi, and
In each case the constructs had acceptable levels of reliabil- Phillipss (1991) recommendation and examined whether a
ity, average variance extracted (AVE), and discriminant one-factor model fit the data better than the hypothesized
validity.5 two-factor model for each pair of constructs involving mar-
Next, we submitted data from the two informant groups ket knowledge breadth. In each case, the chi-square of the
to two multigroup analyses to assess the invariance of the constrained one-factor model was significantly greater than
measurement model. First, we specified a model in which the chi-square for the unconstrained two-factor model, sug-
we constrained factor loadings, variances, and covariances gesting discriminant validity.7
in the two groups to be equal and compared it with a base-
6For the seven-point scale adopted for product innovation
line model with no equality constraints, using a chi-square
difference test (Byrne 2001). The results suggested the performance, 1.25 rwg 1.
7The results were as follows: market knowledge breadth versus
absence of significant differences in factors loadings, vari- market knowledge depth (unconstrained model: 2 [d.f.] = 138.26
ances, and covariances (unconstrained model: 2 = 2153.11, [19]; constrained model: 2 [d.f.] = 279.28 [20]; 2 [d.f.] =
d.f. = 1440; constrained model: 2 = 2218.32, d.f. = 1540; 141.02 [1]), market knowledge breadth versus market knowledge
2 = 65.21, d.f. = 100; p > .10). Second, we assessed the specificity (unconstrained model: 2 [d.f.] = 79.08 [13]; con-
strained model: 2 [d.f.] = 461.05 [14]; 2 [d.f.] = 381.97 [1]),
market knowledge breadth versus market knowledge tacitness
(unconstrained model: 2 [d.f.] = 89.83 [19]; constrained model:
2 [d.f.] = 544.78 [20]; 2 [d.f.] = 454.95 [1]), market knowl-
5The exceptions were the AVEs for KIMs and market knowl- edge breadth versus cross-functional collaboration (unconstrained
edge breadth for the second informant group, which were slightly model: 2 [d.f.] = 96.33 [13]; constrained model: 2 [d.f.] =
below the highest squared correlation with the other latent 250.23 [14]; 2 [d.f.] = 153.90 [1]), market knowledge breadth
variables, indicating weak discriminant validity. For these two versus KIMs (unconstrained model: 2 [d.f.] = 238.20 [26]; con-
constructs, we performed an additional test based on the work of strained model: 2 [d.f.] = 488.29 [27]; 2 [d.f.] = 250.09 [1]),
Bagozzi, Yi, and Phillips (1991). In all cases, the two-factor model market knowledge breadth versus technology uncertainty (uncon-
fit the data better than the constrained one-factor model. Detailed strained model: 2 [d.f.] = 95.89 [19]; constrained model: 2
results of these analyses are available on request. [d.f.] = 525.09 [20]; 2 [d.f.] = 429.20 [1]), market knowledge

102 / Journal of Marketing, January 2007

 TABLE 1
Correlation Matrix and Descriptive Statistics of Measures

1 2 3 4 5 6 7 8 9 10 11 12

1. Market knowledge breadth .70 .70 .09 .07 .57 .58 .47 .23 .17 .45 .54 N.A.
2. Market knowledge depth .56 .71 .01 .04 .62 .53 .41 .19 .06 .43 .35 N.A.
3. Market knowledge tacitness .07 .02 .78 .23 .11 .18 .07 .31 .23 .03 .12 N.A.
4. Market knowledge specificity .06 .04 .24 .77 .20 .02 .15 .17 .27 .31 .20 N.A.
5. KIMs .49 .51 .10 .20 .69 .62 .43 .30 .13 .54 .34 N.A.
6. Cross-functional collaboration .44 .42 .12 .04 .53 .72 .27 .10 .03 .36 .39 N.A.
7. Product innovation performance .41 .41 .04 .11 .39 .26 .86 .23 .17 .35 .24 N.A.
8. Technological uncertainty .19 .19 .28 .14 .28 .12 .20 .75 .58 .19 .26 N.A.
9. Market uncertainty .18 .08 .24 .22 .13 .07 .14 .49 .73 .22 .40 N.A.
10. Organizational slack .38 .38 .02 .25 .45 .34 .31 .18 .19 .80 .36 N.A.
11. Radical innovation .42 .29 .11 .17 .29 .31 .19 .22 .32 .31 .73 N.A.
12. Firm size .07 .12 .04 .17 .06 .02 .24 .01 .02 .07 .03 N.A.

Number of items 4 4 4 3 5 3 5 4 3 3 3 N.A.

M 3.69 3.60 3.39 3.03 3.59 3.65 4.60 3.33 3.33 3.32 3.40 4.73
SD .60 .58 .76 .84 .68 .61 1.08 .74 .69 .76 .71 1.50
Skewness .14 .04 .31 .01 .64 .17 .52 .16 .26 .30 .09 0.55
Kurtosis .61 .00 .05 .45 .69 .02 .22 .20 .01 .12 .06 0.07
Notes: The diagonal elements are square roots of the AVE. The upper-right triangle elements are the correlations among the latent variables
(). The lower-left triangle elements are correlations among the composite measures (unweighted mean of the items for each construct).
N = 363, except for firm size, which is N = 351. N.A. = not applicable.

