Linking supply chain network complexity to interdependence and risk-assessment: scale development and empirical investigation.
Chakraborty, Samyadip
Introduction
As a buyer firm plans of strategizing its relationship with its
major suppliers, a plethora of vital questions peep in the minds of the
decision making managers: what are the pros and cons involved? How
complex is the network environment? What will be the level of
dependence? What are the alternatives? What is the risk involved? While
searching for answers to these frequently asked questions, often one
word keeps flashing in mind, 'network complexity' The word
"network complexity" nowadays has become a very common
vocabulary in supply chain parlance and frequently heard in managerial
discussions on network issues. It has been highlighted as a major
breeding ground of supply chain issues and conflicting network wide
partner relationships (Gilmore 2008; Bode, Wagner 2015). One of the most
vital and omnipresent stumbling blocks that stares glaringly at the
managers before such decision making is, how complex is the system or
the process or the network in the question? Thus, network complexity has
become inherent to today's business scenarios, which cannot be
avoided; rather understanding the complexity involved and quantifying it
becomes vital for making effective and viable decisions aimed at
managing complexity. In modern business setting the competition for
competitive advantage and excellence has progressed to the very next
level where the competition is no longer between firms, rather between
their respective supply chains. However, this does not undermine the
fact that complexity and conflicting interest mark most of the supply
chain network relationships. Thus the challenges that the supply chain
managers face these days are also unique and are actually a mix of three
streams of highly sensitive balancing blocks which the managers should
carefully balance without underestimating any of them. Those three
stumbling blocks that managers face and need to manage and balance are:
complexity management, risk assessment and last but not the least
understanding the interdependence of their own firm's processes on
their partner firm's processes. However the understanding of
network complexity, here after interchangeably also referred to as
supply chain complexity or in a even generic sense supply chain network
complexity, needs revisiting for a better clarity. Thus this paper aims
at first establishing a rational framework linking those three stumbling
blocks that managers face in course of their day to day supply chain
management duties and then progresses to empirically validate the
hypothesized relationships based on perceptions of the industry
practitioners (i.e. the managers); thereby aiming at providing supply
chain managers with a handy empirically proved understanding framework
indicating how perception about supply chain network complexity impacts
inter-dependence and risk assessment between partners in the network.
Extant literature highlights that complexity in the supply chain
network is critical for understanding network transactions and
relationships. Studies highlight that complexity impacts performance of
the chain and also that of the individual network actors (Bode, Wagner
2015; Choi, Krause 2006; Waldrop 1992). Though, different studies have
defined complexity from different perspectives, a clear and easily
comprehendible working definition of supply chain network complexity
remains lacking. So arises the need to propose an acceptable working
definition and also develop a scale dealing with supply chain (SC)
network complexity (Manuj, Sahin 2011; Milgate 2001). Extant studies
took the view of SC network complexity as multi-dimensional and
multi-faceted phenomenon affected by varied sources and described
complexity from different perspectives: number of elements and
subsystems (Bozarth et al. 2009; Manuj, Sahin 2011; Choi, Krause 2006;
Choi, Hong 2002; Handfield, Nichols 1999) and high number of elements
affecting the complexity and the supply chain network functioning in
terms of disruptions (Chopra, Sodhi 2014; Bode, Wagner 2015); quality
and nature of relationship (Lian, Laing 2004; Ferlie, Pettigrew 1996);
Inter-relationship between network elements (Choi, Krause 2006; Wycisk
et al. 2008; Johnsson et al. 2007); frequency of interaction (VanVactor
2011; Noorderhaven, Harzing 2009); degree of differentiation (Choi,
Krause 2006; Burt, Doyle 1993), etc.
Few papers which particularly focus at different aspects of
complexity are: Milgate (2001) describing supply related complexity from
uncertainty, technological intricacy and organizational systems
perspective; Choi, Krause (2006) from number of suppliers,
differentiation among suppliers and relationship among suppliers
perspective; Bozarth et al. (2009) discusses detailed and dynamic
complexity while other contemporary papers have highlighted the
multifacets of quality of relationship, frequency and volume of
interaction, number of entities, degree of differentiation and
inter-relationship among entities while discussing complexity in
healthcare perspective. Thus although much attention has been drawn,
what lacks, is a unified definition of supply chain network complexity,
a convenient consolidation of the complexity aspects. Even more
importantly, the conceptualization of a supply chain network complexity
construct and a multi-item scale for measuring network complexity remain
wanting. Following Gilliam and Voss's (2013) procedure for
construct definition development based on past literature, the
definition for the constructs are developed in this study. Based on
extant literature the preliminary definitions are developed and based on
the definitions, the measurement items are generated, which are in turn
subjected to judging by experts and professionals and also subjected to
validity testing, finally followed by exploratory factor analysis the
details of which are mentioned in methodology section; thereby aiming at
reducing vagueness and ambiguity. In this study supply chain network
complexity (SCNC) is defined as the extent of inter-relationship among
the supply chain network partners/actors, their degree of
differentiation in terms of practices, their frequency of interaction
and their volume (i.e. numbers) in network (Bode, Wagner 2015; Bozarth
et al. 2009; Handfield, Nichols 1999; Simon 1962; Choi, Krause 2006;
VanVactor 2011; Noorderhaven, Harzing 2009; Milgate 2001; Prater et al.
2001; Meepetchdee, Shah 2007; Johnsson et al. 2007; Burt, Doyle 1993).
