You are on page 1of 14

Available online at www.sciencedirect.

com

Journal of Operations Management 26 (2008) 557–570


www.elsevier.com/locate/jom

Supply chain variability, organizational structure, and performance:


The moderating effect of demand unpredictability
Richard Germain a,*, Cindy Claycomb b, Cornelia Dröge c
a
154 College of Business, University of Louisville, Louisville, KY 40292, United States
b
Wichita State University, United States
c
Michigan State University, United States
Received 2 September 2005; received in revised form 1 September 2006; accepted 1 October 2007
Available online 13 October 2007

Abstract
Supply chain process variability is the level of inconsistency, or volatility, in the flow of goods into, through, and out of a firm.
The research investigates the links among organizational structure (formalization and integration), supply chain process variability,
and performance as moderated by environmental uncertainty. We found that in a predictable demand environment, only formal
control affects supply chain process variability, leading to improved financial results; but in an unpredictable demand environment,
only cross-functional integration affects supply chain process variability, leading to improved financial performance. We also
examined whether supply chain process variability is a complete or partial mediator of the relationship between organizational
structure and performance, and found that: (1) in a predictable demand environment, supply chain process variability completely
mediates the relationship between formal control and performance and (2) in an unpredictable demand environment, supply chain
process variability partially mediates the relationship between integration and performance. Supply chain process variability has an
inverse relationship with financial performance, regardless of the demand environment; and organizational structure provides
managers with the mechanisms to mitigate this variability’s detrimental impact on financial performance.
# 2007 Elsevier B.V. All rights reserved.

Keywords: Supply chain; Performance; Demand

1. Introduction et al., 1992; Wacker, 1987). Our focus is broader: supply


chain process variability encompasses the (in)consis-
Firms are continually searching for causes of tency in flows into and out of the firm, as well as internal
financial performance and one source is process variabilities, such as production lead-times and produc-
variability. Research to date has investigated very tion output rates. We investigate two specific organiza-
specific manufacturing process modifications to diag- tional mechanisms that, depending on environmental
nose, manage, and reduce production process variability conditions, may manage and reduce supply chain
(Antony et al., 1999; Bhatnagar and Chandra, 1994; process variability: these mechanisms are formal
Bitran and Sarkar, 1994; Kher and Fredendall, 2004; control and integration.
Lee and Apley, 2004; Lee and Tang, 1998; Melnyk Supply chain process variability, concerned with
material flow (un)evenness inbound, within, and out-
bound, is important for two reasons: (1) variability’s
* Corresponding author. Tel.: +1 502 852 4680;
negative implications for firm performance (due to both
fax: +1 502 852 7672. direct and opportunity costs) and (2) variability’s
E-mail address: richard.germain@louisville.edu (R. Germain). potential amplification, both within a firm and across

0272-6963/$ – see front matter # 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.jom.2007.10.002
558 R. Germain et al. / Journal of Operations Management 26 (2008) 557–570

firms. For a particular firm, supply chain process is accounted for? We examine whether the answer to
variability results in costs associated with overtime, this question also depends on demand unpredictability.
unused capacity, excess or insufficient inventory,
premium freight, changeovers, material handling, 2. Model overview and construct definitions
record keeping, irregularities in quality, dissatisfied
customers, disruptions to throughput improvement, and 2.1. Overview of the model
others. The direct costs of variability have a negative
influence on financial performance, but opportunity Fig. 1 presents the research framework. Our focus is
costs may also be significant. Furthermore, variability on how four constructs relate to one another under
within a firm from one source is often amplified because conditions of low versus high demand unpredictability;
it impacts other types of variability; for example, i.e., we propose that demand unpredictability is a
variability in inbound inventory flows may lead to moderator of model relationships and thus there are two
unplanned downtime (Inman and Gonsalvez, 1997). diagrams in the Figure. We begin by defining all
Finally, looking at all firms in the supply chain, constructs central to our model: formal control and
variability originating in any one firm can have a major integration (which are organizational structure con-
impact on variability in another firm even if they are structs), supply chain process variability and demand
indirectly linked through a channel intermediary. This unpredictability (which is an environmental construct).
amplification is embodied in the ‘‘bullwhip effect,’’
which states that demand variability increases as it 2.2. Organizational structure: definitions of formal
moves up the supply chain from its source (Kahn, 1987; control and integration
Lee et al., 1997a,b; Chen et al., 2000).
The research model incorporates supply chain Formalization refers to the presence of written rules,
process variability with traditional contingency theory procedures, and documents (Miller, 1991). Formal
variables: demand unpredictability (a dimension of control refers to devices used to monitor systems on a
environmental uncertainty); formal control and integra- written, codified, and rational basis (Khandwalla, 1974;
tion (dimensions of organizational structure); and Workman et al., 1998). Developing and enforcing
financial performance. Our first research question is: performance control and behavioral prescriptions
Under which demand unpredictability contingencies do improves decisions and increases the predictability of
formal control and integration reduce or alleviate performance (Pierce and Delbecq, 1977). Formal
supply chain process variability? When demand is control ensures there is a clearly specified body of
relatively predictable, the expectation is that formal rules and control mechanisms. All firms utilize control
control is the organizational mechanism that should devices to monitor performance. If formal control is
control supply chain process variability and lead to limited, then informal control (obtained through
improved financial results; on the other hand, in socialization processes) must be present if anarchy
unpredictable demand environments, cross-functional and chaos are to be avoided.
integration should control supply chain process Cross-functional integration refers to lateral com-
variability and lead to improved financial performance. munication within a firm and signifies the level of
This means that low versus high demand unpredict- coordination among different work units (Galbraith,
ability moderates the relationship between organiza- 1994; Lawrence and Lorsch, 1967; Ruekert and Walker,
tional structure and supply chain process variability. 1987). The hierarchical nature of institutions suggests
Demand unpredictability has been identified as a key that the natural communication flow is vertical: upward
contingency variable in the design of supply chains toward higher level managers and downward toward
(Fine, 1998, 2000) and in whether an efficiency versus subordinates. Effective lateral communication, prere-
flexibility strategy should be followed (Fisher, 1997), quisite for the integration of functional activities, is one
but integrating supply chain process variability into goal of cross-functional teams and cross-functional
contingency theory has yet to be studied in the extant liaison personnel. Enterprise resource systems foster
literature. Our second research question concerns the electronic integration within the firm, a process
mediating effect of supply chain process variability on underway with the transformation of material require-
the relationship between organizational structure and ments planning software since the early 1980s. The
financial performance: Is there a direct effect of formal level of cross-firm integration escalated during the late
control or integration on performance, once their 1990s with the development of the Internet and
indirect impact through supply chain process variability specialized software platforms, especially those related
R. Germain et al. / Journal of Operations Management 26 (2008) 557–570 559