Given these results, we used the average of the The results in Model 2 show that market knowledge
responses of the two informants for all the remaining analy- breadth (b = .26, p < .01), market knowledge depth (b = .18,
ses to reduce the potential for common method variance p < .01), and market knowledge specificity (b = .09, p <
(Slater and Atuahene-Gima 2004). Table 1 reports descrip- .05) have positive and significant effects on product innova-
tive statistics and correlations among the study variables. tion performance, whereas the effect of market knowledge
Note that we measured product innovation performance on tacitness is not significant. These results support H6a, b, d but
a seven-point scale. This provides a different psychological not H6c. Cross-functional collaboration is unrelated to prod-
frame to the informants compared with the other variables uct innovation performance; thus, H7 is not supported.
measured on five-point scales, thus hindering common Among the control variables, organizational slack (b = .16,
method bias. p < .01) and technology uncertainty (b = .16, p < .01) are
positively related to KIMs, but only firm size is related to
product innovation performance (b = .21, p < .01).
Analysis and Results
Mediating Effect of KIMs
Direct Effects
We followed the three-step regression procedure that Baron
We used regression analysis to test the direct effects of mar- and Kenny (1986) recommend to examine the mediating
ket knowledge characteristics and cross-functional collabo- role of KIMs. As we showed previously, market knowledge
ration on KIMs and product innovation performance. As breadth, depth, and specificity have positive and significant
Table 2 (Model 1) shows, market knowledge breadth (b = effects on product innovation performance. Furthermore, all
.16, p < .01), market knowledge depth (b = .20, p < .01), dimensions of market knowledge, except for tacitness, and
market knowledge specificity (b = .13, p < .01), and cross- cross-functional collaboration are positively related to
functional collaboration (b = .31, p < .01) have a positive KIMs. When KIMs are entered into Model 3 (Table 2), it
and significant effect on KIMs. These results support H1, shows a positive and significant effect on product innova-
H2, H4, and H5. The effect of market knowledge tacitness tion performance (b = .14, p < .05), in support of H8. The
on KIMs is not significant; thus, H3 is not supported. inclusion of KIMs leads to a slight decrease in the effect
size of market knowledge breadth (from .26 to .24) and of
breadth versus market uncertainty (unconstrained model: 2 market knowledge depth (from .18 to .15), but both remain
[d.f.] = 86.16 [13]; constrained model: 2 [d.f.] = 355.42 [14]; 2 significant, suggesting partial mediation. The effect of mar-
[d.f.] = 269.26 [1]), market knowledge breadth versus organiza- ket knowledge specificity on product innovation perfor-
tional slack (unconstrained model: 2 [d.f.] = 77.78 [13]; con- mance was not significant, suggesting full mediation.
strained model: 2 [d.f.] = 395.53 [14]; 2 [d.f.] = 317.75 [1]), Recall that market knowledge tacitness and cross-
market knowledge breadth versus innovation radicalness (uncon- functional collaboration are unrelated to product innovation
strained model: 2 [d.f.] = 83.75 [13]; constrained model: 2
[d.f.] = 260.34 [14]; 2 [d.f.] = 176.59 [1]), market knowledge
performance. Testing for mediation for these variables vio-
breadth versus product innovation performance (unconstrained lates Baron and Kennys (1986) first test condition. How-
model: 2 [d.f.] = 193.21 [26]; constrained model: 2 [d.f.] = ever, several recent studies in various fields of research have
535.17 [27]; 2 [d.f.] = 341.96 [1]). argued that this constraint may be relaxed without hamper-

Market Knowledge Dimensions and Cross-Functional Collaboration / 103

 TABLE 2
Results of Regression Analysis: Standardized Path Coefficients (t-Values)

KIMs Product Innovation Performance

Independent Variables Model 1 Model 2 Model 3

Control Variables
Organizational slack .16 .10 .08
(3.38)** (1.94) (1.53)
Radical innovation .02 .05 .04
(.41) (.87) (.83)
Firm size .04 .21 .21
(1.05) (4.52)** (4.42)**
Technology uncertainty .16 .08 .06
(3.52)** (1.43) (1.00)
Market uncertainty .07 .03 .04
(1.45) (.49) (.66)
Main Effects
Market knowledge breadth .16 .26 .24
(3.13)** (4.40)** (3.99)**
Market knowledge depth .20 .18 .15
(3.88)** (3.08)** (2.58)**
Market knowledge tacitness .01 .01 .01
(.13) (.26) (.27)
Market knowledge specificity .13 .09 .07
(3.00)** (1.79)* (1.43)
Cross-functional collaboration .31 .04 .01
(6.73)** (.73) (.07)
Mediating Effect
KIMs .14
(2.19)*

F value 30.82** 14.78** 14.02**

R2 .48 .30 .31
Adjusted R2 .46 .28 .29
R2 .01
F change 4.80*
*p < .05 (one-tailed test for hypotheses, and two-tailed test for control variables).
**p < .01 (one-tailed test for hypotheses, and two-tailed test for control variables).
Notes: N = 351.

ing the validity of the mediation analysis (see Preacher and tion (b = .08, t = 2.06, p < .05), in support of H9a, b, d and
Hayes 2004, p. 719; Shrout and Bolger 2002, pp. 42930; H10 but not of H9c.
Smith, Collins, and Clark 2005, pp. 35455).8 Specifically,
Sobels (1982) test enables the investigation of indirect Structural Equation Modeling
effects for independent variables, regardless of the signifi- We examined the robustness of the preceding results with
cance of their total effects on the dependent variable. With structural equation modeling (SEM). The first model
the exception of market knowledge tacitness, Sobels test (SEM1) examined the direct effects of the independent
indicated significant, indirect effects for market knowledge variables on product innovation performance with the path
breadth (b = .04, t = 1.71, p < .05), market knowledge depth from KIMs constrained to zero. The fit indexes (2 [d.f.] =
(b = .05, t = 1.86, p < .05), market knowledge specificity 1206.80 [751], CFI = .94, and RMSEA = .04) suggested a
(b = .02, t = 1.71, p < .05), and cross-functional collabora- good fit with the data. The second model (SEM2), which
involved a full mediation of the effects of the independent
variables by KIMs, also showed a good fit with the data (2
8For example, Shrout and Bolger (2002, p. 429) support recom- [d.f.] = 1228.10 [755], CFI = .94, and RMSEA = .04).
mendations to set aside the first step of Baron and Kennys (1986) Finally, based on modification indexes, a third partial medi-
approach, arguing that [b]ecause a test of the X Y association ation model (SEM 3), which allowed a direct effect of mar-
may be more powerful when mediation is taken into account, it ket knowledge breadth on product innovation performance,
seems unwise to defer considering mediation until the bivariate showed a good fit (2 [d.f.] = 1203.50 [754], CFI = .94, and
association between X and Y is established. Similarly, Preacher
and Hayes (2004, p. 719) note several errors that could occur by RMSEA = .04). Model comparisons with the chi-square
strict adherence to this condition and argue that it is possible to difference test indicated that SEM3 performed better than
find a significant, indirect effect even when there is no evidence both SEM1 (2 [d.f.] = 3.3 [3], p > .10) and SEM2 (2
for a significant, direct effect. [d.f.] = 24.6 [1], p < .001).

