INFORMATION PROCESSING, KNOWLEDGE DEVELOPMENT, AND
STRATEGIC SUPPLY CHAIN PERFORMANCE
G. TOMAS M. HULT Michigan State University
DAVID J. KETCHEN, JR. Florida State University STANLEY F. SLATER Colorado State University
Little is known about why some supply chains perform well while others do not. Drawing on the knowledge-based view of the firm and theory from the information processing and organizational learning literatures, we devised a model linking knowl-edge development to cycle time in strategic supply chains—chains whose members are strategically, operationally, and technologically integrated. Using data from 58 chains in aFortune500 firm, we found that the knowledge development process explained substantial variance in cycle time.
A supply chain is a network of actors that trans-forms raw materials into distributed products (Bowersox, Closs, & Stank, 1999). Some functions may occur within a single firm, while others cross firm boundaries. Our focus is strategic supply chains: chains whose members are strategically, operationally, and technologically integrated. Long-term relations between members are developed to provide stability in these chains, but such links are severed when needs change. Thus, predictability is sought but not at the expense of creating inflexibility. The allure of such situations has led firms to spend vast sums to improve supply chain process-es; for instance, UPS spent $9 billion in the period between 1986 and the late 1990s (Farhoomand & Ng, 2000). However, managing a strategic supply chain is complex. For example, members’ loyalties may lie with their home organizations or adjoining members rather than with the chain. Challenges such as these often lead to the promise of improved outcomes going unfulfilled (Bowersox et al., 1999). We consider here how knowledge development may enhance supply chain outcomes. Because chain members are not all part of the same organi-zation, knowledge may be an important source of coordination (Hansen, 2002) and thus be central to
chain functioning. Our dependent variable was cy-cle time, the interval between a user’s request for a product or service and its delivery (Hult, Ketchen, & Nichols, 2000). The current environment is char-acterized by time-based competition, wherein in-creasingly impatient customers favor providers (such as Dell and Toyota) that offer speedy delivery of a quality product or service. Thus, fast cycle time facilitates increased market share, as well as lower overhead and inventory costs. These links led Handfield and Nichols (2002: 13) to conclude that cycle time has a “direct linkage to profits,” a central concern of virtually all businesses. Our research questions were, Does the knowledge development process shape supply chain cycle time? and, If so, how?
THEORY AND HYPOTHESES
Figure 1 presents a model of the knowledge de-velopment process in strategic supply chains. Given the early stage of supply chain theory devel-opment and our cross-sectional data, each hypothesis provides a “snapshot” of an element in the knowl-edge development process. The model includes four antecedents—achieved memory, knowledge acquisi-tion activities, informaacquisi-tion distribuacquisi-tion activities, and shared meaning—and cycle time.
The model draws on three complementary per-spectives. Exploiting information to improve pro-cesses and/or outcomes is a focus of each perspec-tive, but each offers different insights about how this exploitation occurs. Taken together, they pro-vide a more holistic view of knowledge develop-We appreciate the financial support of MSU-CIBER
and the FedEx Center for Supply Chain Management, and the constructive input provided during different phases of the research process by Associate Editor Sara Rynes, the five reviewers, Roger J. Calantone, Kenneth K. Boyer, Tunga Kiyak, Ernest L. Nichols, Jr., and K. Michele Kacmar.
FIGURE 1 The Knowledge Development Process and Cycle Time in Supply Chains a aHypothesis 9 predicts that shared meaning has a stronger link with cycle time than knowledge acquisition and information distribution activities. other hypotheses pertain to the relationships identified in the model.
ment than any one view can provide. The first is the knowledge-based view of the firm. According to the knowledge-based view, which builds on tenets of the resource-based view, unique abilities to cre-ate and exploit wisdom enhance outcomes (Grant, 1996). Thus, the knowledge-based view provides a foundation for our expectation that knowledge de-velopment shapes cycle time.
The organizational learning literature is the sec-ond source of grounding. In encapsulating the lit-erature, Huber (1991) identified four key learning elements: knowledge acquisition, information dis-tribution, information interpretation, and organiza-tional memory. Marketing and decision sciences studies have measured these elements, albeit some-times under different labels. We take a logical step forward by examining their relations within the supply chain context.
Our model also overlaps with extant organiza-tional information processing models. In these models, gathering, processing, and acting on data from the environment is a firm’s main task (e.g., Daft & Weick, 1984). Although supply chains’ pri-mary role is as material-processing and product movement systems, information processing is crit-ical to their success (Bowersox et al., 1999). Like organizations, chains may benefit from improved information processing, but the nature of these ac-tivities differs for organizations and chains. More importantly, chains’ ability to leverage such activ-ities into improved outcomes is expected to vary. Links between Memory and Information
Processing
Corner, Keats, and Kinicki (1994) depicted mem-ory as a step between interpretation and decision. These authors defined their model’s boundaries by noting that it applies to “ongoing, established organizations rather than emerging ones” (1994: 296). We suggest that Corner and colleagues’ model also does not apply to supply chains in that the concepts need to reordered in a supply chain set-ting. The ad hoc nature of supply chains means that tapping memory may be difficult. Yet access to memory is vital because a chain lacks many of the formal and informal mechanisms that guide deci-sions in established firms, such as hierarchy (for-mal) and strong values, traditions, and beliefs (in-formal) (Handfield & Nichols, 2002). Thus, memory was our model’s launching point. Our focus was on achieved memory,which we defined as the amount of knowledge, experience, and familiarity with the supply chain process (cf. Moorman & Miner, 1997). Our initial hypotheses build on conclusions drawn by Huber (1991) about organizational
learn-ing, adapted for the supply chain context. First, Huber asserted that memory targets knowledge ac-quisition activities in certain directions. If past suc-cess has been tied to the development of new prod-ucts, for example, knowledge acquisition is often focused on changes in customer tastes. In a strate-gic supply chain, memory does not focus on these activities, but rather encouragesmore acquisition. Members of chains that possess significant memory are aware that knowledge coordination across nodes reduces duplication, waste, and redundancy (Handfield & Nichols, 2002). We posit that this recognition should lead those chains to seek out more knowledge than chains that lack memory. Said differently, the more knowledge chain mem-bers possess, the greater their awareness that addi-tional knowledge can ultimately enhance out-comes. Over time, a deviation-amplifying loop may emerge between memory and knowledge acquisi-tion, enhancing both (cf. Lindsley, Brass, & Thomas, 1995). Indeed, Bowersox and colleagues (1999) found that supply chains that use bench-marks to assess relative performance have more accumulated knowledge than others and also have stronger beliefs that they need additional knowl-edge throughout their systems. Thus, we predict:
Hypothesis 1. A supply chain’s level of achieved memory is positively related to its level of knowledge acquisition activities.
