close

Вход

Забыли?

вход по аккаунту

?

j.dss.2018.08.008

код для вставкиСкачать
Accepted Manuscript
A data-analytics approach to identifying hidden critical suppliers
in supply networks: Development of nexus supplier index
Benjamin B.M. Shao, Zhan (Michael) Shi, Thomas Y. Choi,
Sangho Chae
PII:
DOI:
Reference:
S0167-9236(18)30137-4
doi:10.1016/j.dss.2018.08.008
DECSUP 12982
To appear in:
Decision Support Systems
Received date:
Revised date:
Accepted date:
3 April 2018
15 July 2018
17 August 2018
Please cite this article as: Benjamin B.M. Shao, Zhan (Michael) Shi, Thomas Y. Choi,
Sangho Chae , A data-analytics approach to identifying hidden critical suppliers in supply
networks: Development of nexus supplier index. Decsup (2018), doi:10.1016/
j.dss.2018.08.008
This is a PDF file of an unedited manuscript that has been accepted for publication. As
a service to our customers we are providing this early version of the manuscript. The
manuscript will undergo copyediting, typesetting, and review of the resulting proof before
it is published in its final form. Please note that during the production process errors may
be discovered which could affect the content, and all legal disclaimers that apply to the
journal pertain.
ACCEPTED MANUSCRIPT
A Data-Analytics Approach to Identifying Hidden Critical Suppliers in Supply Networks:
Development of Nexus Supplier Index
Benjamin B.M. Shaoa, Zhan (Michael) Shia,*, Thomas Y. Choia, Sangho Chaeb
T
Tilburg School of Economics and Management
Tilburg University
5000 LE Tilburg, The Netherlands
US
Abstract
CR
b
W. P. Carey School of Business
Arizona State University
Tempe, AZ 85287, USA
IP
a
Recent events involving supplier-caused business disruptions bring to the forefront the issue of managing
AN
hidden yet critical suppliers that may exist deep in the supply network. While managing prominent
strategic suppliers in the top tier is well understood, we have only just begun to recognize a different type
M
of critical suppliers called nexus suppliers. Nexus suppliers are critical because of their structural
positions in the supply network. They can be several tiers removed in the extended supply network and
ED
hence may not have direct contact with, and not be visible to, the focal buying firm. In this study, we
explore the identification and categorization of nexus suppliers. Based on the theory of nexus supplier and
data envelopment analysis (DEA), we propose a data-analytics approach to compute what we call Nexus
PT
Supplier Index (NSI). It is a measure that combines various network centrality measures to capture and
reflect different aspects of a supplier’s structural importance. The contribution of our study is to take the
CE
concept of nexus suppliers that exists only in theory to practice and demonstrate how to look for nexus
suppliers in the real world. To achieve this aim, we develop a mathematical model for NSI, compile a
AC
large data set using Bloomberg Terminal, and engage in computations to identify and categorize nexus
suppliers. The target company is Honda, and we review the results with the top supply management team
at Honda of America. Implications for practice and future research are discussed.
Keywords: Nexus supplier index, data analytics, hidden suppliers, centrality measures, social network
analysis, data envelopment analysis
_________________________
*
Corresponding author. Email address: ZMShi@asu.edu (Z. Shi).
1
ACCEPTED MANUSCRIPT
1. Introduction
Recent events that involved supplier-caused business disruptions bring to the forefront the issue of
managing critical suppliers that are embedded deep in the supply network. When an explosion occurred at
a plant of Evonik Industries, a little-known raw materials supplier, it brought the global automotive
T
supply chain to a halt (Carty, 2012). While the theory and practice of managing prominent strategic
IP
suppliers are well understood (Kaufman et al., 2000; Slobodow et al., 2008; Koufteros et al., 2012),
CR
managers and researchers have only just begun to recognize the potential criticality of these less-known
suppliers. Yan et al. (2015) propose the theory of nexus suppliers to emphasize the importance of such
US
hidden critical suppliers, where nexus supplier is defined as “any supplier in a multi-tiered supply
network that potentially exerts a profound impact on a buyer’s performance due to its network position.”
AN
Traditional strategic suppliers refer to the suppliers in a firm’s supply base whose actions and
outcomes have significant and direct impact on the focal firm’s risk management (Kraljic, 1983). A
M
strategic supplier provides the focal firm with essential input materials and critical products or
ED
technologies (Wagner & Johnson, 2004; Borgatti & Li, 2009). In this regard, strategic suppliers tend to be
top-tier suppliers with high impact on the buying firm’s profitability, and the performance of the focal
PT
firm depends largely on these suppliers’ capabilities and resources. As a result, the focal firm aims to
cultivate a collaborative partnership with strategic suppliers (Bensaou, 1999; Chen et al., 2004).
CE
By contrast, a nexus supplier refers to a supplier that is critical to the focal firm’s operations
AC
because of its structural position in the firm’s supply network, such as how its business relationships are
linked with other firms in the focal firm’s extended supply network. Recent theoretical studies such as
Ang at el. (2016) have shown the topology of top-tier supply relationships can significantly influence the
performance and strategy of downstream buying firms. A nexus supplier may be several tiers removed in
the supply network, and hence may not be immediately visible to the focal buying firm (Yan et al., 2015).
Incidents such as the explosion at Evonik Industries reflect the importance of understanding how a
supplier’s structural embeddedness in the extended supply network (i.e., the extent to which a supplier’s
2
ACCEPTED MANUSCRIPT
criticality depends on the position and structure of its supply network) may influence the performance of
downstream firms (i.e. General Motors, Ford, etc.). This need for understanding prompts firms to find the
nexus suppliers that could potentially affect its performance. Companies with a better understanding of
suppliers’ network positions are likely to outperform those that do not consider them (Choi & Kim 2008).
T
Presently, the concept of nexus supplier exists only in theory. Yet to be answered is the question of
IP
how to identify them in practice. Only then can a buying firm proactively manage its nexus suppliers to
CR
mitigate the associated risks (e.g., materials flow disruption) or take advantage of potential market
benefits (e.g., innovation opportunities). The difficulty of identifying nexus suppliers is further
US
compounded by the fact that many buying firms do not have a global view of their extended supply
networks and thus they do not know a priori which suppliers can be the candidates for nexus suppliers.
AN
For example, Toyota did not know, let alone manage, much of its lower-tier supply chain until the 2011
Tohoku earthquake disrupted its suppliers’ suppliers (Ang et al., 2017). In response, we map out a multi-
M
tier supply network based on the data collected from Bloomberg Terminal and develop a data-analytics
ED
framework to capture a supplier’s network structural importance for identifying nexus suppliers.
We propose a mathematical model called Nexus Supplier Index (NSI). Our NSI model is based on
PT
data envelopment analysis (DEA) and incorporates various network centrality measures (i.e., degree,
betweenness, eigenvector, and closeness), each of which reflects a certain aspect of a supplier’s
CE
importance in the focal buying firm’s supply network. NSI provides a single unified metric that combines
AC
multiple centrality measures to evaluate a supplier’s potential for being a nexus supplier. We take this
position because no single centrality measure is comprehensive enough to reflect the overall criticality of
a supplier. The proposed NSI is designed to reflect the overall structural importance of a supplier and help
a buying firm make informed decisions by qualifying certain suppliers as nexus suppliers for better risk
assessment, evaluation, and portfolio development of suppliers.
To apply and evaluate the proposed NSI model, we empirically examine a real-world supply
network that is constructed for Honda Motor Company. Going beyond top-tier suppliers, we identify
3
ACCEPTED MANUSCRIPT
lower-tier suppliers for Honda beyond its supply base to the fourth-tier to map out a supply network. We
develop a data-analytics system to facilitate the tasks of data storage, visualization, and NSI computation.
Using the system, we compute the NSI scores for all the suppliers in Honda’s supply network to identify
the nexus supplier candidates and then categorize them into specific types. Through this empirical
T
instantiation, we demonstrate the feasibility of putting the nexus supplier concept into action by
IP
leveraging data-analytics methods and exploiting a large volume of business relationship data. We present
CR
and discuss the results with the top supply management team at Honda of America for result validation.
2. Literature Review
US
2.1. Rating of strategic suppliers
Strategic suppliers (Dyer & Singh, 1998; Gadde & Snehota, 2000) and strategic alliances (McCarter
AN
& Northcraft, 2007) have been explored by researchers in operations management and strategic
M
management. Strategic suppliers offer essential products, capabilities and technologies (Olsen & Ellram,
1997; Ellram & Carr, 1994; Monczka et al., 1998; Azadegan et al., 2008) while strategic alliances are
ED
formed to promote interfirm communication and cultural fit (Cavusgil et al., 1995; Arino et al., 1997;
Dacin et al., 1997; Hitt et al., 2000; Zahra et al., 2000; Wu et al., 2009).
PT
Systematic approaches have been proposed to evaluate the performance of these strategic suppliers.
CE
Liu and Hai (2005) present a voting analytic hierarchy process (VAHP) method for ranking and selecting
suppliers. They extend Yahya and Kingsman's (1999) analytic hierarchy process (AHP) method for rating
AC
suppliers by integrating DEA and managers’ supplier evaluations based on multiple criteria. Mafakheri et
al. (2011) also apply AHP to propose a multiple-criteria dynamic programming (MCDP) method for
supplier evaluation and order allocation. Considering production capacity interdependence among
suppliers, Li et al. (2013) develop analytical models to optimize buying firms’ sourcing and pricing
decisions when there are multiple suppliers with different wholesale prices and reliability levels. Hu and
Kostamis (2015) and Yim (2014) also examine the buying firm’s multiple-sourcing strategies to propose
an analytical model that optimizes order quantities for multiple suppliers.