104 / Journal of Marketing, January 2007

 With the exception of a nonsignificant effect of market through KIMs. This is consistent with the structural contin-
knowledge depth, the structural paths in SEM3 were consis- gency theory argument that increased information-
tent with the results we obtained using the regression analy- processing demands that result from the interdependence of
sis, and KIMs (b = .18, t = 1.70, p < .05) and market knowl- functional units determine the degree to which KIMs are
edge breadth (b = .39, t = 3.50, p < .01) were significantly adopted. Although cross-functional collaboration and KIMs
related to product innovation performance.9 Following are important factors in the product innovation process, as is
Browns (1997), Shrout and Bolgers (2002), and others widely reported in the marketing literature, we offer the
(see n. 8) recommendations, we again tested the signifi- new insight that the latter factor is the route that makes the
cance of the specific mediation effects as follows: market former a more valuable resource in product innovation.
knowledge breadth (total effect b = .85, p < .05; direct In summary, our findings imply that market knowledge
effect = .83, p < .05; indirect effect b = .03, not significant specificity and cross-functional collaboration (and, in part,
[n.s.]), market knowledge depth (total effect b = .16, n.s.; market knowledge depth) are only potential resources and
direct effect = .09, n.s.; indirect effect b = .07, p < .10), mar- may not be inherently valuable for product innovation per-
ket knowledge specificity (total effect b = .14, n.s.; direct formance. Being distal antecedents, their potential for
effect = .10, n.s.; indirect effect b = .04, p < .10), and cross- enhancing product innovation performance may be realized,
functional collaboration (total effect b = .05, n.s.; direct totally or in part, through KIMs. However, the situation
effect = .17, n.s.; indirect effect b = .12, p < .10).10 appears to be more complex because the effect of market
Overall, the results of this study suggest four conclu- knowledge breadth on product innovation performance is
sions: First, market knowledge specificity and cross- direct, not through KIMs. Therefore, market knowledge
functional collaboration influence product innovation per- breadth appears to be the most potent driver of product
formance through KIMs. Second, market knowledge depth innovation performance among the different dimensions of
has both a direct and an indirect effect (through KIMs) on market knowledge we examined herein. This finding
product innovation performance. Third, market knowledge implies that types of market knowledge may have different
breadth has a direct, nonmediated effect on product innova- intrinsic properties, a nuanced insight that has not yet been
tion performance. Fourth, product innovation performance recognized in extant contingency theory and in the KBV.
is not influenced either directly or indirectly by market The nonsignificant effect of market knowledge tacitness
knowledge tacitness. on KIMs was a surprise because prior research has found
that it increases the difficulty of knowledge transfer. The
lack of a relationship between market knowledge tacitness
Discussion and product innovation performance echoes the mixed per-
Our goal in this article was to advance the marketing litera- spectives pertaining to the effect of tacitness in extant
ture by untangling the complex relationships among market research (see Galunic and Rodan 1998; Hedlund 1994;
knowledge dimensions, cross-functional collaboration, McEvily and Chakravarthy 2002). We conducted a series of
KIMs, and product innovation performance. We found that post hoc moderating tests with other variables in this study
market knowledge depth and specificity affect product but found no significant nonlinear or moderated effects of
innovation performance through KIMs. Note that though market knowledge tacitness on product innovation perfor-
the direct effect size of KIMs on product innovation perfor- mance. A possible explanation for our finding may be
mance is small relative to the effect of market knowledge gleaned from the work of Nonaka (1994), who argues that
breadth, it plays the additional role of linking not only mar- tacit knowledge is created by individuals and is initially
ket knowledge depth and specificity but also cross- shared within homogeneous communities of interaction,
functional collaboration to product innovation performance. such as functional units. To overcome problems of different
These results echo the view that being a sticky asset, knowl- thought worlds among such units, firms may rely on project
edge is a contingency variable and that the design for KIMs leaders with A-shaped capabilities to craft a unifying
needs to take into account the nature of the knowledge base vision that does justice to all the disciplines represented
of the firm (e.g., Birkinshaw, Nobel, and Ridderstrle 2002; (Madhavan and Grover 1998, p. 4). As Griffin and Hauser
Germain and Droge 1997). (1996, p. 209, emphasis added) note, These successful
We found no support for the direct, positive effect of developers may have internalized a process that allows
cross-functional collaboration on product innovation perfor- them to get product to market successfully. This implies
mance. Instead, we found that cross-functional collabora- that unlike other market knowledge dimensions, integration
tion positively affects product innovation performance of tacit market knowledge may be achieved through the
tacit skills of effective team leaders rather than through
formal KIMs. This appears to explain why firms such as
9Differences between estimated coefficients in regression analy- Toyota employ the integrative leader as one of the key
sis and SEM can be due to different estimation approaches (ordi- KIMs in product development (Sobek, Liker, and Ward
nary least squares versus maximum likelihood) or because regres- 1998, p. 40).
sion analysis does not account for measurement error. As we mentioned previously, in other contexts, scholars
10For indirect effects (both with Sobels test and SEM), we
report unstandardized coefficients and one-tailed significance lev-
have argued convincingly that KIMs are moderators. We
els. For SEM, we obtained total, direct, and indirect effects esti- evaluated this view and found significant interaction effects
mates and significance using the AMOS 6.0 bootstrap estimation only between KIMs and cross-functional collaboration (b =
procedure with 1000 resamples. .11, t = 1.95, p < .05) and market knowledge depth (b =