In this study SCNC is investigated as an antecedent to two vital
relationship and behavioral aspects of the network partners:
interdependence (Kumar et al. 1995; Vijayasarathy 2010) and risk
assessment (Tummala, Schoenherr 2011; Ha et al. 2011; Wagner, Bode 2006;
Ellis et al. 2010). Interdependence and risk assessment are the two
variables being introduced in the study as consequence of SCNC. Figure 1
shows the conceptual model. The rationale behind introducing these two
constructs can be linked to theoretical underpinnings and key extant
literature. Like most studies which have their basis deeply ingrained in
the theoretical underpinnings, this study draws its inspiration from two
related theories: relational exchange theory (Dyer, Singh 1998) and
bilateral deterrence theory (Bacharach, Lawler 1981). Relational
theories sees firms as social entities and thus highlights the vital
influence or linkage between social constructs like trust, commitment,
power, conflict, risk assessment, network complexity etc having
controlling influence on interactions and exchanges. Dyer and
Singh's (1998) relational view theory suggests that firm's
critical resources can span firm boundaries and get embedded in
inter-firm processes(routines) and interaction; thereby shifting the
focus of prime interest to 'relationships'. Relational
exchange theory highlights that cooperation, communication and trust
plays pivotal role in exchanges and even at times surpasses formal
contracts. In the study, the objective is to analyze the perceptual
measures to understand how they affect the transactional norms. Many of
those norms that are mentioned, gets affected by the different aspects
of the complexity variable as discussed earlier and hence finds rational
linkages, needing deeper probing. Herein comes the relevance of the
theory can be linked. Another theory which can be indirectly linked is
the bilateral-deterrence theory (Bacharach, Lawler 1981; Lawler,
Bacharach 1987) which in a nut-shell suggests that one network
entity's desire to engage in conflict depends largely on its
understanding about the retaliation from the entities with which it
involves in conflict and also the available alternatives it has, since
retaliation poses a greater threat. Thus interdependence is an important
measure for understanding each other's position and retaliation
objectives. Greater the measure value, lesser should be the fear of
conflict occurrence. As per the definition of the SCNC and the extant
literature, complexity of the network entails the nature of bonding
between entities, their number and inter-relationship ship is bound to
affect the interdependence and also the eagerness for ensuring risk
assessment in their transaction relationship. Extant literature
highlights that interdependence involves how the network transaction
partners perceive their dependence on each other and hence essentially
involves both the buyer and supplier side (Vijayasarathy 2010; Kumar et
al. 1995). This interdependence may be symmetric where both parties have
equal dependence or asymmetric i.e. unequal level of dependency. In this
study, interdependence is defined as the extent to which the supply
chain network partners/ actors depend on each other and are unable to
replace each other for their transaction relationship (Vij ayasarathy
2010; Kumar et al. 1995). Vijayasarathy (2010) gave an effective way of
measuring interdependence using a summated scale involving respondent
dependence and supplier dependence which also tackled the skewed
dependence aspect by subtracting the absolute difference between the two
dependence variables. This study adapts similar measures with
modifications which are discussed in details in the measures section.
Risk assessment literature is in existence since a long time and
extant conceptual research works highlights its relevance and importance
in management context involving transactions and inter-actor
relationships (Ha et al. 2011; Wagner, Bode 2006). Risk assessment has
been portrayed as a vital building block of interaction among the
network actors (Prahalad, Ramaswamy 2004). Different studies have
discussed about risk in different perspectives. Ellis et al. (2010)
discussed and empirically examined the link between supply disruption
probability and magnitude with that of overall supply disruption risk.
Using a single item scale for disruption risk they highlighted that a
positive relationship existed between magnitude of disruption and
probability of disruption with overall disruption risk. Further in a
recent study Bode and Wagner (2015) linked upstream supply chain
complexity drivers and supply chain disruption. Blackhurst et al. (2008)
in their conceptual risk framework for supplier risk assessment
indicated that risk as a whole is very generic and can be segregated as
risk identification, assessment, decision-making and monitoring.
Prahalad and Ramaswamy (2004) mentioned that risk assessment is one of
the most vital aspects that decides the course of transaction
relationships and is thus an integral building block of network-wide
actor to actor relationship. In complex business scenarios, especially
where options of switching between actors is high, where often frequency
of interaction varies, and where the network entities remain
inter-connected, assessment of risk becomes most vital because the
strategies like transferring risk, taking risk, eliminating risk,
reducing risk, etc that are implemented follow the risk assessment stage
(Hallikas et al. 2004). The need for purchasing organizations to have
superior risk assessment platform and techniques in place for superior
business relationship with network actors have been highlighted by
contemporary studies (Zsidin et al. 2004; Harland et al. 2003). However
risk assessment literature lacks empirical support and also most studies
are conceptual framework driven and mostly prescriptive in nature.
Established scales, giving a fair understanding about risk assessment
remain wanting. This study aims at plugging this gap and proposes a
multi-item risk assessment scale and operationalizes that. Based on
extant literature, in this study risk assessment (RAS) is defined as the
extent to which the supply chain network partners/actors can make
informed decisions by adequately assessing the stakes involved in their
transaction relationship with network partners (Prahalad, Ramaswamy
2004; Tummala, Schoenherr 2011; Ha et al. 2011; Wagner, Bode 2006; Ellis
et al. 2010).
1. Hypothesis development
In this study it is postulated that supply chain network complexity
increases the extent of interdependence among the network
partners/actors and also enhances the extent of risk assessment among
the transacting network actors. The study also postulates that higher
interdependence in turn should also foster greater extent of risk
assessment among the network partners.
1.1. Supply chain network complexity and interdependence
Network Complexity is defined in terms of number of transacting
actors, extent of inter-relationship among actors, degree of
differentiation among them in terms of practices and also their
frequency of interaction (Choi, Krause 2006; VanVactor 2011).
Interdependence signifies the level of dependence of the buyer firm on
the supplier firm and vice versa in terms of alternatives, ease of
shifting from one set of partners to another, the cost involved etc.
Usually when the number of actors on both fronts i.e. buyer and sellers
are high which extant research characterizes as a complex network
criteria, there remains a fear that the buyer or the seller can easily
switch and hence they try to bind each other through firm specific
investments aiming at increasing each other's onus in the
relationship. Also in situation when the actors in the network are
linked or related to one another closely then in case of a break-up in
the transaction relationship, finding an alternative becomes difficult
and this fosters moving into a state of higher dependence from both
sides to secure such fallout. In case the degree of differentiation in
practices is severely disparate the situation is complex as the one
network actor cannot easily go in sync with the other partner with ease
and hence through investments or contracts or binding systems try to
enhance each other's dependence. When the interaction frequency is
less, there remains scope of ambiguity and this leads to a complex
situation and under this circumstances also fall out may occur. Thus
when the supply chain network complexity is higher in order to secure
future fallout in the transaction relationship, which might add high
economic burden or implications, the network partners try to move into a
stable relationship state which is marked by higher levels of
interdependence. So it is hypothesized that:
H1. Supply chain network complexity has a positive effect on
partner interdependence.