A second assertion offered by Huber (1991) is that memory focuses information distribution in particular directions. Memory suggests where a cer-tain piece of information will be likely to have a positive impact on outcomes and so determines which entities will receive the information. In stra-tegic supply chains, the presence of more memory encouragesmoreinformation distribution. Specifi-cally, a strategic supply chain is formed in an effort to create a rare, valuable, and inimitable source of knowledge and coordination (cf. Grant, 1996). A chain’s members have other (perhaps conflicting) roles to fulfill outside of the chain. Thus, chains require continuous information distribution in or-der to maintain strategic, operational, and techno-logical integration. Previous retained learning, such as lessons from success and failure experi-ences, makes clear the need for information to be circulated (Bowersox et al., 1999). Thus, chains with more memory should distribute information more than those with little memory. Stated for-mally:
Hypothesis 2. A supply chain’s level of achieved memory is positively related to its level of information distribution activities.
Huber (1991) also contended that shared mean-ing (the extent to which participants develop com-mon understandings about data and events [Corner et al., 1994]) depends on memory. The underlying causal chain is that memory provides a frame of reference or cognitive map that in turn guides organizational participants toward common under-standings. Other potential sources of frames of ref-erence, such as strong culture (Gioia & Thomas, 1996), are missing from supply chains. In their absence, shared meanings are needed to harness collective action. Memory’s capacity as a repository for shared supply chain experiences suggests it is likely to have a pronounced role in creating shared meanings. In particular, strategic supply chains demand that members balance newly obtained information with time-tested, strategically aligned activities (cf. Bowersox et al., 1999). Effective inte-gration of “new” and “old” knowledge should be more likely when a high level of memory has been established. Thus, we expect:
Hypothesis 3. A supply chain’s level of achieved memory is positively related to its level of shared meaning.
Links among Elements of Information Processing The middle portion of our model focuses on re-lations among the three elements of information processing. The organizational information pro-cessing view is that coping with information is an organization’s main task (Daft & Weick, 1984). The flow of information has a curvilinear link with out-comes; an inflection point can be reached at which dealing with more information is overwhelming (Huber, 1991). Given this potential, information is distributed only to those who need it. Thus, knowl-edge acquisition and information dissemination levels should not be related.
Supply chains are different from organizations, however. Specifically, they are more similar to the functional areas of an organization (for instance, marketing and accounting) in that they are focused on one task. In both cases, all members of an entity (a function or a chain) need the same information. As a result, knowledge acquisition and distribution should be positively related. Indeed, circulating acquired knowledge is vital given the variation in how often supply chains are used. In our data, for example, the frequency with which members initi-ated the supply chain process ranged from 0.5 to 12.5 times per month. New knowledge related to supply chain practices (such as regulations, prod-ucts, and procedures) is acquired frequently. To achieve desired outcomes, this knowledge must be
available to all nodes, even if chain membership is a part-time role (Baker & Sinkula, 1999). Thus, we predict:
Hypothesis 4. A supply chain’s level of knowl-edge acquisition activities is positively related to its level of information distribution activ-ities.
Shared meaning is the third information process-ing variable depicted in our model. Huber (1991) posited that the development of diverse interpreta-tions reflects learning because such diversity in-creases an organization’s repertoire of possible ac-tions. In a related vein, organizational information processing research contains the assertion that peo-ple holding diverse views develop enough overlap in interpretation to act without reaching consensus on meaning. This logic holds for many settings, particularly those in which subunits are relatively independent (Gioia & Thomas, 1996), but diverse interpretations are often a hindrance in the supply chain context. Chain tasks are straightforward, largely consisting of getting a quality product (and/or material) where it needs to be in a timely manner. As such, supply chains are characterized by reciprocal interdependence; each node depends on adjoining nodes to perform its role (Thompson, 1967). Diverse views of concepts such as quality and timeliness need to be resolved so that effort can be focused on necessary activities (Handfield & Ni-chols, 2002). Thus, strategic supply chains benefit when participants learn to think alike.