Market Knowledge Dimensions and Cross-Functional Collaboration / 105

.16, t = 2.93, p < .01).11 These findings are contrary to embrace a more fine-grained notion of market knowledge.
expectations, but they are novel. They suggest that KIMs Without this, marketing theory is unlikely to unearth new
play not only a mediation role but also an unexpected nega- insights into the role of market knowledge in product inno-
tive moderating role. vation, such as those we offer here.
Moorman and Miner (1997, p. 101) find that environ- Second, the results regarding the effects of market
mental turbulence can reduce the value of shared knowl- knowledge dimensions suggest that theoretical exploration
edge within organizations. Although they measure shared of the failure of firms in product innovation should not be
knowledge or organizational memory more broadly, the ascribed mainly to their failure in cross-functional collabo-
routine component of memory and our concept of KIMs ration, as suggested in much of the previous research and in
may overlap to some extent. For this reason, we tested the anecdotal reports. Rather, it may be possible for firms to fail
interaction effects of technology/market turbulence and even when they have effective and efficient cross-functional
KIMs on product innovation performance but found no sig- collaboration, if they do not have broad, deep, and specific
nificant results. A possible reason might be that we focus market knowledge. This study indicates that examining
more narrowly on structured mechanisms for knowledge only the direct effects of market knowledge and cross-
integration, whereas the concept of organizational memory functional collaboration separately may lead to an incorrect
included collective beliefs, physical artifacts, and behav- view of their power and, thus, to erroneous implications
ioral routines, which may interact more strongly with envi- about their role in product innovation.
ronmental turbulence. Third, we clarify how and why KIMs matter in product
innovation performance by showing simultaneously their
Theoretical Contributions mediating and moderating roles. We show that KIMs con-
This study contributes to marketing theory of product inno- vert some market knowledge dimensions and cross-
vation in five main ways. First, the direct effect of market functional collaboration into product innovation perfor-
knowledge breadth sheds new light on the importance mance. This new insight implies that by failing to consider
ascribed to the concept of market orientation in product the mediating role of KIMs, previous research may have
innovation and the importance that marketing theory places assumed away the information-processing demands in
on broad understanding of customers and competitors product innovation and thus may have reached a premature,
(Atuahene-Gima 1995, 2005; Jaworski and Kohli 1993; and perhaps overly optimistic, view of the importance of
Narver and Slater 1990). However, the indirect, positive market knowledge and cross-functional collaboration in
effects of market knowledge depth and specificity also product innovation. More important, these findings suggest
underscore the need for market orientation theory to an important qualification of the dual tenets of KBV,
namely, that knowledge is a necessary but not a sufficient
condition for competitive advantage and that effective inte-
gration of knowledge is the key driver of competitive
advantage. We show that KIMs play an important role in
11To test interactions with SEM, we followed the procedure that product innovation performance, as predicted by the KBV,
Ping (1995) suggests to build the interaction terms. First, for each by partially mediating the effects of market knowledge
pair of constructs X and Z forming an interaction, we mean- depth and by completely mediating the effects of market
centered all their indicators x1xn and z1zm. Second, for each pair knowledge specificity and cross-functional collaboration on
of constructs X and Z, we computed a single indicator that repre- product innovation performance. In other words, and con-
sents the latent product xz as (x1 + xn)(z1 + zm). Third, we sistent with the KBVs contention, market knowledge depth
specified a model including the two focal constructs and their
and specificity as well as cross-functional collaboration
interaction term. The loading and error for the latent product are
given by two equations: may not be intrinsically valuable; their value may be real-
ized through KIMs.
xz = (x1 + xn)(z1 + zm), and Fourth, contrary to the KBV tenet, market knowledge
xz = (x1 + xn)2VAR(X)(z1 + zm) breath is valuable in and of itself because it has a direct,
unmediated effect on product innovation performance.
+ (z1 + zm)2VAR(Z)(x1 + xn) Thus, it is a sufficient condition for product innovation per-
+ (x1 + xn)(z1 + zm), formance. These differential strengths of market knowledge
characteristics suggest that the KBV understates the inher-
for which all the terms on the right-hand side are taken from the ent value of some types of market knowledge. Thus, this
measurement model (Ping 1995). The fit indexes for the five mod-
els containing the relevant interaction terms are as follows: study suggests the need for further research to focus on the
KIMs market knowledge breadth interaction (2 [d.f.] = 1306.49 explicit articulation of the types of market knowledge and
[823], CFI = .94, and RMSEA = .04), KIMs market knowledge the consideration of their different roles in understanding
depth interaction (2 [d.f.] = 1325.09 [823], CFI = .94, and the firms effectiveness in product innovation. Without such
RMSEA = .04), KIMs market knowledge tacitness interaction an approach, a more nuanced understanding of the role of
(2 [d.f.] = 1298.10 [823], CFI = .94, and RMSEA = .04), KIMs market knowledge may be missed.
market knowledge specificity interaction (2 [d.f.] = 1327.92
Fifth, given the complexity of the study context, the
[823], CFI = .94, and RMSEA = .04), and KIMs cross-
functional collaboration interaction (2 [d.f.] = 1284.85 [823], negative moderating effects of KIMs suggest that at high
CFI = .94, and RMSEA = .04). levels, they could stifle the effects of market knowledge

106 / Journal of Marketing, January 2007

depth and cross-functional collaboration on product innova- formal knowledge integration processes may not achieve
tion performance. It appears that though some characteris- their intended objectives in product innovation. Toyota pro-
tics of market knowledge and cross-functional collaboration vides a good example of a company that values cross-
may make KIMs necessary, the degree of KIMs must be functional collaboration but takes the extra step of deploy-
tempered by the amount of contextual complexity the firm ing KIMs (Sobek, Liker, and Ward 1998). Finally, managers
faces and the resulting need for functional flexibility and, must be creative in balancing the need for KIMs engen-
perhaps, autonomy (Eisenhardt and Martin 2000). At high dered by the information-processing demands of cross-
levels, KIMs may stifle the flexibility and creativity of functional collaboration and market knowledge and the
cross-functional interactions and in the use of deep market need for functional flexibility in the collaboration and use
knowledge (Kumar and Seth 1998, p. 583). Although previ- of market knowledge. Our results suggest that KIMs are not
ous studies in other contexts have found positive moderat- without costs of implementation, which could trap the
ing roles for KIMs, the new insight we offer is that there unwary and thus lead to less effective use of market knowl-
may be a threshold of the KIM level beyond which market edge and cross-functional collaboration.
knowledge depth and cross-functional collaboration may
have detrimental effects on performance. Thus, the use of Limitations and Future Research Directions
KIMs appears to involve a trade-off between their necessity This study has several limitations that should be considered
occasioned by stickiness of market knowledge and the in the interpretations of the findings. First, we used cross-
information-processing demands of cross-functional collab- sectional data, which cannot suggest causal relationships. In
oration on the one hand and the implementation costs of particular, if the theoretical lens is organizational learning
KIMs on the other hand. This is a trade-off that has not been and knowledge creation, it could be theorized that KIMs are
uncovered in extant research. antecedents to the dimensions of market knowledge we dis-
cussed here (e.g., Zahra, Ireland, and Hitt 2000). However,
Managerial Implications our focus was on knowledge integration viewed from a
The argument that market knowledge and cross-functional structural contingency theory lens and, more specifically, on
collaboration enhance product innovation performance has the view of knowledge as a contingency variable (Birkin-
gained wide acceptance among practitioners. Our study shaw, Nobel, and Ridderstrle 2002; Germain and Droge
supports this conclusion but also qualifies it in several ways 1997; Olson, Walker, and Ruekert 1995). This perspective
for managers. First, our study calls on managers to consider suggests that the type of knowledge determines the nature
the attributes of the market knowledge they use in new of coordination mechanisms for knowledge integration
product projects to design KIMs properly. Marketing man- (Galbraith 1973). Future studies based on the former lens
agers are fervent adherents of the market orientation tenet could examine how KIMs affect the dimensions of market
with respect to the value of acquiring broad and comprehen- knowledge.
sive knowledge about customers and competitors. This Second, the generalizability of the results is limited
notion is justified given the research on the positive effects because we used data from a sample of firms in China.
of market orientation on product innovation outcomes. Third, extant research (Olson, Walker, and Ruekert 1995;
However, this study digs deeper into the features of market Sobek, Liker, and Ward 1998) and contingency theory
knowledge by finding that though broad market knowledge describe several integrative mechanisms (Galbraith 1973)
is important, managers need to pay equal attention to deep that we did not consider here and should be investigated in
and specific market knowledge. These dimensions of mar- further research for their ability to leverage market knowl-
ket knowledge appear to influence the design of KIMs, edge in product innovation. Fourth, our exclusive focus on
which in turn affect product innovation performance. Thus, market knowledge is a limitation, given the role of techno-
the new insight for managers is that broad market knowl- logical knowledge in product innovation. A more explicit
edge is inherently valuable for product innovation perfor- incorporation of the nature of technological knowledge
mance, but emphasizing breadth over other market knowl- along the dimensions used for market knowledge in future
edge characteristics and integration mechanisms may be studies may provide a better understanding of knowledge
detrimental for the achievement of the full potential of the integration and product innovation performance.
firms new product projects. Managers need to endow new Fifth, although we controlled for several factors that
project teams with the human and financial resources to account for variance in firms growth, market power, and
acquire and apply broad, deep, and specific market knowl- technological skills (i.e., firm size, slack, environmental
edge in new product development activities. In this respect, conditions, and radical innovation), we did not include
our measures of market knowledge characteristics could some potentially influential covariates that have been con-
serve as guides for managers who want to collect and use sidered in previous studies, such as industry maturity
customer and competitor knowledge in line with these (Eisenhardt and Tabrizi 1995), company age (Sinkula
characteristics. 1994), and R&D strength (Li and Calantone 1998). This
Second, the results suggest that it is the structured might limit the definitive evaluation of the relative impor-
knowledge integration processes that enable the translation tance of the antecedents of KIMs and product innovation
of cross-functional collaboration into better product innova- performance in the current study, and it offers a chance for
tion performance. This means that managers who encourage future researchers to take steps in this direction. Finally, two
cross-functional collaboration as an end in itself but neglect of our scales, market knowledge breadth and KIMs, did not