Fig. 1. Theoretical framework.

to collaborative planning, forecasting, and replenish- any process – be it labor productivity, machine
ment (CPFR). In this research however, we focus on productivity, materials productivity, or total supply
integration requiring direct human interaction: cross- chain productivity – increases with the consistency and
functional teams, committees, and liaison personnel. speed by which materials flow through that process, and
productivity decreases with increases in the variability
2.3. The meaning of supply chain process associated with the flow. Process variability is measured
variability by the variance ‘‘of the timing or quantities demanded
or of the time spent in various process steps’’
Supply chain process variability may be defined as (Schmenner and Swink, 1998, pp. 102–103). The
the level of inconsistency in the flow of goods theory of swift, even flow argues for coordination of a
throughout the firm. It is concerned with material flow supply chain: the more integrated the connections
unevenness. There are three main sources of variability between supply chain exchange parties and the faster,
that plague a firm in a supply chain: (1) upstream more even the flow from raw materials to end
sources such as supplier performance (e.g., inconsistent customers, the more productive the supply chain.
or variable delivery performance); (2) internal sources
related to manufacturing such as inconsistent through- 2.4. Environmental context: demand
put and output rates, highly variable inventory levels unpredictability as moderator
(the antithesis of JIT), and inconsistent product quality;
(3) downstream sources related to customers, such as Environmental uncertainty, which firms strive to
changes in orders and inconsistent deliveries across reduce (Beckman et al., 2004), refers to the difficulty
time (Davis, 1993; Germain et al., 2001). Supply chain firms have in predicting the future because of
processes are variable when delivery times, production incomplete information or changing conditions.
lead-times, or throughput rates fluctuate (Schmenner Demand unpredictability is a major contributor to
and Swink, 1998). overall uncertainty (Davis, 1993; Chen et al., 2000).
Understanding the influence of process variability in Demand unpredictability refers to the degree to which a
a supply chain system is a goal of inventory control firm can anticipate or forecast sales and market trends:
theory and is manifested in the theory of swift, even demand unpredictability is low when sales forecasts and
flow (Davis, 1993; Schmenner and Swink, 1998). The market trends are easier to monitor and predict, while in
theory holds that a fast and even flow of materials a highly unpredictable demand environment, sales
through a process, regardless of the type of process, forecasts and market trends are more difficult to monitor
results in a more productive process. Productivity for and predict (Celly and Frazier, 1996). Demand
560 R. Germain et al. / Journal of Operations Management 26 (2008) 557–570

unpredictability can be characteristic of the general One of the principal contingency variables identified
level of demand for an industry’s products or it can early on by contingency theorists was environmental
come from particular customers and not from others. uncertainty. Burns and Stalker (1966) and Lawrence
The impact of demand unpredictability manifests and Lorsch (1967) related environmental uncertainty to
itself in a variety of ways, negatively influencing a organizational structure and concluded that unpredict-
firm’s product offerings, plans, operations, and strate- able environments require more organic structures
gies. High levels of demand unpredictability often arise while predictable environments require more mechan-
from innovative products. Relative to businesses with istic structures. Managing uncertainty through various
predictable demand, these businesses often experience structural mechanisms has been noted as an essential
less time for deliberation and choice, shorter product issue for organizational design (Thompson, 2003;
life cycles and lead-times for made-to-order products, Williamson, 1981; Workman et al., 1998). In particular,
and brief competitive advantages, but also higher formalization (a mechanistic structure element) was
contribution margins, product variety, stock out rates, hypothesized to result in better performance in
customer price-sensitivity, and markdowns for obsolete predictable environments but not necessarily in
inventory (Fine, 1998; Fisher, 1997; Grewal and unpredictable ones; integration (an organic structure
Tansuhaj, 2001). Although firms can respond to demand element) will result in better performance in unpre-
unpredictability, it is a constant source of uncertainty dictable environments. We develop this basic proposal
for firms when customer preferences are variable below in H1 and H2, but note that the two configurations
(Beckman et al., 2004). In unpredictable demand in Fig. 1 are the result of this contingency focus.
environments, it is difficult for firms to precisely
forecast fluctuations in demand attributable to, for 3.2. H1: The predictable demand environment
example, economic variables (e.g., purchase postpone- configuration
ment), marketing mix variables (e.g., price changes), or
market trends. Formal control systems work well in a predictable
Supply chain processes need stability and predict- environment as standards of performance remain
ability for effective coordination. High levels of uniform across time. A manufacturing firm that can
uncertainty in an operating context create a less correctly forecast demand for a particular time period
controlled situation. Operations are ultimately driven can plan or schedule operations at a steady rate during
by market demand patterns, and therefore unpredictable that period (Thompson, 2003). Deviations from opera-
demand implies less streamlined and highly variable tional standards can signal process problems in this
operational processes. Unpredictable demand patterns case. Such deviations indicate inefficient, poorly
mean that firms have to change production schedules, designed, or poorly executed processes. An organiza-
switch products and parts, maintain different levels of tion that faces a predictable environment would rely
inventory, and alter throughput rates and delivery primarily on formal control (i.e., standardized rules) for
schedules more frequently and at irregular intervals adaptation or corrective action in order to eliminate or
(Germain et al., 2001; Jaworski and Kohli, 1993). reduce supply chain process variability and increase
financial performance (Ruekert et al., 1985; Miller,
3. Theoretical model and hypotheses 1997; Thompson, 2003). For example, Germain et al.
(2001) found that formal performance control relates
3.1. Theoretical context: contingency theory inversely to throughput variance (an element of supply
chain process variability) and positively to financial
The fundamental premise of contingency theory is performance.
that the selection of an appropriate organizational In contrast, integration is expected to have little if
structure to ‘‘fit’’ the environment leads to higher any impact on supply chain process variability and
performance (Van de Ven and Drazin, 1985; Walker and financial performance when demand is predictable.
Ruekert, 1987). Rather than accept a deterministic logic When demand is predictable, the need for interfunc-
(e.g., all firms should be formalized) or an ‘‘every-case- tional coordination is lessened, as links between
is-different’’ approach, contingency theorists assert that functions and the functions themselves are relatively
there is a middle ground where it is possible to analyze stable across time. The goal is to design a system for
the differences in organizational structures in an orderly maximum efficiency in the face of predictability, rather
way (Ruekert et al., 1985; Miller, 1997; Birkinshaw than adjustments across functions in the face of
et al., 2002). unpredictability (Fisher, 1997). In summary, we
R. Germain et al. / Journal of Operations Management 26 (2008) 557–570 561