4
ACCEPTED MANUSCRIPT
While these studies provide practical and systematic approaches to rating suppliers, their focus is on
internal capabilities of top-tier suppliers and ranking these top-tier suppliers. In the cases of further
upstream suppliers (i.e. tiers 2 and 3 suppliers), often their internal capabilities are unknown to the buying
firm, making it difficult to apply the above approaches. For instance, how could a final assembler who is
T
at the downstream end of the supply network begin to consider the potential impact of these unknown
IP
suppliers on its performance? To evaluate the unknown suppliers beyond top-tier, we have chosen to take
CR
a network perspective to understand structural characteristics of the firms in supply networks.
2.2. A network perspective
US
Recently, researchers have moved beyond two-tier supply chains (Guo et al., 2010) to evaluate a
supplier’s importance by considering its position in the larger supply network context (Choi et al., 2001;
AN
Andersson et al., 2002; Pathak et al., 2007; Artto et al., 2008; Kim et al., 2011; Basole & Bellamy, 2014;
Bhattacharjee & Cruz, 2015). At the core of the argument is a firm’s structural relationship with other
M
firms in the supply network and its impact on performance. While a supplier’s internal capabilities, its
ED
competitiveness, and direct ties with other firms are crucial factors to consider (Gnyawali & Madhavan,
2001; Echols & Tsai, 2005; Hagedoorn, 2006; Li, 2013), salient network attributes such as centrality,
PT
network density, repeated ties, and common ties also need to be taken into account when evaluating
CE
suppliers (Gulati & Gargiulo, 1999).
Emphasizing the concept of structural embeddedness, Choi and Kim (2008) suggest buying firms
AC
consider network structural characteristics when evaluating suppliers. Structural embeddedness of a
supplier refers to the extent to which a supplier’s criticality depends on the position and structure of its
supply network. They argue that suppliers’ performance is influenced by other companies in the supply
networks, so suppliers’ structural embeddedness can be as important as their internal capabilities. Borgatti
and Li (2009) call for more development of network perspectives and point out how social network
concepts such as ego-network structure, structural holes, node centrality, network cohesion, and structural
equivalence can potentially be applied to supply chain management. Kim et al. (2011) apply the key
5
ACCEPTED MANUSCRIPT
metrics of social network analysis to supply network constructs. They illuminate the importance of
understanding individual supply network members in terms of structural position in the network and
suggest buying firms consider the potential roles of suppliers based on their network centrality measures
such as degree, closeness, betweenness and eigenvector.
Degree centrality in the social graph is defined as the number of links that a node has (Freeman,
IP
T
1978). In the supply network context, degree centrality reflects a firm’s influence over its supply chain
CR
partners (Borgatti & Li, 2009). Closeness centrality describes the extent to which a node is geodesically
close to all the other nodes in the social network (Marsden, 2002), and it indicates reliable access to
US
information from the supply network (Kim et al., 2011). A node with high betweenness centrality plays
the role of intermediary in the social network (Freeman, 1978), and can facilitate or control the flows of
AN
materials and information in supply networks (Kim et al., 2011). Finally, eigenvector centrality is a
measure of the network status in which a node has many links with other central nodes (Bonacich, 1987).
M
A supplier with high eigenvector centrality provides indirect linkages to critical players in supply
ED
networks (Yan et al., 2015). Recent studies by Bellamy et al. (2014) and Dong et al. (2015) apply some of
these centrality measures to study their association with the focal firm’s performance. However, a
PT
systematic approach to integrate different network centrality measures to evaluate direct and indirect
CE
suppliers is yet to be developed.
2.3. Nexus suppliers and data analytics
AC
Reflecting the structural perspective of supply network research, Yan et al. (2015) conceptualize
critical suppliers based on their positions in the supply networks. The nexus suppliers can potentially
exert influence on a buying firm’s performance because they take central positions in the buying firm’s
supply network and can take up a gate-keeping position for materials flow, cost, quality management, and
information access. Nexus suppliers can be categorized into three types: operational, monopolistic and
informational nexus suppliers (Yan et al., 2015). Each type plays a different role with varying degree of
criticality, interdependence, and asymmetry, all of which carry different managerial implications for the
6
ACCEPTED MANUSCRIPT
focal buying firm. An operational nexus supplier has a large number of ties; it organizes and incorporates
parts from multiple suppliers for the focal buying company. A monopolistic nexus supplier exists in the
extended industrial network and has a greater chance of sitting on the shortest paths between pairs of
other nodes; it would thus likely affect supply continuity and assurance. Finally, an informational nexus
T
supplier is connected to a highly diverse group of firms in the supply network; it may serve as the source
IP
of early market information or technological innovations for the focal firm. Because of their different
CR
nature, each type of nexus suppliers is expected to exert different influence on the buying firm’s
performance. While Yan et al. (2015) introduce the concept of nexus supplier, their study is largely
US
theoretical. We are in need of a model that can empirically evaluate and identify nexus suppliers.
Researchers have recently begun to apply data analytics to operations management issues, such as
AN
inventory management, quality management, process design, pricing, and new product development
(Simchi-Levi, 2014). Ferreira et al. (2016) apply data analytics to develop demand forecasting and price
M
optimization models. Basole et al. (2017) demonstrate data-driven approach to visualize innovation-
ED
focused supply networks. Huang and Van Mieghem (2014) propose a framework to convert clickstream
data from non-transactional websites into advance demand information that can be used for inventory
PT
management. The framework complements traditional approaches of inventory management by analyzing
CE
the webpage clicking behavior of customers as an additional variable for demand forecasting. Abrahams
et al. (2015) present a text analytic framework for discovering product defects. Chan et al. (2016)
AC
introduce a mixed-method approach for utilizing social media data in new product development projects.
Our goal then is to adopt a data analytics approach to assess the criticality of nexus suppliers
through a composite of several centrality measures to arrive at index scores, which we call nexus supplier
index (NSI). In the following sections, we first propose the NSI framework and then demonstrate its
feasibility by examining the real-world supply network of the automaker Honda. Our intent is to help
supply managers interpret the results of our data analysis, consider limitations in the model and data
currently available, and generate insights that can help manage nexus suppliers.
7
ACCEPTED MANUSCRIPT
3. Nexus Supplier Index (NSI)
We develop a mathematical model composed of well-defined network constructs to compute the
Nexus Supplier Index (NSI). The purpose of NSI is to provide a single unified metric that combines
multiple centrality measures (e.g., degree, betweenness, eigenvector, etc.) to evaluate the overall
T
criticality of suppliers in the focal firm’s supply network, as no single centrality measure is
IP
comprehensive enough to capture the aggregate importance of a supplier. While it is possible to apply
CR
each centrality measure separately to the task at hand, such an approach would be ad hoc and time
consuming. In addition, the interpretation of those results would be convoluted by the peculiar properties
US
associated with each measure. Our approach, instead, is to develop a composite NSI measure that reflects
the overall importance of a supplier to help the focal firm make an informed decision on whether a
AN
supplier in its supply network could be qualified as a nexus supplier.
Our NSI framework is built based on data envelopment analysis (DEA). Originally intended to
M
construct a production frontier, DEA is a non-parametric, linear programming method for calculating
ED
efficiency. DEA has since been used for developing other advanced performance metrics, such as
Malmquist index that compares production output across different economies, industries and firms (Färe
PT
et al., 1994; Chou & Shao, 2014). Further, DEA has then been extended to resource-allocation rules
CE
(Korhonen & Syrjänen, 2004), quality management (Chin et al., 2009), multi-criteria ranking (Giannoulis
& Ishizaka, 2010) and strategic sourcing decisions (Talluri et al., 2013).
AC
DEA has been used as a benchmarking tool to generate a score that indicates the relative distance of
an entity to the best practices so as to measure its overall performance compared with its peers (Cook et
al., 2014). More importantly and related to our study, such overall performance measured by DEA is
manifested in the form of a composite measure that aggregates individual indicators (Cook et al., 2014).
Following the same logic, we employ DEA as the method for developing our NSI score by aggregating
individual centrality measures to identify nexus suppliers in a supply network.
8
ACCEPTED MANUSCRIPT
Our DEA-based model for computing the nexus supplier index (NSI) of supplier node p in a supply
network is formulated as follows:
Maximize NSIp =
(1)
(i = 1,…, N)
subject to
(2)
α, β, γ, σ ≥ 0
T
(3)
IP
where D is degree centrality, B is betweenness centrality, V is eigenvector centrality, and F is distance
CR
farness measure. The definitions of these centrality measures are given in the network analysis literature
where Xpj = 1 if there is a link between node p and node j,
US
(e.g., Wasserman and Faust, 1994): Dp =
or 0 otherwise; Bp =
where d(i, p) is the number of paths node p is on, and d is the number of
AN
geodesics connecting p; V = δ(I-ωR)RT where δ is the scaling vector for score normalization, I is the
M
identity matrix, ω reflects the extent one weighs the centrality of other nodes p is tied to, R is the
adjacency matrix, and T is the a matrix of 1’s; and F =
is the reciprocal of closeness. Note α, β, γ,
ED
and σ in Eq. (1) are weights to be decided for degree, betweenness, eigenvector, and farness, respectively.