Market Knowledge Dimensions and Cross-Functional Collaboration / 107

meet the .50 threshold for AVE, suggesting the need for fur- functional collaboration involves potential bottlenecks, such
ther scale development. In addition to alleviating these lim- as functional competition, conflicts, and personality differ-
itations, there are other fertile avenues for further research. ences. Although not exhaustive, these aspects point to
First, our study highlights the key role of market knowl- important opportunities for further research to provide a
edge characteristics in enhancing KIMs and product inno- deeper examination of how cross-functional relationships
vation performance and underscores the central importance and knowledge affect innovation.
accorded to knowledge and its integration in the concept of The results of this study, along with those in previous
market orientation. However, despite the numerous studies studies, suggest that there is little understanding of the
on market orientation, scholars are yet to recognize the dis- value of tacit knowledge. McEvily and Chakravarthy (2002)
tinctions among different knowledge characteristics indicate that tacitness may involve two dimensions: charac-
espoused in the KBV. We argue that market-oriented firms teristic ambiguity and linkage ambiguity, the former dimen-
may differ with respect to the type of market knowledge sion being an antecedent of the latter dimension. Given the
they collect and use in their product innovation processes. KBVs arguments that tacitness is a key factor that could
Thus, we suggest that two firms with the same degree of erect imitation barriers to protect the firms advantage,
market orientation that have differential levels of market research is needed on whether these two dimensions apply
knowledge breadth, depth, tacitness, and specificity could to market knowledge and whether they offer similar or dif-
display markedly different effectiveness in product innova- ferential advantages in product innovation. Finally, further
tion. This implies that examining the role of market orienta- research should examine more closely the dynamic inter-
tion in product innovation without isolating and accounting play between market knowledge characteristics and techno-
for the fine details of market knowledge characteristics may logical knowledge characteristics in the linkage between
lead to an incomplete understanding or even misleading cross-functional collaboration and KIMs in product innova-
results. tion. We believe that given the potential synergies between
Second, our findings underscore the need for market and technological knowledge in product innovation,
researchers to examine factors that may be more proximate further research should tease out subtler, but still important,
antecedents of product innovation performance and, thus, combined effects of the different characteristics of these
potential drivers of the effect of cross-functional collabora- kinds of firm knowledge.
tion. Researchers need to move past their exclusive focus on In conclusion, this study challenges researchers and
the direct effects of cross-functional collaboration and begin managers to take a more sophisticated assessment of how
to examine factors that may moderate or mediate its role in and why market knowledge and cross-functional collabora-
product innovation. These include different types of firm tion affect product innovation outcomes. We believe that by
internal and external environments, such as internal com- delineating the differential relative effects of the dimensions
mitment and competitive uncertainty (Song and Parry of market knowledge and cross-functional collaboration and
1997); different managerial styles; different routines in the by showing the dual mediating and moderating roles of
product innovation process, such as product standards and KIMs, this study illuminates in a more systematic way how
market definition (Dougherty 1992); different functional such a goal can be achieved. With additional work that
relationship experiences; and so forth. Indeed, with the builds on this study and incorporates the role of technologi-
exception of Maltz and Kohli (2000), we and other scholars cal knowledge, a more comprehensive account of the com-
in the marketing literature seem to have a positive bias plexity of the processes by which firms develop successful
toward cross-functional collaboration. However, cross- product innovations can be achieved.

108 / Journal of Marketing, January 2007

 APPENDIX
CFA of Measures

Standardized
Factor
Measure and Source Description Loadings t-Value rwg
Product innovation Rate the extent to which your firm has achieved the following
performancea product development objectives:
(Atuahene-Gima, Market share relative to the firms stated objectives. .76 16.72 .91
Slater, and Olson 2005) Sales relative to stated objectives. .81 18.49 .92
AVE = .74 Return on assets relative to stated objectives. .93 23.12 .92
CR = .93 Return on investment related to stated objectives. .93 23.09 .92
= .94 Profitability relative to stated objectives. .87 20.76 .91

Market knowledge Compared to major competitors, our firms knowledge of

breadthb (Zahra, Competitors strategies is narrow vs. broad. .63 12.27 .90
Ireland, and Hitt 2000) Competitors strategies is limited vs. wide ranging. .60 11.55 .87
AVE = .49 Our customers is narrow vs. broad. .78 16.12 .90
CR = .79 Our customers is limited vs. wide ranging. .76 15.80 .91
= .81 Competitors strategies is specialized vs. general.f
Our customers is specialized vs. general.f

