Structural anatomy and evolution of supply chain alliance networks: A multi‐method approach

• Author: Hyunwoo Park,Rahul C. Basole,Marcus A. Bellamy
• DOI: http://doi.org/10.1016/j.jom.2018.09.001
• Date: 01 November 2018
• Published date: 01 November 2018

Contents lists available at ScienceDirect

Journal of Operations Management

journal homepage: www.elsevier.com/locate/jom

Structural anatomy and evolution of supply chain alliance networks: A

multi-method approach

Hyunwoo Park

a,∗

, Marcus A. Bellamy

b

, Rahul C. Basole

c

a

Fisher College of Business, The Ohio State University, USA

b

Questrom School of Business, Boston University, USA

c

College of Computing and Institute for People & Technology, Georgia Institute of Technology, USA

ARTICLE INFO

Keywords:

Supply chain alliance network

Network panel

Exponential random graph model (ERGM)

Multi-method

ABSTRACT

We investigate the evolution of supply chain alliance networks with a focus on the influence of structural, firm-,

and industry-level mechanisms. While several structural supply chain characteristics have been found to be

significant drivers of firm innovation and performance, a dearth of studies exists examining how these char-

acteristics change over time by influencing one another. We develop and empirically test hypotheses on the

impact of prior structural configurations and the moderating roles of absorptive capacity and industry growth on

the temporal trajectory of supply chain alliance network structures. Adopting a multi-method approach, we

jointly use econometric analyses and simulation experiments to examine our hypotheses from complementary

angles. Specifically, we characterize the dynamic relationship between the structural mechanisms on a long-

itudinal dataset of 2221 unique firms and 13,668 firm-year observations spanning 25 years. We find empirical

support for negative crossover effects between two key structural properties of supply chain alliance networks, a

positive moderation of a firm's absorptive capacity, and a negative moderation of industry growth on the

structural reinforcement. We conduct corresponding simulation experiments based on a separable temporal

exponential random graph model (STERGM) to track the temporal changes in the simulated networks' key

measures. The simulation results concur with most of our empirical findings and provide additional insights

complementary to our econometric analysis results. By focusing on the mechanism of temporal changes in

network structural properties, our study contributes to supply chain management research with a supply net-

work perspective and interfirm alliance network research by broadening its scope into structural dynamism. Our

multi-method approach demonstrates how multiple complementary methodologies can foster a more nuanced

understanding of managing supply chain alliance network management.

1. Introduction

Nearly two decades have passed since Choi et al. (2001) introduced

a network-oriented view of supply chain management that spawned a

broad and active area of research ever since (e.g., Choi and Hong, 2002;

Kim et al., 2011;Pathak et al., 2014;Kim et al., 2015). A network-

centric study of supply chain systems is now well established in the

operations management (OM) literature, inspiring many different re-

search avenues associated with these systems’ structural dependencies

and inherent complexity (Choi et al., 2001;Nair et al., 2009;Pathak

et al., 2007,2014;Bellamy and Basole, 2013). Despite important ad-

vances, the majority of empirical studies on supply network phenomena

have been based on case studies or cross-sectional data. Beyond the

progress already made (e.g., Pathak et al., 2014), more insight is needed

to fully address the call for research that accounts for evolutions in

supply networks by emphasizing the temporal element (Pathak et al.,

2007).

Prior empirical research on supply networks has identified various

structural drivers founded in network theory. As a recent example, Yan

et al. (2015) highlighted the role of lower-tier suppliers in a supply

network with no direct ties to focal firms that bring identifiable pro-

ducts to market. These suppliers are nonetheless critical because they

are uniquely positioned to profoundly influence performance at mul-

tiple tiers in the supply network and, ultimately, the performance of the

focal firm. The authors provide an elegant framework for identifying

such critical suppliers, though they do not test the framework with an

empirical or analytical model. As another example, Bellamy et al.