We expect that greater information distribution activities will be related to supply chain members’ ability to arrive at shared meanings. In an organi-zation, a strong culture and a common affiliation can drive members toward shared meaning (Gioia & Thomas, 1996). In contrast, a supply chain partici-pant’s primary loyalty generally lies not with the chain, but with his or her home organization. Building on the organizational information pro-cessing literature, proponents of media richness have argued that communication media that effec-tively transmit varied cues and provide rapid inter-action (such as face-to-face meetings) can facilitate shared understandings (Daft & Lengel, 1986). A chainwide emphasis on discussions, meetings, and information sharing may thus help fill a key role that culture serves for organizations: providing fa-miliarity and the opportunity to develop a common mind-set about issues. One example is the annual conference FedEx holds with its strategic suppliers (Farhoomand & Ng, 2000). On the basis of this reasoning, we predict:
Hypothesis 5. A supply chain’s level of infor-mation distribution activities is positively re-lated to its level of shared meaning.
Links between Information Processing and Supply Chain Outcomes
All three perspectives underlying our model in-clude an emphasis on outcomes. A central tenet of the knowledge-based view is that firms’ relative abilities to build and draw on participants’ knowl-edge differentiates high and low performance (Grant, 1996). The available evidence suggests managers’ information-processing activities are tied to performance (e.g., Thomas, Clark, & Gioia, 1993). Finally, a goal of enhanced outcomes is in-herent in the organizational learning concept be-cause it is through such outcomes that learning’s value is manifested (Huber, 1991).
One benefit of knowledge acquisition activities is that these activities enhance each supply chain node’s knowledge flows. The knowledge-based view suggests such flows will enhance each node’s ability to efficiently perform its role (cf. Grant, 1996). Thus, the more knowledge acquisition activ-ities are emphasized, the greater knowledge flows grow, and the more outcomes can improve. We would expect to observe negative links between knowledge acquisition activities and cycle time be-cause lower cycle time is desirable. Thus, we pre-dict that:
Hypothesis 6. A supply chain’s level of knowl-edge acquisition activities is negatively related to cycle time.
Ideas from organizational learning provide the basis for predicting a link between information dis-tribution activities and cycle time. As information is distributed, as described above, each supply chain node becomes more educated about how to efficiently operate within a chain as well as more educated about other nodes’ needs and preferences (Hult, Ketchen, & Nichols, 2002). These increased skills and awareness should enhance the ability to minimize cycle time (cf. Huber, 1991). Stated for-mally:
Hypothesis 7. A supply chain’s level of infor-mation distribution activities is negatively re-lated to cycle time.
A central premise expressed in the organiza-tional information processing literature is that shared meaning provides a basis for commonly di-rected effort among organization members. This premise is reflected prominently in both theory (e.g., Daft & Weick, 1984) and empirical studies
(e.g., Gioia & Thomas, 1996; Thomas et al., 1993). Adapting organizational information processing theorizing to the strategic supply chain context, we would expect that a chain can achieve better out-comes if the development of shared meaning effec-tively channels chain members’ vision, strategies, and operations in the same direction. Thus, we posit:
Hypothesis 8. A supply chain’s level of shared meaning is negatively related to cycle time. In both the knowledge-based and organizational information processing literatures, authors have posited that assigned meaning is more valuable than information (Daft & Weick, 1984; Grant, 1996). Information is important, but translating informa-tion into knowledge provides the basis for better management. Given the lack of deeply rooted cul-tures in supply chains, shared meaning becomes the critical mechanism to ensure coordination (Handfield & Nichols, 2002). Building on these no-tions, we would expect shared meaning to be more strongly associated with cycle time than the other two antecedents. Stated formally:
Hypothesis 9. Shared meaning has a stronger, negative relationship with cycle time than do knowledge acquisition activities or informa-tion distribuinforma-tion activities.
METHODS Sample
Supply chains of a Fortune 500 transportation firm operating in over 200 countries were studied. In each chain, data were drawn from three types of nodes: internal users, corporate buyers, and exter-nal suppliers. Eachinternal userwas the key pur-chasing contact in a business unit and was regarded in the firm as an end user. Each corporate buyer was part of the firm’s purchasing function (labeled “Strategic Sourcing and Supply” in the firm) and purchased products and services from external suppliers on behalf of the internal users (all pur-chase requests above $1,000 had to go through the buyers; users could independently purchase items below that amount). The external suppliers pro-vided a broad range of products (such as parts for trucks, airplanes, and the firm’s tracking system) and services (such as software, training, and main-tenance) to the users via the corporate buyers.
We conducted pretests with eight academics and seven executives to assess the scale items’ face va-lidity. Next, we did a pilot study of 36 executives to assess the research design. A survey was then ad-ministered to the users, buyers, and suppliers at the
same time during a two-month period in 1998 –99. Respondents were asked to focus on their most recent supply chain process. The decision to use the fine-grained approach of studying one firm’s supply chains (versus the coarse-grained approach of studying chains in diverse firms) allowed us to avoid the potentially confounding effects of varia-tion in company practices.
Company records verified the direct links among users, buyers, and suppliers. To ensure a high de-gree of familiarity with the issues surveyed among the suppliers, we only collected data from individ-uals working within suppliers who could answer a series of qualifying questions (examples: “Do you know your contact person in the Strategic Sourcing and Supply unit at [the company]?”; “Do you know about the category approach of Strategic Sourcing and Supply within [the company]?”; and “Do you know your primary category management group of [the company’s] Strategic Sourcing and Supply?”). In creating the sample, we also verified with the company that these suppliers were viewed as stra-tegic partners. All respondents were asked to iden-tify their corresponding supply chain participants (that is, the buyer identified the supplier and the user, and so on). This procedure allowed us, with company assistance, to link responses from specific buyers and users with the suppliers who were a part of the study. Using these steps, we obtained data from 58 strategic supply chains composed of one user, one buyer, and one supplier (representing 141 of the 346 users, 115 of the 338 corporate buyers, and 58 of the 235 strategic suppliers). Measures
The Appendix lists our measures. We adapted established scales to measure achieved memory, knowledge acquisition activities, information dis-tribution activities, subjective cycle time, and the importance of the supply chain relationship (a con-trol variable). Objective measures of cycle time and the frequency of the supply chain relationship (a control variable) were also included. We based the shared meaning scale on Huber (1991). Subjective cycle time was reverse-coded to correspond to the direction of objective cycle time. Table 1 summa-rizes the average variances extracted, construct re-liabilities, factor “loadings,” and fit indexes from a confirmatory factor analysis of the major study variables. Table 2 reports the means, standard de-viations, and correlations.