Market knowledge depthb Compared to our major competitors, our firms knowledge
(Zahra, Ireland, and Hitt about
2000) Competitors strategies is shallow vs. deep. .68 13.40 .90
AVE = .50 Competitors strategies is basic vs. advanced. .62 11.80 .89
CR = .80 This firms customers is shallow vs. deep. .76 15.58 .90
= .82 This firms customers is basic vs. advanced. .78 16.16 .88

Market knowledge Market knowledge competencies are difficult to

tacitnessc (Szulanski Comprehensively document in manuals or reports. .86 19.29 .84
1996) Comprehensively understand from written documents. .88 19.91 .85
AVE = .61 Identify without personal experience in using them. .68 13.93 .86
CR = .86 Precisely communicate through written documents. .68 13.94 .84
= .86 There is no precise list of market knowledge skills
necessary to successfully perform our activities.f

Market knowledge Please indicate your agreement with each of the following
specificityc (New items statements with respect to your firms market knowledge:
based on Reed and Our knowledge of customers and competitors is quite .66 13.16 .84
DeFillippi 1990) specific to our kind of business.
AVE = .60 It will be very difficult for an employee to transfer market .86 17.74 .84
CR = .82 knowledge acquired in our firm to other business
= .81 environments.
Our market knowledge and skills are tailored to meet the .79 16.23 .83
specific conditions of our business.
Our market knowledge largely depends on the human and
physical assets we have dedicated to acquiring information
about market conditions.f

KIMsd (Zahra, Ireland, To what extent does your firm use each of the following
and Hitt 2000; Zahra activities to capture, interpret, and integrate knowledge and
and Nielsen 2002) information about market and technology conditions?
AVE = .47 Regular formal reports and memos that summarize .72 14.84 .87
CR = .82 learning.
= .83 Information sharing meetings. .77 16.19 .88
Face-to-face discussions by cross-functional teams. .71 14.32 .88
Formal analysis of failing product development projects. .60 11.70 .85
Formal analysis of successful product development .59 11.46 .86
projects.
Use of experts and consultants to synthesize knowledge.f

Cross-functional In this organization different departments

collaboratione (Li and Cooperate fully in generating and screening new ideas for .75 14.81 .88
Calantone 1998) new products.
AVE = .52 Fully cooperate in establishing goals and priorities for our .79 14.96 .86
CR = .76 strategies.
= .75

Market Knowledge Dimensions and Cross-Functional Collaboration / 109

 APPENDIX
Continued

Standardized
Factor
Measure and Source Description Loadings t-Value rwg
Are adequately represented on project teams and other .61 11.52 .87
strategic activities.

Technological Please indicate your agreement with each of the following

uncertaintyc statements with respect to your firms environment:
(Jaworski and Kohli It was very difficult to forecast technology developments in .68 13.99 .83
1993) our industry.
AVE = .56 Technology environment was highly uncertain. .86 19.03 .83
CR = .83 Technological developments were highly unpredictable. .82 17.82 .84
= .83 Technologically, our industry was a very complex .59 11.54 .84
environment.

Market uncertaintyc Please indicate your agreement with each of the following
(Jaworski and Kohli statements with respect to your firms environment:
1993) Customer needs and product preferences changed quite .58 10.91 .88
AVE = .53 rapidly.
CR = .77 Customer product demands and preferences were highly .82 15.20 .87
= .76 uncertain.
It was difficult to predict changes in customer needs and .76 16.57 .85
preferences.
Market competitive conditions were highly unpredictable.f

Organizational slackc Please indicate your agreement with each of the following
(New scale) statements with respect to your firm:
AVE = .64 We have uncommitted resources that can be used to fund .72 14.89 .86
CR = .84 strategic initiatives at short notice.
= .84 We have a large amount of resources available in the short .88 19.57 .83
run to fund our initiatives.
We will have no problems obtaining resources at short .79 16.82 .83
notice to support new strategic initiatives.
We have a large amount of resources at the discretion of
management to fund new strategic initiatives.f

Radical innovatione To what extent does each of the following factors describe this
(McGrath 2001) firms product development?
AVE = .53 The products offered were new to the firm and the industry. .78 15.40 .88
CR = .77 The customer or client needs served were new to the firm. .72 14.04 .88
= .77 The users of the products or services were new to the firm. .69 13.23 .87
The new products were based on revolutionary changes in
technology.f
aSeven-point scale (1 = low, and 7 = high).
bFive-point semantic differential scales (anchoring points in italics).
cFive-point scale (1 = strongly disagree, and 5 = strongly agree).
dFive-point scale (1 = never used, and 5 = widely used).
eFive-point scale (1 = to no extent, and 5 = to great extent).
fItems were dropped from the scale during the measure purification phase.
Notes: CR = composite reliability.

REFERENCES
Anderson, James C. and David W. Gerbing (1988), Structural and Janet Murray (2004), Antecedents and Outcomes of
Equation Modeling in Practice: A Review and Recommended Marketing Strategy Comprehensiveness, Journal of Market-
Two-Step Approach, Psychological Bulletin, 103 (3), 41123. ing, 68 (October), 3346.
Atuahene-Gima, Kwaku (1995), An Exploratory Analysis of the , Stanley Slater, and Eric Oslon (2005), The Contingent
Impact of Market Orientation on New Product Performance: A Value of Responsive and Proactive Market Orientation on New
Contingency Approach, Journal of Product Innovation Man- Product Program Performance, Journal of Product Innovation
agement, 12 (4), 27593. Management, 22 (November), 46482.
(2005), Resolving the CapabilityRigidity Paradox in Bagozzi, Richard P., Youjae Yi, and Lynn W. Phillips (1991),
New Product Innovation, Journal of Marketing, 69 (October), Assessing Construct Validity in Organizational Research,
6183. Administrative Science Quarterly, 36 (3), 42158.