1.2. Supply chain network complexity and risk assessment
In the presence of complex network relationships an overall
environment of uncertainty prevails where the network actors are often
confused to open up and do business and often deter themselves from
cooperating each other and try to play safe. Risk assessments (Prahalad,
Ramaswamy 2004; Blackhurst et al. 2008; Ellis et al. 2010) plays a
pivotal role in deciding the course of action for the downstream
activities and concerns about informed decision making so as to
completely understand the stakes involved in the transaction
relationship. In scenarios of supply chain network complexity being on
the higher side, the inherent uncertainty in the system remains at a
higher level and gives more importance to even thorough understanding
about the risks involved and assessing the consequences in the case of
relationship catastrophe. Thus increased complexity levels should foster
increased onus of risk assessment. So it is hypothesized that:
H2. Supply chain network complexity has a positive effect on
partner risk assessment.
[FIGURE 1 OMITTED]
1.3. Interdependence and risk assessment
Informed decision making is the key aspect of risk assessment. The
salient attributes that characterize risk assessment are making informed
decisions about the transactions (i.e. the partners are aware about the
details of the decision they are making), understanding the stakes
involved in the relationship (what to gain and what might be lost),
remaining aware about the implications/outcomes of the ensuing
transaction relationship (i.e. aware about the expected and worst case
unexpected outcomes) and most importantly understanding the
responsibilities/liabilities involved with the transaction relationship.
Ellis et al. (2010) highlighted that perceptual assessment of risk
guides decision making. As the notion of loss is inherent in risk,
understanding of vulnerability is critical in deciding the course of
relationship as it gives a feeling of being exposed while including an
element of uncertainty or risk (Handfield, Bechtel 2002; 2004). In the
presence of high level of interdependence, both the transacting partners
are at higher risk of incurring losses in case they fallout and hence
often enter long-term binding contracts (Casciaro, Piskorski 2005),
which necessitates a deeper understanding about what is at stake. Thus
higher level of interdependence should foster higher degree of risk
assessment as they need to understand the level of vulnerability/risk
involved in the relationship. But when a relationship is marked by high
degree of interdependence and the fallout of a relationship breakup
might be catastrophic for both partners, the urgency of risk assessment
increases since both are aware about the cost of breakup, keeping chance
of opportunism remains low (Kumar et al. 1995; Wagner, Bode 2006). So,
it is hypothesized that:
H3. Partner interdependence has a positive effect on partner risk
assessment in supply chain network.
2. Methodology
The current research is aimed at deeply understanding the supply
chain network complexity (SCNC) and risk-assessment (RAS) constructs and
thereafter developing relevant scales for detailed empirical study. In
the process, a survey was prepared using two new scales and another
existing multi-item scale. There was the need for developing and using
the new scales for SCNC and RAS, in the absence of relevant multi-item
scales. For the inter-dependence construct, despite a pre-existing scale
was identified as mutual dependence (Vijayasarathy 2010); it was further
revalidated in the course of the study for better adaptability to the
sector-context and country-context. The methodology steps followed are
as follows.
2.1. Pretesting and scale development
In this study, Churchill's method for developing and testing
reflective scales was followed (Churchill 1979). The entire method can
be segregated into four broad steps: first, construct development
followed by checking its content and face validity; second,
dimensionality testing; third, checking for internal consistency;
fourth, checking and thus ensuring that convergent, discriminant, and
nomological validity of the measures are fulfilled (Anderson, Gerbing
1988; Churchill 1979). In the first stage, based on the literature and
existing definitions, pools of items were generated through thorough
probing of the literature for the two constructs. After the successful
generation of the initial pool, a substantive validity test was carried
out followed by scale purification (Anderson, Gerbing 1991).
10 items for SCNC and 9 items for risk assessment were there in the
initial pool which was given to two senior professors from the faculty
of management of two reputed Indian business-schools and four industry
experts from well known firms. Based on their feedbacks, necessary
modifications were done to the initial pool items. This pool was given
to 68 industry experts in the pretesting stage. Following Lawshe (1975),
a substantive validity testing was carried out based on those expert
responses. Such techniques had been used in another contemporary study
for newly developed construct-scale validation (Ambulkar et al. 2015).
As part of the validity testing, the experts were asked to rate if
the items were essential or non-essential in the construct and study
contexts and were asked to give their responses
(satisfied/dissatisfied). The coefficient ([C.sub.sv]) for measuring
substantive validity was calculated as [C.sub.sv] = ([n.sub.c] -
[n.sub.0])/N (Lawshe 1975; Anderson, Gerbing 1991) where nc represented
number of respondents who assigned the items as essential and satisfied,
which [n.sub.0] indicated respondents who marked the item as
non-essential. In this way the [C.sub.sv] value varied between -1 to +1.
Based on the highest and most appropriate values the items were chosen.
Next, in consultation with 2 academic researchers who knew about the
aforesaid technique, finally 4 items were chosen for SCNC and RAS
constructs each. In the following steps the reliability and validity
testing were done. Using SPSS software, the two constructs, based on the
collected 68 responses in the pilot stage, were separately subjected to
exploratory factor analysis and the results gave clear single factors
explaining 73% of the variance in case of SCNC and 70% in case of RAS.
The Keyser-Meyer-Oklin (KMO) indicating sample adequacy was 0.80 and
0.76 respectively. Also the chi-sq values were significant for both
indicating single-factor solution to be significant. Convergent validity
and reliability were also established from the cronbach alpha values of
0.81 and 0.87 respectively for SCNC and RAS and the range of factor
loadings were all above 0.6. The discriminant validity was also checked
using chi-square difference test (Stratman, Roth 2002). In the first
confirmatory factor analysis (CFA), the two latent constructs were
allowed to freely correlate and in the second CFA they were constrained
to one. The difference in the chi-square between the unconstrained and
constrained models involving the SCNC and RAS construct pair was
calculated and it emerged significant, establishing discriminant
validity. The predictive validity was established during the course of
large scale survey study. It was established for the RAS construct,
however for the SCNC construct it could not be established as it lacked
an antecedent variable in the study conceptualization. However the
nomological validity for both the studies were established as the final
questionnaire before sending for survey was given to 10 new experts and
2 other academicians (different from those involved before) who agreed
with the conceptualization of the constructs.