Interrater reliability.To examine interrater reli-ability, we first used intraclass correlation (ICC) analysis, via analysis of variance (ANOVA), to as-sess the ratio of between-targets variance to total
(between-targets plus within-target) variance. Fol-lowing suggestions by Shrout and Fleiss (1979), we examined ICC(1,1) and ICC(1,3). ICC(1,1) is the re-liability of a rating by one rater (either an internal user, corporate buyer, or external supplier) and ICC(1,3) is the reliability of the average rating across three raters. In our study, ICC(1,1) ranged between .17 and .36 and ICC(1,3) ranged between .38 and .63. Overall, these results indicate a reason-able level of interrater reliability in a supply chain setting (e.g., Boyer & Verma, 2000).
Second, we assessed each item’s robustness via multigroup analysis using LISREL 8.54. Sets of beta estimates were constrained, one parameter at a time, to be equal and different across the three nodes. The significance of the resulting change in chi-square was examined (Anderson, 1987). The chi-square changes ranged from 0.03 to 5.76 with a change in the degrees of freedom of 2 (1,696 ⫺ 1,694) for all items. Values were lower than the recommended cutoff of 5.99, except for item 2 of the subjective cycle time scale, which was different across nodes (⌬2 ⫽10.58,⌬df ⫽2,p⬍ .05); this item was omitted from further analysis.
Dimensionality, composite reliability, and va-lidity. The dimensionality, composite reliability, and validity of the six latent constructs involving 22 items were evaluated via confirmatory factor analysis (CFA) using LISREL 8.54. We aggregated the data for the users, buyers, and suppliers who made up the 58 chains (thus,n⫽174 for the CFA) to allow for examination of all items in one mea-surement model. Model fit was assessed with five indexes: the DELTA2 index, the relative noncen-trality index (RNI), the comparative fit index (CFI), the Tucker-Lewis index (TLI), and the root-mean-square error of approximation index (RMSEA) (Gerbing & Anderson, 1992; Hu & Bentler, 1999).
TABLE 1
Summary Statistics, Confirmatory Factor Analysis of Major Constructsa
Construct Average Variance Extracted Composite Reliability Range of Parameter Estimates Achieved memory 87.00% .95 .88–.98 Knowledge acquisition 55.60 .86 .69–.81 Information distribution 77.25 .93 .87–.90 Shared meaning 89.00 .94 .93–.95
Subjective cycle time 69.33 .87 .57–.96
Importance 72.80 .93 .58–.94
an⫽174. Fit statistics are as follows:2⫽475.33;df⫽194;
DELTA2 ⫽.97; RNI (relative noncentrality index)⫽.97; CFI (comparative fit index)⫽.97; TLI (Tucker-Lewis index)⫽.97; RMSEA (root-mean-square residual)⫽.06.
The CFA resulted in an excellent fit to the data, with the DELTA2, RNI, CFI, and TLI all at .97, and the RMSEA at .06.
Composite reliability was calculated via the pro-cedures suggested by Fornell and Larcker (1981). The parameter estimates and their associated t -values were also examined, along with the average variance extracted for each construct (Anderson & Gerbing, 1988). The composite reliabilities, which ranged from .86 to .95 (Table 1), were excellent. The factor “loadings” ranged from .57 to .98 (p ⬍ .01), and the average variances extracted ranged from 55.60 to 89.00 percent (Table 1), exceeding the cutoff of 50 percent suggested by Fornell and Larcker (1981) to achieve convergent validity. The 22 items were also reliable and valid when evalu-ated on the basis of each item’s error variance, modification index, and residual covariation (item skewness was less than兩.69兩, and kurtosis was less than 兩.96兩, indicating that each item was normally distributed).
We assessed discriminant validity by analyzing all possible pairs of constructs in a series of two-factor CFA models (Anderson, 1987). Each model was run twice; once constraining the phi () coef-ficient to unity and once freeing this parameter. A chi-square difference test was then performed on the nested models to assess if the unconstrained models fit better. The critical value (⌬2
1 ⬎ 3.84) was exceeded in all cases (the chi-square difference ranged from 20.15 for the “pairwise” test of knowl-edge acquisition and information distribution to 246.26 for subjective cycle time and information distribution). Thus, the six scales and their 22 items are reliable and valid in this study.
ANALYSIS AND RESULTS
The hypotheses were tested using hierarchical regression analysis (least squares) with controls en-tered in step 1 and predictors enen-tered in step 2. The
unit of analysis was the supply chain; we did not use structural equation modeling because the sam-ple was small (n⫽58). To create scores at the level of the supply chain, we standardized (mean-centered) the scores for each construct at the node level (user, buyer, and supplier) and averaged them to form an index for each construct at the supply chain level. The results are presented in Table 3 and depicted in Figure 2. Each regression model also included a number of control variables.