110 / Journal of Marketing, January 2007

Barki, Henri and Alain Pinsonneault (2005), A Model of Organi- (1996a), Prospering in Dynamically-Competitive Envi-
zational Integration, Implementation Effort and Performance, ronments: Organizational Capability as Knowledge Integra-
Organization Science, 16 (2), 16579. tion, Organization Science, 7 (4), 37587.
Baron, Reuben and David A. Kenny (1986), The Moderator- (1996b), Toward a Knowledge-Based Theory of the
Mediator Variable Distinction in Social Psychological Firm, Strategic Management Journal, 17 (Winter Special
Research: Conceptual, Strategic, and Statistical Considera- Issue), 109122.
tions, Journal of Personality and Social Psychology, 51 (6), Griffin, Abbie and John R. Hauser (1996), Integrating R&D and
117382. Marketing: A Review and Analysis of the Literature, Journal
Bierly, Paul and Alok Chakrabarti (1996), Generic Knowledge of Product Innovation Management, 13 (3), 191215.
Strategies in the U.S. Pharmaceutical Industry, Strategic Man- Hedlund, Gunnar (1994), A Model of Knowledge Management
agement Journal, 17 (Winter Special Issue), 12335. and the N-Form Corporation, Strategic Management Journal,
Birkinshaw, Julian, Robert Nobel, and Jonas Ridderstrle (2002), 15 (Summer Special Issue), 7390.
Knowledge as a Contingency Variable: Do the Characteristics Hoopes, David G. and Steven Postrel (1999), Shared Knowledge,
of Knowledge Predict Organizational Structure? Organization Glitches, and Product Development Performance, Strategic
Science, 13 (3), 27489. Management Journal, 20 (9), 83765.
Brown, R.L. (1997), Assessing Specific Mediational Effects in James, Lawrence R., Robert G. Demaree, and Gerrit Wolf (1984),
Complex Theoretical Models, Structural Equation Modeling, Estimating Within-Group Interrater Reliability With and
4 (2), 14256. Without Response Bias, Journal of Applied Psychology, 69
BusinessWeek (2006), The Worlds Most Innovative Companies, (1), 8598.
(April 24), 6374. Jaworski, Bernard J. and Ajay K. Kohli (1993), Market Orienta-
Byrne, Barbara M. (2001), Structural Equation Modeling with tion: Antecedents and Consequences, Journal of Marketing,
AMOS. Hillsdale, NJ: Lawrence Erlbaum Associates. 57 (January), 48189.
Chandy, Rajesh K. and Gerard J. Tellis (1998), Organizing for Kahn, Kenneth B. (1996), Interdepartmental Integration: A Defi-
Radical Product Innovation: The Overlooked Role of Willing- nition with Implications for Product Development Perfor-
ness to Cannibalize, Journal of Marketing Research, 34 mance, Journal of Product Innovation Management, 13 (2),
(November), 47487. 13751.
Day, George S. (1994), The Capabilities of Market-Driven Orga- and John T. Mentzer (1998), Marketings Integration with
nizations, Journal of Marketing, 58 (October), 37 52. Other Departments, Journal of Business Research, 41 (1),
and Prakash Nedungadi (1994), Managerial Representa- 5362.
tions of Competitive Advantage, Journal of Marketing, 58 Kogut, Bruce and Udo Zander (1992), Knowledge of the Firm,
(April), 3144.
Combinative Capabilities, and the Replication of Technology,
Dougherty, Deborah (1992), Interpretive Barriers to Successful
Organization Science, 3 (3), 38397.
Product Innovation in Large Firms, Organization Science, 3
Kohli, Ajay K. and Bernard J. Jaworski (1990), Market Orienta-
(2), 179202.
tion: The Construct, Research Propositions, and Managerial
Drazin, Robert and Andrew H. Van de Ven (1985), Alternative
Implications, Journal of Marketing, 54 (April), 118.
Forms of Fit in Contingency Theory, Administrative Science
Kumar, Sanjiv and A. Anju Seth (1998), The Design of Coordina-
Quarterly, 30 (4), 51439.
Eisenhardt, Kathleen M. and Jeffrey A. Martin (2000), Dynamic tion and Control Mechanisms for Managing Joint Venture-
Capabilities: What Are They? Strategic Management Journal, Parent Relationships, Strategic Management Journal, 19
21 (1011), 110521. (June), 57999.
and Behnam N. Tabrizi (1995), Accelerating Adaptive Leonard-Barton, Dorothy (1992), Core Capabilities and Core
Processes: Product Innovation in the Global Computer Indus- Rigidities: A Paradox in Managing New Product Develop-
try, Administrative Science Quarterly, 40 (1), 84110. ment, Strategic Management Journal, 13 (Summer Special
Fisher, Robert J., Elliot Maltz, and Bernard J. Jaworski (1997), Issue), 11125.
Enhancing Communication Between Marketing and Engi- Li, Haiyang and Kwaku Atuahene-Gima (2001), Product Innova-
neering: The Moderating Role of Relative Functional Identifi- tion Strategy and Performance of New Technology Ventures in
cation, Journal of Marketing, 61 (July), 5470. China, Academy of Management Journal, 44 (6), 112334.
Fornell, Claes and David F. Larcker (1981), Evaluating Structural Li, Tiger and Roger J. Calantone (1998), The Impact of Market-
Equation Models with Unobservable Variables and Measure- Knowledge Competence on New Product Advantage: Concep-
ment Error, Journal of Marketing Research, 18 (February), tualization and Empirical Examination, Journal of Marketing,
3950. 62 (October), 1329.
Galbraith, J. (1973), Designing Complex Organizations. Reading, Luo, Xueming, Rebecca J. Slotegraaf, and Xing Pan (2006),
MA: Addison-Wesley. Cross-Functional Collaboration: The Simultaneous Role of
Galunic, Charles D. and Simon Rodan (1998), Resource Recom- Cooperation and Competition Within Firms, Journal of Mar-
binations in the Firm: Knowledge Structures and the Potential keting, 70 (April), 6780.
for Schumpeterian Innovation, Strategic Management Madhavan, Ravindranath and Rajiv Grover (1998), From Embed-
Journal, 19 (12), 11931201. ded Knowledge to Embodied Knowledge: New Product Devel-
Garud, Raghu and Praveen R. Nayyar (1994), Transformative opment as Knowledge Management, Journal of Marketing, 62
Capacity: Continual Structuring by Intertemporal Technology (October), 112.
Transfer, Strategic Management Journal, 15 (5), 36585. Maltz, Elliot and Ajay K. Kohli (2000), Reducing Marketings
Germain, Richard and Cornelia Droge (1997), An Empirical Conflict with Other Functions: The Differential Effects of Inte-
Study of the Impact of Just-in-Time Task Scope Versus Just-in- grating Mechanisms, Journal of the Academy of Marketing
Time Workflow Integration on Organizational Design, Deci- Science, 28 (4), 47992.
sion Sciences, 28 (3), 61535. McEvily, Susan K. and Bala Chakravarthy (2002), The Persis-
Grant, Robert M. (1991), The Resource-Based Theory of tence of Knowledge-Based Advantage: An Empirical Test for
Competitive Advantage: Implications for Strategy Formula- Product Performance and Technological Knowledge, Strategic
tion, California Management Review, 33 (3), 11435. Management Journal, 23 (4), 285305.