PT
In essence, they are decision variables and the only restriction is their positivity of Constraint (3). In our
NSI context, these weights provide information on how a supplier firm’s structural embeddedness can be
CE
changed to make it more critical (e.g., through the focal firm’s direct sourcing from a lower-tier supplier).
We choose these centrality measures because they capture different aspects of node importance at
AC
the node level (degree), neighborhood level (eigenvector), and network level (betweenness and farness).
Degree refers to the number of links a node has. A node with more links and higher degree is deemed
more important. However, relying on degree alone can be misleading as it is a local measure, where one
simply counts the links attached to a specific node. Eigenvector measures the prestige of a node based on
the importance of the other nodes it is connected to. Nodes that are connected to other important nodes
should have higher prestige and influence. Betweenness is measured based on the entire network under
9
ACCEPTED MANUSCRIPT
consideration. It counts the number of shortest paths a certain node resides on between any other two
nodes, and is developed based on the idea that a node lying on the shortest paths controls communication
flows. A node with high degree, betweenness and eigenvector is thus considered more critical, so these
centrality measures ought to be maximized and hence placed in the numerator of Eq. (1).
T
We also have distance farness in the denominator of Eq. (1), which measures the average distance
IP
of a node to every other node and can be viewed the reciprocal of the closeness centrality. A node is
CR
considered less important if it is relatively farther from all other nodes. The smaller the value of farness a
node has, the more important it is deemed to be. The goal is to minimize the distance farness measure,
US
placing it in the denominator of Eq. (1). Constraint (2) ensures the NSI score for supplier node p be less
than 1, where a higher NSI score indicates greater importance of a supplier node and hence its higher
AN
propensity to be a nexus supplier. Our NSI model thus includes the four essential centrality measures
most frequently used in social network analysis (Kilduff & Tsai, 2003) and combines them into one
M
composite index that reflects the various aspects of node importance (i.e., supplier criticality). In theory,
ED
NSI reflects a supplier’s structural embeddedness in a supplier network using an integrative and
comprehensive approach that considers four essential dimensions of node criticality.
PT
Huang et al. (2014) identify the node role in a network using multiple indicators. Similarly, we use
multiple centrality measures to evaluate the importance of a supplier node and indicate its criticality in a
CE
supply network. However, instead of using normalized simple averages of the multiple indicators (i.e., α
AC
= β = γ = σ = 1) as in Huang et al. (2014), we find appropriate weights α, β, γ and σ for their respective
centrality measures to derive the NSI score. In doing so, we address the open problem of automatic
optimization of weights previous researchers have put forth (Huang et al., 2014). As noted by Sherman
and Zhu (2006), DEA gives each unit the benefit by making itself look as good as possible when
compared with other units through its own weight selection. Our DEA-based NSI approach differs from a
simple ratio measure by treating weights as independent variables that are objectively decided when
optimizing the NSI score. This weight solution represents a departure from previous studies, which either
10
ACCEPTED MANUSCRIPT
use normalized weights to assume equal importance of the components or ask the decision maker to
assign subjective weights to reflect the perceived importance of each component.
One potential problem with the model (1)-(3) is that it has an infinite number of solutions: if (α*,
β*, γ* and σ*) is a solution, then (cα*, cβ*, cγ* and cσ*) is another solution for any constant c. To address
T
this problem, the constraint σFp = 1 can be added. Moreover, the model (1)-(3) assumes constant returns
IP
to scale (CRS) (Charnes et al., 1978). Thus, to allow for the more flexible case of variable returns to scale
CR
(VRS) (Banker et al., 1984), we incorporate a scale factor μ (Coelli et al., 2005; Sherman & Zhu, 2006).
Thus, our complete NSI model is as follows:
US
Maximize NSIp =
(i = 1,…, N)
subject to
=1
AN
α, β, γ, σ ≥ 0
(5)
(6)
(7)
(8)
M
μ unrestricted
(4)
Using the duality in linear programming, we derive an equivalent envelopment form of the primal
ED
model (4)-(8) that involves fewer constraints and hence is preferred (Coelli, et al., 2005):
PT
(10)
CE
subject to
(9)
(11)
AC
(12)
(13)
(14)
(15)
(16)
11
ACCEPTED MANUSCRIPT
In the model (9)-(16), θp is the NSI score to be optimized. Parameter ε represents an infinitely small
number that has no impact on the optimal NSI score θp;
and
represents the slack for the farness measure;
indicate the slack for degree, betweenness, and eigenvector centrality measures,
respectively. Our NSI model consists of the following notable characteristics. First, it provides a
T
comprehensive way to evaluate the criticality of a supplier node in the focal firm’s supply network and
IP
hence its likelihood of being a nexus supplier. Second, through the aggregate NSI score that combines
CR
degree, betweenness, eigenvector, and distance farness measures, our objective function incorporates the
practical meanings of each measure and reflects them in the functional form (i.e., by maximizing degree,
US
betweenness, and eigenvector, while minimizing distance farness). Third, the NSI model produces a score
θp between 0 and 1, a range that is similar to a ratio measure, and the NSI score follows our intuitive
AN
understanding of how NSI scores should behave: a greater score would mean higher criticality, and
numerically it would stay between 0 and 1. In sum, our DEA-based NSI model represents a composite
M
measure of supplier node importance that allows managers to better assess each supplier’s criticality
ED
against other nodes in the supplier network (Cook et al., 2014).
Once a supplier is found to possess a high NSI score, it becomes a candidate for nexus supplier. We
PT
can then examine its individual centrality measures to determine its type as operational, monopolistic,
CE
and/or informational (Yan et al., 2015). We carry out this task by using the categorization model depicted
in Table 1 to help associate a nexus supplier with its type. This categorization scheme serves as a useful
AC
mechanism for classifying the nexus suppliers identified by the NSI model (9)-(16). We consider degree,
betweenness, eigenvector, and diversity measures, as specified in Table 1. Of particular interest is the
concept of diversity, as it captures a supplier’s exposure to varied ideas and propensity of being an
informational nexus supplier. Diversity is measured by counting the number of unique industries a node’s
suppliers belong to, and it reflects the extent to which a supplier is connected to a range of different
organizations across different industries. A supplier with high diversity can be critical, as it is exposed to
12
ACCEPTED MANUSCRIPT
heterogeneous experiences and contextually varied information about market opportunities and
technology innovations (Mackelprang et al., 2017).
Table 1. Characteristics of the Three Types of Nexus Suppliers
Operational Nexus
Monopolistic Nexus
Informational Nexus
Supplier
Supplier
Supplier
High betweenness
centrality
Criticality
Significant impacts on the
operational performance
of the end product
Significant impacts on
supply continuity due to
low substitutability
High diversity
T
High degree, betweenness
and eigenvector centrality
Sources of early market
and technological
information
CR
IP
Structural
Characteristics
4. Data and Analytics System
AN
4.1. Data: Honda Motor Company’s supply network
US
Source: Adapted from Yan et al. (2015)
We use Honda’s supply network primarily because the company has been featured extensively in a
M
stream of studies (e.g., Choi & Hong 2002, Choi & Linton 2011, Kim et al. 2011) that we have chosen to
ED
build on. In general, industry analysts consider Honda’s supply chain management in high regard, which
increases the practical value of this study’s outcome. Their supply chain would also tend to be relatively
PT
stable, and that helps justify taking a cross-sectional view of its supply network based on publiclyavailable data. Mapping out the multi-tier supply network for a global industrial company is a challenging
CE
task. The main difficulty has been the lack of a reliable and comprehensive dataset that covers the flow of
AC
sales from suppliers to customers. In practice, such information is revealed in many different sources yet
each source tends to be incomplete. For example, in the United States, public firms disclose their
customer information in the filings to the Securities and Exchange Commission (SEC), but the disclosure
is mandatory only if 10% or more of their revenue is derived from sales to a particular customer (a public
firm can also voluntarily report other customers to which the sales are below the 10% threshold if the firm
determines that such disclosure is in its own best interest). In addition to public filings, supplier-customer
13
ACCEPTED MANUSCRIPT
information is also scattered in other sources such as earnings call transcripts, industry reports, press
releases, firm websites, etc.
Only recently have financial data vendors started to collect supplier-customer-relationship data
systematically from these different places and integrate them into a coherent database. For this research,
T
we utilize the Bloomberg Supply Chain (SPLC) database, which is available on Bloomberg Terminal.
IP
Bloomberg SPLC keeps track of about 28,000 companies worldwide. For each of these companies, the
CR
database provides a list of suppliers based on the flow of sales information it has aggregated from a
variety of sources. Though there is no guarantee that these supplier lists are exhaustive, to the best of our
US
knowledge, SPLC is comparatively comprehensive in the coverage. The resolution of SPLC data is at the
company level, so the plant-level or component-level supplier data is unavailable. Also, SPLC provides
AN
only cross-sectional, not longitudinal, data on firms’ suppliers.
We compile Honda’s supply network data one link at a time from the SPLC database and our data
M
collection strategy reflects the multi-tier nature of the supply network. Figure 1 helps illustrate the data
ED
collection process. With Honda as the focal buying firm (heavy solid oval in Figure 1), we first collect all
the companies listed in SPLC that have a direct sales flow to Honda and we call these companies Honda’s
PT
top-tier suppliers (illustrated as lighter ovals in Figure 1 with arrows representing the direction of sales).