Table 1
Summary of the hypotheses
Independent constructs Dependent constructs
Predictable demand environment Unpredictable demand environment
Supply chain process Financial performance Supply chain process Financial performance
variability variability
Formal control H1a () H1b (+) H2a (null) H2b (null)
Integration H1c (null) H1d (null) H2c () H2d (+)
Supply chain process H3 () and equal to the H3 () and equal to
variability path in the unpredictable the path in the predictable
demand environment demand environment

propose a contingency effect: formal control, but not of operations (Thompson, 2003; Ruekert and Walker,
integration, affects supply chain process variability 1987; Workman et al., 1998). With cross-functional
when demand is predictable (first part of Fig. 1). H1 a–d integration, units become aware of how processes in one
is summarized in Table 1 (along with the other function affect another. This allows firms to create
hypotheses): unified forecasts (which are more accurate than
individual functions’ forecasts) and to make cost
H1. When demand is predictable, formal control (a) trade-offs across units. In effect, mutual adjustment is
inversely associates with supply chain process varia- accomplished by cross-functional integrative devices
bility and (b) positively associates with financial per- (Ruekert and Walker, 1987; Thompson, 2003).
formance; however, there is no association of An unpredictable environment creates administra-
integration with (c) supply chain process variability tive and operational challenges for a firm not found in
and (d) financial performance when low demand unpre- predictable environments where exceptions do not
dictability exists. dominate (Burns and Stalker, 1966; Lawrence and
Lorsch, 1967). Tasks become more complex and non-
3.3. H2: The unpredictable demand environment routine, causing emphasis to shift from formal control
configuration to exception processing (Miller and Dröge, 1986;
Miller, 1997). To handle contingencies, firms select
When an environment becomes unpredictable, new particular organizational structures to increase adapt-
complications arise for the firm. Formal control devices ability (Miller, 1987). The traditional argument is that
are inadequate because an organization faces environ- adaptability to demand unpredictability is found in
mental contingencies. Contingencies are external more ‘‘organic’’ structures, with less formal control and
uncertainties ‘‘in which the outcomes of organizational greater cross-functional integration (Burns and Stalker,
action are in part determined by the actions of elements 1966; Miles and Snow, 1978; Miller, 1991). Key
of the environment’’ (Thompson, 2003, p. 159). A firm decisions such as where to position inventory to protect
must determine when and how to act, and its cues must against stock outs when demand is unpredictable are
be taken primarily from the environment. Formal better made by organic, integrative, outward looking
control is thus expected not to affect supply chain structures (Fisher, 1997), and breaking down barriers
process variability or financial performance when between departments allows a firm to foresee and
demand is unpredictable. handle process variability (Deming, 1986). This overall
When an environment is unpredictable, the firm picture is shown in the bottom configuration of Fig. 1
should pursue cross-functional integration to ‘‘cluster and is summarized in Table 1:
capacities into self-sufficient units,’’ each equipped
with a full array of resources requisite for an H2. When demand is unpredictable, there are no asso-
organization to meet contingencies (Thompson, 2003, ciations of formal control with (a) supply chain process
p. 78). Firms faced with uncertain environments variability and (b) financial performance; however,
establish homogenous groups of specialists (e.g., in integration is (c) inversely associated with supply chain
finance, in human resources, in marketing) for ‘‘house- process variability and (d) positively associated with
keeping’’ purposes, but develop interdepartmental financial performance when high demand unpredict-
committees, teams, and task forces for coordination ability exists.
562 R. Germain et al. / Journal of Operations Management 26 (2008) 557–570

3.4. H3: The effect of supply chain process H3. Supply chain process variability and financial
variability on performance performance are inversely associated; the strengths of
this relationship (i.e., the betas) are equal across pre-
There has long been theorized a relationship between dictable and unpredictable demand environments.
variability and performance (Bhatnagar and Chandra,
1994). Deming (1986) went so far as to propose that the 3.5. Summary
key aim of management is to control variability in order
to ultimately lower costs. The general operations Our contingency theory model as a whole (see Fig. 1
management theory of ‘‘swift, even flow’’ argues that and Table 1) proposes that (1) formal control is the chief
reduced variance in materials flow and operational organizational mechanism by which supply chain
activities contribute to improved performance process variability can be reduced in predictable
(Schmenner and Swink, 1998). The more consistent demand environments (while integration plays little
the flow of materials, the more productive processes or no role) and (2) integration is the chief organizational
should be, and this implies improved financial mechanism by which supply chain process variability
performance (Germain et al., 2001). can be reduced in unpredictable demand environments
Researchers have connected financial performance (while formal control plays little or no role). It is also
inversely to variability, but any particular research has a clear that the direct impact of formal control on
narrowly defined domain of variability. For example, financial performance is proposed only in predictable
production schedule instability or variability (Inman demand environments, while the direct impact of
and Gonsalvez, 1997) and manufacturing bottlenecks integration on financial performance is proposed only in
(Davis, 1993) are treated as significant sources of unpredictable demand environments. In addition, the
excessive cost. Comparing world class manufacturers to slope of the relationship between supply chain process
other manufacturers, researchers found that stable variability and financial performance is hypothesized
schedule policies (Harrison, 1996) and process dis- equal across these different environments.
cipline and control (Oliver et al., 1994) are exhibited by
world-class performers. The logistics literature is 4. Method
particularly concerned with lead-time variance, which
drives up buffer stocks and associated holding costs 4.1. Sample
(e.g., Bowersox and Closs, 1996). Late delivery may
lead to expediting costs, a search for substitute sources The ‘‘executive list’’ of the Institute of Supply
of supply, and plant slowdown or shutdown, while early Management’s (ISM) manufacturer’s membership was
delivery may lead to de-expediting costs (e.g., storage obtained. A member having reached the senior most
costs). purchasing title in their firm or division obtains the
Fluctuations in production schedules or customer ‘‘executive’’ distinction. This sampling frame contained
delivery cycles increase inventory levels and associated the name and contact information of 1264 managers.
holding costs, time spent on set-ups and changeovers, as For the purpose of statistical power, a target sample
well as transportation costs, thus negatively influencing size of 200 was set. Prior experience with the sampling
financial performance. Low process variability and processes described below indicated an expected
reliable and stable delivery schedules associate with response rate in the range of 50% from subjects pre-
high-performing manufacturing plants (Mapes et al., qualified by telephone. We therefore set a target of
2000). Process variance control provides additional obtaining agreements of participation from about 400
power and economic value to a company’s processes respondents. Individuals were randomly selected from
through increased productivity and lower costs (Bohn, the sampling frame and contacted by telephone. The
1994; Deming, 1986). Germain et al. (2001) found a conversation that ensued confirmed employment by a
negative association between throughput variance and manufacturer (as a check against frame misclassifica-
performance. By analogy, all these examples suggest tions). Then, a key informant check was undertaken. If
that supply chain process variability, a broader they possessed sufficient knowledge to complete the
variability construct than most researchers have studied, survey, they were asked to participate. In 159 cases, one
adversely influences costs or capacity utilization, and of the criteria was not met and these managers were
therefore overall productivity. Thus supply chain randomly replaced with others from the ISM list. Thus,
process variability inversely relates to financial perfor- in the end, 402 ‘‘executive’’ purchasing managers were
mance. contacted who were employed by a manufacturer, who
R. Germain et al. / Journal of Operations Management 26 (2008) 557–570 563