Market Knowledge Dimensions and Cross-Functional Collaboration / 111

McGrath, Rita G. (2001), Exploratory Learning, Innovative Sobek, Durward K., Jeffrey K. Liker, and Allen Ward (1998),
Capacity, and Managerial Oversight, Academy of Manage- Another Look at How Toyota Integrates Product Develop-
ment Journal, 44 (1), 11831. ment, Harvard Business Review, 76 (4), 3649.
Moorman, Christine and Anne S. Miner (1997), The Impact of Sobel, Michael E. (1982), Asymptotic Confidence Intervals for
Organizational Memory on New Product Performance and Cre- Indirect Effects in Structural Equation Models, in Sociological
ativity, Journal of Marketing Research, 34 (February), Methodology, S. Leinhardt, ed. San Francisco: Jossey-Bass,
91106. 290312.
Narver, John C. and Stanley F. Slater (1990), The Effect of a Song, Michael, Mitzi M. Montoya-Weiss, and Jeffrey B. Schmidt
Market Orientation on Business Performance, Journal of Mar- (1997), Antecedents and Consequences of Cross-Functional
keting, 54 (October), 2035. Cooperation: A Comparison of R&D, Manufacturing, and Mar-
Nonaka, Ikujiro (1994), A Dynamic Theory of Organizational keting Perspectives, Journal of Product Innovation Manage-
Knowledge Creation, Organization Science, 5 (1), 1437. ment, 14 (1), 3547.
Olson, Eric M., Orville C. Walker Jr., and Robert W. Ruekert and Mark E. Parry (1997), A Cross-National Compara-
(1995), Organizing for Effective New Product Development: tive Study of New Product Development Processes: Japan and
The Moderating Role of Product Innovativeness, Journal of the United States, Journal of Marketing, 61 (April), 118.
Marketing, 59 (January), 4862. Srbom, Dag (1974), A General Method for Studying Differ-
, , , and Joseph M. Bonner (2001), Patterns ences in Factors Means and Factor Structure Between Groups,
of Cooperation During New Product Development Among British Journal of Mathematical and Statistical Psychology,
Marketing, Operations and R&D: Implications for Project Per- 27, 22939.
formance, Journal of Product Innovation Management, 18 (4), Srinivasan, Raji, Gary L. Lilien, and Arvind Rangaswamy (2002),
25871. Technological Opportunism and Radical Technology Adop-
Ping, Robert A., Jr. (1995), A Parsimonious Estimating Tech- tion: An Application to E-Business, Journal of Marketing, 66
nique for Interaction and Quadratic Latent Variables, Journal (July), 4760.
of Marketing Research, 32 (August), 33647. Steenkamp, Jan-Benedict E.M. and Hans Baumgartner (1998),
Prabhu, Jaideep C., Rajesh K. Chandy, and Mark E. Ellis (2005), Assessing Measurement Invariance in Cross-National Con-
The Impact of Acquisitions on Innovation: Poison Pill, sumer Research, Journal of Consumer Research, 25 (June),
Placebo, or Tonic? Journal of Marketing, 69 (January), 7890.
11430. Subramani, Mani R. and N. Venkatraman (2003), Safeguarding
Prahalad, C.K. and Richard A. Bettis (1986), The Dominant Investments in Asymmetric Interorganizational Relationships:
Logic: A New Linkage Between Diversity and Performance, Theory and Evidence, Academy of Management Journal, 46
Strategic Management Journal, 7 (6), 485501. (1), 4662.
Preacher, Kristopher J. and Andrew F. Hayes (2004), SPSS and Szulanski, Gabriel (1996), Exploring Internal Stickiness: Impedi-
SAS Procedures for Estimating Indirect Effects in Simple ments to the Transfer of Best Practice Within the Firm, Strate-
Mediation Models, Behavior Research Methods, Instruments gic Management Journal, 17 (Winter Special Issue), 2743.
and Computers, 36 (4), 71731. Thompson, James D. (1967), Organizations in Action. New York:
Reed, Richard and Robert J. DeFillippi (1990), Causal Ambigu- McGraw-Hill.
ity, Barriers to Imitation, and Sustainable Competitive Advan- Turner, Karynne L. and Mona V. Makhija (2006), The Role of
tage, Academy of Management Review, 15 (1), 88102. Organizational Controls in Managing Knowledge, Academy of
Ruekert, Robert W. and Orville C. Walker (1987), Marketings Management Review, 31 (1), 197217.
Interaction with Other Functional Units: A Conceptual Frame- Venkatraman, N. (1989), The Concept of Fit in Strategy
work and Empirical Evidence, Journal of Marketing, 51 (Jan- Research: Toward a Verbal and Statistical Correspondence,
uary), 119. Academy of Management Review, 14 (3), 42344.
Sheremata, Willow A. (2000), Centrifugal and Centripetal Forces Xie, Jinhong, Michael Song, and Anne Stringfellow (2003),
in Radical New Product Development Under Time Pressure, Antecedents and Consequences of Goal Incongruity on Prod-
Academy of Management Review, 25 (2), 389408. uct Development in Five Countries: A Marketing View, Jour-
Shrout, Patrick E. and Niall Bolger (2002), Mediation in Experi- nal of Product Innovation Management, 20 (May), 23350.
mental and Nonexperimental Studies: New Procedures and Zahra, Shaker A., R. Duane Ireland, and Michael A. Hitt (2000),
Recommendations, Psychological Methods, 7 (4), 42245. International Expansion by New Venture Firms: International
Sinkula, James M. (1994), Market Information Processing and Diversity, Mode of Market Entry, Technological Learning, and
Organizational Learning, Journal of Marketing, 58 (January), Performance, Academy of Management Journal, 43 (5),
3545. 92550.
Slater, Stanley F. and Kwaku Atuahene-Gima (2004), Conducting and Anders P. Nielsen (2002), Sources of Capabilities,
Survey Research in Strategic Management, in Research Integration and Technology Commercialization, Strategic
Methodology in Strategy and Management, Vol. 1, David J. Management Journal, 23 (5), 37798.
Ketchen Jr. and Don D. Bergh, eds. Amsterdam: Elsevier/JAI Zeithaml, Valarie A., P. Rajan Varadarajan, and Carl P. Zeithaml
Press, 22750. (1988), The Contingency Approach: Its Foundations and
Smith, Ken G., Christopher J. Collins, and Kevin D. Clark (2005), Relevance to Theory Building and Research in Marketing,
Existing Knowledge, Knowledge Creating Capability, and European Journal of Marketing, 22 (7), 3764.
Rate of New Product Introduction in High-Technology Firms,
Academy of Management Journal, 48 (April), 34657.

112 / Journal of Marketing, January 2007