2.2. Sample and data collection for final study
The survey was distributed to professionals holding important
positions and having responsible designations in the supply chain of
reputed Indian firms belonging to 7 different sectors like
manufacturing, retail, food processing, telecommunication, leather,
textiles and chemical. However, this choice of sectors was based on a
leading Indian business magazine's report for the financial year
ending March 31st, 2014, where they highlighted these 7 sectors as top
seven performing sectors. The database for the survey was obtained from
two sources; first, a professional membership list of a regional supply
chain professional body (the list was purchased from a market survey
firm); second, list of alumni of two reputed Indian b-schools (obtained
from known sources to the author), who graduated between 2000 and 2012
and had their alumni profile updated in their respective institutes with
roles which fall within the generic purview of supply chain management.
The survey was conducted using online mode only and the survey was
hosted by using online platform. The respondents were sent online survey
links along with a forwarding email explaining about the survey in
details, its purpose and also the mandatory disclaimer. First, all the
three databases were combined and subjected to a careful sorting
exercise, looking for supply chain related roles and a consolidated list
of 1914 unique individuals was generated. But out of those in the list,
120 entries had missing email IDs. Second, the online survey was
administered to 1794 potential respondents, of which 64 emails had
delivery failure issues and the respondents were not reachable through
any other means, bringing the final effective distributed survey
questionnaire number to 1730. Out of the 1730 potential respondents, 203
completed surveys were obtained at a response rate of 11.73% which is
quite acceptable as per online survey response rate standards (Ambulkar
et al. 2015). Table 1 provides the demographic details of the
respondents. Majority of the respondents were Supply chain managers
(53.7%) and most respondents were from manufacturing (38.4%) and retail
(21.2%) sectors. The obtained responses can be considered to be quite
representative keeping into consideration that the final consolidated
list was prepared by combining the three databases giving specific
attention to the roles and the original consolidated list had 49% names
with roles as supply chain managers /executives /specialists, nearly 30%
as vice-president/director/partner, around 17% as Sr.
Executive/Buyer/Analyst, while the rest fell in the category of logistic
managers. The respondents' profile of the final survey indicated
proportionate responses which were within +/-5% of that of the original
list. Again, looked from the sector perspective, the manufacturing
sector accounted for around 41% of respondents in the original
consolidated list, whereas textile and leather industry accounted for 3%
each. In the final survey, the distributions of the respondents were
acceptably proportionate because manufacturing sector accounted for 38%
while textile and leather accounted for around 2% each. Thus the
obtained response and the sample can well be generalized as
representative.
However one important note should be made when studying supply
chain professionals in the Indian context, especially their sectors and
job-roles. In India manufacturing, retail, telecommunication, chemicals,
etc. account for lion's share of the organized industrial sector,
whereas very few leather and textile firms belong to that category which
can be considered under organized sector. Now, that all the databases
represented executives and managers who were either affiliated to a
reputed professional body or graduated from two reputed b-schools, it
can be rationally assumed that most of them went into organized sectors
and very few got opportunities to venture into textile and leather
industry as those firms recruit very few people from established
b-school campuses (though exceptions are there and this comment should
not be always generalized). Thus keeping all these factors into
consideration, while understanding the study's sample profile and
hence its possible scope of generalization will help managers and
academicians in analyzing the study outcomes.
The respondents came from firms having sales figures as high as
well above 100 million USD (converted @ 1 USD = 60 INR) and also those
below 10 million USD. The experience level of the respondents also
varied and most respondents (63%) belonged to between 5 years to 10
years experience band. A reminder email was sent to the respondents for
bettering the response rate. Thus two waves of responses were received
(before and after the reminder). Non-response bias testing was performed
comparing the early responses (before reminder) with the later responses
(after reminder) (Armstrong, Overton 1977; Ambulkar et al. 2015).
Chi-square test results showed no significant difference between the
first-wave and second wave along two categories of firm size by number
of employees and firm revenue at level of 0.1; thereby assuring an
unbiased sample.
Common method bias was also checked using Harman's single
factor test (Harman 1976). In the current study the largest variance
explained by a single factor was 32.17% which is not the majority of the
total variance. Also further absence of common method bias was checked
following latent factor test (Podsakoff et al. 2003). With introduction
of a latent factor to the main model, no significant loss in the factor
loadings was observed, indicating minimal common method bias in the
current study.
2.3. Measures and data analyses
There are three overall research variables in the model: Supply
Chain Network Complexity (SCNC), Risk Assessment (RAS) and
Interdependence (ID). Supply Chain Network Complexity has been
operationalized using four items, measured on a seven-point Likert scale
(1 = strongly disagree and 7 = strongly agree). The items described
different aspects of SCNC, varying levels of which should represent
varying levels of network complexity. The items that characterized SCNC:
high number of first tier suppliers, strong inter-relationship
relationship among the tier-one suppliers, high degree of
differentiation in terms of levels of operational practices among the
actors, and Low frequency and the volume of interaction among the actors
(Bode, Wagner 2015; Bozarth et al. 2009; Handfield, Nichols 1999; Simon
1962; Choi, Krause 2006; VanVactor 2011; Noorderhaven, Harzing 2009;
Milgate 2001; Prater et al. 2001; Meepetchdee, Shah 2007; Johnsson et
al. 2007; Burt, Doyle 1993).
Risk Assessment too has been operationalized using four items,
measured on a seven-point Likert scale (1 = strongly disagree and 7 =
strongly agree). The items characterizing RAS were: ability to make
informed decisions while transacting, ability to understand the stakes
i.e. risk involved in the relationship, awareness about the implications
of the transaction relationship, and understanding the
responsibilities/liabilities involved in the transaction relationship
(Prahalad, Ramaswamy 2004; Tummala, Schoenherr 2011; Ha et al. 2011;
Wagner, Bode 2006; Ellis et al. 2010; Zsidin et al. 2004; Harland et al.
2003; Blackhurst et al. 2008).
Interdependence scale measure was adapted with necessary
modifications from Vijayasarathy (2010) where it was referred to as
mutual dependence. However as the entire scale was being tested in a
completely different context, the same steps were repeated. Totally six
items represented buyer dependence (BD) and supplier dependence (SD). In
order to ensure that the respondents had a similar understanding of the
two different constructs as in Vijayasarathy (2010) study, the responses
for the six items were subj ected to principal component analysis
(factor analysis) using varimax rotation. The results supported a clear
two-factor structure explaining 67.3% of the variance and had eigen
values above unity. The factor structure is provided in table 2 and all
the six items show good loadings (above 0.6) and all cross-loadings were
below 0.2. Also the reliability was established by calculating Cronbach
alpha which came to 0.76 and 0.81 for BD and SD respectively. The
complete wording of the scale items are provided in the appendix
section.