CE
We then iterate through the top-tier suppliers: For each of them (e.g., Bridgestone Corp), we treat it as the
focal company and collect all of its top-tier suppliers that are not currently in our dataset. These newly
AC
added companies are each two tiers removed from Honda and we thus label them tier 2 suppliers of
Honda (dashed ovals in Figure 1). Note that a top-tier supplier of Honda can have a direct sales flow to
another top-tier supplier of Honda, as Bridgestone does in Figure 1. In this case, we use the length of the
shortest path to Honda for tier labeling. Then, we repeat this iterative process further upstream. By the
end of the data collection period, we have gathered all the supplier information for companies in the top,
second, and third tiers of Honda’s supply network; we collect data to the fourth-tier companies but not
their suppliers. It should be noted that Bloomberg does not allow downloading the SPLC data in bulk nor
14
ACCEPTED MANUSCRIPT
does it provide an interface for retrieving the data programmatically. Therefore, we used a research
assistant to manually collect the data and save it in plain text format. The data collection process took 15
weeks and over 300 person-hours.
US
CR
IP
T
Figure 1. Data Collection Procedure
AN
Up to tier 4, the final data set includes the network consisting of 10,833 companies and 47,183
supplier-customer relationships (these supplier-customer relationships represent uni-directional
M
relationships rather than bi-directional, so if firm A supplies to firm B and firm B also supplies to firm A,
ED
then they are represented as two distinct relationships in our statistics). Of the 10,832 companies
excluding Honda, 245 are Honda’s top-tier suppliers, 1,643 are tier 2 suppliers, 4,605 are tier 3 suppliers,
PT
and 4,339 are tier 4 suppliers. SPLC also provides information about the country of a company’s
headquarters and the industry sector it belongs to using the Global Industry Classification Standard
CE
(GICS) (see https://www.msci.com/gics). The collected supply network spans 83 countries and 66
AC
industry sectors. Tables 2 and 3 list the top 10 countries and top 10 GICS industries represented in the
dataset. Not surprisingly, the suppliers in the United States, Honda’s home country of Japan, and East
Asian countries are ranked high in the country list. In the industry list, the electronic components and
semiconductor sectors are ranked high. This finding is consistent with the observed industry trend of
“drive-by-wire” where many of the mechanical auto parts are being replaced with high-tech components.
Table 2. Breakdown of Companies by Country
Rank
1
2
Country/Region
U.S.
Japan
15
# of Companies
1,595
1,226
ACCEPTED MANUSCRIPT
3
4
5
6
7
8
9
10
South Korea
China
Taiwan
India
United Kingdom
France
Germany
Australia
876
822
603
269
259
190
142
128
Table 3. Breakdown of Companies by GICS
T
# of Companies
678
513
473
437
411
344
338
314
277
241
IP
Industry
Electronic Equipment & Components
Semiconductors & Semiconductor Equipment
Machinery
Chemicals
Software
Metals & Mining
Auto Components
IT Services
Electrical Equipment
Communications Equipment
CR
GICS Code
452030
453010
201060
151010
451030
151040
251010
451020
201040
452010
US
Rank
1
2
3
4
5
6
7
8
9
10
AN
As for the number of top-tier suppliers each company in the network has (the number of companies
from which there is a direct sales flow), the number ranges from 0 to 433 and is highly skewed to the
M
right. We further calculate the mean number of top-tier suppliers for companies in each tier (away from
ED
Honda) and the result shows the mean number displays a decreasing pattern, from 19 for top-tier, to 13
for tier 2, and to 5 for tier 3. It suggests that as we move from the center of the Honda supply network to
PT
its peripheral, the number of suppliers per company decreases on average. This result is consistent with
our intuition that companies further upstream in a supply chain in general have fewer suppliers.
CE
4.2. Data-analytics system: Storage, visualization, and computation of network data
To facilitate the analysis of Honda data and to potentially generalize to other supply chain datasets,
AC
we build an integrated data-analytics system. It assists in streamlining the tasks of supply network data
storage, visualization, and NSI computation. Utilizing the raw data exported from sources such as
Bloomberg SPLC, the system comprises three main components (solid boxes) and three ancillary
modules (dashed boxes), as shown in Figure 2.
16
ACCEPTED MANUSCRIPT
CR
IP
T
Figure 2. Architecture of Data-Analytics System
US
The first ancillary module, implemented in scripting language, pre-processes the raw data by
transforming it to a format that can be loaded into the storage component of our system. We manage the
AN
collected supply network data in a dedicated graph database, which is an emerging data storage
technology (see https://en.wikipedia.org/wiki/Graph_database and http://neo4j.org). Unlike traditional
M
database which represents data in tabular format, a graph database represents data as nodes, edges, and
ED
properties. Graph database is easily scalable and has demonstrated advantages in the analytics of big,
networked data. Using a graph to represent the supply network is natural, as companies, their
PT
characteristics, supply relationships, and details of the supply relationships can be represented as nodes,
nodes’ properties, directed edges between nodes, and edges’ properties, respectively. More importantly,
CE
using the graph database technology renders graph-like analytical queries much more efficient. It would
AC
take a significantly less amount of time to find out, for instance, who are the tier 3 suppliers to a particular
company, or what is the shortest path between any two companies in the supply network.
The supply network database then feeds the data to two components in the system that handle the
analytics workflow. The first is a user interface for visualizing and navigating the supply network. Taking
the Honda dataset, for example, once it is loaded into the graph database, we can visualize the whole
supply network. Using the interface, researchers can also execute queries or interactively zoom into a
particular local segment of the supply network. Figure 3 is such an example where the focal buying firm
17
ACCEPTED MANUSCRIPT
Honda and 25 randomly selected top-tier suppliers are visualized. The second component that handles the
analytics is a computing engine implemented in statistical programming language R. The component
takes supply network data as input, calculates various network centralities measures, and carries out the
optimization algorithm as specified in the model to compute NSI scores. Lastly, based on the scores, it
T
generates a list of candidates for researchers and managers to conduct further analysis.
ED
M
AN
US
CR
IP
Figure 3. Honda's Local Supply Network (Partially Shown)
PT
The supply network database then feeds the data to two components in the system that handle the
analytics workflow. The first is a user interface for visualizing and navigating the supply network. Taking
CE
the Honda dataset, for example, once it is loaded into the graph database, we can visualize the whole
supply network. Using the interface, researchers can also execute queries or interactively zoom into a
AC
particular local segment of the supply network. Figure 3 is such an example where the focal buying firm
Honda and 25 randomly selected top-tier suppliers are visualized. The second component that handles the
analytics is a computing engine implemented in statistical programming language R. The component
takes supply network data as input, calculates various network centralities measures, and carries out the
optimization algorithm as specified in the model to compute NSI scores. Lastly, based on the scores, it
generates a list of candidates for researchers and managers to conduct further analysis.
18
ACCEPTED MANUSCRIPT
5. Results and Evaluation
5.1. Analysis results
Using the data-analytics system, we compute NSI scores for all the suppliers in the second and third
tiers of Honda’s supply network. We skip the top-tier suppliers because they are all known to Honda and
T
our focus is on identifying unknown but critical suppliers. We do not compute NSI scores for the fourth-
IP
tier suppliers because we lack data on those in tier 5. In Figure 4, we plot the distribution of NSI scores.
CR
We observe a bell-shaped curve with a thin long right tail. The suppliers located at this tail are those
identified to be of great importance and, hence, candidates for nexus suppliers. The inflection point on the
US
right side of the curve occurring pretty close to the long tail is encouraging from the management
perspective because that means we have only a handful of potential nexus suppliers to consider.
CE
PT
ED
M
AN
Figure 4. Distribution of NSI Scores in the 2nd and 3rd Tiers of Honda Supply Network
AC
In Table 4, we list the top 50 suppliers, ranked in the descending order of NSI scores. We find
major multinational companies in the list, such as Siemens, SAP, Hewlett-Packard, and General Electric,
across diverse industry sectors such as electronics, chemicals, and software development. We observe an
overwhelming majority from the second tier, suggesting a systematic difference between the scores of the
tiers 2 and 3 suppliers. Hence, we investigate the respective distributions of NSI scores in the second and
third tiers. In Figure 5, the solid and dotted distributions correspond to the second and third tiers,
respectively. We find the distribution for the second tier is shifted towards the right, suggesting the tier 2
19
ACCEPTED MANUSCRIPT
suppliers on average have a higher NSI score than tier 3 suppliers. The shape of the two curves appears
similar, which suggests lack of underlying systematic bias. In Tables 5 and 6, we list the top 50 suppliers
by NSI score from the second and third tiers.