Hypothesis 1 predicts a positive link between achieved memory and knowledge acquisition activ-ities. The relationship was positive and significant ( ⫽ .22, p ⬍ .10), supporting Hypothesis 1. Hy-potheses 2 and 4 state that achieved memory and knowledge acquisition activities have a positive link with information distribution activities. Knowledge acquisition had a significant, positive link ( ⫽ .75, p⬍ .01), but achieved memory did not ( ⫽ .05, p ⫽ .64); thus, Hypothesis 4 was supported, but not Hypothesis 2. Hypotheses 3 and 5 state that achieved memory and information dis-tribution activities have a positive link with shared meaning. Information distribution had a signifi-cant, positive link (⫽.61,p⬍.01), but achieved memory did not (⫽ .19,p⫽.12). Thus, Hypoth-esis 5 was supported, but not HypothHypoth-esis 3.
Hypotheses 6, 7, and 8 predict that knowledge acquisition, information distribution, and shared meaning will have negative links with cycle time. Hypothesis 9 predicts that shared meaning will have a stronger, negative link to cycle time than the other two antecedents. We tested these hypotheses using subjective and objective cycle time measures. Knowledge acquisition ( ⫽ ⫺.84, p ⬍ .01) and shared meaning ( ⫽ ⫺.46,p ⫽ .06) were signifi-cantly and negatively related to subjective cycle time, while information distribution (⫽.79,p ⬍ .05) was significantly but positively related (ad-justedR2⫽.38). We found no significant difference between the knowledge acquisition and shared TABLE 2
Means, Standard Deviations, and Correlationsa
Variable Mean s.d. 1 2 3 4 5 6 7
1. Achieved memory 4.73 0.97
2. Knowledge acquisition 3.31 0.62 .66
3. Information distribution 3.00 0.72 .60 .86
4. Shared meaning 3.15 0.63 .61 .70 .77
5. Subjective cycle time 3.88 0.85 ⫺.32 ⫺.55 ⫺.38 ⫺.49
6. Objective cycle time 33.49 53.83 .11 ⫺.09 .14 ⫺.36 .39
7. Control variable: Importance 4.99 0.91 .69 .79 .68 .72 ⫺.46 .36
8. Control variable: Frequency 2.72 2.62 .02 ⫺.14 ⫺.12 ⫺.28 .20 .07 ⫺.17
aAll correlations greater than or equal to .20 are significant (atp⬍.05). The standardized values were used for the correlation analysis.
TABLE 3
Results of Regression Analysesa
Model and Variables  b s.e. t p Results of Hypothesis Test
Knowledge acquisition Importance of supply chain relationship
.65 0.46 0.09 5.09 ⬍.01
Frequency of supply chain relationship ⫺.02 ⫺0.00 0.00 0.23 .82
Achieved memory .22 0.14 0.08 1.75 ⬍.10 Hypothesis 1 supported
R2 .67 AdjustedR2 .65 ⌬R2 .02 Information distribution Importance of supply chain relationship
.11 0.09 0.12 0.79 .44
Frequency of supply chain relationship .01 0.00 0.00 0.05 .96
Achieved memory .05 0.05 0.09 0.48 .64 Hypothesis 2 not supported
Knowledge acquisition .75 0.90 0.16 5.80 ⬍.01 Hypothesis 4 supported
R2 .77 AdjustedR2 .74 ⌬R2 .21 Shared meaning Knowledge acquisition ⫺.13 ⫺0.14 0.20 0.68 .50
Importance of supply chain relationship
.21 0.16 0.11 1.41 .17
Frequency of supply chain relationship ⫺.19 ⫺0.01 0.00 2.13 ⬍.05
Achieved memory .19 0.13 0.08 1.58 .12 Hypothesis 3 not supported
Information distribution .61 0.54 0.15 3.63 ⬍.01 Hypothesis 5 supported
R2 .73 AdjustedR2 .69 ⌬R2 .11 Subjective cycle timeb
Achieved memory ⫺.00 ⫺0.00 0.18 0.00 .99
Importance of supply chain relationship
⫺.13 ⫺0.14 0.25 0.57 .58
Frequency of supply chain relationship .05 0.00 0.01 0.34 .74
Knowledge acquisition ⫺.84 ⫺1.22 0.43 2.81 ⬍.01 Hypothesis 6 supported
Information distribution .79 0.91 0.37 2.44 ⬍.05 Hypothesis 7 not supported
Shared meaning ⫺.46 ⫺0.62 0.32 1.95 .06 Hypothesis 8 supported
R2 .48 AdjustedR2 .38 ⌬R2 .18 Objective cycle timec
Achieved memory .20 6.07 8.01 0.76 .34
Importance of supply chain relationship
.16 5.56 10.62 0.52 .62
Frequency of supply chain relationship .04 0.08 0.42 0.20 .82
Knowledge acquisition ⫺.08 ⫺3.44 18.33 0.19 .85 Hypothesis 6 not supported
Information distribution .70 25.43 12.62 2.02 ⬍.05 Hypothesis 7 not supported
Shared meaning ⫺.93 ⫺37.75 12.67 2.98 ⬍.01 Hypothesis 8 supported
R2
.35 AdjustedR2
.19
⌬R2 .29
aR2is the total variance explained for an equation. AdjustedR2is the total variance explained for the equation with the number of
predictor variables and the sample size taken into account.⌬R2refers to the variance explained by the predictor variables in each equation
above that explained by the control variables included.
bWe found no significant difference between the effects of knowledge acquisition and shared meaning on subjective cycle time (t⫽
1.31,df⫽51). Thus, Hypothesis 9 is not supported in this analysis.
cWe found a significant difference between the effects of knowledge acquisition and shared meaning on objective cycle time (t⫽3.11, df⫽51,p⬍.01). Thus Hypothesis 9 is supported in this analysis.