(2014) developed an empirical model to highlight two structural supply

network characteristics as significant drivers of firm innovation. One is

focused on a firm's effectiveness in accessing knowledge and

https://doi.org/10.1016/j.jom.2018.09.001

Received 6 April 2017; Received in revised form 9 September 2018; Accepted 12 September 2018

∗

Corresponding author.

E-mail addresses: park.2706@osu.edu (H. Park), bellamym@bu.edu (M.A. Bellamy), basole@gatech.edu (R.C. Basole).

Journal of Operations Management 63 (2018) 79–96

Available online 28 September 2018

0272-6963/ © 2018 Elsevier B.V. All rights reserved.

T

information flows in its supply network (termed accessibility), while

the other centers on the shared ties among a firm's supply network

partners (termed interconnectedness). Their results highlighted the

importance of the interaction among structural drivers of supply net-

works as well as the moderating roles of internal and external factors.

Pathak et al. (2014) also explored this interaction in examining co-

opetition dynamics arising from the formation and dissolution of supply

network ties. Among other things, the authors emphasize how in-

dividual firms' efforts to impede or facilitate ties among its partners also

shape the structural form of the overall supply network. Taking stock of

these prior findings, it is evident that structural drivers of supply net-

works are not necessarily independent of one another over time. In-

stead, one may foster or hinder the growth of another. Thus, networks

may not fully realize the advantages of one structural driver in the

absence of other complementary ones. For this reason, we argue for the

importance of studying how such structural characteristics interact and

influence each other over time.

The network-related literature on supply chains has long recognized

that supply networks are complex adaptive systems and thus exhibit

dynamism (Choi et al., 2001;Choi and Hong, 2002;Pathak et al.,

2007). Researchers and practitioners have identified the complexity

and dynamism inherent in supply networks, in particular, as an urgent

and ongoing issue in modern supply networks (Bode and Wagner, 2015;

Choi and Krause, 2006). Case studies such as Pathak et al. (2014) have

contributed to the literature by developing key theoretical and man-

agerial insights based on real-world supply chains. These nonetheless

are fundamentally limited in studying dynamism, given the difficulty of

tracking the same set of firms and mapping their corresponding supply

networks over regular time intervals. Most large-scale empirical studies

using archival data have relied on cross-sectional analysis, with limited

attention to the structural dynamism of supply networks (Bellamy et al.,

2014). Moreover, the prior literature has emphasized absorptive capa-

city (Tsai, 2001;Bellamy et al., 2014) and industry growth (Kiss and

Barr, 2015;Zhang and Li, 2010) as principal internal and external

factors, respectively; both have been shown to moderate structural

change in networks. We seek to fill this void by combining both a dy-

namic panel analysis and empirically informed simulation experiments

to trace supply networks’ structural anatomy (i.e., their structural form)

and evolution (i.e., structural changes over time), while also in-

corporating factors internal and external to a firm. The combination of

empirical analysis and a simulation study has been particularly valu-

able in understanding the dynamism of complex systems such as pro-

duct and process innovation (Adner and Levinthal, 2001) and market

dynamism (Davis et al., 2009). Our study uses the supply chain alliance

network as the research context in which we seek to further bridge

existing supply chain alliance research (e.g., Monczka et al., 1998;

Perry et al., 1999;Yang et al., 2008) with the existing supply network

literature. To further contribute to the broader alliance literature, we

take a multi-method approach, combining large-scale empirical analysis

and simulation experiments.