The validity and reliability of the scales were established,
following Vijayasarathy (2010) and Casciaro and Piskorski (2005), but
with certain key modifications to suit the study requirements,
interdependence (ID) was calculated as follows: ID = [(BD + SD) -
Absolute (BD-SD)]/2. This division by 2 was necessary to bring the ID
items to the same 7-points scale being used for the rest of the
variables. Uniqueness of this ID variable measures happen to be that,
all the three interdependence items were calculated separately
corresponding to the BD and SD item scores and also the ID score was
adjusted for the skewed dependencies (by subtracting the absolute
differences between BD and SD) to capture appropriate magnitude of it.
Subsequently again the SCNC, RAS and ID items were subj ected to
exploratory factor analysis (EFA) to ensure that the factor structures
hold well.
The EFA results involving the perception measures for SCNC, RAS and
ID items, indicated an acceptable KMO value of 0.868 and also the
Bartlett's coefficient got significant at 0.1%. Three well laid and
rotated factor structures emerged from the principal component analysis
results using varimax rotation. The loadings were all good (above 0.6)
and most importantly all the three variables (SCNC, RAS and ID) showed
satisfactory Cronbach alpha values above 0.8. The complete results are
provided in Table 3.
2.4. The measurement model
After successful EFA analysis, confirmatory factor analysis (CFA)
was carried out to assess the reliability, validity and dimensionality
of the constructs. The CFA results indicated acceptable values for CFI
(comparative fit index): 0.981; TLI (Tucker-Lewis Index): 0.976; GFI
(Goodness of Fit Index): 0.882 (close to 0.9); IFI (Incremental Fit
Index): 0.978; [CFI, GFI, IFI and TLI should be near 0.9 and above
(Anderson, Gerbing, 1988)]; CMIN/df i.e. [chi square]/d.f. = 1.69
(preferably should be below 2) and RMSEA (root mean square error of
approximation): 0.027 (preferably below 0.05) (Anderson, Gerbing 1988;
Hu, Bentler 1999). Table 4 shows the CFA results. The measurement items
are provided in Appendix. Factor loadings, composite reliabilities (CR),
average variance extracted (AVE) and squared multiple correlation (SMC)
were examined to assess convergent validity. All the factor loadings
came out to be above 0.6 and significant at p < 0.001, suggestive of
high levels of convergence (Hair et al. 2010). Composite reliability of
all three factors came out to be greater than 0.7, further supporting
convergent validity and internal consistency (Hair et al. 2010). The
AVEs for all the three constructs were each greater than 0.5, supporting
convergent validity. Details of the measurement model have been provided
in Table 4. The AVEs for each of the constructs also sufficiently
exceeded the squared correlations with the other constructs, indicating
support for discriminant validity (Table 5).
3. Analysis and results
3.1. The structural model
In this study structural equation modeling (SEM) using Amos was
used to test the hypothesized relationships, indicated in the study
model (Fig. 1). The model fit indices for the structural model appears
acceptable and satisfying as per the acceptable practices (Anderson,
Gerbing 1998; Hu, Bentler 1999; Iacobucci 2010). The SEM results yielded
acceptable fit statistics: [chi square]/d.f = 1.44; CFI = 0.947; IFI =
0.956; TLI = 0.942; GFI = 0.863; RMSEA = 0.043.
Amos output indicated that the standardized path coefficients
between supply chain network complexity and interdependence (c = 0.229,
p < 0.001) was highly significant supporting hypothesis 1 (H1);
between supply chain network complexity and risk assessment (c = 0.029)
was found to be insignificant rejecting hypothesis 2 (H2); between
interdependence and risk assessment (c = 0.231, p < 0.001) was highly
significant too, thereby supporting hypothesis 3 (H3). Please refer to
Figure 2. The details of the structural model output are provided in
Table 6.
The study findings thus supported H1, that greater SCNC was
positively linked with greater levels of interdependence and again it
was found that higher interdependence between partners in a supply
network led to greater support for higher degrees of risk assessment,
supporting H3. However the non-significance of the path linking SCNC and
RAS refuted H2 and threw up findings contrary to rational understanding.
Thus the outcomes indicated an indirect effect of network complexity on
risk assessment through interdependence.
3.2. Discussion and implications
This study presents a clear definition of SCNC and RAS including
operationalizing of the constructs. The newly developed 4 item scale for
SCNC examines the impact of network complexity on interdependence and
risk assessment constructs. As noted from the extant literature,
complexity happens to be a multi-faceted aspect and hence its presence
and perception needed a suitable instrument. This study attempted at
bridging that long awaited gap by providing a handy scale for measuring
network complexity and empirically examining the impact of SCNC on the
transaction relationship variables of interdependence and risk
assessment. The 4 item risk assessment scale also comes in handy and
helps in quantification of the long discussed conceptual risk assessment
construct. Network complexity as discussed in this study indicates a
business environment which is marked by presence of high number of
suppliers and hence the inter-relationship and frequency of interaction
between the network actors becomes critical. When the complexity is high
then the chance of shifting or switching actors poses a threat because
the actors tend to remain inter-related. The cost involved in switching
may become high for both the supplier and the buyer. Moreover the fear
remains about the fact that the switching partner may involve in
spilling over the crucial key information when moving over to other
network actor for transaction and also it entails cost.
3.3. Managerial and academic implication
The findings from this study should be of immense importance to the
business community and especially the managers in understanding network
complexity and its impact on the relationship and transaction between
actors in a supply chain networks. This study discusses and develops new
multi-item scales to measure network complexity and risk assessment
constructs, both of which existed as conceptual constructs in the extant
literature and were lacked a direct measurement instrument. Often these
were measured by other reflective measures and their discussions
remained at abstract levels without detailed quantification. This study
also validated the measures for interdependence scale and adapted it, so
as to use it with other Likert scales in similar studies. Previous
studies were predominantly conceptual and lacked quantification. This
will help the managers in quantifying their business network complexity
and in the process aid in decision making. This research highlights that
contrary to rational understanding, managers are more concerned about
the level of dependence before probing for assessing the risk rather
than just the network complexity. From academic perspective, this study
opens up new direction of management research in the field of network
complexity and marks the shift from predominantly conceptual studies to
those of empirical investigations.