FIGURE 2 Results for Hypothesized Predictor Variables and Control Variables a a“O” is objective cycle time; “S” is subjective cycle time. Standardized coefficients are shown in the figure for the significant paths. Please see Tab le 3 for complete
meaning effects (t ⫽ 1.31, df ⫽ 51) using the method suggested by Cohen, Cohen, Aiken, and West (2003) of assessing differences between par-tial regression coefficients. Thus, Hypothesis 6 and Hypothesis 8 were supported, but not Hypothesis 7 and Hypothesis 9.
In the objective model, shared meaning ( ⫽
⫺.93,p⬍.01) had a significant and negative effect, as predicted, while knowledge acquisition had a negative and insignificant link (⫽ ⫺.08,p⫽.85), and information distribution (⫽.70,p⬍.05) had a significant, positive effect (adjustedR2⫽.19). We found that shared meaning had a stronger relation with cycle time than knowledge acquisition using the Cohen et al. (2003) test (t⫽3.11,df⫽ 51,p ⬍ .01). As such, Hypothesis 8 and Hypothesis 9 were supported, but not Hypothesis 6 and Hypothesis 7. Finally, we found that almost all of our equations and predictors had adequate statistical power. A method suggested by Cohen and coauthors (2003) indicated the probability of finding the sample sta-tistics (R2s) we achieved (see Table 3) was at least .80 for the regressionR2s if alpha equaled .05 (that is, power was greater than or equal to .80 at a significance level of .05). One predictor variable had a power of .30 (␣ ⫽ .05) in one equation; all other predictors had power above .80 in all equa-tions (cf. Lenth, 2001).
DISCUSSION
Our results should be interpreted in the context of the study’s limitations. Given the early stage of literature development, we studied one firm’s sup-ply chains in order to avoid the potentially con-founding effects of variation in company practices. In particular, we focused on strategic supply chains that involved suppliers who were quite familiar with the focal firm’s supply chains. Such strategic chains are increasingly important to organizations (Bowersox et al., 1999), but our focus on them created a selection bias and restriction of range vis-a`-vis supply chains in general. Thus, inquiry that captures chains from multiple organizations and incorporates a greater variety of suppliers is needed to establish generalizability. Our cross-sectional design prevented us from studying possi-ble feedback loops among our variapossi-bles. Also, our emphasis on cycle time alone is limiting. Future studies will benefit from including other supply chain outcomes, such as quality, cost, and flexibil-ity. Additionally, the use of objective measures for achieved memory and shared meaning would have strengthened our design.
Our study’s main contribution lies in providing insight into how the knowledge development
pro-cess shapes supply chain outcomes. Our models explained 38 percent of the variance in subjective cycle time and 19 percent of the variance in objec-tive cycle time. The support found for four predic-tions provides evidence in favor of much of our theorizing, while the mixed evidence for two pre-dictions and lack of support for three others high-lights the need for further inquiry. Below, we discuss our study’s theoretical and practical impli-cations.
Memory and Information Processing
Hypothesis 1 was supported (at the .10 level), indicating that chains possessing more memory tend to seek new knowledge more than chains pos-sessing less memory. This is counterintuitive be-cause knowledge seeking is often viewed as a means to close gaps between what decision makers know and what they need to know (e.g., Daft & Lengel, 1986). Thus, entities lacking knowledge should seek information more than those already possessing it. To the contrary, our results suggest that, in strategic supply chains, memory and knowledge acquisition may enter a deviation-amplifying loop wherein higher levels of each en-courage the other to grow (cf. Lindsley et al., 1995). If longitudinal studies were to support this idea, it would suggest that managers may need to monitor knowledge acquisition activities to ensure they do not distract chain members from the movement of products. Such studies would also assess memory’s role as an outcome—a role we could not examine with our cross-sectional data.
We did not find support for our prediction that achieved memory is linked to information distribu-tion activities (Hypothesis 2). Given this finding, we examined post hoc the possibility that there might be a curvilinear relationship. Perhaps at high levels of memory, previously circulated informa-tion is adequate, and distribuinforma-tion activities decline. We ran a regression model with achieved memory, knowledge acquisition, our two control variables, and squared terms for achieved memory and knowledge acquisition as the predictors. Neither squared term was significant; thus, there is no evi-dence of a curvilinear relationship. These results imply that memory’s influence on information dis-tribution is indirect, working via knowledge acqui-sition (the relationship depicted in Hypothesis 4). One practical implication is that if changes in dis-tribution patterns are needed, memory’s direct role may be minimal, but it can initiate useful changes in knowledge acquisition activities.
The predicted link between achieved memory and shared meaning (Hypothesis 3) rested on
seem-ingly solid conceptual grounds. Borrowing from Huber (1991), we posited that memory provides a common frame of reference that steers participants in supply chains toward shared understandings. The relationship was in the predicted direction but was not statistically significant. Perhaps studies of other supply chain contexts will find support for a link between achieved memory and shared meaning.
Links among Elements of Information Processing We found support for Hypothesis 4’s contention that knowledge acquisition activities shape infor-mation distribution activities. In organizations, concern about information overload leads organi-zations to direct information only to those who need it (Huber, 1991). In contrast, supply chain members all need the same information. The sup-port for our prediction highlights the imsup-portance of knowledge creation in this context. Although a supply chain is traditionally viewed as a material-or product-processing system, it is perhaps equally valuable to view it as an information-processing and interpretation system (cf. Daft & Weick, 1984). Thus, future efforts aimed at further articulating the confluence of materials/products and information within the supply chain context should prove fruit-ful. Assessing not just knowledge development, but also the amount of knowledge created, would be valuable.