Table 4. Top 50 Suppliers Ranked by NSI Score in the 2nd and 3rd Tiers of Honda Supply Network
2
2
0.9701
0.9657
28
29
5
6
7
8
9
10
11
12
Samsung Electronics Co Ltd
Microsoft Corp
LG Electronics Inc
Apple Inc
Ford Motor Co
General Motors Co
Intel Corp
Toyota Motor Corp
2
2
2
3
2
2
2
2
0.9598
0.9327
0.9319
0.9267
0.9263
0.9229
0.9209
0.9199
30
31
32
33
34
35
36
37
13
14
15
16
Accenture PLC
Akzo Nobel NV
Sony Corp
Flextronics International
Ltd
Infineon Technologies AG
Zebra Technologies Corp
Volkswagen AG
Fujitsu Ltd
VeriSign Inc
ANSYS Inc
ABB Ltd
Linde AG
2
2
3
2
0.9168
0.9135
0.9132
0.9132
38
39
40
41
2
2
2
2
2
2
2
2
0.9130
0.9119
0.9119
0.9111
0.9100
0.9097
0.9052
0.9012
42
43
44
45
46
47
48
49
Honeywell International
Inc
2
0.9002
50
25
AN
M
ED
PT
CE
17
18
19
20
21
22
23
24
Tier
3
2
NSI
0.8998
0.8977
2
3
0.8976
0.8969
2
2
2
2
2
2
2
2
0.8955
0.8954
0.8953
0.8946
0.8938
0.8934
0.8922
0.8915
3
2
2
2
0.8910
0.8908
0.8907
0.8903
2
2
2
2
2
3
2
3
0.8902
0.8901
0.8899
0.8894
0.8856
0.8844
0.8840
0.8824
2
0.8820
T
Hewlett-Packard Co
General Electric Co
Name
Amazon.com Inc
Silicon Laboratories
Inc
Texas Instruments Inc
Bayerische Motoren
Werke AG
Dow Chemical Co/The
ON Semiconductor Corp
BT Group PLC
Daimler AG
Ingram Micro Inc
Cisco Systems Inc
Nissan Motor Co Ltd
Telefonaktiebolaget LM
Ericsson
Airbus Group NV
Caterpillar Inc
Arrow Electronics Inc
Avago Technologies Ltd
IP
Rank
26
27
CR
NSI
1.0000
0.9711
US
Tier
2
2
3
4
Name
Siemens AG
SAP SE
AC
Rank
1
2
Vodafone Group PLC
Atmel Corp
Open Text Corp
Adobe Systems Inc
Avnet Inc
Boeing Co/The
Polycom Inc
Fiat Chrysler
Automobiles NV
Danaher Corp
Table 5. Top 50 Suppliers Ranked by NSI Score in the 2nd Tier of Honda Supply Network
Rank
1
2
3
4
5
6
7
8
Name
Siemens AG
SAP SE
Hewlett-Packard Co
General Electric Co
Samsung Electronics Co Ltd
Microsoft Corp
LG Electronics Inc
Ford Motor Co
NSI
1.0000
0.9711
0.9701
0.9657
0.9598
0.9327
0.9319
0.9263
Rank
26
27
28
29
30
31
32
33
20
Name
Dow Chemical Co/The
ON Semiconductor Corp
BT Group PLC
Daimler AG
Ingram Micro Inc
Cisco Systems Inc
Nissan Motor Co Ltd
Telefonaktiebolaget Ericsson
NSI
0.8955
0.8954
0.8953
0.8946
0.8938
0.8934
0.8922
0.8915
ACCEPTED MANUSCRIPT
Caterpillar Inc
Arrow Electronics Inc
Avago Technologies Ltd
Vodafone Group PLC
Atmel Corp
Open Text Corp
Adobe Systems Inc
Avnet Inc
Polycom Inc
Danaher Corp
WPP PLC
MicroStrategy Inc
salesforce.com inc
Teradata Corp
Rio Tinto PLC
Smiths Group PLC
Renault SA
T
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
IP
0.9229
0.9209
0.9199
0.9168
0.9135
0.9132
0.9130
0.9119
0.9119
0.9111
0.9100
0.9097
0.9052
0.9012
0.9002
0.8977
0.8976
CR
General Motors Co
Intel Corp
Toyota Motor Corp
Accenture PLC
Akzo Nobel NV
Flextronics International Ltd
Infineon Technologies AG
Zebra Technologies Corp
Volkswagen AG
Fujitsu Ltd
VeriSign Inc
ANSYS Inc
ABB Ltd
Linde AG
Honeywell International Inc
Silicon Laboratories Inc
Texas Instruments Inc
US
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
0.8908
0.8907
0.8903
0.8902
0.8901
0.8899
0.8894
0.8856
0.8840
0.8820
0.8816
0.8800
0.8791
0.8758
0.8753
0.8752
0.8751
Table 6. Top 50 Suppliers Ranked by NSI Score in the 3rd Tier of Honda Supply Network
PT
CE
Rank
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
M
AN
NSI
0.9267
0.9132
0.8998
0.8969
0.8910
0.8844
0.8824
0.8790
0.8765
0.8756
0.8749
0.8738
0.8658
0.8643
0.8605
0.8559
0.8555
0.8552
0.8531
0.8527
0.8481
0.8471
0.8464
0.8464
0.8460
ED
Name
Apple Inc
Sony Corp
Amazon.com Inc
Bayerische Motoren Werke AG
Airbus Group NV
Boeing Co/The
Fiat Chrysler Automobiles NV
Royal Dutch Shell PLC
Hyundai Motor Co
Verizon Communications Inc
Lockheed Martin Corp
Koninklijke Philips NV
Nestle SA
Unilever NV
Peugeot SA
Telefonica SA
Bombardier Inc
Exxon Mobil Corp
Deutsche Lufthansa AG
HTC Corp
AT&T Inc
Northrop Grumman Corp
Berkshire Hathaway Inc
Mitsubishi Motors Corp
Kia Motors Corp
AC
Rank
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
21
Name
Procter & Gamble Co/The
Rolls-Royce Holdings PLC
China Mobile Ltd
RadioShack Corp
Asustek Computer Inc
Delta Electronics Inc
Byd Co Ltd
China Unicom Hong Kong Ltd
Huawei Technologies Co Ltd
Areva SA
Safran SA
Koninklijke KPN NV
Deutsche Post AG
General Dynamics Corp
China Telecom Corp Ltd
Walgreens Boots Alliance Inc
Petroleo Brasileiro SA
Robert Bosch GmbH
Staples Inc
International Paper Co
Hyundai Heavy Industries Co
Textron Inc
Redington India Ltd
WPG Holdings Ltd
Sprint Corp
NSI
0.8445
0.8423
0.8383
0.8379
0.8371
0.8364
0.8358
0.8352
0.8337
0.8331
0.8315
0.8284
0.8272
0.8266
0.8264
0.8260
0.8247
0.8240
0.8236
0.8234
0.8229
0.8217
0.8209
0.8205
0.8193
ACCEPTED MANUSCRIPT
CR
IP
T
Figure 5. Distribution of NSI Scores by Tier in Honda Supply Network
We classify the top 50 nexus supplier candidates in the second tier (Table 5) and those in the third
US
tier (Table 6) into three groups: operational, monopolistic, and informational nexus suppliers, according
to their characteristics in network structure. We apply the following decision criteria per Table 1:
A supplier is an operational nexus supplier if its degree, betweenness, and eigenvector centrality
AN

M
values are all in the top 20 percentile among the top 50 nexus suppliers in its own tier. They are
expected to have significant influence on the operational performance of Honda’s end products.
A supplier is a monopolistic nexus supplier if its betweenness centrality value is in the top 10
ED

PT
percentile among the top 50 nexus suppliers in its tier. They are expected to have significant
impacts on Honda’s supply continuity.
A supplier is an informational nexus supplier if the number of unique industries its suppliers
CE

encompass is in the top 10 percentile among the top 50 nexus suppliers in its own tier. Due to
AC
their diverse ties in the extended supply network, these nexus suppliers are expected to be
Honda’s important sources of early market and technological information.
In choosing the cutoff percentage points (i.e., 20% for operational nexus supplier, 10% for
monopolistic nexus supplier, and 10% for informational nexus supplier), our goal is to identify the top 5
companies from the 50 nexus suppliers for each category. Supply chain managers should choose the
cutoff points based on their specific business requirement and constraints. Applying our cutoff points to
the data, we identify the operational, monopolistic, and informational nexus suppliers in the second and
22
ACCEPTED MANUSCRIPT
third tiers of the Honda supply network, as shown in Tables 7, 8, and 9, respectively. We observe there is
a significant overlap between top operational, monopolistic, and informational nexus suppliers. In Tier 2,
Siemens, Hewlett-Packard, General Electric, and Samsung are among the top 5 in all three lists, and in
Tier 3, Amazon and Koninklijke Philips appear in all three lists. We stress that since our data is on the
T
business relationships between companies, the index captures the criticality of companies in the network
IP
of business relationships, so it does not necessarily reflect their importance in terms of material flow.
CR
Table 7. Operational Suppliers
Betweenness
0.0594
0.0246
0.0414
0.0568
0.0414
Eigenvector
0.0991
0.1067
0.1652
0.1025
0.1362
NSI
1.0000
0.9711
0.9701
0.9657
0.9598
Degree
0.0327
0.0272
0.0253
0.0269
0.0281
Betweenness
0.0274
0.0131
0.0178
0.0126
0.0169
Eigenvector
0.1015
0.1064
0.0855
0.0619
0.0652
NSI
0.9267
0.9132
0.8998
0.8844
0.8738
M
AN
US
Degree
0.0479
0.0281
0.0514
0.0503
0.0407
ED
Tier 2
Company
Siemens AG
SAP SE
Hewlett-Packard Co
General Electric Co
Samsung Electronics Co
Tier 3
Company
Apple Inc
Sony Corp
Amazon.com Inc
Boeing Co/The
Koninklijke Philips NV
Table 8. Monopolistic Suppliers
AC
CE
PT
Tier 2
Company
Siemens AG
General Electric Co
LG Electronics Inc
Hewlett-Packard Co
Samsung Electronics Co
Tier 3
Company
Apple Inc
Nestle SA
Unilever NV
Amazon.com Inc
Koninklijke Philips NV
Betweenness
0.0594
0.0568
0.0505
0.0414
0.0414
NSI
1.0000
0.9657
0.9319
0.9701
0.9598
Betweenness
0.0274
0.0246
0.0229
0.0178
0.0169
NSI
0.9267
0.8658
0.8643
0.8998
0.8738
Table 9. Informational Suppliers
Tier 2
Company
General Electric Co
Siemens AG
Hewlett-Packard Co
Industries
41
38
36
23
NSI
0.9657
1.0000
0.9701
ACCEPTED MANUSCRIPT
0.8915
0.9598
0.9119
0.8955
Industries
45
43
40
36
33
NSI
0.8643
0.8658
0.8738
0.8998
0.8260
T
32
29
29
29
IP
Telefonaktiebolaget LM Ericsson
Samsung Electronics Co Ltd
Volkswagen AG
Dow Chemical Co/The
Tier 3
Company
Unilever NV
Nestle SA
Koninklijke Philips NV
Amazon.com Inc
Walgreens Boots Alliance Inc
CR
5.2. Evaluation and validation: Visit to Honda
One member of our research team met with the supply chain managers of Honda North America to
US
discuss the results. The Honda supply chain managers recognized that the data analytics approach
underlying this study of nexus supplier index provides a novel perspective and could be complementary
AN
to the current practice of strategic supplier management which is largely material-flow based and risk
oriented. Through carefully examining our results (e.g., Tables 4-9), Honda managers found that the
M
proposed NSI analytical framework could contribute to the practice in a number of ways. The lists of
ED
nexus supplier candidates based on NSI scores could help managers to look for key companies that they
were not aware of (i.e., identifying blind spots in their supply networks). For instance, there were
PT
suppliers in the lists we compiled whose names they had not heard before, and seeing them on the list
CE
would motivate the Honda managers to begin getting better acquainted with these suppliers.