[FIGURE 2 OMITTED]
Conclusions
The aim of this study was to understand how firms perceive network
complexity in the supply chain and explore how network complexity
subsequently impacts the understanding of interdependence and risk
assessment in the transaction relationship. The findings from this paper
help in establishing a rational framework linking three stumbling blocks
that managers face in course of executing their day to day supply chain
management duties. This study empirically validates the hypothesized
relationships; thereby providing supply chain managers with a handy
empirically proved understanding framework. This study extends the
extant literature, develops and operationalizes two constructs called
supply chain network complexity and risk assessment, besides validating
the interdependence construct in its new usable form derived from the
summated composite score of two constructs: supplier dependence and
buyer dependence. This study also examined the relationship between the
constructs and demonstrated that network complexity has a positive and
direct impact on interdependence, which in turn directly and positive
affects risk assessment; however supply chain network complexity
directly do not influence the risk assessment aspect in a firm network.
The studies contribution lies in providing a consolidated definition of
Supply chain network complexity and development of SCNC and risk
assessment scales besides revalidating the interdependence construct and
its scale elements. This study will open newer horizon of supply chain
management research and add to its body of knowledge. It can be viewed
as a unique attempt to link three somewhat rationally related but
empirically unlinked streams of literature.
As with all social science research, there are some limitations of
this study too. First, the study took into consideration the perceptions
of the managers without segregating them as per the complexity levels of
the network they belong to. All the respondents may not have given their
responses with similar network complexity into consideration. Second,
the study used cross-sectional data, which limits the study's
ability to draw causal linkage, especially because interdependence and
complexity may have some effects which lingers over time and might
reveal newer aspects when looked over time. Third, the study considered
mostly respondents from non-service sectors and service sector networks
were grossly ignored. Therefore, future studies may segregate network
complexity into groups depending on different level of complexity and
study that impact on the consequent variables. Also the study findings
may suffer in terms of limitation of generalization, especially to
service sector scenarios. Future probing can be done keeping the service
sector into consideration. Finally the supply chain network complexity
construct is nascent and may be extended to take a more generic sense
and also explore firm relationships involving dyadic relationships which
might throw up different yet interesting findings. Moreover due to
lacking in the extended conceptualization of what can be antecedents to
the SCNC construct, all validity tests were done expect predictive
validity testing. Thus there remains a scope of introducing a rational
antecedents and then trying to establish the predictive validity.
Samyadip CHAKRABORTY
Department of Operations and IT, ICFAI Business School (IBS)
Hyderabad, IFHE University, Dontanapalli, Shankarpally Road, Hyderabad -
501504, Telangana State, India
E-mail: samyadip@ibsindia.org
Received 08 December 2014; accepted 25 May 2015
http://dx.doi.org/10.3846/btp.2015.620
References
Ambulkar, S.; Blackhurst, J.; Grawe, S. 2015. Firm's
resilience to supply chain disruptions: Scale development and empirical
examination, Journal of Operations Management 33: 111-122.
http://dx.doi.org/10.1016/j.jom.2014.11.002
Anderson, J. C.; Gerbing, D. W. 1988. An updated paradigm for scale
development incorporating uni-dimensionality and its assessment, Journal
of Marketing Research 25: 186-192. http://dx.doi.org/10.2307/3172650
Anderson, J. C.; Gerbing, D. W. 1991. Predicting the performance of
measures in a confirmatory factor analysis with a pretest assessment of
their substantive validities, Journal of Applied Psychology 76
(October): 732-740. http://dx.doi.org/10.1037/0021-9010.76.5.732
Armstrong, J. S.; Overton, T. S. 1977. Estimating nonresponse bias
in mail surveys, Journal of Marketing Research 14(3): 396-402.
http://dx.doi.org/10.2307/3150783
Bacharach, S. B.; Lawler, E. J. 1981. Bargaining: power, tactics
and outcomes. San Francisco: Jossey-Bass Inc.
Blackhurst, J. V.; Scheibe, K. P.; Johnson, D. J. 2008. Supplier
risk assessment and monitoring for the automotive industry,
International Journal of Physical Distribution & Logistics
Management 38(2): 143-165. http://dx.doi.org/10.1108/09600030810861215
Bode, C.; Wagner, S. M. 2015. Structural drivers of upstream supply
chain complexity and the frequency of supply chain disruptions, Journal
of Operations Management 36: 215-228.
http://dx.doi.org/10.1016/j.jom.2014.12.004
Bozarth, C.; Warsing, D.; Flynn, B.; Flynn, E. J. 2009. The impact
of supply chain complexity on manufacturing performance, Journal of
Operations Management 27(1): 78-93.
http://dx.doi.org/10.1016/j.jom.2008.07.003
Burt, D. N.; Doyle, M. F. 1993. The American Keiretsu: A strategic
weapon for global competitiveness. Homewood: Business One.
Casciaro, T.; Piskorski, M. J. 2005. Power imbalance, mutual
dependence, and constraint absorption: A closer look at resource
dependence theory, Administrative Science Quarterly 50(2): 167-199.
Choi, T. Y.; Hong, Y. 2002. Unveiling the structure of supply
networks: case studies in Honda, Acura, and Daimler Chrysler, Journal of
Operations Management 20(5): 469-493.
http://dx.doi.org/10.1016/S0272-6963(02)00025-6
Choi, T.; Krause, D. R. 2006. The supply base and its complexity:
implications for transaction costs, risks, responsiveness, and
innovation, Journal of Operations Management 24(5): 637-652.
http://dx.doi.org/10.1016/j.jom.2005.07.002
Chopra, S.; Sodhi, M. 2014. Reducing the risk of supply chain
disruptions, MIT Sloan Management Review 55: 73-80.
Churchill Jr., G. A. 1979. A paradigm for developing better
measures of marketing constructs, Journal of Marketing Research 16:
64-73. http://dx.doi.org/10.2307/3150876
Dyer, J. H.; Singh, H. 1998. The relational view: Cooperative
strategy and sources of inter-organizational competitive advantage,
Academy of Management Review 23(4): 660-679.
Ellis, S. C.; Henry, R. M.; Shockley, J. 2010. Buyer perceptions of
supply disruption risk: a behavioral view and empirical assessment,
Journal of Operations Management 28(1): 34-46.
http://dx.doi.org/10.1016/j.jom.2009.07.002
Ferlie, E.; Pettigrew, A. 1996. Managing through networks: some
issues and implications for the NHS, British Journal of Management
7(s1): 81-99. http://dx.doi.org/10.1111/j.1467-8551.1996.tb00149.x
Gilliam, D. A.; Voss, K. 2013. A proposed procedure for construct
definition in marketing, European Journal of Marketing 47(1/2): 5-26.