The expectation stated in Hypothesis 5, that in-formation distribution activities shape shared meaning, was supported. In organizations, strong cultural elements such as an agreed-upon vision foster shared meanings (Gioia & Thomas, 1996). Drawing on the organizational information process-ing literature, we argued that a chainwide emphasis on information exchange might fulfill this role. Specifically, common understandings are more likely to arise when members are provided with a forum in which to trade ideas. A pragmatic impli-cation is that chains struggling to arrive at shared meanings might rely more on rich media, such as discussions. Such media can facilitate the resolu-tion of varied perspectives by effectively transmit-ting emotions and subtleties in meaning (Daft & Lengel, 1986).
Information Processing and Outcomes
We posited that more knowledge acquisition tivities (Hypothesis 6), information distribution ac-tivities (Hypothesis 7), and shared meaning (Hy-pothesis 8) would be associated with faster cycle time. Knowledge acquisition had a beneficial link
with subjective but not objective cycle time. Infor-mation distribution did not have a beneficial link with either measure. Shared meaning was benefi-cial for both measures. Viewed alongside our ear-lier findings, these results suggest a trail of links from memory to knowledge acquisition to informa-tion distribuinforma-tion to shared meaning to outcomes. Such a flow of links would suggest shared meaning facilitates connections between other information-processing aspects on one hand and cycle time on the other (cf. Thomas et al., 1993). A sample large enough that longitudinally based causal modeling could be confidently used is needed to confirm this contention. Also, it is possible that our respondents overestimated both knowledge acquisition and sub-jective cycle time, a possibility that future studies should consider.
Hypothesis 9 predicts that shared meaning would have the strongest favorable relations with cycle time of the information-processing elements. This prediction was supported for objective cycle time, but not for subjective cycle time. These re-sults suggest that shared meaning is more vital to cycle time than chain members believe it is. Thus, educating chain members about the potential value of shared meaning may be a managerial imperative (Handfield & Nichols, 2002).
Overall, our models explained substantial vari-ance in subjective and objective cycle time. Our results indicate that, at least in the supply chains examined, the knowledge development process is an important antecedent to supply chain efficiency. More broadly, our results vis-a`-vis cycle time aug-ment a small body of research linking knowledge and outcomes in support of the knowledge-based view’s central tenet (cf. Grant, 1996). As such, the results highlight the potential value of future knowledge-based studies for explaining outcomes of interest.
Conclusion
Drawing on three perspectives—the knowledge-based view, organizational information processing, and organizational learning—we developed a model that explained substantial cycle time vari-ance over 58 strategic supply chains. Thus, our study helps close the gap between what scholars know about shaping chain performance and what they need to know. More broadly, our findings show the value of juxtaposing the three tives. Rather than basing studies on single perspec-tives, perhaps scholars should view each as a necessary but not sufficient component for under-standing knowledge development. For managers, our study highlights the differences in managing
organizations versus supply chains and suggests tools for better supply chain management. Diverse inter-pretations often lead to more creative solutions in organizations. In contrast, we found a strong link between shared meaning and reduced cycle time, suggesting that leading members to think alike about concepts such as quality and speed (perhaps through emphasizing information sharing and face-to-face discussions) may improve chain performance.
REFERENCES
Anderson, J. C. 1987. An approach for confirmatory mea-surement and structural equation modeling of organ-izational properties. Management Science, 33: 525–541.
Anderson, J. C., & Gerbing, D. W. 1988. Some methods for respecifying measurement models to obtain unidi-mensional construct measurement.Journal of Mar-keting Research,19: 453– 460.
Anderson, E., & Weitz, B. 1992. The use of pledges to build and sustain commitment in distribution chan-nels.Journal of Marketing Research,29: 18 –34. Baker, W. E., & Sinkula, J. M. 1999. The synergistic effect
of market orientation and learning orientation on organizational performance. Journal of the Acad-emy of Marketing Science,27: 411– 427.
Bowersox, D. J., Closs, D. J., & Stank, T. P. 1999. 21st century logistics: Making supply chain integration a reality.Oak Brook, IL: Council of Logistics Manage-ment.
Boyer, K. K., & Verma, R. 2000. Multiple raters in survey-based operations management research: A review and tutorial. Production and Operations Manage-ment,9(2): 128 –140.
Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. 2003. Applied multiple regression/correlation analysis for the behavioral sciences(3rd ed.). Hillsdale, NJ: Erlbaum.
Corner, P. D., Kinicki, A. J., & Keats, B. W. 1994. Integrat-ing organizational and individual information pro-cessing perspectives on choice. Organization Sci-ence,5: 294 –308.
Daft, R. L., & Lengel, R. H. 1986. Organizational informa-tion requirements, media richness, and structural design.Management Science,32: 554 –571. Daft, R. L., & Weick, K. L. 1984. Toward a model of
organizations as interpretation systems.Academy of Management Review,9: 284 –295.
Farhoomand, A. F., & Ng, P. 2000.FedEx Corp.: Struc-tural transformation through e-business. Case no. HKU098, Harvard Business School, Boston.
Fornell, C., & Larcker, D. F. 1981. Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Re-search,18: 39 –50.
Gerbing, D. W., & Anderson, J. C. 1992. Monte Carlo evaluations of goodness of fit indices for structural equation models. Sociological Methods and Re-search,21: 132–160.
Gioia, D. A., & Thomas, J. B. 1996. Identity, image, and issue interpretation: Sensemaking during strategic change in academia.Administrative Science Quar-terly,41: 370 – 403.