In this paper, we seek to address the following research questions:

(1) How does the prior structural form of a firm's supply chain alliance

network affect its subsequent structure and evolution over time? and

(2) How do a firm's internal knowledge capabilities and the external

growth of its industry moderate this effect? Specifically, we investigate

how two important structural characteristics of a firm's supply chain

alliance network—accessibility and interconnectedness—change over

time. We then test how two principal factors internal and external to a

firm—absorptive capacity and industry growth—moderate the impact

of those structural characteristics over time. For the empirical part, we

analyze a panel dataset to construct supply chain alliance networks,

augmented by firms' financial information. Our empirical findings in-

dicate that accessibility and interconnectedness negatively influence

each other. A firm with high accessibility tends to be in a network

position that leads to a lower level of interconnectedness, and vice

versa. We also find evidence that absorptive capacity positively

moderates the reinforcement of accessibility, while industry growth

negatively moderates the reinforcement of interconnectedness. For our

simulation, we build a dynamic network simulation model using a se-

parable temporal exponential random graph model (STERGM)

(Krivitsky and Handcock, 2014). The STERGM is used to produce si-

mulated networks over time that mimic real-world supply chain alli-

ance networks. Our simulation model is complementary to the em-

pirical analysis in that it more directly incorporates the

interdependencies between two key node attributes that drive the

evolution of the network structure and potential nonlinear effects that

were neither hypothesized nor identified in the empirical model on its

own. Simulation results provide additional nuances not found in the

empirical results, notably contingencies demonstrating a positive

crossover effect from accessibility to interconnectedness. Evidence of

this is most profound when absorptive capacity decreases from its ob-

served value (leaving industry growth unchanged) and when industry

growth increases from its observed value (leaving absorptive capacity

unchanged).

Our study provides theoretical, methodological, and managerial

insights for supply network researchers and practitioners. In terms of

theoretical contributions, we characterize the direct impact of prior

network structure and the moderating effects of firm- and industry-level

characteristics on the evolution of supply chain alliance networks. Our

study contributes to the OM field by improving our systematic under-

standing of forces driving structural change in supply chain alliance

networks and, more broadly, interfirm alliance research.

Methodologically, our paper is a multi-method study combining large-

scale empirical analysis with archival data and empirically informed

simulation modeling. Several OM scholars have recognized the im-

portance of multi-method papers because of the potential value in their

complementary angles (e.g., Boyer and Swink, 2008;Choi et al., 2016;

Chandrasekaran et al., 2016). We believe our work enriches the re-

pertoire of multi-method OM studies. For managers and practitioners,

our findings suggest the potentially hidden trade-off in different supply

chain alliance network strategies shown to influence firm performance.

By pursuing one such strategy, a manager may be unknowingly

blocking the development of others. For example, a network strategy

focused on building more ties between direct partners can actually

negatively influence a manager's ability to develop ties that increase the

firm's accessibility to the broader supply chain alliance network. Ad-

ditionally, strategies that ensure a high level of accessibility in sub-

sequent periods may not persist if not properly matched with a con-

siderable capability to leverage external knowledge. Further, while the

growth of a manager's particular industry generally hinder develop-

ment of ties with direct partners in subsequent periods, we observed

that high enough levels of industry growth actually flipped the re-

lationship between accessibility and its subsequent interconnectedness

from negative to positive. Thus, managers of supply chain alliance

networks in growing industries may be able to reconcile having to

choose between either an accessibility- or interconnectedness-focused

strategy to compete.

Because this paper uses multiple methodologies, it is worth out-

lining the structure in more detail. Authors of multi-method studies

generally have two options for delineating the paper. One is sequential

narration, reporting the method and results with one methodology,

then the other; this is natural when the results of one methodology lead

or motivate the use of another methodology (Chandrasekaran et al.,

2016). The other option is juxtaposition or parallel presentation of both

methods. Here, the method section first outlines multiple methodolo-

gies and the results section then reports all findings. We find the first

approach more suitable for our study because our simulation modeling

with STERGM truly is an empirically informed simulation model. Ac-

cordingly, our paper is organized as follows: §2 provides an overarching

theoretical theme for the paper. §3 develops and grounds hypotheses to

be tested. §4 describes our empirical analysis, including data collection

process, variable construction, and model specification. §5 presents the

H. Park et al. Journal of Operations Management 63 (2018) 79–96

80