Gilmore, D. 2008. The war on supply chain complexity, Supply Chain
Digest [online], [cited 06 June 2014]. Available from Internet:
http://www.scdigest.com/assets/FirstThoughts/08-08-21.php
Ha, B. C.; Park, Y. K.; Cho, S. 2011. Suppliers' affective
trust and trust in competency in buyers: its effect on collaboration and
logistics efficiency, International Journal of Operations &
Production Management 31(1): 56-77.
http://dx.doi.org/10.1108/01443571111098744
Hair, J. F.; Black, W. C.; Babin, B. J.; Anderson, R. E. 2010.
Multivariate data analysis. 7th ed. Upper Saddle River: Prentice-Hall,
NJ.
Hallikas, J.; Karvonen, I.; Pulkkinen, U.; Virolainen, V. M.;
Tuominen, M. 2004. Risk management processes in supplier networks,
International Journal of Production Economics 90(1): 47-58.
http://dx.doi.org/10.1016/jijpe.2004.02.007
Handfield, R. B.; Bechtel, C. 2004. Trust, power, dependence, and
economics: can SCM research borrow paradigms?, International Journal of
Production Economics 90(1): 47-58.
http://dx.doi.org/10.1504/ijism.2004.004595
Handfield, R. B.; Bechtel, C. 2002. The role of trust and
relationship structure in improving supply chain responsiveness,
Industrial Marketing Management 31(4): 367-382.
http://dx.doi.org/10.1016/S0019-8501(01)00169-9
Handfield, R. B.; Nichols, E. L. 1999. Introduction to supply chain
management. Englewood Cliffs: Prentice Hall.
Harland, C.; Brenchley, R.; Walker, H. 2003. Risk in supply
networks, Journal of Purchasing and Supply Management 9(2): 51-62.
http://dx.doi.org/10.1016/S1478-4092(03)00004-9
Harman, H. H. 1976. Modern factor analysis. Chicago: The University
of Chicago Press.
Hu, L. T.; Bentler, P. M. 1999. Cutoff criteria for fit indexes in
covariance structure analysis: Conventional criteria versus new
alternatives, Structural Equation Modeling: a Multidisciplinary Journal
6(1): 1-55. http://dx.doi.org/10.1080/10705519909540118
Iacobucci, D. 2010. Structural equations modeling: Fit indices,
sample size, and advanced topics, Journal of Consumer Psychology 20(1):
90-98. http://dx.doi.org/10.1016Zj.jcps.2009.09.003
Johnsson, P.; Kjellsdotter, L.; Rudberg, M. 2007. Applying advanced
planning systems for supply chain planning: three case studies,
International Journal of Physical Distribution & Logistics
Management 37(10): 816-834. http://dx.doi.org/10.1108/09600030710848932
Kumar, N.; Scheer, L. K.; Steenkamp, J. B. E. 1995. The effects of
perceived interdependence on dealer attitudes, Journal of Marketing
Research 32(3): 348-356. http://dx.doi.org/10.2307/3151986
Lawler, E. J.; Bacharach, S. B. 1987. Comparison of dependence and
punitive forms of power, Social Forces 66(2): 446-462.
http://dx.doi.org/10.1093/sf/66.2.446
Lawshe, C. H. 1975. A quantitative approach to content validity,
Personnel Psychology 28(4): 563-575.
http://dx.doi.org/10.1111/j.1744-6570.1975.tb01393.x
Lian, P. C. S.; Laing, A. W. 2004. Public sector purchasing of
health services: a comparison with private sector purchasing, Journal of
Purchasing & Supply Management 10(6): 247-256.
http://dx.doi.org/10.1016/j.pursup.2004.11.005
Manuj, I.; Sahin, F. 2011. A model of supply chain and supply chain
decision-making complexity, International Journal of Physical
Distribution & Logistics Management 41(5): 511-549.
http://dx.doi.org/10.1108/09600031111138844
Meepetchdee, Y.; Shah, N. 2007. Logistical network design with
robustness and complexity considerations, International Journal of
Physical Distribution & Logistics Management 37(3): 201-222.
http://dx.doi.org/10.1108/09600030710742425
Milgate, M. 2001. Supply chain complexity and delivery performance:
an international exploratory study, International
Journal of Supply Chain Management 6(3): 106-118.
http://dx.doi.org/10.1108/13598540110399110
Noorderhaven, N.; Harzing, A. W. 2009. Knowledge-sharing and social
interaction within MNEs, Journal of International Business Studies
40(5): 719-741. http://dx.doi.org/10.1057/jibs.2008.106
Podsakoff, P. M.; MacKenzie, S. B.; Lee, J. Y.; Podsakoff, N. P.
2003. Common method biases in behavioral research: a critical review of
the literature and recommended remedies, Journal of Applied Psychology
88(5): 879-903. http://dx.doi.org/10.1037/0021-9010.88.5.879
Prahalad, C. K.; Ramaswamy, V. 2004. Co-creation experience: the
next practice in value creation, Journal of Interactive Marketing 18(3):
5-14. http://dx.doi.org/10.1002/dir.20015
Prater, E.; Biehl, M.; Smith, M. A. 2001. International supply
chain agility: tradeoffs between flexibility and uncertainty,
International Journal of Operations & Production Management 21(5/6):
823-839. http://dx.doi.org/10.1108/01443570110390507
Simon, H. A. 1962. The architecture of complexity, Proceedings of
the American Philosophical Society 106(6): 467-482.
Stratman, J. K.; Roth, A. V. 2002. Enterprise resource planning
(ERP) competence constructs: two-stage multi-item scale development and
validation, Decision Sciences 33(4): 601628.
http://dx.doi.org/10.1111/j.1540-5915.2002.tb01658.x
Tummala, R.; Schoenherr, T. 2011. Assessing and managing risks
using the supply chain risk management process (SCRMP), Supply Chain
Management 16(6): 474-483. http://dx.doi.org/10.1108/13598541111171165
VanVactor, J. D. 2011. A case study of collaborative communications
within healthcare logistics, Leadership in Health Services 24(1): 51-63.
http://dx.doi.org/10.1108/17511871111102526
Vijayasarathy, L. R. 2010. Supply integration: an investigation of
its multi-dimensionality and relational antecedents, International
Journal of Production Economics 124(2): 489-505.
http://dx.doi.org/10.1016/j.ijpe.2010.01.010
Wagner, S. M.; Bode, C. 2006. An empirical investigation into
supply chain vulnerability, Journal of Purchasing and Supply Management
12(6): 301-312. http://dx.doi.org/10.1016/j.pursup.2007.01.004
Waldrop, M. M. 1992. Complexity: the emerging science at the edge
of chaos. New York: Simon & Schuster.