Grant, R. M. 1996. Toward a knowledge-based theory of the firm.Strategic Management Journal,17(winter special issue): 109 –122.
Handfield, R. B., & Nichols, E. L. 2002. Supply chain redesign: Transforming supply chains into inte-grated value systems, Upper Saddle River, NJ: Prentice-Hall.
Hansen, M. T. 2002. Knowledge networks: Explaining effective knowledge sharing in multiunit companies. Organization Science,13: 232–249.
Hu, L., & Bentler, P. M. 1999. Cutoff criteria for fit in-dexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equa-tion Modeling,6: 1–55.
Huber, G. P. 1991. Organizational learning: The contrib-uting processes and the literatures. Organization Science,2: 88 –115.
Hult, G. T. M., Ketchen, D. J., & Nichols, E. L. 2000. Measuring cycle time in organizational processes. Cycle Time Research,6(1): 13–27.
Hult, G. T. M., Ketchen, D. J., & Nichols, E. L. 2002. An examination of cultural competitiveness and order fulfillment cycle time within supply chains. Acad-emy of Management Journal,45: 501–511. Jaworski, B. J., & Kohli, A. K. 1993. Market orientation:
Antecedents and consequences.Journal of Market-ing,52(July): 53–70.
Lenth, R. V. 2001. Some practical guidelines for effective sample size determination. American Statistician, 55(3): 187–193.
Lindsley, D., Brass, D., & Thomas, J. 1995. Efficacy-performance spirals: A multi-level view.Academy of Management Review,20(3): 645– 678.
Moorman, C., & Miner, A. S. 1997. The impact of organ-izational memory on new product performance and creativity.Journal of Marketing Research, 34(Feb-ruary): 91–106.
Shrout, P. E., & Fleiss, J. L. 1979. Intraclass correlations: Uses in assessing rater reliability. Psychological Bulletin,86: 420 – 428.
Thomas, J. B., Clark, S. M., & Gioia, D. A. 1993. Strategic sensemaking and organizational performance: Link-ages among scanning, interpretation, action, and out-comes.Academy of Management Journal,36: 239 – 270.
Thompson, J. D. 1967. Organizations in action. New York: McGraw Hill.
APPENDIX Measurement Scales
Respondents were asked to relate their answers to the order fulfillment process of the supply chain involving end users, corporate buyers, and external suppliers of the transportation firm studied here.
For the constructs ofachieved memory, memory ori-entation, and subjective cycle time,and for the control variable ofimportance,response options ranged from 1, “strongly disagree,” to 7, “strongly agree.”
For knowledge acquisition, information distribution, andinformation interpretation,response options ranged from 1, “strongly disagree,” to 5, “strongly agree.” Achieved Memory (adapted from Moorman & Miner [1997])
• We have a great deal of knowledge about the supply chain.
• We have a great deal of experience with the supply chain.
• We have a great deal of familiarity with the supply chain.
Knowledge Acquisition Activities (adapted from Jawor-ski and Kohli [1993])
• We meet regularly to find out what products we need in the future.
• We do a lot of in-house research on products we may need.
• We poll participants once a year to assess the quality of our supply chain services.
• We periodically review the likely effect of changes in the supply chain environment.
• Formal routines exist to uncover faulty assumptions about the supply chain.
Information Distribution Activities (adapted from Jawor-ski and Kohli [1993])
• We frequently have interdepartmental meetings to dis-cuss trends in our supply chain.
• We spend time discussing future supply chain needs. • We share data on participant satisfaction in the supply
chain on a regular basis.
• We alert participants when something important hap-pens in the supply chain.
Shared Meaning (new scale, based on Huber [1991]) • We develop a shared understanding of the available
supply chain information.
• We develop a shared understanding of the implica-tions of a supply chain activity.
Subjective Cycle Time (reverse-coded; adapted from Hult, Ketchen, and Nichols [2002])
• We are satisfied with the speediness of the supply chain (order fulfillment) process.
• Involving the corporate buyers of the XYZ corporation in decision-making shortens the supply chain (order fulfillment) process.a
• Based on our knowledge of the supply chain (order fulfillment) process, we think it is short and efficient. • The length of the supply chain (order fulfillment)
pro-cess could not be much shorter than it is today. Objective Cycle Time
• The cycle time of the supply chain (order fulfillment) process in number of days.
Control Variable 1: Importance of the Supply Chain Re-lationship (adapted from Anderson and Weitz [1992]) • We have a strong sense of loyalty to the supply chain. • We defend the supply chain when others criticize it. • Our supply chain relationship is a long-term alliance. • We are committed to each other in the supply chain. • We are patient with each other in the supply chain
when someone makes mistakes.
Control Variable 2: Frequency of the Supply Chain Rela-tionship
• Frequency of the supply chain relationship per month (i.e., order fulfillment activity).
aThis item was deleted in the purification process.
G. Tomas M. Hult([email protected])is the director of the Center for International Business Education and Re-search (MSU-CIBER) and an associate professor of mar-keting and supply chain management at Michigan State University. He received his Ph.D. from the University of Memphis. His research focuses on global strategy, supply chain management, knowledge management, and firm-level cultural competitiveness.
David J. Ketchen, Jr.(Ph.D., the Pennsylvania State sity) is a professor of management at Florida State Univer-sity. His research focuses on the multilevel determinants of superior performance, entrepreneurial strategies, the inter-section of strategy and supply chain management, and methodological issues in management research.
Stanley F. Slater(Ph.D., University of Washington) is a professor of marketing at Colorado State University. His primary research interests are marketing’s role in the implementation of business strategy and the nature and benefits of being market oriented.