Further, while Honda managers anticipated Siemens and SAP would come out on top in Table 4,
AC
new observation was made that Honda and VW are connected in the tier 2 level (see Table 5). Given that
VW was currently experiencing difficulties and might reduce its production, Honda could look at which
of the Honda’s top-tier suppliers might end up suffering and take proactive measures to mitigate any
potential negative impact. Other novel findings include that Apple is the most important tier 3 supplier to
Honda (see Table 6). This was a surprise to Honda managers. It is possible that the suppliers in the second
tier may be buying a lot of hardware products from Apple and that is how Apple appears as the most
significant nexus supplier at the third tier. However, Apple also sells software and it is possible software
24
ACCEPTED MANUSCRIPT
produced by Apple may end up in a component that goes into a Honda product. In a sense, our results
offer what to look for and where to start looking.
For the suppliers not well-known to Honda, managers expressed that they could start with a small
contract to begin the process of discovery and relationship building. Honda can also introduce them to its
T
other supplier companies. For instance, if the NSI list identifies a tier 3 supplier Honda is interested in
IP
working with more closely, Honda could introduce this company to other top tier and tier 2 suppliers for
CR
potential business relationships. These new findings make it possible for Honda to determine which
suppliers and/or their extended business relationships to monitor more closely when making policy or
US
investment decisions. These lists could offer the first cut at which supplier companies to consider.
6. Discussion and Implications
AN
Our study attempts to identify critical suppliers that may be embedded deep in a supply chain. A
focal buying firm such as Honda may not be aware of these hidden yet critical suppliers called nexus
M
suppliers. We have taken a step toward looking beyond a buying firm’s typical supply chain management
ED
space. Since it is impossible to actively manage all suppliers in the supply network, we posit that these
nexus suppliers offer a good starting point. To that end, we articulate a way of computing the NSI and
PT
show how to use the results from the focal buying firm’s perspective.
CE
To manage a multi-tier supply chain, one needs to engage in the mapping of the supply network
(e.g., Choi & Hong, 2002; Pathak et al., 2007). This mapping can be difficult and time-consuming. Often,
AC
plant-level supply networks are constructed through the use of bill of materials (BOM) (they identify all
suppliers attached to each part and then start building the network one tier at a time). A Japanese
semiconductor company mapped their supply networks after the 2011 tsunami and the complete mapping
of their supply networks took one year and involved more than 100 people. A pharmaceutical company
also revealed to us that when it mapped the supply network for one of its drug products, they had to
devote 400 to 500 people and a total of 18 months. Thus, in comparison, even with its granularity
25
ACCEPTED MANUSCRIPT
shortcomings (i.e. the data exist at the company level), the use of Bloomberg Terminal data still
represents a viable approach to developing the NSI.
According to a Honda top executive, “If we can glean one or two or 10 key hints on the list, these
might be suppliers we need to dig into more, partner with, mitigate risk with; there is huge value to that.”
T
NSI provides a company like Honda with insights into which suppliers to partner with. It can help the
IP
company look beyond the top-tier level and outside its normal supply chain scope. It can also help the
company look at its supply chain differently, not just to manage risks such as natural disasters but also to
CR
identify certain suppliers as “nuggets” to facilitate closer partnerships and seek more information sharing.
US
NSI identifies a new kind of critical suppliers, different from strategic suppliers that a buying
company is used to working with (Yan et al., 20015). While a strategic supplier is critical due to its
AN
internal attributes, a nexus supplier is critical because of its ties with other suppliers in the supply
network. Typically, the focal buying firm and its strategic suppliers have high co-dependence and often
M
undertake costly integration projects (Salvador & Villena, 2013). In contrast, the focal firm and its nexus
ED
suppliers are not necessarily connected with each other. Instead of direct connections, they are likely
connected over multiple links in an indirect way (Yan et al., 20015). Nexus suppliers may not possess
PT
superior capabilities and resources that can have immediate impact on the buying firm. The benefits of
CE
managing nexus suppliers consist in the portfolio of ties within the broader supply network. To that end,
the focal firm should examine the network positions and embeddedness of its suppliers and evaluate their
AC
overall criticality from a network perspective (Choi & Kim, 2008). In essence, the buying firm should try
to identify nexus suppliers through the complex interrelations (i.e., links) with its other suppliers in the
extended supply network.
Due to the time and resource constraints, many buying firms focus on their top-tier, strategic
suppliers. This is understandable, but giving exclusive attention to top-tier suppliers can increase
susceptibility to risk. Suppliers further upstream in the supply chain can cause unexpected disruptions to a
firm’s operations. In this light, it is crucial for the focal firm to identify its nexus suppliers and seek ways
26
ACCEPTED MANUSCRIPT
to manage them proactively. These suppliers may have a profound impact on its performance, depending
on how they are embedded in the supply network and how an unforeseen risk event takes place. The
influence of a nexus supplier on the focal firm can lag in time, depending again on how many tiers they
are removed from the buying firm. Still, the impact of nexus suppliers can be real and significant,
T
especially in the areas of materials flow, emerging ideas, and market information.
IP
6.1. Research implications
CR
Implications can be drawn from our data analytics approach for academic research that also uses
network data. First, our study showcases the feasibility of identifying potentially important nodes in a
US
network by integrating various centrality measures. Instead of using each measure individually, our DEAbased model combines them into one aggregate measure that reflects the overall criticality of a certain
AN
node in the network. Researchers in other disciplines like computer science and psychology can adopt the
same approach for developing their own methodologies if a similar need arises from a specific context
M
(e.g., relationship cultivations in social media).
ED
Second, the aggregate measure makes it possible to highlight a node of significance through data
visualization and facilitate the drill down capability in data analytics. That is, data scientists can use a
PT
number like NSI score to visually rank order the nodes or display them in different sizes and colors to
CE
help users focus on most important areas. It is also noted that our DEA-based model takes into account
the node importance at multiple levels—at the node level (degree), neighborhood level (eigenvector), and
AC
network level (betweenness and farness) (Marsden, 2002; Wasserman & Faust, 1994). Therefore,
compared with the individual approaches that rely on a particular centrality measure, our NSI approach is
theoretically more comprehensive.
Finally, by letting the data speak for itself, our DEA-based model suggests an objective approach to
deciding the weights when combining individual measures for the optimization of NSI scores. Our
approach avoids the equal and subjective weighting schemes commonly used in existing studies (e.g.,
Huang et al., 2014). In other words, our DEA-based model does not assume equal importance for the
27
ACCEPTED MANUSCRIPT
centrality measures involved; neither does it assign subjective weights based on the user’s perception
about their importance. Instead, our model treats weights as decision variables and objectively decides
their values in optimizing NSI scores. In the future, researchers in this genre of studies should look for
alternate ways of weighting approaches beyond what we have done here.
T
6.2. Practical implications
IP
Several practical implications can also be drawn from our findings. First, part of the importance of
CR
NSI derives from nexus suppliers’ potential to provide strategic information and help enhance the
performance of operations at the buying firm. This potential can be only realized if the buying firm would
US
recognize and monitor its nexus suppliers to leverage such potentials. Second, nexus suppliers differ from
strategic suppliers in the way they are embedded in the extended supply network (Yan et al., 20015). It is
AN
noted that nexus suppliers may not necessarily have superior internal capabilities in terms of operational
capability, technological leadership, or strategic cultural congruence with the buying company. As such,
M
the focal firm may easily overlook these critical yet hidden suppliers. Third, the extended supply network
ED
is the context through which one identifies nexus suppliers. Managers need to consider a supplier’s
relationships not only within the buying firm’s supply base but also with firms outside the base in other
PT
related industries.
CE
Our NSI study provides guidelines for supply managers to better understand the concept of nexus
suppliers and identify and categorize them into the respective types. As the potential risks incurred by
AC
failure of nexus suppliers rise, firms are in need to learn more about nexus suppliers but they have
difficulty identifying these critical suppliers. A better understanding of nexus suppliers should help
supply managers take a multi-tier network view to look beyond strategic suppliers in the immediate range
of its supply network (Craighead et al., 2007; Yan et al., 2015). In a changing business environment,
many innovative ideas will come from non-traditional suppliers of parts and components or from start-up
enterprises in emerging markets. These suppliers can more appropriately be qualified as nexus suppliers
but less likely as strategic suppliers.