possessed sufficient knowledge to complete the survey, Table 2 provides the 2-digit SIC classification of the
and who agreed to participate. Although initial contact firms in the sample. For the purpose of x2-testing
was by phone, all respondents were promised anon- (described below), industries with fewer than 10 firms
ymity and did not have to identify themselves when were consolidated into an ‘‘other’’ category. The most
returning the survey instrument. frequent SIC group is 28 (chemicals, n = 34), followed
In 78 cases, we were able to acquire the name of a by SIC 34 (fabricated metal products, n = 21), SIC 30
second qualified person who was willing to complete the (rubber, n = 18), SIC 20 (food, n = 17), SIC 35
survey. They were identified by the previously qualified (industrial machinery, n = 17), and SIC 36 (electronic
ISM member. Multiple respondents per firm were and electrical equipment, n = 14).
obtained to assess inter-rater agreement and to thereby
lessen concern with mono-respondent bias. A total of 480 4.2. Measurement of the moderator: demand
surveys were sent to 402 firms. Of these, 227 useable unpredictability
surveys were returned from 210 firms for response rates
of 47% for surveys (or 227/480) and 52% for firms (or Three items were used to measure demand unpre-
210/402). However, two surveys were later dropped dictability (Celly and Frazier, 1996) for the SBU or
because they had too many missing values, and thus the division. On 7-point semantic differential scales,
sample size for the subsequent analyses was 208. respondents were asked whether: (1) ‘‘sales are
Respondent title levels included directors (66%), predictable . . . unpredictable,’’ (2) ‘‘sales forecasts
managers (16%), and vice-presidents (14%) and their are likely to be accurate . . . inaccurate,’’ and (3)
functional areas included purchasing (72%), materials ‘‘market trends are easy to monitor . . . difficult to
management (10%), and manufacturing (5%). The firms monitor.’’ The mean of these three items was taken and
in the sample are large, with average annual sales of the sample was then split as close as possible to the
US$ 1.406 billion. They ranged from US$ 1.25 million median of the composite score (Calantone et al., 2002).
in sales to US$ 42 billion, with a standard deviation of This led to n = 112 in the predictable demand group
US$ 4.188 billion. The average number of employees (mean of 3.06 for unpredictability) and n = 96 in the
was 4573, ranging from 15 to 122,000, with a standard unpredictable demand group (mean of 4.97 for
deviation of 13,132. The respondents were asked to unpredictability). The option of dividing the sample
answer the questionnaire for their SBU or division into three and then comparing the ‘‘high’’ versus ‘‘low’’
however, not for the entire firm (or for the entire firm if it groups (and discarding the middle) was not imple-
was the SBU). mented. This option, which costs sample size, is usually

Table 2
Industry classification, total and by demand unpredictability group
2-Digit SIC Industry description Demand environment Total Percentage
of total (%)
Predictable Unpredictable
20 Food & kindred products 9 8 17 8
26 Paper and allied products 6 6 12 6
28 Chemicals and allied products 18 16 34 16
30 Rubber and miscellaneous plastics products 12 6 18 9
33 Primary metal industries 7 3 10 5
34 Fabricated metal products except machinery 12 9 21 10
and transportation equipment
35 Industrial and commercial machinery and 3 14 17 8
computer equipment
36 Electronic and electrical equipment/components, 7 7 14 7
except computer equipment
37 Transportation equipment 4 6 10 5
39 Miscellaneous manufacturing industries 7 5 12 6
All other 2-digit SIC groups 26 16 42 20
Missing 1 – 1 –
Total 112 96 208 100
2 2
x -test of independence between 2-digit SIC group and demand unpredictability group: x = 13.420; d.f. = 10; p = .201 (n = 207 due to one firm not
providing their 2-digit SIC group).
564 R. Germain et al. / Journal of Operations Management 26 (2008) 557–570

Table 3
Measurement model and two-group confirmatory factor analysis results
Constructs and measurement items Loading t-Value
Formal control (r = .77; rwg = .77): Rate the extent to which the following control devices used to gather information about the performance of your
firm. 7-point scales, endpoints of ‘‘rarely used’’ and ‘‘often used.’’
x1 A comprehensive management control & information system .652 9.636
x2 Use of cost centers for cost control .747 11.066
x3 Use of profit centers and profit targets .706 10.316
x4 Quality control of operations using sampling & other methods .582 8.159
Integration (r = .81; rwg = .76): In assuring the compatibility among decision in one area (e.g., purchasing) with those in other areas (e.g.,
production), to what extent are the following integrative mechanisms used? 7-point scales with endpoints of ‘‘rarely used’’ and ‘‘often used.’’
x5 Interdepartmental committees that are set up to .810 12.755
allow departments to engage in joint decision-
making on an ongoing basis
x6 Cross-functional teams that are temporary bodies set up .849 13.536
to facilitate inter-departmental
collaboration on a specific project control
x7 Liaison personnel whose specific job it is to coordinate .658 9.829
the efforts of several departments for the purposes
of a specific project
Supply chain process variability (r = .63; rwg = .81): Rate the following on consistency. 7-pt scales, endpoints of ‘‘always the same, very consistent,
low variance’’ and ‘‘rarely the same, very inconsistent, high variance.’’
y1 Amount of time for shipments to arrive from key suppliers .657 8.047
y2 Amount of time for shipments to reach key customers .613 7.545
y3 Production lead-time (fixed production schedule) .436 5.323
y4 Daily production output rate .528 9.829
Financial performance (r = .95; rwg = .93): Rate performance in each of the following areas. 7-point scales with endpoints of ‘‘well below industry
average’’ and ‘‘well above industry average.’’
x5 Average return on investment over the past 3 years .912 16.847
x6 Average profit over the past 3 years .956 18.272
x7 Profit growth over the past 3 years .922 17.175
Notes: Loadings are common metric completely standardized estimates; all t-values significant at p < .01. r = scale composite reliability;
rwg = interrater agreement score. CFA fit statistics: x2 = 242.591; d.f. = 170; p < .01; RMSEA = .065; CFI = .941.