Wycisk, C.; McKelvey, B.; Hulsmann, M. 2008. Smart parts, supply
networks as complex adaptive systems: analysis and implications,
International Journal of Physical Distribution & Logistics
Management 38(2): 108-125. http://dx.doi.org/10.1108/09600030810861198
Zsidisin, G. A.; Ellram, L. M.; Carter, J. R.; Cavinato, J. L.
2004. An analysis of supply risk assessment techniques, International
Journal of Physical Distribution & Logistics Management 34(5):
397-413. http://dx.doi.org/10.1108/09600030410545445
APPENDIX
Construct Scale items used in
the large-scale
study
Supply Chain Our firm's supply
Network Complexity chain network is
characterized by:
SCNC1 Large number of
first tier
suppliers.
SCNC2 Strong inter-
relationship among
the tier-one
suppliers.
SCNC3 High degree of
differentiation in
terms of levels of
operational
practices among the
actors.
SCNC4 Low frequency of
interaction among
the actors.
Risk Assessment While transacting
with major
suppliers:
RAS1 Our firm can make
informed decisions.
RAS2 Our firm understands
the stakes involved
in the relationship.
RAS3 Our firm remains
aware about the
implications of the
transaction
relationship.
RAS4 Our firm understands
the
responsibilities/
liabilities involved
in the transaction
relationship.
Interdependence
Buyer Dependence BD1 There are other
major suppliers
capable of providing
us with comparable
orders.
BD2 Total cost of
switching to a
different set of
major suppliers
would be
prohibitive.
BD3 It is difficult to
replace our major
suppliers while
maintaining
comparable profit
margins.
Supplier Dependence SD1 Other buying firms
can provide our
major suppliers with
comparable orders.
SD2 Total cost of
switching to a
different set of
buying firms would
be prohibitive for
our major suppliers.
SD3 It is difficult for
our major suppliers
to replace us with
other buying firms
and maintain
comparable profit
margins.
Samyadip CHAKRABORTY is an Assistant Professor in the department of
operations and IT at IBS Hyderabad, IFHE University, Hyderabad, India.
He completed his PhD from ICFAI Business School, IFHE University,
Hyderabad, India. He has been a visiting doctoral scholar at the
University of Toledo, College of Business and Innovation, Toledo, Ohio.
He holds an MBA degree from the University of Liverpool, UK. His area of
research interest and focus is healthcare operations and healthcare
supply chain management. He has publications in: Business Theory and
Practice, International Journal of Applied Engineering Research, IUP
Journal of Supply Chain Management, Knowledge Management Journal
(India), SS International Journal of Economics and Management,
International Journal of Business Economics and Management Research,
Macmillan Advanced research series, ECONSPEAK, among other scholarly
outlets.
Caption: Fig. 1. Conceptual model
Caption: Fig. 2. Path model
Table 1. Demographic characteristics of the sample
Job title Number of % of total
respondents responses
Supply Chain Manager/ 109 53.69
Executive/Specialist
Vice-president/director/partner 54 26.6
Sr. Executive/Buyer/Analyst 32 15.76
Logistic Manager 6 2.96
Others 2 1
Sector/type of firm Number of
respondents
Manufacturing 78 38.42
Retail 43 21.18
Food processing 19 9.35
Telecommunication 21 10.34
Leather 03 1.48
Textile 03 1.48
Chemical (including 36 17.73
pharmaceuticals)
Firm Sales (data in INR Number of
converted to US$) respondents
$10 million or less 45 22.17
More than $10 million to $50 million 62 30.54
More than $50 million to $100 million 81 39.9
Above $100 million 15 7.39
Experience (in years)
Less than 5 years 39 19.22
More than 5 to 10 years 128 63.05
Above 10 years 36 17.73
Total 203 100
Table 2. Results of the principal component
analysis of the factors contributing to
interdependence measure
Items Factors
Buyer dependence Supplier Dependence
BD1 0.617 0.036
BD2 0.795 0.119
BD3 0.679 0.101
SD1 0.064 0.718
SD2 0.112 0.632
SD3 0.023 0.761
Table 3. EFA findings
KMO: 0.868 Bartlett: significant at 0.1%
Total variance explained: 69.7%
Factors Measurement Item Cronbach's
Items loadings Alpha
Supply SCNC1 0.889 0.906
Chain SCNC2 0.670
Network SCNC3 0893
Complexity SCNC4 0.705
Risk RAS1 0.865 0.892
Assessment RAS2 0.755
RAS3 0.806
RAS4 0.872
Interdependence ID1 0.852 0.850
ID2 0.795
ID3 0.863
Table 4. Measurement model: CFA results
Construct Items Std Estm. p-Value SMC AVE CR
Supply SCNC1 0.97 * 0.94 0.741 0.91
Chain SCNC2 0.67 * 0.45
Network SCNC3 0.94 * 0.89
Complexity SCNC4 0.71 * 0.50
Risk RAS1 0.81 * 0.63 0.502 0.80
Assessment RAS2 0.79 * 0.61
RAS3 0.71 * 0.51
RAS4 0.93 * 0.92
Interdependence ID1 0.81 * 0.67 0.549 0.79
ID2 0.75 * 0.55
ID3 0.88 * 0.76
Note: * significant at p < 0.001
Table 5. Discriminant validity table
SCN RAS ID
Supply Chain 0.741
Network Complexity
Risk Assessment 0.094 0.502
Interdependence 0.166 0.025 0.549
Table 6. Hypotheses testing results
Hypotheses Paths Standardized p-Value Results
path
coefficients
H1 SCNC[right arrow]ID 0.229 *** Accepted
H2 SCNC[right arrow]RAS 0.029 0.638 Rejected
(NS)
H3 ID[right arrow]RAS 0.231 *** Accepted
Note: *** Implies significant at p < 0.001 & NS implies
"not significant"