28
ACCEPTED MANUSCRIPT
NSI can also help firms manage supply risks. For instance, the buying firm could develop a supplier
development program, which would require the commitment of significant financial, capital, and
personnel resources. It is critical for the buying firm to select the right portfolio of suppliers to be
included in the program. However, selecting the right suppliers is not straightforward, as it requires the
T
assessment of a supplier’s true value, determined not only by its internal qualities but also by its network
IP
positions. Analysis of the extended supplier network through our NSI model can help in this regard. A
CR
supplier can also benefit from a better understanding of how its position in the broader interorganizational networks may affect its value to other customers and stakeholders. It can proactively
US
cultivate the relationships with other firms in its extended supply network, and by doing so, it can gain
competitive advantage. For example, once a supplier learns that a particular customer is seeking ideas for
AN
product innovation, it can diversify its inter-organizational links to help capture early market signals and
share relevant information with that customer. By the same token, upon recognizing its potential value as
M
a nexus supplier to its customers, a supplier can approach them directly as a more attractive candidate for
ED
supplier development or investment and seek to develop mutually beneficial long-term relationships.
6.3. Limitations
PT
Like any research, our study has limitations. First, we recognize the exploratory nature of our NSI
CE
study. The DEA model will need to be further refined as we continue to gather more insights on nexus
suppliers. The data-analytics framework is general and the best analytic components can be adapted and
AC
adopted on a case-by-case basis. When applying our framework to a different company or dataset, nonDEA approaches (e.g., AHP, MCDP, etc.) can be considered to implement the NSI. Future studies can
investigate the advantages and disadvantages of different approaches. Second, the issue of how a nexus
supplier may impact the performance of the focal company needs to be further investigated. Our NSI
study takes the first step toward a model by which nexus suppliers can be identified and categorized into
different types. However, the specific patterns through which the impact on the focal firm’s performance
manifests itself still need to be empirically investigated. Third, while the Bloomberg SPLC data provides
29
ACCEPTED MANUSCRIPT
an extensive coverage of business relationships among firms, its lack of data resolution (e.g., lack of
plant- or product-level information, lack of temporal information, etc.) has limited our empirical analysis
and interpretation. As another avenue for future research, additional data sources, contexts, and firms
(e.g., firms in the service industry) can be sought to further validate and improve the NSI model and the
T
data analytics approach to assess if they can be generalized to other settings. Finally, the focus of our
IP
paper is on the identification of nexus suppliers, but each type of nexus suppliers may exhibit different
CR
network behaviors and hence have different impacts on the performance of the focal buying firm. Future
research can look into such performance impacts in terms of operations, continuity and technological
US
innovations of the focal buying firm based on the types of nexus suppliers.
7. Conclusion
AN
In this study, we propose Nexus Supplier Index (NSI), a data-analytics framework for identifying
M
and classifying nexus suppliers. We assess the method’s effectiveness through an empirical application to
the real-world supply network of Honda. The novelty and potential of our findings are positively
ED
evaluated by Honda managers. To streamline the analytics workflow for NSI, we also build an integrated
information system to handle the tasks of data storage, management, navigation, and analytics. Through
PT
this research, we put the theory of nexus suppliers into action and demonstrate the value of data analytics
CE
in providing supply managers with a better understanding of nexus suppliers both as a management tool
and as a practical procedure that can help them manage the risks and harness market opportunities. Our
AC
research takes the first step toward the practical use of the nexus supplier concept by identifying and
categorizing nexus suppliers with publicly available data (e.g., Bloomberg SPLC). Many firms have
difficulty identifying nexus suppliers, let alone knowing how to develop effective strategies for managing
relationships with them. Our research provides supply managers with insights into this critical issue and
assists them in developing a roadmap to manage nexus supplier relationships.
Using data analytics, we offer a roadmap for the focal buying firm to identify and categorize nexus
suppliers. We discuss practical implications and lessons learned from the findings of our study. This
30
ACCEPTED MANUSCRIPT
research provides supply managers with insights into the presence of nexus suppliers and assists them in
developing potential strategies to better manage this new type of critical suppliers whose importance is
grounded in their network structural positions in the supply networks.
Acknowledgments
T
The authors thank the anonymous reviewers and the editor-in-chief for their constructive comments and
IP
insightful suggestions. Any errors that remain are the sole responsibility of the authors.
CR
References
AC
CE
PT
ED
M
AN
US
Ang, E., lancu, D.A., & Swinney, R. (2017). Disruption Risk and Optimal Sourcing in Multitier Supply
Networks. Management Science, 63(8), 2397-2419.
Abrahams, A.S., Fan, W., Wang, G.A., Zhang, Z., & Jiao, J. (2015). An integrated text analytic
framework for product defect discovery. Production and Operations Management, 24, 975–990.
Andersson, U., Forsgren, M., & Holm, U. (2002). The strategic impact of external networks: Subsidiary
performance and competence development in the multinational corporation. Strategic Management
Journal, 23, 979–996.
Arino, A., Abramov, M., Skorobogatykh, I., Rykounina, I., & Vila, J. (1997). Partner selection and trust
building in West European-Russian joint ventures: A western perspective. International Studies of
Management & Organization, 27, 19–37.
Artto, K., Eloranta, K., & Kujala, J. (2008). Subcontractors’ business relationships as risk sources in
project networks. International Journal of Managing Projects in Business, 1, 88–105.
Azadegan, A., Dooley, K.J., Carter, P.L., & Carter, J.R. (2008). Supplier innovativeness and the role of
interorganizational learning in enhancing manufacturer capabilities. Journal of Supply Chain
Management, 44, 14–35.
Basole, R.C., & Bellamy, M.A. (2014). Visual analysis of supply network risks: Insights from the
electronics industry. Decision Support Systems, 67, 109-120.
Basole, R.C., Bellamy, M.A., & Park, H. (2017). Visualization of innovation in global supply chain
networks. Decision Sciences, 48(2), 288-306.
Bellamy, M.A., Ghosh, S., & Hora, M. (2014). The influence of supply network structure on firm
innovation. Journal of Operations Management, 32, 357–373.
Bensaou, M. (1999). Portfolios of buyer-supplier relationships. Sloan Management Review, 40(4), 35–44.
Bhattacharjee, S., & Cruz, J. (2015). Economic sustainability of closed loop supply chains: A holistic
model for decision and policy analysis. Decision Support Systems, 77, 67-86.
Bonacich, P. (1987). Power and centrality: A family of measures. American Journal of Sociology, 92(5),
1170-1182.
Borgatti, S.P., & Li, X. (2009). On social network analysis in a supply chain context. Journal of Supply
Chain Management, 45, 5–22.
Carty, S.S. (2012). PA-12 resin shortages from German supplier fire could disrupt auto industry. The
Huffington Post, April 17, 2012.
Cavusgil, S.T., Yeoh, P.-L., & Mitri, M. (1995). Selecting foreign distributors: An expert systems
approach. Industrial Marketing Management, 24, 297–304.
Chan, H.K., Wang, X., Lacka, E., & Zhang, M. (2016). A mixed-method approach to extracting the value
of social media data. Production and Operations Management, 25(3), 568-583.
31
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
AN
US
CR
IP
T
Charnes, A., Cooper, W.W., & Rhodes, E. (1978). Measuring the efficiency of decision making units,
European Journal of Operational Research, 2(6), 429–444.
Chen, I.J., Paulraj, A., & Lado, A.A. (2004). Strategic purchasing, supply management, and firm
performance. Journal of Operations Management, 22, 505–523.
Chin, K.S., Wang, Y.M., Poon, G.K.K., & Yang, J.B. (2009). Failure mode and effects analysis by data
envelopment analysis. Decision Support Systems, 48(1), 246-256.
Choi, T.Y., Dooley, K.J., & Rungtusanatham, M. (2001). Supply networks and complex adaptive
systems: Control versus emergence. Journal of Operations Management, 19, 351–366.
Choi, T.Y., & Hong, Y. (2002). Unveiling the structure of supply networks: Case studies in Honda,
Acura, and DaimlerChrysler. Journal of Operations Management, 20, 469–493.
Choi, T.Y., & Kim, Y. (2008). Structural embeddedness and supplier management: A network
perspective. Journal of Supply Chain Management, 44, 5–13.
Choi, T.Y., Linton, T. (2011). Don’t let your supply chain control your business. Harvard Business
Review 89. 112–117.
Chou, Y.C., & Shao, B.B.M. (2014). Total factor productivity growth in information technology services
industries: A multi-theoretical perspective. Decision Support Systems, 62(1), 106-118.
Coelli, T.J., Rao, D.S.P., O’Donnell, C.J., & Battese, G.E. (2005). An Introduction to Efficiency and
Productivity Analysis. New York, NY: Springer Science & Business Media.
Cook, W.D., Tone, K., & Zhu, J. (2014). Data envelopment analysis: Prior to choosing a model. Omega,
44, 1-4.
Craighead, C. W., Blackhurst, J., Rungtusanatham, M. J., & Handfield, R. B. (2007). The severity of
supply chain disruptions: Design characteristics and mitigation capabilities. Decision Sciences,
38(1), 131-156.