chosen to accentuate differences, thereby making it respect to the number of employees (t = .934; p > .10),
more likely that significant differences or models will annual sales (t = 1.380; p > .10), or production tech-
be found. In our case, the common median split nology routineness (t = 1.040; p > .10; Khandwalla,
supported most of the hypothesized differences (as will 1974). Production technology routineness was mea-
be shown in Section 4), and thus eliminating the sured on five items (i.e., custom, small batch, large
‘‘muddled middle’’ was not necessary. batch, mass assembly, and continuous process technol-
We examined whether these two groups differ in ogy) on 7-point scale with endpoints of ‘‘1 = used
significant ways. It is conceivable that these two rarely’’ and 7 = used almost exclusively.’’ The items
demand unpredictability groups are disproportionately were given weights of 1 through 5 and the sum was
populated by different industries, for example. To assess taken. The results suggest that differences in demand
this possibility, a x2-test of independence between the unpredictability groups are not being confounded by
SIC groups and the demand unpredictability groups was industry type, firm size, or production technology
examined. Table 2 presents the 2-digit SIC distribution routineness.
by demand unpredictability group. To lessen the
number of cells with expected values below five in 4.3. Measurement of model constructs
the x2-test, all 2-digit SIC groups with fewer than 10
firms were combined into an ‘‘other’’ category. The x2- Table 3 shows instructions to respondents and the
test is not significant (x2 = 13.420, d.f. = 10; p = .201). exact scale items for the remaining constructs. Formal
Furthermore, independent sample t-tests revealed that control (Khandwalla, 1974) and integration (Miller and
the demand unpredictability groups did not differ with Dröge, 1986) were measured using established scales.
R. Germain et al. / Journal of Operations Management 26 (2008) 557–570 565

In the case of formal control, respondents rated items 1993). In the initial 2-group confirmatory factor analysis
related to use of cost/profit centers, quality methods, (CFA), factor loadings and errors were estimated freely
and management information and control systems on 7- across groups (baseline x2 = 209.060; d.f. = 143;
point scales with endpoints of ‘‘rarely used’’ and ‘‘often p < .01; this is equivalent to estimating each separately).
used.’’ For integration, respondents rated reliance on The loadings and errors were then declared invariant, or
cross-functional teams, liaison personal, and temporary equal, across groups (x2 = 242.591; d.f. = 170; p < .01).
bodies brought together for specific projects on seven- The difference in x2 between the two models is not
point scales with endpoints ‘‘rarely used’’ and ‘‘often significant (Dx2 = 37.571; Dd.f. = 28; p > .10). This
used.’’ means that no asymmetries exist in the loadings or errors
For supply chain process variability, respondents across the two groups; i.e., the loadings and errors can be
rated consistency in inbound lead-times, outbound modeled as equal across the two groups in all subsequent
delivery cycles, production lead-times, and daily analyses. This is a critical result because such measure-
production output rates (Germain et al., 2001), thus ment equality greatly facilitates maximum likelihood
covering the domain of ‘‘incoming, within, outgoing.’’ estimation in multi-group structural equations modeling
Seven point scales were used with endpoints of ‘‘always (SEM) (Hair et al., 1998).
the same, very consistent, low variance’’ and ‘‘rarely the Several measures of overall fit were then assessed.
same, very inconsistent, high variance.’’ Note that the The root mean square error of approximation (RMSEA)
scale endpoints are very descriptive so that respondents equaled .065 and is below the maximum acceptable
can understand the variability construct. High ratings on value of .08. The confirmatory fit index (CFI) equaled
these scales suggest that the firm unevenly processes .941 and is above the minimum acceptable value of .90.
materials and that the flow of goods is relatively The goodness of fit index (GFI) was low (.821) and
inconsistent. below the acceptable cut-off of .90. However, the non-
Financial performance of the SBU was measured normed fit index (NNFI) value of .923 exceeds the
using Miller’s (1991) scale. The three-item perceptual acceptable minimum of .90. Based on these mixed
construct tapped return on investment, profit, and profit results, we tentatively conclude an adequate measure-
growth on seven-point scales with endpoints of ‘‘well ment model fit exists.
below industry average’’ and ‘‘well above industry The results of the 2-group CFA with loadings and
average.’’ Respondents rated performance over the past errors set equal across groups are summarized in
3 years to offset particularly good or bad years due to Table 3. All standardized loadings exceed .40 ( p < .01)
unusual circumstances. We did not ask for ‘‘hard’’ and the scale composite reliabilities exceed .60. This
performance measures because usually only about a suggests adequate reliability and convergent validity.
third of respondents provide them (it is often against One final model restriction was examined: the correla-
company policy), while an unknown number refuse to tion between formal control and integration was set
participate at all when they see such questions. These equal across groups. Model fit was not impaired
perceptual performance measures cannot be verified (Dx2 = .09; Dd.f. = 1; p > .10). The correlation
against secondary data because they pertain to SBUs between the two exogenous constructs was declared
and ‘‘hard’’ data for this unit of analysis is very difficult invariant across groups in the SEM.
or impossible to obtain.
5. Results
4.4. Reliability and validity
The first 2-group SEM model tested (Fig. 2)
Interrater agreement scores (rwg) were estimated for corresponds directly to the hypothesized model shown
the 17 instances where two respondents were obtained in Fig. 1. Paths from formal control to supply chain
per firm. Values over .70 are acceptable and justify using process variability and to financial performance were
the mean of the two respondents (James et al., 1993). estimated in the ‘‘low’’ demand unpredictability group
The interrater agreement scores, provided in Table 3, all only; paths from integration to supply chain process
exceed .70, the acceptable minimum. The data was variability and financial performance were estimated in
collapsed down to 208 firms by taking the average the ‘‘high’’ demand unpredictability group only; and the
across respondents when two existed. path from supply chain process variability to financial
Testing of the measurement model and hypotheses performance was set equal across groups. This
was conducted using 2-group LISREL with the specification directly tests the model as stated in the
covariance matrices as input (Jöreskog and Sörbom, hypotheses. Furthermore, factor loadings, errors, and
566 R. Germain et al. / Journal of Operations Management 26 (2008) 557–570

Fig. 2. Empirical model.

the non-causal correlation between formal control and supporting H2c and H2d. The path from supply chain
integration were set equal across groups (as justified by process variability to financial performance is significant
the measurement model results). With the exception of (b21 = .210; p < .05; this was set equal across groups).
the low GFI (.826), the remaining fit indices were The remaining hypotheses were tested by selectively
adequate: RMSEA = .062; CFI = .944; NNFI = .942. estimating paths in specific groups. The results of these
The x2 = 245.464 (d.f. = 176) was significant ( p < .01), single degree of freedom Dx2-tests are summarized in
providing strong support for the model. Table 5. The first row of Table 5 shows that no
Table 4 summarizes the results of testing the hypo- significant improvement in fit was observed when the
thesized model. In the predictable demand group, the path path from integration to supply chain process variability
from formal control to supply chain process variability is was estimated in the predictable demand group
significant (g11 = .295; p < .05), but the path to (Dx2 = .128; p > .10). This path is not significant and
financial performance is not (g21 = .136; n.s.). H1a, thus H1c is supported. The next three rows in Table 5
but not H1b, is supported. In the unpredictable demand can be similarly interpreted: the inclusion and estima-
group, the paths from integration to supply chain process tion of paths that had been set to zero in the
variability (g12 = .337; p < .01) and to financial hypothesized model led to no significant improvement
performance (g22 = .239; p < .05) are significant, in model fit. This supports H1d, H2a, and H2b.

Table 4
Summary of hypothesized Two-group LISREL model results
Parameter (hypothesis) Parameter estimate (t-value)
Demand environment
Predictable Unpredictable
g11: Formal control ! Supply chain process variability .295 (2.239b) Set to zero
g21: Formal control ! Financial performance .136 (1.249) Set to zero
g12: Integration ! Supply chain process variability Set to zero .337 (2.403a)
g22: Integration ! Financial performance Set to zero .239 (2.027b)
b21: Supply chain process variability ! Financial performance (set equal) .210 (2.214b)
Notes: Parameter estimates are common metric completely standardized estimates. Fit statistics: x2 = 245.464; d.f. = 176; p = <.01;
RMSEA = .062; CFI = .944; GFI = .826; NNFI = .942. a, p < .01; b, p < .05.
R. Germain et al. / Journal of Operations Management 26 (2008) 557–570 567

Table 5
Summary of Dx2 (Dd.f. = 1) tests
Model Model x2 (d.f.) Dx2 between hypothesized Conclusion
and freed model ( p-value)
Integration ! supply chain process 245.336 (175) .128 (Dd.f. = 1; p > .10) No significant change in fit.
variability estimated in predictable H1c is supported
demand environment
Integration ! financial performance 245.433 (175) .031 (Dd.f. = 1; p > .10) No significant change in fit.
estimated in predictable H1d is supported
demand environment
Formal control ! supply chain 244.735 (175) .729 (Dd.f. = 1; p > .10) No significant change in
process variability estimated in fit. H2a is supported
unpredictable demand environment
Formal control ! financial 243.430 (175) 2.034 (Dd.f. = 1; p > .10) No significant change
performance estimated in in fit. H2b is supported
unpredictable demand
environment
Supply chain process 245.361 (175) .083 (Dd.f. = 1; p > .10) No significant change in fit.
variability ! financial The effect is equal across
performance estimated groups: H3 is supported
freely in each group
All of the above 241.346 (171) 4.118 (Dd.f. = 5; p > .10) Simultaneous support
of the above

The next row of Table 5 shows what happens when variability and influencing performance. The research
the path between supply chain process variability and contributes to contingency theory and the theory of
financial performance is estimated freely in each group. swift, even flow in several ways.
The paths in the predictable (b2,1 = .225; t = 1.702; First, from a contingency perspective, managers
p < .10) and in the unpredictable demand groups should understand that: (1) in a predictable demand
(b2,1 = .208; t = 1.698; p < .10) are both signifi- environment, formal control is a driver of performance
cant, however, separate estimations did not lead to a (total effect = .201; p < .05), but its impact is com-
significant change in fit (Dx2 = .083; p > .10). The pletely mediated by supply chain process variability;
paths are not different across groups and H3 is and (2) in an unpredictable demand environment,
supported. integration is a driver of performance (total
A final model simultaneously incorporating the effect = .305; p < .01) and its impact is partially
above findings was examined. The null paths were mediated by supply chain process variability.
estimated and the path from supply chain process Second, supply chain process variability is equally
variability to financial performance was estimated detrimental to financial performance in unpredictable
freely in each group. The last row in Table 5 shows and predictable demand environments. One way to
that the difference between this model and the interpret supply chain process variability is as a proxy
hypothesized one is not significant (Dx2 = 4.118; for depth and breadth of knowledge, which is a key
Dd.f. = 5; p > .10). This suggests that the original resource to engender long-lived competitive advantage
hypothesized model is a parsimonious representation of (Roth, 1996a,b). Deming (1986) long argued that
the data (with the exception that the hypothesized path variability and knowledge are inversely related; indeed,
from formal control to financial performance was found the quest for knowledge is driven by the existence of
to be nonsignificant in the predictable demand group). unexplained variability (Anderson et al., 1994). Mana-
ging supply chain process variability successfully
6. Discussion: implications and limitations through organizational mechanisms – which ultimately
represents managerial ‘‘knowing how’’ (Grant, 1996) –
A summary of the research findings is presented in is based on deeply rooted knowledge, and thus can be an
Fig. 3. Our research theme is supported: demand intangible core competency. For example, General
uncertainty is a key contingency variable that firms Motors was able to closely imitate the lean, low
should take into account when selecting an organiza- variance operations of Toyota only after many years of
tional structure for managing supply chain process experimentation, joint ventures, and worker/manager
568 R. Germain et al. / Journal of Operations Management 26 (2008) 557–570

Fig. 3. Summary of findings.