Dacin, M.T., Ventresca, M.J., & Beal, B.D. (1999). The embeddedness of organizations: Dialogue &
directions. Journal of Management, 25, 317–356.
Dong, M.C., Liu, Z., Yu, Y., & Zheng, J.-H. (2015). Opportunism in distribution networks: The role of
network embeddedness and dependence. Production and Operations Management, 24, 1657–1670.
Dyer, J.H., & Singh, H. (1998). The relational view: Cooperative strategy and sources of
interorganizational competitive advantage. Academy of Management Review, 23, 660–679.
Echols, A., & Tsai, W. (2005). Niche and performance: The moderating role of network embeddedness.
Strategic Management Journal, 26, 219–238.
Ellram, L.M., & Carr, A. (1994). Strategic purchasing: A history and review of the literature. Journal of
Supply Chain Management, 30(1), 9–19.
Färe, R., Grosskopt, S., Norris, M., & Zhang, Z. (1994). Productivity growth, technical progress, and
efficiency change in industrialized countries. American Economic Review, 84(1), 66-83.
Ferreira, K.J., Lee, B.H.A., & Simchi-Levi, D. (2016). Analytics for an online retailer: Demand
forecasting and price optimization. Manufacturing & Service Operations Management, 18(1), 69-88.
Freeman, L. C. (1978). Centrality in social networks conceptual clarification. Social Networks, 1(3), 215239.
Gadde, L.-E., & Snehota, I. (2000). Making the most of supplier relationships. Industrial Marketing
Management, 29, 305–316.
Giannoulis, C., & Ishizaka, A. (2010). A Web-based decision support system with ELECTRE III for a
personalised ranking of British universities. Decision Support Systems, 48(3), 488-497, 2010.
Gnyawali, D.R., & Madhavan, R. (2001). Cooperative networks and competitive dynamics: A structural
embeddedness perspective. Academy of Management Review, 26, 431–445.
Gulati, R., & Gargiulo, M. (1999). Where do interorganizational networks come from? American Journal
of Sociology, 104, 177–231.
Guo, P., Song, J.S., & Wang, Y. (2010). Outsourcing structures and information flow in a three-tier
supply chain. International Journal of Production Economics, 128, 175-187.
32
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
AN
US
CR
IP
T
Hagedoorn, J. (2006). Understanding the cross-level embeddedness of interfirm partnership formation.
Academy of Management Review, 31, 670–680.
Hitt, M.A., Dacin, M.T., Levitas, E., Arregle, J.-L., & Borza, A. (2000). Partner selection in emerging and
developed market contexts: Resource-based and organizational learning perspectives. Academy of
Management Journal, 43, 449–467.
Hu, B., Kostamis, D. (2015). Managing supply disruptions when sourcing from reliable and unreliable
suppliers. Production and Operations Management, 24, 808–820.
Huang, S., Lv, T., Zhang, X., Yang, Y., Zheng, W., & Wen, C. (2014). Identifying node role in social
network based on multiple indicators. PLOS ONE, 9(8), 1–16.
Huang, T., & Van Mieghem, J.A. (2014). Clickstream data and inventory management: Model and
empirical analysis. Production and Operations Management, 23, 333–347.
Kaufman, A., Wood, C.H., & Theyel, G. (2000). Collaboration and technology linkages: A strategic
supplier typology. Strategic Management Journal, 21, 649–663.
Kilduff, M. & Tsai, W. (2003). Social Networks and Organizations. Sage Publications, London.
Kim, Y., Choi, T.Y., Yan, T., & Dooley, K. (2011). Structural investigation of supply networks: A social
network analysis approach. Journal of Operations Management, 29, 194–211.
Korhonen, P., & Syrjänen, M. (2004). Resource allocation based on efficiency analysis. Management
Science, 50(8), 1134-1144.
Koufteros, X., Vickery, S.K., & Dröge, C. (2012). The effects of strategic supplier selection on buyer
competitive performance in matched domains: Does supplier integration mediate the relationships?
Journal of Supply Chain Management, 48, 93–115.
Kraljic, P. (1983). Purchasing must become supply management. Harvard Business Review, 61(5), 109–
117.
Li, C. (2013). Sourcing for supplier effort and competition: Design of the supply base and pricing
mechanism. Management Science, 59, 1389-1406.
Li, T., Sethi, S.P., & Zhang, J. (2013). Supply diversification with responsive pricing. Production and
Operations Management, 22, 447–458.
Liu, F.-H.F., & Hai, H.L. (2005). The voting analytic hierarchy process method for selecting supplier.
International Journal of Production Economics, 97, 308–317.
Mackelprang, A. W., Bernardes, E., Burke, G. J., & Welter, C. (2017). Supplier innovation strategy and
performance: A matter of supply chain market positioning. Decision Sciences (forthcoming).
Mafakheri, F., Breton, M., & Ghoniem, A. (2011). Supplier selection-order allocation: A two-stage
multiple criteria dynamic programming approach. International Journal of Production Economics,
132, 52–57.
Marsden, P. V. (2002). Egocentric and sociocentric measures of network centrality. Social
Networks, 24(4), 407-422.
McCarter, M.W., & Northcraft, G.B. (2007). Happy together? Insights and implications of viewing
managed supply chains as a social dilemma. Journal of Operations Management, 25, 498–511.
Monczka, R.M., Petersen, K.J., Handfield, R.B., & Ragatz, G.L. (1998). Success factors in strategic
supplier alliances: The buying company perspective. Decision Sciences, 29, 553–577.
Olsen, R.F., & Ellram, L.M. (1997). A portfolio approach to supplier relationships. Industrial Marketing
Management, 26, 101–113.
Pathak, S.D., Day, J.M., Nair, A., Sawaya, W.J., & Kristal, M.M. (2007). Complexity and adaptivity in
supply networks: Building supply network theory using a complex adaptive systems perspective.
Decision Sciences, 38, 547–580.
Salvador, F., & Villena, V.H. (2013). Supplier integration and NPD outcomes: Conditional moderation
effects of modular design competence. Journal of Supply Chain Management, 49(1), 87-113.
Sherman, H.D., & Zhu, J. (2006). Service Productivity Management: Improving Service Performance
Using Data Envelopment Analysis (DEA). New York, NY: Springer Science & Business Media.
33
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
AN
US
CR
IP
T
Simchi-Levi, D. (2014). OM research: From problem-driven to data-driven research. Manufacturing &
Service Operations Management, 16, 2–10.
Slobodow, B., Abdullah, O., & Babuschak, W.C. (2008). When supplier partnerships aren’t. Sloan
Management Review, 49(2), 77-83.
Talluri, S., DeCampos. H.A., & Hult, G.T.M. (2013). Supplier rationalization: A sourcing decision
model. Decision Sciences, 44(1), 57-86.
Wagner, S.M., & Johnson, J.L. (2004). Configuring and managing strategic supplier portfolios. Industrial
Marketing Management, 33, 717–730.
Wasserman, S., & Faust, K. (1994). Social Network Analysis. New York: Cambridge University Press.
Wu, W.Y., Shih, H.-A., & Chan, H.-C. (2009). The analytic network process for partner selection criteria
in strategic alliances. Expert Systems with Applications, 36, 4646–4653.
Yahya, S., & Kingsman, B. (1999). Vendor rating for an entrepreneur development programme: A case
study using the analytic hierarchy process method. Journal of the Operational Research Society, 50,
916–930.
Yan, T., Choi, T.Y., Kim, Y., & Yang, Y. (2015). A theory of the nexus supplier: A critical supplier from
a network perspective. Journal of Supply Chain Management, 51, 52–66.
Yim, A. (2014). Failure risk and quality cost management in single versus multiple sourcing
decision. Decision Sciences, 45(2), 341-354.
Zahra, S.A., Ireland, R.D., Gutierrez, I., & Hitt, M.A. (2000). Introduction to special topic forum
privatization and entrepreneurial transformation: Emerging issues and a future research agenda.
Academy of Management Review, 25, 509–524.
34
ACCEPTED MANUSCRIPT
Biographic Note
T
Benjamin Shao is an associate professor of information systems and the co-director of the Digital Society
Initiative in the W. P. Carey School of Business at Arizona State University. His research interests include
IT impacts, business analytics, IT supply chain interface, and healthcare IT.
CR
IP
Zhan (Michael) Shi is an assistant professor of information systems at W.P. Carey School of Business,
Arizona State University. He uses economic models and machine learning methods to study digital
product markets, social and organization networks, and entrepreneurship in the tech industry.
AN
US
Thomas Choi is Harold E. Fearon Chair of Purchasing Management at W. P. Carey School of Business,
Arizona State University. He has been studying the upstream side of supply chains, where a buying
company interfaces with many suppliers organized in various forms of networks.
AC
CE
PT
ED
M
Sangho Chae is an assistant professor of supply chain management in Tilburg School of Economics and
Management at Tilburg University. His research interests include supply network structure, multi-tier
supply chain management, and behavioral aspects of supply chain decision-making.
35
ACCEPTED MANUSCRIPT
Highlights
We propose a data-analytic approach for identifying hidden critical suppliers in supply
networks.

We construct a large dataset of the multi-tier supply network of Honda Motors.

We apply our data-analytic method to analyze Honda’s supply network and identify the
nexus suppliers.

We review the method and results with the supply management team at Honda America.
AC
CE
PT
ED
M
AN
US
CR
IP
T

36
Документ
Категория
Без категории
Просмотров
0
Размер файла
1 558 Кб
Теги
008, 2018, dss
1/--страниц
Пожаловаться на содержимое документа