education and training. However, with rapid supply Further research should address the antecedents of both
chain process change, effective managerial knowledge speed and evenness, whether they are related to one
regarding managing supply chain process variability another, and whether both predict financial performance
deteriorates. According to Bohn (1994, p. 71), ‘‘one of when modeled together. Managers should also under-
the most painful forms of ignorance is false knowl- stand that formal control and integration are not the only
edge’’; hence managers must continually evaluate mechanisms that the firm may use to control supply
supply chain process variability and ways to alleviate chain process variability. Others include: (1) just-in-
it. Managers should understand that supply chain time, when conceived of as a variance reducing set of
process variability associates with poorer performance practices; (2) every day low pricing, which reduces the
regardless of the demand unpredictability environment. effects of forward buying that result from trade and
We should also note that scope of our supply chain retail promotions; (3) CPFR, which reduces supply
process variability construct is focused. The construct chain surprises through the cross-firm exchange of
may be expanded to: (1) process variability as spanning planned promotional activities and joint forecast
the entire supply chain (e.g., incorporating second tier development (Wisner et al., 2005); (4) concurrent
suppliers) and (2) process variability in domains other engineering or product design for manufacturability/
than flow evenness (e.g., product quality variability). distribution (Fixson, 2005).
Variability is important to sets of firms in a supply chain Third, while the research supported the effect of
because it is amplified as it moves upstream from its environment on how organizational structure associates
source (i.e., the ‘‘bullwhip effect;’’ see Kahn, 1987; Lee with supply chain process variability and financial
et al., 1997a,b; Chen et al., 2000). The bullwhip effect in performance, managers should understand that addi-
the entire supply chain is an important area for further tional environmental moderators and dimensions of
research. For example, does supply chain process organizational structure exist. Environmental modera-
variability, operationalized either as in our research or tors include the industry’s pace of technological change,
in an expanded fashion, affect the performance of the exchange rate variability, regulation variability, and
supply chain as a whole? seasonality. The latter may associate with supply chain
While the theory of swift, even flow is supported, we process variability, yet this uncontrollable but partially
did not model the ‘‘swiftness’’ component of the theory. predictable contextual factor was not one we considered
R. Germain et al. / Journal of Operations Management 26 (2008) 557–570 569

separately. Additional dimensions of organizational flexibility and an ability to innovate (Benner and
structure include decentralization and complexity or Tushman, 2003).
specialization (Miller and Dröge, 1986). It may be that Finally, some additional limitations should be noted.
decentralization in the supply chain domain (e.g., These include the lack of ‘‘hard’’ performance measures
decentralization of transportation and production (e.g., actual return on investment, which is difficult to
scheduling decisions) associates with lower supply obtain for SBUs), the lack of a rigorous test for non-
chain process variability only when demand is response bias, and the use of purchasing executives as
unpredictable. Uncertain environments require flex- survey respondents. These shortcomings aside, the
ibility, and decentralization may provide managers the research was successful in casting supply chain process
leeway necessary to adapt and control process variability within a contingency theory framework.
variability. However, when the environment is pre-
dictable, complexity and centralization (typically
associated with mechanistic structures) may foster less References
process variability. Further research should thus
Anand, G., Ward, P.T., 2004. Fit, flexibility and performance in
integrate decentralization and complexity into the manufacturing: Coping with dynamic environments. Production
theoretical framework. and Operations Management 13 (4), 369–385.
Fourth, managers should consider better organizing Anderson, J.C., Rungtusanatham, M., Schroeder, R.G., 1994. A theory
their firms’ suppliers, customers, and partners; this of quality management underlying the Deming management
implies that cross-firm organization and governance method. Academy of Management Review 19 (3), 472–509.
Antony, J., Hughes, M., Kaye, M., 1999. Reducing manufacturing
systems are important areas for further research. For process variability using experimental design technique: a case
instance, which cross-firm governance systems reduce study. Integrated Manufacturing Systems 10 (3), 162–169.
the firm’s variability and/or chain wide variability? To Beckman, C.M., Haunschild, P.R., Phillips, D.J., 2004. Friends or
give a more specific example, researchers could look at strangers? Firm-specific uncertainty, market uncertainty,
and network partner selection. Organization Science 15 (3),
the effect of network partner selection as a way to
259–275.
control supply chain process variability in predictable Benner, M.J., Tushman, M.L., 2003. Exploitation, exploration, and
versus unpredictable environments, based on the work process management: The productivity dilemma revisited. Acad-
of Beckman et al. (2004). It could be that firms form emy of Management Review 28 (3), 238–252.
new relationships with new partners for exploration Bhatnagar, R., Chandra, P., 1994. Variability in assembly and compet-
purposes and/or additional relationships with existing ing systems: Effect on performance and recovery. IIE Transactions
26 (5), 18–31.
partners for exploitation (March, 1991), depending on Birkinshaw, J., Nobel, R., Ridderstråle, J., 2002. Knowledge as a
the predictability of the demand environment. Beckman contingency variable: Do the characteristics of knowledge predict
et al. (2004, p. 273) found ‘‘that firms use networks organization structure? Organization Science 13 (3), 274–289.
primarily as a means to exploit under conditions of Bitran, G.R., Sarkar, D., 1994. Throughput analysis in manufacturing
market uncertainty,’’ supporting the well-established networks. European Journal of Operational Research 74 (3), 448–
465.
idea that networks provide access to knowledge (which Bohn, R.E., 1994. Measuring and managing technological knowledge.
reduces uncertainty). If supply chain process variability Sloan Management Review 36 (1), 61–73.
is inversely related to knowledge, then the more Bowersox, D., Closs, D., 1996. Logistical Management. McGraw-
networking, the less variability there should be. Hill, New York.
Burns, T., Stalker, G.M., 1966. The Management of Innovation,
Fifth, further research should examine the relation-
second ed. Tavistock, London.
ship between supply chain process variability and Calantone, R., Dröge, C., Vickery, S., 2002. Investigating the man-
agility. Supply chain agility concerns flexibility ufacturing–marketing interface in new product development: Does
(Swafford et al., 2006) and agility may thus be a context affect the strength of relationships? Journal of Operations
way of coping with supply chain process variability. In Management 20 (3), 273–287.
other words, agility may not eliminate variability but Celly, K.S., Frazier, G., 1996. Outcome-based and behavior-based
coordination efforts in channel relationships. Journal of Marketing
rather mitigate variability’s impact. It may be that Research 33 (2), 200–210.
investments in agility really pay off only when demand Chen, F., Drezner, Z., Ryan, J., Simchi-Levi, D., 2000. Quantifying the
is unpredictable, for example (Anand and Ward, 2004; bullwhip effect in a simple supply chain: The impact of forecast-
Fisher, 1997). A predictable environment may require ing, lead times and information. Management Science 46 (3), 436–
supply chain efficiency, while an unpredictable 443.
Davis, T., 1993. Effective supply chain management. Sloan Manage-
demand environment may require flexibility or agility. ment Review 34 (2), 35–46.
Focusing only on productivity gains may lead to long- Deming, W.E., 1986. Out of the Crisis. Massachusetts Institute of
term organizational difficulties if the firm forgoes Technology, Cambridge, MA.
570 R. Germain et al. / Journal of Operations Management 26 (2008) 557–570

Fine, C.H., 1998. Clockspeed Winning Industry Control in the Age of March, J.G., 1991. Exploration and exploitation in organizational
Temporary Advantage. Perseus Book, Reading, MA. learning. Organization Science 2 (1), 71–87.
Fine, C.H., 2000. Clockspeed-based strategies for supply chain Melnyk, S.A., Denzler, D.R., Fredendall, L., 1992. Variance control
design. Production and Operations Management 9 (3), 213–221. vs. dispatching efficiency. Production and Inventory Management
Fisher, M.L., 1997. What is the right supply chain for your product? Journal 33 (3), 6–13.
Harvard Business Review 75 (2), 105–117. Miles, R., Snow, C., 1978. Organizational Strategy, Structure and
Fixson, S.K., 2005. Product architecture assessment: A tool to link Process. McGraw-Hill, New York.
product, process, and supply chain design decisions. Journal of Miller, D., 1987. Strategy making and structure: Analysis and impli-
Operations Management 23 (3–4), 345–369. cations for performance. Academy of Management Journal 30 (1),
Galbraith, J.R., 1994. Competing with Flexible Lateral Organizations, 7–32.
second ed. Addison-Wesley, Reading, MA. Miller, D., 1991. Stale in the saddle: CEO tenure and the match
Germain, R., Dröge, C., Christensen, W., 2001. The mediating role of between organization and environment. Management Science 37
operations knowledge in the relationship of context with perfor- (1), 34–52.
mance. Journal of Operations Management 19 (ER4), 453–469. Miller, D., 1997. Configurations revisited. Strategic Management
Grant, R.M., 1996. Toward a knowledge-based theory of the firm. Journal 11 (Summer), 111–125.
Strategic Management Journal 17 (10), 109–122. Miller, D., Dröge, C., 1986. Psychological and traditional determi-
Grewal, R., Tansuhaj, P., 2001. Building organizational capabilities nants of structure. Administrative Science Quarterly 31 (4), 539–
for managing economic crisis: The role of market orientation and 560.
strategic flexibility. Journal of Marketing 65 (2), 67–80. Oliver, N., Delbridge, R., Jones, D., Lowe, J., 1994. World class
Hair Jr., J.F., Tatham, R.L., Anderson, R.E., Black, W., 1998. Multi- manufacturing: Further evidence in the lean production debate.
variate Data Analysis, fifth ed. Prentice-Hall, Upper Saddle River, British Journal of Management 5 (Special Issue), S53–S63.
NJ. Pierce, J.L., Delbecq, A.L., 1977. Organization structure, individual
Harrison, A., 1996. An investigation of the impact of schedule stability attitudes and innovation. Academy of Management Review 2 (1),
on supplier responsiveness. International Journal of Logistics 27–37.
Management 7 (1), 83–91. Roth, A.V., 1996a. Achieving strategic agility through economies of
Inman, R.R., Gonsalvez, D.J.A., 1997. The causes of schedule knowledge. Strategy and Leadership 24 (2), 30–37.
instability in an automotive supply chain. Production and Inven- Roth, A.V., 1996b. Neo-operations strategy: Linking capabilities-
tory Management Journal 38 (2), 26–31. based competition to technology. In: Gaynor, G.H. (Ed.), Neo-
James, L.R., Demaree, R.G., Wolf, G., 1993. rwg: an assessment of Operations Strategy: Linking Capabilities-Based Competition to
within-group interrater agreement. Journal of Applied Psychology Technology. McGraw-Hill, New York, pp. 38.1–38.44.
78 (2), 306–309. Ruekert, R., Walker, O., 1987. Marketing’s interaction with other
Jaworski, B.J., Kohli, A.K., 1993. Market orientation: Antecedents functional units: A conceptual framework and empirical evidence.
and consequences. Journal of Marketing 57 (3), 53–70. Journal of Marketing 51 (1), 1–19.
Jöreskog, K.G., Sörbom, D., 1993. LISREL 8: User’s Reference Ruekert, R., Walker, O., Roering, K., 1985. The organization of
Guide. Scientific Software International, Chicago, IL. marketing activities: A contingency theory of structure and per-
Kahn, J., 1987. Inventories and the volatility of production. The formance. Journal of Marketing 49 (Winter), 13–25.
American Economic Review 77, 667–679. Schmenner, R.W., Swink, M.L., 1998. On theory in operations
Khandwalla, P.N., 1974. Mass output orientation of operations tech- management. Journal of Operations Management 17 (1), 97–113.
nology and organizational structure. Administrative Science Quar- Swafford, P.M., Ghosh, S., Murthy, N., 2006. The antecedents of
terly 19 (1), 74–97. supply chain agility of a firm: Scale development and model
Kher, H.V., Fredendall, L.D., 2004. Comparing variance reduction to testing. Journal of Operations Management 27 (2), 170–188.
managing system variance in a job shop. Computers and Industrial Thompson, J.D., 2003. Organizations in Action, second ed. Transac-
Engineering 46 (1), 101–120. tion Publishers, New Brunswick.
Lawrence, P.R., Lorsch, J.W., 1967. Organization and Environment: Van de Ven, A.H., Drazin, R., 1985. The concept of fit in contingency
Managing Differentiation and Integration. Richard D. Irwin, theory. Research in Organizational Behavior 7, 333–365.
Homewood, IL. Wacker, J.G., 1987. The complementary nature of manufacturing
Lee, H.Y., Apley, D.W., 2004. Diagnosing manufacturing variation goals by their relationship to throughput time: A theory of internal
using second-order and fourth-order statistics. International Jour- variability of production systems. Journal of Operations Manage-
nal of Flexible Manufacturing Systems 16 (1), 45–64. ment 7 (1–2), 91–106.
Lee, H.L., Tang, C.S., 1998. Variability reduction through operations Walker, O., Ruekert, R., 1987. Marketing’s role in the implementation
reversal. Management Science 44 (2), 162–172. of business strategies: A critical review and conceptual frame-
Lee, H., Padmanabhan, P., Whang, S., 1997a. The bullwhip effect in work. Journal of Marketing 51 (July), 15–33.
supply chains. Sloan Management Review 38 (3), 93–102. Williamson, O.E., 1981. The economics of organization: The transac-
Lee, H., Padmanabhan, P., Whang, S., 1997b. Information distortion in tion cost approach. American Journal of Sociology 87 (3), 548–577.
a supply chain: The bullwhip effect. Management Science 43 (4), Wisner, J.D., Keong Leong, G., Tan, K., 2005. Principles of Supply
546–558. Chain Management. Thompson, New York.
Mapes, J., Szwejczewski, M., New, C., 2000. Process variability and Workman, J., Homburg, C., Gruner, K., 1998. Marketing organization:
its effect on plant performance. International Journal of Opera- An integrative framework of dimensions and determinants. Jour-
tions and Production Management 20 (7), 792–808. nal of Marketing 62 (July), 21–41.

You might also like